characterization of the ductility of bolted end plate - CMM
characterization of the ductility of bolted end plate - CMM
characterization of the ductility of bolted end plate - CMM
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
CHARACTERIZATION OF THE DUCTILITY OF<br />
BOLTED END PLATE BEAM-TO-COLUMN<br />
STEEL CONNECTIONS<br />
Ana Margarida Girão Coelho<br />
Thesis presented in fulfilment <strong>of</strong> <strong>the</strong> requirements for <strong>the</strong> degree<br />
<strong>of</strong> Doctor <strong>of</strong> Philosophy in Civil Engineering under <strong>the</strong><br />
scientific advising <strong>of</strong> Pr<strong>of</strong>. Dr. Luís Simões da Silva and<br />
Pr<strong>of</strong>. Ir. Frans S. K. Bijlaard.<br />
Tese apresentada para obtenção do grau de doutor em<br />
Engenharia Civil sob orientação científica do Pr<strong>of</strong>. Dr. Luís<br />
Simões da Silva e do Pr<strong>of</strong>. Frans S. K. Bijlaard.<br />
Universidade de Coimbra<br />
July 2004
O trabalho apresentado nesta tese de doutoramento foi financiado pelo<br />
Ministério da Ciência e Ensino Superior, ao abrigo do programa PRODEP<br />
(Concurso Público 4/5.3/PRODEP/2000) e com apoio da Fundação para a<br />
Ciência e Tecnologia (Bolsa de Doutoramento SFRH/BD/5125/2001).<br />
Coimbra, 2004.
To Miguel, my son<br />
and Encarnação and Hermínio, my parents.
ACKNOWLEDGEMENTS<br />
The author would like to express her sincere gratitude to Pr<strong>of</strong>. Dr. Luís A. P.<br />
Simões da Silva (University <strong>of</strong> Coimbra) and Pr<strong>of</strong>. Ir. Frans S. K. Bijlaard (Delft<br />
University <strong>of</strong> Technology). Pr<strong>of</strong>. Simões da Silva and Pr<strong>of</strong>. Bijlaard are a model<br />
for <strong>the</strong>ir technical expertise, pr<strong>of</strong>essionalism, scientific knowledge and ethics.<br />
Financial support from <strong>the</strong> Portuguese Ministry <strong>of</strong> Science and Higher Education<br />
(Ministério da Ciência e Ensino Superior) under contract grants from<br />
PRODEP (Concurso Público 4/5.3/PRODEP/2000) and Fundação para a<br />
Ciência e Tecnologia (Grant SFRH/BD/5125/2001) is gratefully acknowledged.<br />
The assistance provided by Mr. Nol Gresnigt, Mr. Henk Kolstein and Mr.<br />
Edwin Scharp from <strong>the</strong> Department <strong>of</strong> Steel and Timber Structures <strong>of</strong> <strong>the</strong> Delft<br />
University <strong>of</strong> Technology is most appreciated. To Corrie van der Wouden and Jan<br />
Willem van de Kuilen, thank you for your fri<strong>end</strong>ship. This research project was<br />
also made possible by <strong>the</strong> assistance <strong>of</strong> several people at <strong>the</strong> Department <strong>of</strong> Civil<br />
Engineering <strong>of</strong> <strong>the</strong> Faculty <strong>of</strong> Science and Technology <strong>of</strong> <strong>the</strong> University <strong>of</strong> Coimbra.<br />
Thank you Aldina Santiago, Luciano Lima, Luís Borges, Luís Neves, Pedro<br />
Simão, Rui Simões and Sandra Jordão.<br />
The fri<strong>end</strong>ship and support <strong>of</strong> my sister Rita and all my fri<strong>end</strong>s is also very<br />
much appreciated. Thank you all. To Carina, a special word <strong>of</strong> appreciation for <strong>the</strong><br />
works with <strong>the</strong> cover <strong>of</strong> this <strong>the</strong>sis.<br />
To Cláudio, thank you for your patience, love and understanding.
TABLE OF CONTENTS<br />
ABSTRACT<br />
RESUMO (Portuguese abstract)<br />
NOTATION<br />
PART I<br />
STATE-OF-THE-ART AND LITERATURE REVIEW<br />
1 MODELLING OF THE MOMENT-ROTATION CHARACTERISTICS OF<br />
BOLTED JOINTS: BACKGROUND REVIEW<br />
1.1 General introduction 3<br />
1.1.1 Literature review 4<br />
1.1.2 Scope <strong>of</strong> <strong>the</strong> work, objectives and research approach 7<br />
1.1.3 Outline <strong>of</strong> <strong>the</strong> dissertation 9<br />
1.2 Definitions 10<br />
1.3 Methods for modelling <strong>the</strong> rotational behaviour <strong>of</strong> beam-tocolumn<br />
joints<br />
12<br />
1.3.1 Generality 12<br />
1.3.2 The component method 12<br />
1.4 Characterization <strong>of</strong> basic components <strong>of</strong> <strong>bolted</strong> joints in terms<br />
<strong>of</strong> plastic resistance and initial stiffness<br />
14<br />
1.4.1 T-stub model for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> tension zone<br />
<strong>of</strong> <strong>bolted</strong> joints<br />
15<br />
1.4.1.1 Plastic resistance <strong>of</strong> single T-stub connections<br />
15<br />
1.4.1.2 Initial stiffness <strong>of</strong> single T-stub connections<br />
19<br />
1.4.2 Characterization <strong>of</strong> <strong>the</strong> several joint components 24<br />
1.5 Characterization <strong>of</strong> <strong>the</strong> post-limit behaviour <strong>of</strong> basic components<br />
<strong>of</strong> <strong>bolted</strong> joints<br />
29<br />
1.5.1 Column web in shear (component with high <strong>ductility</strong>)<br />
30<br />
1.5.2 Column flange in b<strong>end</strong>ing, <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing<br />
and bolts in tension (T-stub idealization)<br />
32<br />
1.5.3 Column web in compression (component with<br />
limited <strong>ductility</strong>)<br />
32<br />
1<br />
3
1.5.4 Column web in tension (component with limited<br />
<strong>ductility</strong>)<br />
34<br />
1.6 Evaluation <strong>of</strong> <strong>the</strong> moment-rotation response <strong>of</strong> <strong>bolted</strong> joints<br />
by means <strong>of</strong> component models<br />
34<br />
1.6.1 Eurocode 3 component model 37<br />
1.6.1.1 Model for stiffness evaluation 37<br />
1.6.1.2 Model for resistance evaluation 38<br />
1.6.1.3 Idealization <strong>of</strong> <strong>the</strong> moment-rotation curve 39<br />
1.6.2 Guidelines for evaluation <strong>of</strong> <strong>the</strong> <strong>ductility</strong> <strong>of</strong> <strong>bolted</strong><br />
joints<br />
39<br />
1.7 References 44<br />
App<strong>end</strong>ix A: Design provisions for <strong>characterization</strong> <strong>of</strong> resistance<br />
and stiffness <strong>of</strong> T-stubs<br />
50<br />
A.1 Basic formulations for prediction <strong>of</strong> plastic resistance <strong>of</strong><br />
<strong>bolted</strong> T-stubs<br />
50<br />
A.1.1 Type-1 mechanism 50<br />
A.1.2 Type-2 mechanism 50<br />
A.1.3 Type-3 mechanism 51<br />
A.1.4 Supplementary mechanism 51<br />
A.2 Influence <strong>of</strong> <strong>the</strong> moment-shear interaction on resistance formulations<br />
51<br />
A.2.1 Type-1 mechanism 52<br />
A.2.2 Type-2 mechanism 53<br />
A.3 Influence <strong>of</strong> <strong>the</strong> bolt dimensions on resistance formulations 54<br />
A.4 Formulations for prediction <strong>of</strong> elastic stiffness <strong>of</strong> <strong>bolted</strong> Tstubs<br />
56<br />
A.4.1 Elastic <strong>the</strong>ory for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> elastic stiffness<br />
<strong>of</strong> a <strong>bolted</strong> T-stub<br />
56<br />
A.4.2 Simplification <strong>of</strong> <strong>the</strong> stiffness coefficients for inclusion<br />
in design codes<br />
57<br />
PART II<br />
FURTHER DEVELOPMENTS ON THE T-STUB MODEL<br />
2 IMPROVEMENTS ON THE T-STUB MODEL: INTRODUCTION 61<br />
2.1 Introduction 61<br />
2.2 Failure modes 62<br />
2.3 References 65<br />
3 EXPERIMENTAL ASSESSMENT OF THE BEHAVIOUR OF T-STUB CON-<br />
NECTIONS<br />
3.1 Introduction 67<br />
3.2 Description <strong>of</strong> <strong>the</strong> experimental programme 67<br />
3.2.1 Geometrical properties <strong>of</strong> <strong>the</strong> specimens 67<br />
3.2.2 Mechanical properties <strong>of</strong> <strong>the</strong> specimens 69<br />
59<br />
67
3.2.2.1 Tension tests on <strong>the</strong> bolts 69<br />
3.2.2.2 Tension tests on <strong>the</strong> structural steel 73<br />
3.2.3 Testing procedure 75<br />
3.2.4 Aspects related to <strong>the</strong> welding procedure 78<br />
3.3 Experimental results 82<br />
3.3.1 Reference test series WT1 82<br />
3.3.2 Failure modes and general characteristics <strong>of</strong> <strong>the</strong> 87<br />
overall behaviour <strong>of</strong> <strong>the</strong> test specimens<br />
3.4 Concluding remarks 90<br />
3.5 References 92<br />
4 NUMERICAL ASSESSMENT OF THE BEHAVIOUR OF T-STUB CONNEC-<br />
TIONS<br />
4.1 Introduction 93<br />
4.2 Previous research 94<br />
4.3 Description <strong>of</strong> <strong>the</strong> model 96<br />
4.4 Calibration <strong>of</strong> <strong>the</strong> finite element model 99<br />
4.4.1 Geometry 100<br />
4.4.2 Boundary and load conditions 101<br />
4.4.3 Mechanical properties <strong>of</strong> steel components 102<br />
4.4.4 Specimen discretization 102<br />
4.4.5 Contact analysis 104<br />
4.5 Failure criteria 104<br />
4.6 Numerical results for HR T-stub T1 106<br />
4.7 Numerical results for WP T-stub WT1 110<br />
4.8 Considerations on <strong>the</strong> numerical modelling <strong>of</strong> <strong>the</strong> heat affected<br />
zone in WP T-stubs<br />
113<br />
4.9 Concluding remarks 115<br />
4.10 References 116<br />
App<strong>end</strong>ix B: Preliminary study for calibration <strong>of</strong> <strong>the</strong> finite element<br />
model (e.g. HR-T-stub T1)<br />
119<br />
B.1 Mesh convergence study 119<br />
B.2 Influence <strong>of</strong> <strong>the</strong> definition <strong>of</strong> <strong>the</strong> constitutive law and element<br />
formulation on <strong>the</strong> overall behaviour<br />
121<br />
B.3 Calibration <strong>of</strong> <strong>the</strong> joint element stiffness 121<br />
App<strong>end</strong>ix C: Stress and strain numerical results for HR-T-stub T1 123<br />
C.1 Load steps for stress and strain contours 123<br />
C.2 Von Mises equivalent stresses, σeq 123<br />
C.3 Stresses σxx and strains εxx 124<br />
C.4 Stresses σyy 126<br />
C.5 Stresses σzz 128<br />
C.6 Principal stresses and strains, σ11 and ε11 128<br />
C.7 Displacement results in xy cross-section 132<br />
93
5 PARAMETRIC STUDY 135<br />
5.1 Description <strong>of</strong> <strong>the</strong> specimens 135<br />
5.2 Influence <strong>of</strong> <strong>the</strong> assembly type and <strong>the</strong> weld throat thickness 135<br />
5.3 Influence <strong>of</strong> geometric parameters 147<br />
5.3.1 Gauge <strong>of</strong> <strong>the</strong> bolts 149<br />
5.3.2 Pitch <strong>of</strong> <strong>the</strong> bolts and <strong>end</strong> distance 149<br />
5.3.3 Edge distance and flange thickness 151<br />
5.4 Influence <strong>of</strong> <strong>the</strong> bolt and flange steel grade 158<br />
5.5 Experimental results for <strong>the</strong> stiffened test specimens and <strong>the</strong> 169<br />
rotated configurations<br />
5.5.1 Influence <strong>of</strong> a transverse stiffener 169<br />
5.5.2 Influence <strong>of</strong> <strong>the</strong> T-stub orientation 174<br />
5.6 Summary <strong>of</strong> <strong>the</strong> parametric study and concluding remarks 175<br />
5.7 References 178<br />
6 SIMPLIFIED METHODOLOGIES FOR ASSESSMENT OF THE BEHAVIOUR<br />
OF SINGLE T-STUB CONNECTIONS<br />
6.1 Introduction 179<br />
6.2 Previous research 179<br />
6.2.1 Jaspart proposal (1991) 180<br />
6.2.2 Faella and co-workers model (2000) 181<br />
6.2.3 Swanson model (1999) 182<br />
6.2.4 Beg and co-workers proposals for evaluation <strong>of</strong><br />
<strong>the</strong> deformation capacity (2002)<br />
185<br />
6.2.5 Examples 186<br />
6.2.5.1 Evaluation <strong>of</strong> initial stiffness 186<br />
6.2.5.2 Evaluation <strong>of</strong> plastic resistance 187<br />
6.2.5.3 Piecewise multilinear approximation <strong>of</strong> <strong>the</strong><br />
overall response and evaluation <strong>of</strong> <strong>the</strong><br />
deformation capacity and ultimate resistance<br />
187<br />
6.2.5.4 Summary 193<br />
6.3 Proposal and validation <strong>of</strong> a beam model for <strong>characterization</strong><br />
<strong>of</strong> <strong>the</strong> force-deformation response <strong>of</strong> T-stubs<br />
194<br />
6.3.1 Description <strong>of</strong> <strong>the</strong> model 194<br />
6.3.1.1 Fracture conditions 196<br />
6.3.1.2 Bolt deformation behaviour 196<br />
6.3.1.3 Flange constitutive law 197<br />
6.3.2 Analysis <strong>of</strong> <strong>the</strong> model in <strong>the</strong> elastic range 199<br />
6.3.3 Analysis <strong>of</strong> <strong>the</strong> model in <strong>the</strong> elastoplastic range 204<br />
6.3.4 Sophistication <strong>of</strong> <strong>the</strong> proposed method: modelling<br />
<strong>of</strong> <strong>the</strong> bolt action as a distributed load<br />
214<br />
6.3.5 Influence <strong>of</strong> <strong>the</strong> distance m for <strong>the</strong> WP T-stubs 215<br />
6.4 Concluding remarks 216<br />
6.5 References 218<br />
179
App<strong>end</strong>ix D: Detailed results obtained from application <strong>of</strong> <strong>the</strong> simplified<br />
methods for assessment <strong>of</strong> <strong>the</strong> force-deformation response <strong>of</strong><br />
single T-stub connections<br />
219<br />
D.1 Geometrical and mechanical characteristics <strong>of</strong> <strong>the</strong> specimens 219<br />
D.2 Previous research: exemplification 219<br />
D.2.1 Evaluation <strong>of</strong> initial stiffness 219<br />
D.2.2 Piecewise multilinear approximation <strong>of</strong> <strong>the</strong> overall<br />
response and evaluation <strong>of</strong> <strong>the</strong> deformation capacity<br />
and ultimate resistance<br />
219<br />
D.3 Application <strong>of</strong> <strong>the</strong> proposed model: results for HR-T-stub T1 235<br />
D.4 Application <strong>of</strong> <strong>the</strong> proposed model: results for WP-T-stub<br />
WT1<br />
239<br />
D.5 Prediction <strong>of</strong> <strong>the</strong> nonlinear response <strong>of</strong> <strong>the</strong> above connections<br />
using <strong>the</strong> nominal stress-strain characteristics<br />
241<br />
D.6 Comparative graphs: simple beam model and sophisticated<br />
beam model accounting for <strong>the</strong> bolt action<br />
255<br />
D.7 Comparative graphs: influence <strong>of</strong> <strong>the</strong> distance m for <strong>the</strong> WP-<br />
T-stubs<br />
264<br />
PART III<br />
MONOTONIC BEHAVIOUR OF BEAM-TO-COLUMN BOLTED END<br />
PLATE CONNECTIONS<br />
7 EXPERIMENTAL TESTS ON BOLTED END PLATE CONNECTIONS 275<br />
7.1 Introduction 275<br />
7.2 Description <strong>of</strong> <strong>the</strong> test programme 275<br />
7.2.1 Test details 275<br />
7.2.2 Geometrical properties 277<br />
7.2.3 Mechanical properties 277<br />
7.2.3.1 Tension tests on <strong>the</strong> bolts 277<br />
7.2.3.2 Tension tests <strong>of</strong> <strong>the</strong> structural steel 278<br />
7.2.4 Test arrangement and instrumentation 280<br />
7.2.5 Testing procedure 284<br />
7.3 Test results 284<br />
7.3.1 Moment-rotation curves 288<br />
7.3.2 Behaviour <strong>of</strong> <strong>the</strong> tension zone 294<br />
7.3.2.1 End <strong>plate</strong> deformation behaviour 294<br />
7.3.2.2 Yield line patterns 299<br />
7.3.2.3 Bolt elongation behaviour 299<br />
7.3.2.4 Strain behaviour 300<br />
7.4 Discussion <strong>of</strong> test results 302<br />
7.4.1 Plastic flexural resistance 303<br />
7.4.2 Initial rotational stiffness 304<br />
7.4.3 Rotation capacity 304<br />
7.5 Concluding remarks 305<br />
273
7.6 References 306<br />
8 DUCTILITY OF BOLTED END PLATE CONNECTIONS 307<br />
8.1 Introduction 307<br />
8.2 Modelling <strong>of</strong> bolt row behaviour through equivalent T-stubs 310<br />
8.3 Application to <strong>bolted</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections 310<br />
8.3.1 Component <strong>characterization</strong> 310<br />
8.3.2 Evaluation <strong>of</strong> <strong>the</strong> nonlinear moment-rotation response<br />
318<br />
8.3.3 Evaluation <strong>of</strong> <strong>the</strong> rotation capacity according to 326<br />
o<strong>the</strong>r authors<br />
8.3.4 Characterization <strong>of</strong> <strong>the</strong> joint <strong>ductility</strong> 328<br />
8.4 Discussion 330<br />
8.5 References 332<br />
9 CONCLUSIONS AND RECOMMENDATIONS 333<br />
9.1 Conclusions 333<br />
9.2 Future research 336<br />
9.3 References 338<br />
LIST OF REFERENCES 339
ABSTRACT<br />
The analysis <strong>of</strong> steel-framed building structures with full strength beam-tocolumn<br />
joints is quite standard nowadays. Buildings utilizing such framing<br />
systems are widely used in design practice. However, <strong>the</strong>re is growing recognition<br />
<strong>of</strong> significant benefits in designing joints as partial strength, semi-rigid.<br />
The design <strong>of</strong> joints within this partial strength/semi-rigid approach is becoming<br />
more and more popular. It requires however <strong>the</strong> knowledge <strong>of</strong> <strong>the</strong> full<br />
nonlinear moment-rotation behaviour <strong>of</strong> <strong>the</strong> joint, which is also a design parameter.<br />
Additionally, <strong>the</strong> joint failure must be ductile, i.e. <strong>the</strong> joint must have<br />
sufficient rotation capacity as <strong>the</strong> first plastic hinges occur in <strong>the</strong> joints ra<strong>the</strong>r<br />
than in <strong>the</strong> connected members. The research work reported in this <strong>the</strong>sis deals<br />
with this issue and gives particular attention to <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> joint<br />
<strong>ductility</strong>, which is particularly important in <strong>the</strong> partial strength/semi-rigid joint<br />
scenario.<br />
The experimental and numerical results <strong>of</strong> sixty one individual T-stub tests<br />
and eight full-scale <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connection tests are presented and assessed<br />
based on <strong>the</strong>ir resistance, stiffness and <strong>ductility</strong> characteristics. The results<br />
are used to compare existing resistance and stiffness models and to develop<br />
a simple methodology for evaluation <strong>of</strong> <strong>ductility</strong> properties.<br />
The T-stub model has been used for many years to model <strong>the</strong> tension zone<br />
<strong>of</strong> <strong>bolted</strong> joints. Previous research was mainly concentrated on rolled pr<strong>of</strong>iles<br />
as T-stub elements. In <strong>the</strong> case <strong>of</strong> <strong>end</strong> <strong>plate</strong> connections, <strong>the</strong> T-stub on <strong>the</strong> <strong>end</strong><br />
<strong>plate</strong> side comprises welded <strong>plate</strong>s as T-stub elements. This research also provides<br />
insight into <strong>the</strong> behaviour <strong>of</strong> this different type <strong>of</strong> assembly, in terms <strong>of</strong><br />
resistance, stiffness, deformation capacity and failure modes, in particular. It<br />
also explores <strong>the</strong> main features <strong>of</strong> <strong>the</strong> individual T-stub as a standalone configuration<br />
and evaluates quantitatively and qualitatively <strong>the</strong> influence <strong>of</strong> <strong>the</strong><br />
main geometrical and mechanical parameters on <strong>the</strong> overall behaviour.<br />
A simplified two-dimensional beam model for <strong>the</strong> assessment <strong>of</strong> <strong>the</strong> deformation<br />
response <strong>of</strong> individual T-stubs is developed based on <strong>the</strong> experimental<br />
observations and <strong>the</strong> results <strong>of</strong> <strong>the</strong> finite element investigation. The model is<br />
based on <strong>the</strong> Eurocode 3 approach and includes <strong>the</strong> deformations from tension<br />
bolt elongation and b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> T-stub flange. It is able to predict <strong>the</strong> deformation<br />
capacity <strong>of</strong> a T-stub with a satisfactory degree <strong>of</strong> accuracy.<br />
This study on individual T-stubs is part <strong>of</strong> <strong>the</strong> investigation <strong>of</strong> <strong>end</strong> <strong>plate</strong><br />
behaviour. The outcomes are used to validate a methodology based on <strong>the</strong> socalled<br />
component model to determine <strong>the</strong> rotational behaviour <strong>of</strong> <strong>bolted</strong> <strong>end</strong><br />
<strong>plate</strong> connections. Since most <strong>of</strong> <strong>the</strong> joint rotation in thin <strong>end</strong> <strong>plate</strong>s comes
from <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformation, <strong>the</strong> <strong>characterization</strong> <strong>of</strong> bolt row behaviour<br />
through equivalent T-stubs is <strong>of</strong> <strong>the</strong> utmost importance. A spring model that<br />
includes <strong>the</strong> T-stub idealization <strong>of</strong> <strong>the</strong> tension zone is used to derive <strong>the</strong><br />
nonlinear moment-rotation <strong>of</strong> <strong>the</strong> joint. Special attention is given to <strong>the</strong> <strong>characterization</strong><br />
<strong>of</strong> <strong>the</strong> joint <strong>ductility</strong>. Comparisons <strong>of</strong> <strong>the</strong> joint <strong>ductility</strong> and <strong>the</strong> corresponding<br />
equivalent T-stub for <strong>the</strong> <strong>end</strong> <strong>plate</strong> side are drawn. Finally, some<br />
recomm<strong>end</strong>ations for <strong>the</strong> required <strong>ductility</strong> expressed in terms <strong>of</strong> a <strong>ductility</strong><br />
index are given.
RESUMO<br />
O projecto de estruturas metálicas para edifícios porticados com ligações<br />
viga-pilar de resistência total é relativamente comum. No entanto, tem-se<br />
vindo a reconhecer os benefícios que decorrem da modelação semi-rígida e de<br />
resistência parcial das ligações. Esta abordagem tem-se generalizado no<br />
dimensionamento das ligações metálicas. Para o efeito, é necessário avaliar o<br />
comportamento momento-rotação real das ligações. Adicionalmente, a rotura<br />
das ligações tem de ser dúctil, isto é, as ligações têm de exibir capacidade de<br />
rotação suficiente, uma vez que as primeiras rótulas plásticas se formam no nó<br />
de ligação e não nos elementos (viga ou pilar). O trabalho de investigação<br />
apresentado nesta tese foca este aspecto e dá ênfase à caracterização da<br />
ductilidade das ligações, que é particularmente relevante na modelação semirígida/resistência<br />
parcial.<br />
Descrevem-se e discutem-se os resultados experimentais e numéricos de<br />
sessenta e um testes em ligações em duplo T (T-stubs) individuais e oito<br />
ligações viga-pilar aparafusadas com placa de extremidade. A análise destes<br />
resultados inclui a caracterização das propriedades de resistência, rigidez e<br />
ductilidade das ligações e a sua confrontação com modelos correntes de avaliação<br />
de resistência e rigidez. Em termos de ductilidade, é proposta uma metodologia<br />
simplificada para a caracterização desta propriedade das ligações.<br />
O modelo do T-stub é utilizado na idealização da zona traccionada de<br />
ligações aparafusadas. Os trabalhos de investigação anteriores centraram a<br />
análise desta ligação mais simples em elementos que utilizam perfis laminados<br />
a quente. No caso de ligações com placa de extremidade, os T-stubs equivalentes<br />
na zona da placa englobam elementos soldados. Neste trabalho<br />
procura-se descrever o comportamento deste tipo de T-stub, focando os modos<br />
de rotura, a resistência, a rigidez e a ductilidade, em particular. Exploram-se<br />
também as principais características do T-stub isolado e avalia-se<br />
qualitativa e quantitativamente a influência dos principais parâmetros geométricos<br />
e mecânicos no comportamento global.<br />
Com base nos resultados experimentais e numéricos (elementos finitos)<br />
propõe-se um modelo de viga bidimensional simplificado para caracterização<br />
do comportamento força-deformação de T-stubs. O modelo baseia-se na<br />
abordagem do Eurocódigo 3 e inclui a deformação do parafuso traccionado e<br />
do banzo do T-stub em flexão e permite prever a capacidade de deformação<br />
com um grau de precisão satisfatório.<br />
Este estudo em T-stubs isolados constitui uma parte do trabalho de investigação<br />
do comportamento da placa de extremidade. As conclusões deste es-
tudo são utilizadas na validação de uma metodologia baseada no método das<br />
componentes para avaliação do comportamento rotacional de ligações aparafusadas<br />
com placa de extremidade. Uma vez que a rotação da ligação provém<br />
essencialmente da deformação da placa de extremidade, no caso de placas<br />
finas, a idealização do seu comportamento por meio de T-stubs equivalentes<br />
é particularmente relevante. Um modelo mecânico de molas e bielas rígidas<br />
que inclui a idealização da zona traccionada por intermédio de T-stubs<br />
é utilizado para a caracterização da resposta momento-rotação da ligação,<br />
com particular ênfase na avaliação da sua ductilidade. Estabelecem-se comparações<br />
entre a ductilidade da ligação e os correspondentes T-stubs equivalentes<br />
na zona da placa de extremidade. Finalmente, são propostas algumas<br />
recom<strong>end</strong>ações para a ductilidade mínima da ligação, expressa em termos de<br />
um índice de ductilidade.
NOTATION<br />
Lower cases<br />
a’ Effective edge distance according to <strong>the</strong> Kulak’s prying model<br />
aep Throat thickness <strong>of</strong> a fillet weld at <strong>the</strong> <strong>end</strong> <strong>plate</strong> side<br />
aw Throat thickness <strong>of</strong> a fillet weld<br />
a Total displacements<br />
b Width; actual width <strong>of</strong> a T-stub tributary to a bolt row<br />
b’ Distance between <strong>the</strong> inside edge <strong>of</strong> <strong>the</strong> bolt shank to 50% distance<br />
into pr<strong>of</strong>ile root<br />
beff Effective width; effective width <strong>of</strong> a T-stub tributary to a bolt row<br />
for resistance calculations<br />
b′ eff Effective width <strong>of</strong> a T-stub tributary to a bolt-row for stiffness<br />
calculations<br />
d Length between <strong>the</strong> bolt axis and <strong>the</strong> face <strong>of</strong> <strong>the</strong> T-stub web<br />
dc Clear depth <strong>of</strong> <strong>the</strong> column web<br />
dw Bolt head, nut or washer diameter, as appropriate<br />
d0 Bolt hole clearance<br />
e Edge distance<br />
ecomp End <strong>plate</strong> distance<br />
e1 End distance (from <strong>the</strong> centre <strong>of</strong> <strong>the</strong> bolt hole to <strong>the</strong> adjacent edge)<br />
fu Ultimate or tensile stress<br />
fy Yield stress<br />
h Depth<br />
hmrn Height <strong>of</strong> <strong>the</strong> resultant tension force above <strong>the</strong> neutral axis at<br />
maximum strain<br />
hr Distance <strong>of</strong> bolt row r to <strong>the</strong> centre <strong>of</strong> compression<br />
hyfn Height <strong>of</strong> <strong>the</strong> flush bolt row above <strong>the</strong> neutral axis at yield<br />
k Empirical factor<br />
ke Initial axial stiffness <strong>of</strong> a spring-component<br />
keff.r Effective stiffness coefficient for bolt row r<br />
ki Auxiliary length values for definition <strong>of</strong> <strong>the</strong> bolt conventional<br />
(i=1→4) length, according to Aggerskov<br />
ki Joint element stiffness modulus (1: normal direction; 2,3: tangen-<br />
(i=1→3) tial direction)<br />
kp-l Post-limit axial stiffness <strong>of</strong> a spring-component<br />
lHAZ Width <strong>of</strong> <strong>the</strong> heat affected zone<br />
m Distance from bolt centre to 20% distance into pr<strong>of</strong>ile root or weld<br />
mf Average distance from each bolt to <strong>the</strong> adjacent web and flange<br />
welds below <strong>the</strong> tension flange
mpl Plastic moment <strong>of</strong> a <strong>plate</strong> per unit length<br />
n Effective edge distance; number <strong>of</strong> bolt rows in tension; ratio between<br />
<strong>the</strong> axial force in <strong>the</strong> column and <strong>the</strong> corresponding plastic<br />
level<br />
nth Number <strong>of</strong> threads per unit length <strong>of</strong> <strong>the</strong> bolt<br />
p Pitch <strong>of</strong> <strong>the</strong> bolts<br />
pi-j Distance between bolt rows i and j<br />
q Parameter<br />
qb Uniformly distributed bolt action, statically equivalent to B<br />
qee Initial prying gradient<br />
qij,k Prying gradient<br />
r Fillet radius <strong>of</strong> <strong>the</strong> flange-to-web connection<br />
s Length<br />
sp Length obtained by dispersion at 45º through <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
sx Ratio transverse stress/yield stress in <strong>the</strong> column web<br />
t Thickness<br />
u Degree-<strong>of</strong>-freedom<br />
v Degree-<strong>of</strong>-freedom<br />
w Horizontal distance between bolt axis centrelines (gauge); degree<strong>of</strong>-freedom<br />
x Cartesian axis; distance<br />
xi Distance <strong>of</strong> <strong>the</strong> joint row to <strong>the</strong> tip <strong>of</strong> <strong>the</strong> flanges<br />
y Cartesian axis<br />
z Lever arm; cartesian axis<br />
zi Distance between <strong>the</strong> ith bolt row to <strong>the</strong> centre <strong>of</strong> compression<br />
z1 Distance in [mm] between <strong>the</strong> first bolt row from <strong>the</strong> tension<br />
flange and <strong>the</strong> centre <strong>of</strong> compression<br />
Upper cases<br />
Ab Nominal area <strong>of</strong> <strong>the</strong> bolt shank<br />
As Bolt tensile stress area<br />
Avc Shear area <strong>of</strong> a column pr<strong>of</strong>ile<br />
B Bolt force<br />
E Young modulus<br />
Eh Strain hardening modulus<br />
Eu Modulus <strong>of</strong> <strong>the</strong> stress-strain curve before collapse<br />
F Force; resistance; load; applied load per bolt row in a T-stub<br />
FQi Contact force associated to a joint row<br />
FRd Full “plastic” (design) resistance<br />
Fti Potential resistance <strong>of</strong> bolt row i in <strong>the</strong> tension zone<br />
Fu Ultimate resistance<br />
Fv Vertical forces<br />
F 1. Rd .0 Ratio between <strong>the</strong> design resistance <strong>of</strong> mechanism type-1 accounting<br />
for shear and that corresponding to <strong>the</strong> basic formulation<br />
G Tangential modulus <strong>of</strong> elasticity<br />
Height <strong>of</strong> <strong>the</strong> column below <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
Hc.low
Hc.up Height <strong>of</strong> <strong>the</strong> column above <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
I Moment <strong>of</strong> inertia<br />
K Spring axial stiffness (generic)<br />
Kb Bolt elastic stiffness according to <strong>the</strong> Swanson’s bolt model<br />
Kcws.h Residual stiffness (Krawinkler et al. model for <strong>characterization</strong> <strong>of</strong><br />
<strong>the</strong> behaviour <strong>of</strong> <strong>the</strong> “column web in shear”)<br />
K(flex) Stiffness for <strong>the</strong> flexible beam approach<br />
K(rig) Stiffness for <strong>the</strong> rigid beam approach<br />
L Length; cantilever length<br />
Lb Bolt conventional length<br />
*<br />
L b<br />
Clamping length <strong>of</strong> <strong>the</strong> bolts<br />
Lbeam Length <strong>of</strong> <strong>the</strong> beam<br />
Lcomp Length <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> below <strong>the</strong> compression beam flange<br />
Lg Grip length<br />
Linfluence.i Influence length <strong>of</strong> a joint row<br />
Lload Distance between <strong>the</strong> load application point and <strong>the</strong> face <strong>of</strong> <strong>the</strong><br />
<strong>end</strong> <strong>plate</strong><br />
Ls Bolt shank length<br />
Ltg Bolt threaded length included in <strong>the</strong> grip length<br />
M B<strong>end</strong>ing moment<br />
Mj.Ed<br />
Mj.Rd<br />
N<br />
B<strong>end</strong>ing moment (lower than Mj.Rd) acting in <strong>the</strong> joint<br />
Joint flexural plastic (design) resistance<br />
Axial force<br />
P Concentrated force<br />
Q Prying force<br />
R Norm <strong>of</strong> external forces<br />
Sj.ini Initial rotational stiffness <strong>of</strong> a joint<br />
S0 Bolt preload<br />
V Shear force<br />
Parameter<br />
Zf<br />
Greek letters<br />
α Coefficient obtained from an abacus provided in Eurocode 3;<br />
parameter that represents a <strong>ductility</strong> limit<br />
αf Parameter<br />
β Transformation parameter; ratio flexural resistance <strong>of</strong> flanges /axial<br />
resistance <strong>of</strong> <strong>the</strong> bolts; parameter that represents a <strong>ductility</strong> limit<br />
βa; βb Coefficients that account for <strong>the</strong> shear deformations<br />
βu.lim Limit value for <strong>the</strong> β-ratio to have a collapse failure mode governed<br />
by cracking <strong>of</strong> <strong>the</strong> flange material<br />
χ Curvature<br />
δ Relative displacement; elongation<br />
δ(flex) Displacement <strong>of</strong> a flexible beam at mid-span<br />
δ(rig) Displacement <strong>of</strong> a rigid beam at mid-span<br />
Deformation capacity <strong>of</strong> half T-stub<br />
δu
δ Non-dimensional displacement<br />
δ a Norm <strong>of</strong> <strong>the</strong> iterative displacements<br />
∆ Axial deformation; elongation<br />
∆ a Total displacements for a certain increment<br />
∆ FRd<br />
Deformation corresponding to <strong>the</strong> component plastic resistance<br />
∆u Deformation capacity<br />
ε<br />
ε<br />
Strain; engineering strain; parameter<br />
e<br />
Elastic deformation (strain)<br />
εh Strain at <strong>the</strong> strain hardening point<br />
εhs Engineering strain at which <strong>the</strong> maximum engineering stress is<br />
reached<br />
εn<br />
ε<br />
Natural or logarithmic strain<br />
p<br />
Plastic deformation (strain)<br />
εu Ultimate strain<br />
εuni Uniform strain<br />
ε0 Ultimate transverse strain acting in <strong>the</strong> column web in <strong>the</strong> case<br />
that <strong>the</strong> axial force in <strong>the</strong> column is absent<br />
φ Connection rotational deformation; bolt diameter<br />
Φ Joint rotation<br />
φCd Rotation capacity <strong>of</strong> a connection<br />
ΦCd Rotation capacity <strong>of</strong> a joint<br />
Φ Joint rotation at which <strong>the</strong> moment deteriorates back to Mj.Rd after<br />
*<br />
Cd<br />
reaching a moment above Mj.Rd through deformation beyond ΦXd<br />
φ Rotation <strong>of</strong> <strong>the</strong> connection at maximum load<br />
M max<br />
Φ Rotation <strong>of</strong> <strong>the</strong> joint at maximum load<br />
M max<br />
φXd<br />
ΦXd<br />
Connection rotation value at which <strong>the</strong> moment resistance first<br />
reaches Mj.Rd<br />
Joint rotation value at which <strong>the</strong> moment resistance first reaches<br />
Mj.Rd<br />
γ Shear deformation <strong>of</strong> <strong>the</strong> column web panel<br />
γd Euclidean displacement norm<br />
γdt Euclidean iterative displacement norm<br />
Coefficients<br />
γi<br />
(i=1→3)<br />
γM<br />
Partial safety coefficients (γM0, γM1, γM2)<br />
γw Work norm<br />
γψ Euclidean residual norm<br />
η Stiffness modification factor<br />
ϕi Component <strong>ductility</strong> index<br />
ϑj Joint <strong>ductility</strong> index<br />
λ Ratio between n and m<br />
λ Plate sl<strong>end</strong>erness<br />
p<br />
κN<br />
κwc<br />
Parameter that reflects <strong>the</strong> influence <strong>of</strong> <strong>the</strong> level <strong>of</strong> axial force in<br />
<strong>the</strong> column<br />
Reduction factor to account for <strong>the</strong> effect <strong>of</strong> axial force in <strong>the</strong> col-
umn<br />
µ Friction coefficient; ratio between characteristic strain values;<br />
stiffness ratio<br />
θ Rotation<br />
ρ Reduction factor for <strong>plate</strong> buckling<br />
ρy Yield ratio<br />
*<br />
ρ Alternative definition <strong>of</strong> <strong>the</strong> yield ratio<br />
y<br />
σ Stress; nominal or conventional stress<br />
σeq Von Mises equivalent stress<br />
σn True stress<br />
σx Transverse stress acting in <strong>the</strong> column web<br />
τ Shear stress<br />
Γ Parameter<br />
τy Yield shear stress<br />
υ Poisson’s ratio<br />
ω Reduction factor to allow for possible effects <strong>of</strong> interaction with<br />
shear in <strong>the</strong> column web panel (ω 1, ω2: parameters for computation<br />
<strong>of</strong> ω)<br />
ξ Coefficient<br />
ψ Norm <strong>of</strong> residuals<br />
ψi Component <strong>ductility</strong> index<br />
ζ Coefficient taken as 0.8 in Eurocode 3<br />
Subscripts<br />
av average<br />
b Beam; bolt<br />
bfc Beam web and flange in compression<br />
bot Bottom T-stub<br />
bt Bolts in tension<br />
bwt Beam web in tension<br />
c Column; compression<br />
cfb Column flange in b<strong>end</strong>ing<br />
cp Circular yield line patterns<br />
cwc Column web in compression<br />
cws Column web in shear<br />
cwt Column web in tension<br />
e/el Elastic<br />
ep End <strong>plate</strong><br />
epb End <strong>plate</strong> in b<strong>end</strong>ing<br />
f Flange<br />
fract Fracture<br />
h Bolt head; strain hardening<br />
j Joint<br />
l Lower T-stub element<br />
m Strain hardening range and before collapse<br />
max Maximum
min Minimum<br />
n Bolt nut<br />
nc Non-circular yield line patterns<br />
p/pl Plastic<br />
p-l Post-limit<br />
red Reduced<br />
ri Bolt row i<br />
Rd Pure plastic conditions; design conditions<br />
s Stiffener<br />
t Tension<br />
top Top T-stub<br />
T T-stub element<br />
u Ultimate conditions; upper T-stub element<br />
w Web; weld<br />
wp Web panel<br />
wsh Washer<br />
X Extension <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> above <strong>the</strong> tension beam flange<br />
y Yield<br />
0 T-stub component<br />
1 Type-1 plastic failure mechanism <strong>of</strong> a T-stub; bolt row 1<br />
11 Principal direction 1 for a stress state<br />
2 Type-2 “plastic” failure mechanism <strong>of</strong> a T-stub; bolt row 2<br />
3 Type-3 “plastic” failure mechanism <strong>of</strong> a T-stub; bolt row 3<br />
* Supplementary plastic failure mechanism <strong>of</strong> a T-stub<br />
Abbreviations<br />
B Back (from eye position)<br />
BF Basic formulation <strong>of</strong> resistance<br />
BM Base metal<br />
DTi Reference to a LVDT i<br />
F Front (from eye position)<br />
FBA Resistance formulation accounting for <strong>the</strong> bolt action<br />
FE Finite element<br />
FT Full-threaded bolt<br />
HAZ Heat affected zone<br />
HR Hot-rolled pr<strong>of</strong>ile<br />
L Left (from eye position)<br />
LVDT Linear variable displacement transducer<br />
K-R Knee-range <strong>of</strong> a deformability curve<br />
R Right (from eye position)<br />
SG Strain gauge<br />
ST Short-threaded bolt<br />
WM Weld metal<br />
WP Welded <strong>plate</strong>s<br />
HP Reference to a specific LVDT
PART I: STATE-OF-THE-ART AND LITERATURE REVIEW<br />
1
1 MODELLING OF THE MOMENT-ROTATION CHARACTER-<br />
ISTICS OF BOLTED JOINTS: BACKGROUND REVIEW<br />
1.1 GENERAL INTRODUCTION<br />
Structural joints, particularly <strong>bolted</strong> and welded connections found in common<br />
steel constructions, exhibit a distinctively nonlinear behaviour. This nonlinearity<br />
arises because a joint is an assemblage <strong>of</strong> several components that interact<br />
differently at distinct levels <strong>of</strong> applied loads. The interaction between <strong>the</strong> elemental<br />
parts includes elastoplastic deformations, contact, slip and separation<br />
phenomena. The analysis <strong>of</strong> this complex behaviour is usually approximate in<br />
nature with drastic simplifications. Tests (both experimental and numerical) are<br />
frequently carried out to obtain <strong>the</strong> actual response, which is <strong>the</strong>n modelled approximately<br />
by ma<strong>the</strong>matical expressions that relate <strong>the</strong> main structural joint<br />
properties.<br />
Beam-to-column joints in steel-framed building structures have to transfer<br />
<strong>the</strong> beam and floor loads to <strong>the</strong> columns. Generally, <strong>the</strong> forces transmitted<br />
through <strong>the</strong> joints can be axial and shear forces, b<strong>end</strong>ing and torsion moments.<br />
The b<strong>end</strong>ing deformations are predominant in most cases, when compared to<br />
axial and shear deformations that are hence neglected. The effect <strong>of</strong> torsion is<br />
also negligible in planar frames. Typical beam-to-column moment-resisting<br />
joints in steel-framed structures include <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections, <strong>bolted</strong><br />
connections with (flange and/or web) angle cleats and welded connections.<br />
Their behaviour is represented by a moment vs. rotation curve (M-Φ) that describes<br />
<strong>the</strong> relationship between <strong>the</strong> applied b<strong>end</strong>ing moment, M and <strong>the</strong> corresponding<br />
rotation between <strong>the</strong> members, Φ. This curve defines three main<br />
structural properties: (i) moment resistance, (ii) rotational stiffness and (iii) rotation<br />
capacity. Historically, moment-resisting joints have been designed for<br />
strength and stiffness with little regard to rotational capacity. There is growing<br />
recognition that in many situations this practice is questionable and so guidance<br />
is urged to help designers.<br />
Joints can be grouped according to <strong>the</strong>ir structural properties. The European<br />
code <strong>of</strong> practice for <strong>the</strong> design <strong>of</strong> structural steel joints in buildings, Eurocode 3<br />
[1.1], classifies joints by strength (full strength, partial strength or nominally<br />
pinned) and stiffness (rigid, semi-rigid or nominally pinned). A full strength<br />
joint exhibits a moment resistance at least equal to that <strong>of</strong> <strong>the</strong> connected members<br />
whilst partial strength joints have lower strength than <strong>the</strong> members. Nominally<br />
pinned joints are sufficiently flexible to be regarded as a pin for analysis<br />
purposes, i.e. <strong>the</strong>y are not moment resisting and have no rotational stiffness. A<br />
rigid joint is stiff enough for <strong>the</strong> effect <strong>of</strong> its deformation on <strong>the</strong> distribution <strong>of</strong><br />
3
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
internal forces and b<strong>end</strong>ing moments in <strong>the</strong> structure to be neglected. A semirigid<br />
joint does not meet <strong>the</strong> criteria for a rigid joint or a pin. Naturally, nominally<br />
pinned joints have to be ductile, i.e. <strong>the</strong>y have to rotate plastically at some<br />
stage <strong>of</strong> <strong>the</strong> loading cycle without failure. The semi-rigid/partial strength design<br />
philosophy <strong>of</strong> joints usually leads to more economic and simple solutions.<br />
The use <strong>of</strong> this joint category in steel frames, however, is only feasible if <strong>the</strong>y<br />
develop sufficient rotation capacity in order that <strong>the</strong> int<strong>end</strong>ed failure mechanism<br />
<strong>of</strong> <strong>the</strong> whole structure can be formed prior to fracture <strong>of</strong> <strong>the</strong> joint.<br />
End <strong>plate</strong> <strong>bolted</strong> connections that are widely used in steel-frames as moment-resistant<br />
connections between steel members usually fall in <strong>the</strong> semirigid/partial<br />
strength category. The simplicity and economy associated to <strong>the</strong>ir<br />
fabrication and erection made this joint typology quite popular in steel-framed<br />
structures. In Europe, steel <strong>bolted</strong> partial strength ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections<br />
are typical for low-rise buildings erected using welding at <strong>the</strong> shop and<br />
bolting on site. Rules for prediction <strong>of</strong> strength and stiffness <strong>of</strong> this joint configuration<br />
have been included in modern design codes as <strong>the</strong> Eurocode 3. Yet,<br />
no quantitative guidance for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> <strong>ductility</strong> is available.<br />
The main topics <strong>of</strong> this research work are moment-resisting <strong>bolted</strong> (major<br />
axis) connections joining I-sections in steel-framed structures and <strong>the</strong> <strong>characterization</strong><br />
<strong>of</strong> <strong>the</strong>ir rotational behaviour. Special emphasis is given on ext<strong>end</strong>ed<br />
<strong>end</strong> <strong>plate</strong> connections similar to that shown in Fig. 1.1. The main source <strong>of</strong> deformability<br />
<strong>of</strong> this connection type is <strong>of</strong>ten <strong>the</strong> tension zone that can be modelled<br />
with <strong>the</strong> T-stub approach [1.1-1.5]. The evaluation <strong>of</strong> <strong>the</strong> deformation behaviour<br />
<strong>of</strong> single T-stubs is <strong>the</strong>refore very important and is also focused on in<br />
this work.<br />
(a) Three-dimensional view. (b) Section. (c) Elevation.<br />
Fig. 1.1 Unstiffened <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connection.<br />
1.1.1 Literature review<br />
Bolted <strong>end</strong> <strong>plate</strong> beam-to-column steel connections have been widely studied<br />
over <strong>the</strong> years. The emphasis in most <strong>of</strong> <strong>the</strong> previous research on this subject<br />
4
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
was mainly placed on full strength <strong>end</strong> <strong>plate</strong> connections and <strong>the</strong>refore only <strong>the</strong><br />
resistance and stiffness properties were fully characterized. Thoroughly conducted<br />
experimental tests were carried out for prediction <strong>of</strong> <strong>the</strong> M-Φ curve.<br />
However, <strong>the</strong> information extracted from those experiments was limited to <strong>the</strong><br />
joint typology that had been tested and could not be extrapolated to o<strong>the</strong>r joint<br />
configurations. Analytical methodologies based on finite element (FE) analyses<br />
can be regarded as an alternative tool for investigation and understanding <strong>of</strong><br />
joint behaviour, provided that <strong>the</strong> requirements for a reliable simulation are totally<br />
fulfilled. Many researchers used both approaches in conjunction.<br />
Douty and McGuire [1.6] conducted monotonic experimental tests on <strong>end</strong><br />
<strong>plate</strong> connections to study <strong>the</strong>ir performance, design and use in plastically designed<br />
structures. They identified <strong>the</strong> effect <strong>of</strong> prying action in increasing <strong>the</strong><br />
tension bolt force and recognized <strong>the</strong> importance <strong>of</strong> material strain hardening.<br />
The effect <strong>of</strong> prying action was fur<strong>the</strong>r investigated by Aggerskov [1.4,1.7]<br />
who carried out a series <strong>of</strong> fifteen tests on ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong>s.<br />
Zoetemeijer [1.2,1.8-1.10] reported on detailed series <strong>of</strong> tests performed at<br />
<strong>the</strong> Delft University <strong>of</strong> Technology to propose and validate yield line models<br />
for <strong>the</strong> strength design <strong>of</strong> <strong>the</strong> tension region. This zone <strong>of</strong> <strong>end</strong> <strong>plate</strong> connections<br />
includes <strong>the</strong> following basic elemental parts: column flange, <strong>end</strong> <strong>plate</strong> and <strong>the</strong><br />
bolts in tension. Zoetemeijer also proposed some criteria and simple empirical<br />
expressions for <strong>the</strong> estimation <strong>of</strong> a joint deformation capacity based on a series<br />
<strong>of</strong> experiments described in [1.10]. Packer and Morris [1.3] and Mann and<br />
Morris [1.11] focused on this subject too. Similar to Zoetemeijer, <strong>the</strong>y also idealized<br />
<strong>the</strong> tension region as a T-stub. Fig. 1.2 identifies <strong>the</strong> T-stub which accounts<br />
for <strong>the</strong> deformation <strong>of</strong> <strong>the</strong> column flange and <strong>the</strong> <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing in<br />
<strong>the</strong> particular case <strong>of</strong> an ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> <strong>bolted</strong> connection. In this particular<br />
case, since <strong>the</strong> column flange is unstiffened, <strong>the</strong> T-stub on <strong>the</strong> column side is<br />
orientated at right angles to <strong>the</strong> <strong>end</strong> <strong>plate</strong> T-stub [1.5]. Different investigators<br />
also carried out various studies focusing on mechanisms in T-stubs ra<strong>the</strong>r than<br />
whole <strong>plate</strong>s, particularly to assess <strong>the</strong> resistance properties <strong>of</strong> this simple connection<br />
[1.5,1.12-1.15].<br />
Jenkins et al. [1.16] contributed to a better understanding <strong>of</strong> <strong>end</strong> <strong>plate</strong> behaviour<br />
and proposed standardized <strong>end</strong> <strong>plate</strong> connection types to permit a generalization<br />
<strong>of</strong> joint characteristics obtained from numerical modelling. They<br />
performed FE analysis to determine <strong>the</strong> complete M-Φ curve <strong>of</strong> some joints<br />
that was compared with experimental results. This experimental programme<br />
included eighteen tests. The principal objective <strong>of</strong> <strong>the</strong> programme was to obtain<br />
M-Φ relationships but <strong>the</strong>y also directed attention at o<strong>the</strong>r features as <strong>the</strong><br />
evaluation <strong>of</strong> <strong>the</strong> axial forces in <strong>the</strong> bolts.<br />
The <strong>characterization</strong> <strong>of</strong> <strong>the</strong> initial rotational stiffness <strong>of</strong> beam-to-column<br />
joints was <strong>the</strong> main research topic <strong>of</strong> Davison et al. [1.17] who did various tests<br />
on <strong>end</strong> <strong>plate</strong> connections with different thickness and identical beam and column<br />
sizes. The researchers also investigated <strong>the</strong> effect <strong>of</strong> lack <strong>of</strong> fit [1.18] and<br />
concluded that it was negligible.<br />
5
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
6<br />
Equivalent<br />
T-stub<br />
M<br />
T-stub<br />
T-stub<br />
(a) Unstiffened ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connection: T-stub identification and orientation.<br />
External load<br />
External load<br />
Stiffener<br />
Weld toe<br />
External load<br />
External load<br />
(b) Model for <strong>the</strong> column flange side. (c) Model for <strong>the</strong> <strong>end</strong> <strong>plate</strong> side.<br />
Fig. 1.2 T-stub identification and representation.<br />
Janss and co-workers [1.19] completed a series <strong>of</strong> tests that were later used<br />
by Jaspart [1.20] to propose a methodology for evaluation <strong>of</strong> plastic resistance<br />
and initial rotational stiffness <strong>of</strong> moment joints.<br />
Aggarwal [1.21] and Bose et al. [1.22] carried out comparative tests on <strong>end</strong><br />
<strong>plate</strong> connections for which <strong>the</strong>y characterized <strong>the</strong> moment carrying behaviour.<br />
In particular, Bose et al. [1.22] described <strong>the</strong> observed failure modes that involved<br />
<strong>end</strong> <strong>plate</strong> failure, bolt fracture, bolt stripping, weld fracture and column<br />
web buckling. They used <strong>the</strong>se test results to validate finite element models for<br />
<strong>the</strong> analysis <strong>of</strong> this joint type. In <strong>the</strong>ir tests, most <strong>of</strong> <strong>the</strong> specimens were full<br />
strength joints but <strong>the</strong>y also tested partial strength joints.<br />
More recently, Adegoke and Kemp [1.23] reported on three tests on thin<br />
<strong>end</strong> <strong>plate</strong> partial strength joints that use a similar column/beam set and different<br />
<strong>plate</strong> thickness. These tests provide insight into <strong>the</strong> joint resistance and <strong>ductility</strong><br />
properties. The observed failure modes included failure <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> and<br />
bolt, development <strong>of</strong> cracks in <strong>the</strong> <strong>end</strong> <strong>plate</strong> along <strong>the</strong> weld to <strong>the</strong> beam web in<br />
<strong>the</strong> tension zone that led to fracture <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> in <strong>the</strong> thinner <strong>plate</strong>s [1.23].<br />
Fracture <strong>of</strong> <strong>the</strong> bolt in tension below <strong>the</strong> tension flange determined collapse for
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
<strong>the</strong> thicker <strong>plate</strong> [1.23]. The test results were compared with a bilinear M-Φ relationship<br />
proposed by <strong>the</strong> authors. They also identified <strong>the</strong> influence <strong>of</strong> <strong>the</strong><br />
<strong>plate</strong> thickness on <strong>the</strong> membrane effect and material strain hardening.<br />
Amongst those researchers focusing exclusively on <strong>the</strong> <strong>end</strong> <strong>plate</strong> behaviour,<br />
Zandonini and Zanon [1.24] performed five static tests on ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong><br />
connections with four bolts in tension and different <strong>plate</strong> thickness. In order to<br />
isolate <strong>the</strong> <strong>end</strong> <strong>plate</strong> behaviour, <strong>the</strong> specimens were connected to a counter<br />
beam with minor deformability. The moment capacity <strong>of</strong> <strong>the</strong> connection in all<br />
tests was greater than <strong>the</strong> plastic moment <strong>of</strong> <strong>the</strong> beam. Bursi [1.25] used <strong>the</strong>se<br />
test results to evaluate <strong>the</strong> plastic failure moment capacity <strong>of</strong> <strong>the</strong> tested connections<br />
by means <strong>of</strong> numerical modelling. He compared <strong>the</strong> yield line paths and<br />
failure mechanism models defined numerically with experimental evidence and<br />
found a good agreement between both.<br />
The numerical simulation <strong>of</strong> <strong>bolted</strong> connections also represents a significant<br />
part <strong>of</strong> <strong>the</strong> research work devoted to <strong>end</strong> <strong>plate</strong> behaviour. Krishnamurthy and<br />
co-workers [1.26-1.28] carried out a comprehensive research programme to investigate<br />
<strong>the</strong> rotational response <strong>of</strong> this joint type by means <strong>of</strong> FE analyses.<br />
The objective <strong>of</strong> <strong>the</strong>ir research was <strong>the</strong> development <strong>of</strong> rotational design criteria<br />
applicable to <strong>end</strong> <strong>plate</strong> connections. They performed three-dimensional FE<br />
analyses on <strong>bolted</strong> connections and correlated <strong>the</strong> results to previous twodimensional<br />
analyses to enable <strong>the</strong> prediction <strong>of</strong> <strong>the</strong> more accurate threedimensional<br />
values from <strong>the</strong> less expensive two-dimensional results. Having<br />
validated <strong>the</strong> computer analyses, <strong>the</strong>y proposed equations to predict <strong>the</strong> general<br />
rotational behaviour. However, <strong>the</strong>y overlooked some important phenomena as<br />
<strong>the</strong> flexibility <strong>of</strong> <strong>the</strong> column flange, bolt head and nut, or plasticity <strong>of</strong> <strong>the</strong> material.<br />
Kukreti and collaborators [1.29-1.31] focused on a similar topic. They developed<br />
an analytical methodology based on FE results to characterize <strong>the</strong> M-Φ<br />
behaviour <strong>of</strong> this joint type. Experimental tests were also carried out to verify<br />
<strong>the</strong> methodology. Bahaari and Sherbourne [1.32-1.36] did a series <strong>of</strong> FE analyses<br />
to propose analytical expressions for <strong>the</strong> design <strong>of</strong> <strong>end</strong> <strong>plate</strong> connections. In<br />
<strong>the</strong>ir models <strong>the</strong>y considered all major influences on <strong>the</strong> overall response, including<br />
column, beam, bolt components, material plasticity, strain hardening<br />
and contact phenomena. Bursi and Jaspart [1.37-1.38] gave some recomm<strong>end</strong>ations<br />
on FE modelling <strong>of</strong> <strong>end</strong> <strong>plate</strong> behaviour. Choi and Chung [1.39] developed<br />
a refined three-dimensional FE model for <strong>the</strong> detailed investigation <strong>of</strong> <strong>the</strong><br />
behaviour <strong>of</strong> <strong>end</strong> <strong>plate</strong> connections. Their model accounted for different types<br />
<strong>of</strong> nonlinearities, such as elastoplasticity and contact. They made a thorough<br />
description <strong>of</strong> <strong>the</strong> contact regions <strong>of</strong> <strong>the</strong> joints with increasing loading.<br />
1.1.2 Scope <strong>of</strong> <strong>the</strong> work, objectives and research approach<br />
The research work reported herein focuses on <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> rotational<br />
behaviour <strong>of</strong> <strong>bolted</strong> beam-to-column joints with an ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong>,<br />
7
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
similar to that shown in Fig. 1.1. In this joint type, <strong>the</strong> main source <strong>of</strong> deformability<br />
is <strong>the</strong> tension zone that can be idealized by means <strong>of</strong> equivalent T-stubs,<br />
which correspond to two T-shaped elements connected through <strong>the</strong> flanges by<br />
means <strong>of</strong> one or more bolt rows. This idealization is also adopted in modern<br />
design codes, as <strong>the</strong> Eurocode 3. The models for <strong>the</strong> column and <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
sides are different. The T-stub elements on <strong>the</strong> column flange side are generally<br />
hot rolled pr<strong>of</strong>iles, whilst on <strong>the</strong> <strong>end</strong> <strong>plate</strong> side such elements comprise two<br />
welded <strong>plate</strong>s, <strong>the</strong> <strong>end</strong> <strong>plate</strong> and <strong>the</strong> beam flange, and a fur<strong>the</strong>r additional stiffener<br />
that corresponds to <strong>the</strong> beam web (Fig. 1.2c). The first model (HR-T-stub)<br />
has been extensively studied over <strong>the</strong> past years and was <strong>the</strong> aim <strong>of</strong> several research<br />
programmes that are reported in technical literature. The current approach<br />
to account for <strong>the</strong> behaviour <strong>of</strong> T-stubs made up <strong>of</strong> welded <strong>plate</strong>s (WP-<br />
T-stub) consists in a mere extrapolation <strong>of</strong> <strong>the</strong> existing rules for <strong>the</strong> o<strong>the</strong>r assembly<br />
type. This assumption can be erroneous and can lead to unsafe estimations<br />
<strong>of</strong> <strong>the</strong> characteristic properties. To deal with this problem, a research project<br />
was devised to increase <strong>the</strong> knowledge and understanding <strong>of</strong> <strong>end</strong> <strong>plate</strong> behaviour<br />
and contribute towards <strong>the</strong> improvement <strong>of</strong> its design.<br />
Simultaneously, <strong>the</strong> issue <strong>of</strong> available <strong>ductility</strong> is also addressed in this<br />
work. The knowledge <strong>of</strong> <strong>the</strong> plastic rotation capacity <strong>of</strong> beams is particularly<br />
important in <strong>the</strong> case <strong>of</strong> full strength beam-to-column joints, because yielding<br />
occurs at <strong>the</strong> member <strong>end</strong>s. In <strong>the</strong> case <strong>of</strong> partial strength joints, <strong>the</strong>re is significant<br />
yielding <strong>of</strong> <strong>the</strong> connection and <strong>the</strong> evaluation <strong>of</strong> its <strong>ductility</strong> becomes<br />
crucial. The <strong>ductility</strong> <strong>of</strong> a joint reflects <strong>the</strong> length <strong>of</strong> <strong>the</strong> yield <strong>plate</strong>au <strong>of</strong> <strong>the</strong> M-<br />
Φ response and is intrinsically linked to <strong>the</strong> rotational capacity <strong>of</strong> <strong>the</strong> joint.<br />
The research described in this dissertation is divided into experimental,<br />
numerical (FE modelling) and analytical works. Reliable test results are essential<br />
and support <strong>the</strong> validity <strong>of</strong> analytical and numerical work. Numerical<br />
analyses are important as <strong>the</strong>y provide a means <strong>of</strong> carrying out wide-ranging<br />
parametric studies to complement existing experimental results. Analytical<br />
work allows <strong>the</strong> development <strong>of</strong> relatively simple design models that can be<br />
used in practice.<br />
The experimental programme was conducted at <strong>the</strong> Delft University <strong>of</strong><br />
Technology and included <strong>the</strong> (monotonic) testing <strong>of</strong> thirty-two individual Tstub<br />
connections made up <strong>of</strong> welded <strong>plate</strong>s and eight full-scale single sided<br />
beam-to-column joints. The primary intent <strong>of</strong> <strong>the</strong> first series <strong>of</strong> tests on isolated<br />
T-stubs was to provide insight into <strong>the</strong> actual behaviour <strong>of</strong> this type <strong>of</strong> connection,<br />
failure modes and deformation capacity. The parameters affecting <strong>the</strong> deformation<br />
response <strong>of</strong> <strong>bolted</strong> T-stubs were identified and <strong>the</strong>ir influence on <strong>the</strong><br />
overall behaviour <strong>of</strong> <strong>the</strong> connection was qualitatively and quantitatively assessed.<br />
In addition, <strong>the</strong> role <strong>of</strong> <strong>the</strong> welding and <strong>the</strong> presence <strong>of</strong> transverse stiffeners<br />
were tackled. For <strong>the</strong> follow up study on ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections,<br />
<strong>the</strong> main objective was <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> ultimate behaviour <strong>of</strong> <strong>the</strong> assembly<br />
<strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing-bolts and eventually <strong>the</strong> proposal <strong>of</strong> sound design rules for<br />
this elemental part within <strong>the</strong> framework <strong>of</strong> <strong>the</strong> so-called component method<br />
[1.1,1.40].<br />
8
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
The numerical part <strong>of</strong> <strong>the</strong> work included <strong>the</strong> assessment <strong>of</strong> <strong>the</strong> loadcarrying<br />
behaviour <strong>of</strong> single T-stubs and <strong>the</strong> exploration <strong>of</strong> o<strong>the</strong>r model features,<br />
namely <strong>the</strong> prying effect and <strong>the</strong> variation <strong>of</strong> contact flange surfaces<br />
within <strong>the</strong> course <strong>of</strong> loading. In this context, two T-stub connections representative<br />
<strong>of</strong> HR- and WP-T-stubs were modelled and calibrated against experimental<br />
results. Having validated a three-dimensional FE model for <strong>the</strong> individual<br />
T-stubs, a parametric study was conducted in order to provide a better understanding<br />
<strong>of</strong> <strong>the</strong> overall behaviour and to evaluate <strong>the</strong> influence <strong>of</strong> <strong>the</strong> main<br />
parameters on <strong>the</strong> connection deformability.<br />
The analytical approach <strong>of</strong> <strong>the</strong> research involved:<br />
(i) The proposal <strong>of</strong> a simplified beam model for <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> T-stub<br />
response. With this simplification some information and features <strong>of</strong> <strong>the</strong> T-stub<br />
model may be lost. However, this methodology overcomes <strong>the</strong> complexity <strong>of</strong><br />
<strong>the</strong> above approaches and is less time-consuming.<br />
(ii) The assessment <strong>of</strong> <strong>the</strong> global M-Φ response <strong>of</strong> an <strong>end</strong> <strong>plate</strong> connection based<br />
on <strong>the</strong> component method. A s<strong>of</strong>tware tool developed at <strong>the</strong> University <strong>of</strong><br />
Coimbra [1.41] was used for this assessment. The outcomes were validated<br />
through comparison with experimental evidence.<br />
1.1.3 Outline <strong>of</strong> <strong>the</strong> dissertation<br />
The dissertation is divided into three parts.<br />
Part 1 (Chapter 1) presents background material on ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections.<br />
References to previous research work on <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong><br />
rotational behaviour <strong>of</strong> this joint type are made. Special emphasis is given to<br />
<strong>the</strong> component method for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> M-Φ response.<br />
Part 2 contains five chapters and includes fur<strong>the</strong>r developments on <strong>the</strong> Tstub<br />
model. Chapter 2 is a brief introduction. Chapter 3 describes <strong>the</strong> experimental<br />
programme on isolated T-stub connections made up <strong>of</strong> welded <strong>plate</strong>s.<br />
Chapter 4 includes <strong>the</strong> numerical evaluation <strong>of</strong> <strong>the</strong> force-deformation (F-∆) response<br />
<strong>of</strong> T-stubs. A three-dimensional FE model is recomm<strong>end</strong>ed for that<br />
purpose. In both chapters, detailed results are given for benchmark specimens<br />
and <strong>the</strong> approaches are validated. A parametric study is described in Chapter 5.<br />
It provides insight into <strong>the</strong> main behavioural features <strong>of</strong> T-stub connections and<br />
highlights <strong>the</strong> parameters that affect <strong>the</strong>ir deformability. A two-dimensional<br />
simplified model that provides analytical solutions for <strong>the</strong> F-∆ response is proposed<br />
in Chapter 6. Because <strong>ductility</strong> is such an important characteristic <strong>of</strong><br />
connection performance, this chapter emphasises <strong>the</strong> evaluation <strong>of</strong> deformation<br />
capacity <strong>of</strong> isolated T-stubs.<br />
Part 3 contains Chapters 7 and 8. Chapter 7 is entirely dedicated to <strong>the</strong> experiments<br />
on ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong>s connections. All <strong>the</strong> test details are provided<br />
and <strong>the</strong> results are thoroughly analysed. Chapter 8 presents a <strong>ductility</strong> analysis<br />
where <strong>the</strong> experimental results for <strong>the</strong> overall <strong>end</strong> <strong>plate</strong> connection are con-<br />
9
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
fronted with <strong>the</strong> component tests. For that purpose, a procedure based on <strong>the</strong><br />
component methodology is recomm<strong>end</strong>ed. Comparisons with o<strong>the</strong>r proposals<br />
from <strong>the</strong> literature are also drawn.<br />
Finally, conclusions and recomm<strong>end</strong>ations are summarized in Chapter 9.<br />
1.2 DEFINITIONS<br />
Beam-to-column joints consist <strong>of</strong> a web panel and one or two connections (single-<br />
or double-sided joint configuration) – Fig. 1.3. The web panel zone includes<br />
<strong>the</strong> column web and <strong>the</strong> flange(s) <strong>of</strong> <strong>the</strong> column for <strong>the</strong> height <strong>of</strong> <strong>the</strong><br />
connected beam pr<strong>of</strong>ile(s). The connection is <strong>the</strong> location where two members<br />
are interconnected and <strong>the</strong> means <strong>of</strong> interconnection, i.e. <strong>the</strong> set <strong>of</strong> physical<br />
components that mechanically fasten <strong>the</strong> connected elements.<br />
The behaviour <strong>of</strong> a steel beam-to-column joint is represented by a M-Φ<br />
curve, as already explained. The rotational deformation <strong>of</strong> a joint, Φ, results<br />
from <strong>the</strong> in-plane b<strong>end</strong>ing, M, and is <strong>the</strong> sum <strong>of</strong> <strong>the</strong> shear deformation <strong>of</strong> <strong>the</strong><br />
column web panel zone, γ, and <strong>the</strong> connection deformation, φ. The deformation<br />
<strong>of</strong> <strong>the</strong> connection includes <strong>the</strong> deformation <strong>of</strong> <strong>the</strong> fastening elements (bolts, <strong>end</strong><br />
<strong>plate</strong>, etc.) and <strong>the</strong> load-introduction deformation <strong>of</strong> <strong>the</strong> column web. It results<br />
in a relative rotation between <strong>the</strong> beam and column axes, θb and θc, which is<br />
equal to:<br />
φ = θb − θc<br />
(1.1)<br />
according to Fig 1.4, and provides a flexural deformability curve M-φ. This deformability<br />
is only due to <strong>the</strong> couple <strong>of</strong> forces Fb transferred by <strong>the</strong> flanges <strong>of</strong><br />
<strong>the</strong> beam that are statically equivalent to <strong>the</strong> b<strong>end</strong>ing moment M acting on <strong>the</strong><br />
beam. In this figure, z is <strong>the</strong> lever arm.<br />
The shear deformation <strong>of</strong> <strong>the</strong> column web panel is associated with <strong>the</strong> force<br />
Vwp acting in this panel and leads to a relative rotation γ between <strong>the</strong> beam and<br />
column axes. A shear deformability curve Vwp-γ may <strong>the</strong>n be established. For a<br />
Joint<br />
Fig. 1.3 Parts <strong>of</strong> a beam-to-column joint (single-sided configuration).<br />
10<br />
Web panel<br />
Connection
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
single-sided joint configuration (see Fig. 1.5), <strong>the</strong> shear action in <strong>the</strong> panel is<br />
related to <strong>the</strong> internal actions on <strong>the</strong> joint as follows:<br />
M ⎡ z ⎤ M<br />
Vwp = ⎢1+ ( Vc1 + Vc2)<br />
= β<br />
2<br />
⎥<br />
(1.2)<br />
z ⎣ M ⎦ z<br />
The transformation parameter β relates <strong>the</strong> web panel shear force, Vwp, with <strong>the</strong><br />
internal actions. Conservative values for <strong>the</strong> transformation parameter β, neglecting<br />
<strong>the</strong> effect <strong>of</strong> <strong>the</strong> shear force in <strong>the</strong> column, are suggested in Eurocode<br />
3: (i) β = 1, in <strong>the</strong> case <strong>of</strong> single-sided joints, (ii) β = 2, in <strong>the</strong> case <strong>of</strong> doublesided<br />
joints with equal but unbalanced <strong>end</strong> b<strong>end</strong>ing moments and (iii) β = 0, in<br />
<strong>the</strong> case <strong>of</strong> double-sided joints with balanced <strong>end</strong> b<strong>end</strong>ing moments.<br />
The global M-Φ response <strong>of</strong> <strong>the</strong> joint is obtained by summing <strong>the</strong> contribu-<br />
θc<br />
Fig. 1.4 Sources <strong>of</strong> connection deformability.<br />
Mc1<br />
Nc2<br />
Nc1<br />
Vc2<br />
Mc2<br />
Vc1<br />
θb<br />
Fb<br />
Fb<br />
Vb<br />
M<br />
z<br />
Nb Mb = M<br />
Fig. 1.5 Internal forces acting on <strong>the</strong> joint (single-sided configuration).<br />
11
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
tions <strong>of</strong> rotation <strong>of</strong> <strong>the</strong> connection (φ) and <strong>of</strong> <strong>the</strong> shear panel (γ), as illustrated<br />
in Fig. 1.6. The M-γ curve is obtained from <strong>the</strong> Vwp-γ by means <strong>of</strong> <strong>the</strong> transformation<br />
parameter β.<br />
M<br />
φi φ<br />
γi γ<br />
Fig. 1.6 Global moment-rotation response <strong>of</strong> a joint.<br />
12<br />
Mi<br />
+<br />
M<br />
Mi<br />
=<br />
M<br />
Mi<br />
Φi (= φi + γi)<br />
1.3 METHODS FOR MODELLING THE ROTATIONAL BEHAVIOUR OF BEAM-<br />
TO-COLUMN JOINTS<br />
1.3.1 Generality<br />
The <strong>characterization</strong> <strong>of</strong> <strong>the</strong> M-Φ curve can be ascertained by experimental testing<br />
or ma<strong>the</strong>matical models based on <strong>the</strong> geometrical and mechanical properties<br />
<strong>of</strong> <strong>the</strong> joint. Full-scale experimental tests are naturally <strong>the</strong> most reliable<br />
method <strong>of</strong> description <strong>of</strong> <strong>the</strong> rotational behaviour <strong>of</strong> structural joints. However,<br />
<strong>the</strong>y are time consuming, expensive and cannot certainly be regarded as a design<br />
tool. In addition, <strong>the</strong> data ga<strong>the</strong>red from tests <strong>of</strong> prototype joints are few<br />
and generally limited to displacement and surface measurements, as strain<br />
measurements, for instance. Therefore <strong>the</strong> results cannot be ext<strong>end</strong>ed to different<br />
joint configurations. None<strong>the</strong>less, tests provide accurate information on <strong>the</strong><br />
joint response that is used to validate ma<strong>the</strong>matical models <strong>of</strong> prediction <strong>of</strong> <strong>the</strong><br />
M-Φ curve. Ma<strong>the</strong>matical models for representation <strong>of</strong> <strong>the</strong> curve include: (i)<br />
curve fitting to test results by regression analysis, (ii) simplified analytical<br />
models, (iii) mechanical models that take into account <strong>the</strong> various sources <strong>of</strong><br />
joint deformability and (iv) numerical models. For a review <strong>of</strong> different methods,<br />
<strong>the</strong> reader should refer to Ne<strong>the</strong>rcot and Zandonini [1.42].<br />
Mechanical models are <strong>the</strong> most effective solution for an accurate description<br />
<strong>of</strong> <strong>the</strong> complex nature <strong>of</strong> <strong>bolted</strong> joint behaviour. These models use a set <strong>of</strong><br />
rigid and flexible elements to simulate <strong>the</strong> overall joint. The interplay between<br />
<strong>the</strong>se elements results in different mechanical models, as explained below.<br />
1.3.2 The component method<br />
Current design practice adopts <strong>the</strong> so-called component method for <strong>the</strong> prediction<br />
<strong>of</strong> <strong>the</strong> rotational behaviour <strong>of</strong> beam-to-column joints. For <strong>the</strong> purposes <strong>of</strong><br />
Φ
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
simplicity, any joint can be subdivided into three different zones: tension,<br />
compression and shear. Within each zone, several sources <strong>of</strong> deformability can<br />
be identified, which are simple elemental parts (or “components”) that contribute<br />
to <strong>the</strong> overall response <strong>of</strong> <strong>the</strong> joint. From a <strong>the</strong>oretical point <strong>of</strong> view, this<br />
methodology can be applied to any joint configuration and loading conditions<br />
provided that <strong>the</strong> basic components are properly characterized. Essentially, <strong>the</strong><br />
method comprises three basic steps: (i) identification <strong>of</strong> <strong>the</strong> active components<br />
for a given structural joint, (ii) <strong>characterization</strong> <strong>of</strong> <strong>the</strong> individual component F-<br />
∆ response and (iii) assembly <strong>of</strong> those elements into a mechanical model made<br />
up <strong>of</strong> extensional springs and rigid links. This spring assembly is treated as a<br />
structure, whose F-∆ behaviour is used to generate <strong>the</strong> M-Φ curve <strong>of</strong> <strong>the</strong> full<br />
joint.<br />
The method is illustrated in Fig. 1.7 for <strong>the</strong> particular case <strong>of</strong> a <strong>bolted</strong> ext<strong>end</strong>ed<br />
<strong>end</strong> <strong>plate</strong> connection (with two bolt rows in tension). For <strong>the</strong> computation<br />
<strong>of</strong> <strong>the</strong> joint rotational stiffness, <strong>the</strong> active joint components for this configuration,<br />
according to Eurocode 3, are: column web in shear (cws), column<br />
web in compression (cwc), column web in tension (cwt), column flange in<br />
b<strong>end</strong>ing (cfb), <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing (epb), and bolts in tension (bt). The welds<br />
connecting <strong>the</strong> <strong>end</strong> <strong>plate</strong> and <strong>the</strong> beam are not taken into account for computation<br />
<strong>of</strong> <strong>the</strong> rotational stiffness, as well as components beam web and flange in<br />
compression (bfc) and beam web in tension (bwt). Each component is characterized<br />
by a nonlinear F-∆ response, which can be obtained by means <strong>of</strong> experimental<br />
tests or analytical models. These individual components are assembled<br />
into a mechanical model in order to evaluate <strong>the</strong> M-Φ response <strong>of</strong> <strong>the</strong><br />
whole joint. The Eurocode 3 spring model is represented in Fig. 1.7 [1.40]. Alternative<br />
spring and rigid link models are proposed in literature, as <strong>the</strong> “Innsbruck<br />
model” proposed by Huber and Tschemmernegg [1.43]. Essentially, <strong>the</strong>y<br />
share <strong>the</strong> same basic components but assume different component interplay.<br />
M<br />
(cws)<br />
(cwc)<br />
(cwt.1) (cfb.1) (epb.1) (bt.1)<br />
(cwt.2) (cfb.2) (epb.2) (bt.2)<br />
Fig. 1.7 Component method: active components and mechanical model adopted<br />
by Eurocode 3 for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> joint rotational stiffness.<br />
Φ<br />
M<br />
13
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
1.4 CHARACTERIZATION OF BASIC COMPONENTS OF BOLTED JOINTS IN<br />
TERMS OF PLASTIC RESISTANCE AND INITIAL STIFFNESS<br />
Within <strong>the</strong> framework <strong>of</strong> <strong>the</strong> component method, <strong>the</strong> basic joint components<br />
are modelled by means <strong>of</strong> nonlinear extensional springs (Fig. 1.8a; K: spring<br />
axial stiffness). This complex behaviour can be approximated with simple relationships<br />
without significant loss <strong>of</strong> accuracy. The elastic-perfectly plastic response<br />
is one <strong>of</strong> <strong>the</strong> simplest possible idealizations. Following <strong>the</strong> Eurocode 3<br />
approach for idealization <strong>of</strong> <strong>the</strong> flexural joint spring nonlinear behaviour, this<br />
response is characterized by a secant stiffness, ke/η, and a full plastic resistance,<br />
FRd (Fig. 1.8b). ke is <strong>the</strong> initial stiffness <strong>of</strong> <strong>the</strong> component and η is a<br />
stiffness modification coefficient. Eurocode 3 defines this coefficient for different<br />
types <strong>of</strong> connections. For a single component, similar values can be<br />
adopted. The post-limit stiffness, kp-l is taken as zero, which means that strain<br />
hardening and geometric nonlinear effects are neglected. Regarding <strong>the</strong> component<br />
<strong>ductility</strong>, i.e. <strong>the</strong> extension <strong>of</strong> <strong>the</strong> plastic <strong>plate</strong>au, <strong>the</strong> code [1.1] presents<br />
some qualitative principles that are however insufficient. For instance, <strong>the</strong><br />
component column web in shear has very high <strong>ductility</strong> and <strong>the</strong>refore <strong>the</strong> deformation<br />
capacity is taken as infinite; on <strong>the</strong> o<strong>the</strong>r hand, <strong>the</strong> bolts in tension<br />
are brittle components with no plastic <strong>plate</strong>au.<br />
The following sections present <strong>the</strong> formulations adopted in Eurocode 3 for<br />
prediction <strong>of</strong> <strong>the</strong> plastic resistance and initial stiffness <strong>of</strong> <strong>the</strong> basic components<br />
<strong>of</strong> <strong>bolted</strong> joints. Particular attention is devoted to <strong>the</strong> T-stub model that is used<br />
to idealize <strong>the</strong> tension zone <strong>of</strong> this joint typology. Then, <strong>the</strong> remaining components<br />
are briefly analysed.<br />
F<br />
(a) Extensional spring representing a generic component.<br />
F<br />
14<br />
FRd<br />
ke/η<br />
K<br />
∆<br />
(b) Actual behaviour and elastic-plastic response.<br />
Fig. 1.8 Modelling <strong>of</strong> a component subjected to compression.<br />
∆<br />
Actual behaviour<br />
Elastic-plastic approximation
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
1.4.1 T-stub model for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> tension zone <strong>of</strong> <strong>bolted</strong> joints<br />
The equivalent T-stub corresponds to two T-shaped elements connected<br />
through <strong>the</strong> flanges by means <strong>of</strong> one or more bolt rows as depicted in Fig. 1.9.<br />
The main behavioural aspects <strong>of</strong> <strong>the</strong> T-stub as a standalone configuration have<br />
been widely investigated over <strong>the</strong> past thirty years, both experimentally and<br />
<strong>the</strong>oretically. As a result, <strong>the</strong> structural response <strong>of</strong> this kind <strong>of</strong> connection is<br />
thoroughly known in elastic and plastic ranges, and appropriate design rules for<br />
prediction <strong>of</strong> <strong>the</strong> elastic-plastic F-∆ curve have been assessed.<br />
e<br />
Transverse view<br />
Fig. 1.9 Equivalent T-stub (one bolt row).<br />
w<br />
bf<br />
e<br />
e1<br />
b<br />
0.5p<br />
1.4.1.1 Plastic resistance <strong>of</strong> single T-stub connections<br />
Lateral view<br />
Plan<br />
The evaluation <strong>of</strong> <strong>the</strong> plastic (design) resistance <strong>of</strong> <strong>bolted</strong> T-stub connections is<br />
based on <strong>the</strong> well-known yield line principle. The works <strong>of</strong> Zoetemeijer [1.2],<br />
Packer and Morris [1.3] and Mann and Morris [1.11] form <strong>the</strong> basis <strong>of</strong> <strong>the</strong> procedure<br />
presented below. Zoetemeijer suggests that <strong>the</strong> determination <strong>of</strong> <strong>the</strong><br />
plastic resistance <strong>of</strong> such a connection type is based on <strong>the</strong> plastic behaviour <strong>of</strong><br />
15
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
<strong>the</strong> flanges and <strong>the</strong> bolts, and assumes that <strong>the</strong> yielding is large enough to allow<br />
<strong>the</strong> adoption <strong>of</strong> <strong>the</strong> most favourable static equilibrium [1.2]. For <strong>the</strong> purposes<br />
<strong>of</strong> simplicity, consider a <strong>bolted</strong> T-stub with one bolt row only. This simple connection<br />
can fail according to three possible “plastic” collapse mechanisms, as<br />
illustrated in Fig. 1.10. Type-1 mechanism is characterized by <strong>the</strong> formation <strong>of</strong><br />
four plastic hinges: two hinges are located at <strong>the</strong> bolt axes, due to <strong>the</strong> b<strong>end</strong>ing<br />
moment caused by <strong>the</strong> prying forces, Q and <strong>the</strong> o<strong>the</strong>r two hinges are located at<br />
<strong>the</strong> flange-to-web connection. The formation <strong>of</strong> two plastic hinges at <strong>the</strong><br />
flange-to-web connection and <strong>the</strong> failure <strong>of</strong> <strong>the</strong> bolts typify type-2 mechanism.<br />
The third collapse mechanism involves bolt failure only. A fourth supplementary<br />
mechanism corresponds to <strong>the</strong> metal shear tearing around <strong>the</strong> bolt head or<br />
washer but is not relevant in most cases. The resistance corresponding to each<br />
collapse mechanism is easily computed by establishing <strong>the</strong> equilibrium equations<br />
in <strong>the</strong> plastic conditions (cf. App<strong>end</strong>ix A). The plastic (design) resistance<br />
<strong>of</strong> <strong>the</strong> T-stub, FRd.0, corresponds to <strong>the</strong> smallest value among <strong>the</strong> examined<br />
“plastic” modes, i.e. FRd.0 = min (F1.Rd.0, F2.Rd.0, F3.Rd.0), where:<br />
4M f . Rd<br />
F1.<br />
Rd .0<br />
=<br />
m<br />
(1.3)<br />
2M f. Rd+ 2BRdn 2M f . Rd⎡ ( 2 − β Rd ) λ ⎤<br />
F2.<br />
Rd .0 = = ⎢1+ ⎥<br />
m+ n m ⎢⎣ βRd ( 1+<br />
λ ) ⎥⎦<br />
(1.4)<br />
F3. Rd.0 = 2BRd<br />
(1.5)<br />
The plastic flexural resistance <strong>of</strong> <strong>the</strong> T-flanges, Mf.Rd, is given by:<br />
2<br />
t f<br />
M f . Rd= 4<br />
fy. fbeff (1.6)<br />
where beff is <strong>the</strong> effective width tributary to one bolt row, tf is <strong>the</strong> flange thick-<br />
Q<br />
16<br />
B<br />
F1.Rd.0<br />
B<br />
n m m n<br />
(a) Type-1:<br />
⎛ 2λ<br />
⎞<br />
⎜βRd ≤ ⎟<br />
⎝ 2λ+ 1⎠<br />
.<br />
Q<br />
b<br />
(=F1.Rd.0/2+Q)<br />
Mf.Rd<br />
Mf.Rd<br />
Q<br />
BRd<br />
F2.Rd.0<br />
n m m n<br />
BRd<br />
Q<br />
(b) Type-2 (ξ ≤ 1.0):<br />
⎛ 2λ<br />
⎞<br />
⎜ < βRd<br />
≤ 2⎟<br />
⎝2λ+ 1 ⎠ .<br />
b<br />
Mf.Rd<br />
ξMf.Rd<br />
Fig. 1.10 “Plastic” collapse mechanisms <strong>of</strong> <strong>bolted</strong> T-stubs.<br />
BRd<br />
F3.Rd.0<br />
n m m n<br />
BRd<br />
b<br />
ξMf.Rd<br />
(c) Type-3 (ξ ≤ 1.0):<br />
β > 2 .<br />
( )<br />
Rd
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
ness and fy.f is <strong>the</strong> yield stress <strong>of</strong> <strong>the</strong> flanges. The length m represents <strong>the</strong> distance<br />
between <strong>the</strong> bolt axis and <strong>the</strong> section corresponding to <strong>the</strong> plastic hinge at<br />
<strong>the</strong> flange-to-web connection. According to Eurocode 3, m= d − ζ s,<br />
where d<br />
represents <strong>the</strong> length between <strong>the</strong> bolt axis and <strong>the</strong> face <strong>of</strong> <strong>the</strong> T-stub element<br />
web, ζ is a coefficient taken as 0.8 and s = r or s = 2aw , for hot rolled pr<strong>of</strong>iles<br />
or welded <strong>plate</strong>s as T-stub, respectively; r is <strong>the</strong> fillet radius <strong>of</strong> <strong>the</strong> flangeto-web<br />
connection and aw is <strong>the</strong> throat thickness <strong>of</strong> <strong>the</strong> fillet weld. The geometrical<br />
parameter λ is defined as <strong>the</strong> ratio n/m, being n <strong>the</strong> effective edge distance.<br />
In Eurocode 3, n is taken as <strong>the</strong> minimum value <strong>of</strong> e (distance between<br />
<strong>the</strong> bolt axis and <strong>the</strong> tip <strong>of</strong> <strong>the</strong> flanges) and 1.25m, i.e. n = min (e,1.25m). BRd is<br />
<strong>the</strong> “plastic” (design) resistance <strong>of</strong> a single bolt in tension.<br />
The β-ratio is <strong>the</strong> relation between <strong>the</strong> flexural resistance <strong>of</strong> <strong>the</strong> flanges and<br />
<strong>the</strong> axial resistance <strong>of</strong> <strong>the</strong> bolts and governs <strong>the</strong> occurrence <strong>of</strong> a given (“plastic”)<br />
collapse mode (Fig. 1.11). At plastic conditions, this parameter, βRd, is <strong>the</strong><br />
ratio between <strong>the</strong> plastic resistances corresponding to type-1 mechanism and<br />
that corresponding to a type-3 mechanism:<br />
2M f. Rd<br />
β Rd = (1.7)<br />
BRd m<br />
The basic formulations presented above do not cater for <strong>the</strong> influence <strong>of</strong> <strong>the</strong><br />
moment-shear interaction on <strong>the</strong> resistance <strong>of</strong> <strong>bolted</strong> T-stubs that can lead to a<br />
decrease in <strong>the</strong> plastic resistance. Faella et al. [1.44] assume that such interaction<br />
can be taken into account under <strong>the</strong> Von Mises yield criterion. Analytical<br />
expressions for type-1 and type-2 mechanisms allowing for moment-shear interaction<br />
are derived in App<strong>end</strong>ix A and are reproduced below:<br />
F<br />
2<br />
8⎛m⎞ ⎡ 3 ⎤ M<br />
= ⎜ 1+ −1<br />
⎜<br />
⎟<br />
t ⎟ ⎢ ⎥<br />
⎝ ⎠ ⎢ ( mtf<br />
) ⎥ m<br />
⎣ ⎦<br />
1. Rd .0 2<br />
3 f<br />
F<br />
2B<br />
Rd<br />
1<br />
2λ<br />
2λ+ 1<br />
Type-1<br />
2λ<br />
2λ+ 1<br />
Type-2<br />
f. Rd<br />
1 2<br />
Non-circular yield line patterns<br />
Circular yield line patterns<br />
Type-3<br />
βRd<br />
(1.8)<br />
Fig. 1.11 Influence <strong>of</strong> βRd on <strong>the</strong> “plastic” collapse mechanism <strong>of</strong> <strong>bolted</strong> T-stubs.<br />
17
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
and:<br />
2<br />
16 ⎛m⎞ ⎡ ⎤<br />
3 2λβRd + 1 M f. Rd<br />
F2.<br />
Rd .0 = ⎜ ⎟ ( 1+ λ ) ⎢ 1+ −1⎥<br />
(1.9)<br />
2 2<br />
3 ⎜t ⎟ ⎢<br />
f<br />
4 ( mtf<br />
) ( 1+<br />
λ ) ⎥<br />
⎝ ⎠ m<br />
⎢⎣ ⎥⎦<br />
Naturally, <strong>the</strong> “plastic” resistance for mechanism type-3 is not affected by this<br />
interaction.<br />
Regarding mechanism type-1, a significant increase in resistance can be expected<br />
due to <strong>the</strong> influence <strong>of</strong> <strong>the</strong> bolt action on a finite contact area. Jaspart<br />
[1.20] suggests an alternative formulation to cater for this effect (App<strong>end</strong>ix A):<br />
( 32n−2dw) M f . Rd 32λm−2d M w f . Rd<br />
F1.<br />
Rd .0 = =<br />
(1.10)<br />
8mn − dw( m + n) 8λm− dw( 1+<br />
λ)<br />
m<br />
whereby dw is <strong>the</strong> bolt head, nut or washer diameter, as appropriate.<br />
By combining both effects for type-1 mechanism, <strong>the</strong> plastic resistance can<br />
be expressed as:<br />
18<br />
2<br />
16 ⎛ m ⎞<br />
F1.<br />
Rd .0 = ⎜<br />
3 ⎜<br />
⎟<br />
t ⎟<br />
⎝ f ⎠<br />
⎡<br />
Γ ⎢<br />
⎢<br />
⎣<br />
3<br />
1+ 2<br />
2<br />
4Γ<br />
( mtf<br />
)<br />
⎤ M<br />
−1⎥<br />
⎥ m<br />
⎦<br />
λm<br />
8 − ( 1+<br />
λ )<br />
dw<br />
Γ=<br />
λm<br />
16 −1<br />
d<br />
w<br />
f. Rd<br />
(1.11)<br />
(1.12)<br />
as derived in App<strong>end</strong>ix A.<br />
The effective width <strong>of</strong> <strong>the</strong> T-element flange, beff, that appears explicitly in<br />
<strong>the</strong> above formulae is a notional width and does not necessarily represent any<br />
physical length <strong>of</strong> <strong>the</strong> flange. beff represents <strong>the</strong> width <strong>of</strong> <strong>the</strong> flange <strong>plate</strong> that<br />
contributes to load transmission. Zoetmeijer has successfully introduced this<br />
concept in [1.2]. It accounts for all possible yield line mechanisms <strong>of</strong> <strong>the</strong> Tstub<br />
flange and cannot exceed <strong>the</strong> actual flange width. This effective length has<br />
to be defined by establishing <strong>the</strong> equivalence, in <strong>the</strong> plastic collapse condition,<br />
between <strong>the</strong> beam model and <strong>the</strong> actual <strong>plate</strong> behaviour where collapse occurs<br />
due to <strong>the</strong> development <strong>of</strong> a yield line mechanism [1.44].<br />
In <strong>the</strong> case <strong>of</strong> a <strong>bolted</strong> T-stub with one bolt row, three possible yield line<br />
mechanisms are considered: (i) circular pattern (Fig. 1.12a): beff.1 = 2πm, (ii)<br />
non-circular pattern (Fig. 1.12b): beff.2 = 4m + 1.25e and (iii) beam pattern (Fig.<br />
1.12c): beff.3 = b.<br />
Regarding <strong>the</strong> circular pattern, beff is determined from <strong>the</strong> equivalence between<br />
<strong>the</strong> failure load that corresponds to <strong>the</strong> collapse mechanism <strong>of</strong> a simply<br />
supported <strong>plate</strong> (P = 2πmpl, mpl = tf 2 fy/4) and that <strong>of</strong> <strong>the</strong> equivalent beam model.<br />
By equating both relationships, <strong>the</strong> following expression is determined for beff.1:<br />
2<br />
t f 4m1 beff .1 = 2πfy= 2πm<br />
(1.13)<br />
2<br />
4 t f<br />
f<br />
y
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
(a) Circular pattern. (b) Non-circular pattern. (c) Beam pattern.<br />
Fig. 1.12 Yield line mechanisms <strong>of</strong> <strong>bolted</strong> T-stubs with one bolt row.<br />
(a) Circular pattern. (b) Non-circular pattern. (c) Beam pattern.<br />
(d) “Circular” pattern. (e) “Non-circular” pattern.<br />
Fig. 1.13 Yield line mechanisms <strong>of</strong> <strong>bolted</strong> T-stubs with two bolt rows.<br />
Referring now to <strong>the</strong> non-circular pattern, Zoetemeijer provides a simplified<br />
expression for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> effective width associated to this mechanism<br />
[1.2]. For <strong>the</strong> beam pattern, <strong>the</strong> computation <strong>of</strong> this length is quite<br />
straightforward. The effective width <strong>of</strong> <strong>the</strong> equivalent T-stub corresponds to<br />
<strong>the</strong> smallest value <strong>of</strong> <strong>the</strong> above, i.e., beff = min (beff.1, beff.2, beff.3).<br />
Now, consider <strong>the</strong> case <strong>of</strong> multiple bolt rows. Dep<strong>end</strong>ing on <strong>the</strong> pitch <strong>of</strong> <strong>the</strong><br />
bolts, p, <strong>the</strong>y may behave as a single bolt row or as a bolt group. For <strong>the</strong> particular<br />
case <strong>of</strong> two bolt rows illustrated in Fig. 1.13, <strong>the</strong> behaviour is such <strong>of</strong> a<br />
group in cases c, d and e and <strong>of</strong> an individual bolt in <strong>the</strong> remaining. The effective<br />
width <strong>of</strong> each bolt row is taken as <strong>the</strong> minimum among <strong>the</strong> five cases: (i)<br />
individual bolt (Fig. 1.13a): beff.1 = 2πm, (ii) individual bolt (Fig. 1.13b): beff.2 =<br />
4m + 1.25e, (iii) bolt group (Fig. 1.13c): beff.3 = b, (iv) bolt group, (Fig. 1.13d):<br />
beff.4 = πm + 0.5p and (v) bolt group (Fig. 1.13e): beff.5 = 2m + 0.625e + 0.5p.<br />
Again, beff = min (beff.1, beff.2, beff.3, beff.4, beff.5).<br />
1.4.1.2 Initial stiffness <strong>of</strong> single T-stub connections<br />
The evaluation <strong>of</strong> <strong>the</strong> initial stiffness <strong>of</strong> a T-stub, ke.0, is based on <strong>the</strong> analysis<br />
19
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
<strong>of</strong> <strong>the</strong> elastic response <strong>of</strong> <strong>the</strong> connection, which has been analysed for <strong>the</strong> first<br />
time by Aggerskov [1.7] and later by Holmes and Martin [1.45] to accommodate<br />
<strong>the</strong> effect <strong>of</strong> <strong>the</strong> prying forces on <strong>the</strong> bolt behaviour. Yee and Melchers<br />
[1.5] adopted a similar procedure for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> elastic deformation<br />
<strong>of</strong> this type <strong>of</strong> connection. The single T-stub element is modelled as a simply<br />
supported beam, <strong>the</strong> supports corresponding to <strong>the</strong> location <strong>of</strong> <strong>the</strong> prying<br />
forces. This system is loaded by a concentrated force applied at <strong>the</strong> mid-span<br />
section, equivalent to <strong>the</strong> force applied on <strong>the</strong> T-stub through <strong>the</strong> web and <strong>the</strong><br />
bolt force acting at <strong>the</strong> bolt axes (Fig. 1.14). The analysis <strong>of</strong> <strong>the</strong> T-stub is carried<br />
out by taking <strong>the</strong> interaction <strong>of</strong> <strong>the</strong> two T-stub elements and <strong>the</strong> bolts into<br />
consideration as well as <strong>the</strong> compatibility requirements to cater for <strong>the</strong> bolt deformation.<br />
Jaspart [1.20] applies <strong>the</strong> same approach with a slight modification<br />
concerning <strong>the</strong> position <strong>of</strong> <strong>the</strong> prying forces. This location dep<strong>end</strong>s on <strong>the</strong> relative<br />
stiffness <strong>of</strong> <strong>the</strong> flange and <strong>the</strong> bolts, i.e. <strong>the</strong> flange cross-section dimensions<br />
and <strong>the</strong> bolt diameter, as well as <strong>the</strong> degree <strong>of</strong> bolt preloading. Yee and<br />
Melchers [1.5] assume that <strong>the</strong> prying forces are located at <strong>the</strong> edge <strong>of</strong> <strong>the</strong><br />
flange (n = e ≤ 1.25m). Jaspart [1.20] uses <strong>the</strong> distribution proposed by Douty<br />
and McGuire [1.6] (n = 0.75e ≤ 0.75×1.25m = 0.9375m).<br />
The elastic deformation <strong>of</strong> <strong>the</strong> <strong>bolted</strong> T-stub is determined from <strong>the</strong> following<br />
expressions (subscripts u and l refer to <strong>the</strong> upper and lower T-stub element,<br />
respectively):<br />
Z fu . / l⎡1 qu/<br />
l⎛3<br />
3 ⎞⎤<br />
∆ e.0. u/ l = α f . u/ l 2α<br />
f . u/ l F<br />
E<br />
⎢ − ⎜ − ⎟<br />
4 2 2<br />
⎥<br />
(1.14)<br />
⎣ ⎝ ⎠⎦<br />
E is <strong>the</strong> Young modulus <strong>of</strong> steel. The o<strong>the</strong>r parameters that appear explicitly in<br />
<strong>the</strong> above expression are defined as follows:<br />
3<br />
21.5 ( α f − 2α<br />
f ) Z f<br />
q =<br />
(1.15)<br />
2 3 Lb<br />
26 ( α f − 8α<br />
f ) Z f +<br />
2A<br />
Z<br />
f<br />
20<br />
( ) 3<br />
m+ n<br />
⎡2⎤ =<br />
⎣ ⎦<br />
b′ t<br />
3<br />
eff f<br />
s<br />
(1.16)<br />
n<br />
α f =<br />
(1.17)<br />
2(<br />
m+ n)<br />
b′ eff is <strong>the</strong> effective width for stiffness calculations, computed per bolt row. As<br />
is <strong>the</strong> bolt tensile stress area and Lb is <strong>the</strong> conventional bolt length. Eurocode 3<br />
defines this length as:<br />
Lb = tf. u + tf. l + 2twsh + 0.5(<br />
tn + th)<br />
(1.18)<br />
where th, twsh and tn represent <strong>the</strong> bolt head, washer and nut thickness, respectively<br />
(Fig. 1.15). Aggerskov [1.7] defines a different conventional bolt length<br />
and distinguishes between <strong>the</strong> cases <strong>of</strong> snug-tightened and preloaded bolts. According<br />
to <strong>the</strong> author:
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
B<br />
F<br />
n m m<br />
Fig. 1.14 Equivalent (half-) model for <strong>the</strong> flange flexural elastic behaviour.<br />
L s<br />
L tg<br />
B<br />
n<br />
th<br />
twsh<br />
tf.u<br />
tf.l<br />
twsh<br />
tn Fig. 1.15 Bolt geometrical properties (including washer).<br />
⎧k1<br />
+ 2k4 ⇐snug-tightened<br />
bolts<br />
⎪<br />
Lb= ⎨ k2k3 ⎪<br />
⇐ preloaded bolts<br />
⎩(<br />
k2 + k3)<br />
where (see Fig. 1.15):<br />
(1.19)<br />
k1 = Ls + 1.43Ltg + 0.71tn<br />
tfu . + tfl<br />
.<br />
k3<br />
=<br />
5<br />
(1.20)<br />
k2 = Ls+ 1.43Ltg + 0.91tn+ 0.8twsh k4 = 0.1tn+ 0.4twsh<br />
The initial stiffness coefficients ke.0.u and ke.0.l, which include <strong>the</strong> bolt deformation,<br />
are defined as <strong>the</strong> ratio between <strong>the</strong> applied force F and <strong>the</strong> corresponding<br />
deformation:<br />
F<br />
ke.0.<br />
u/ l =<br />
∆e.0. u/ l<br />
E<br />
=<br />
⎡1 qu/<br />
l⎛3<br />
3 ⎞⎤<br />
Z fu . / l⎢ − ⎜ α fu . / l−2α fu . / l ⎟<br />
4 2 2<br />
⎥<br />
⎣ ⎝ ⎠⎦<br />
(1.21)<br />
The initial stiffness <strong>of</strong> <strong>the</strong> <strong>bolted</strong> T-stub is <strong>the</strong>n given by:<br />
1<br />
ke.0<br />
=<br />
1 1<br />
+<br />
ke.0. u ke.0.<br />
l<br />
(1.22)<br />
The expressions presented above are lengthy and <strong>the</strong>refore <strong>the</strong>y are not<br />
suitable for practical design. Jaspart proposes a simplified approach for <strong>the</strong><br />
prediction <strong>of</strong> <strong>the</strong> axial stiffness <strong>of</strong> <strong>bolted</strong> T-stubs in [1.46]. This approach relies<br />
21
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
on two major assumptions: (i) <strong>the</strong> distance n is taken as 1.25m (see Fig. 1.16b)<br />
and (ii) <strong>the</strong> bolt deformability is dissociated from that <strong>of</strong> <strong>the</strong> T-stub (Fig.<br />
1.16c).<br />
Under <strong>the</strong>se assumptions, <strong>the</strong> initial stiffness coefficients <strong>of</strong> <strong>the</strong> single Tstub<br />
elements may be simplified to <strong>the</strong> following expressions (cf. App<strong>end</strong>ix A;<br />
subscripts f and b refer to <strong>the</strong> flange and <strong>the</strong> bolt, respectively):<br />
3<br />
Eb′ eff t f . u / l<br />
keT<br />
. . u/ l = (1.23)<br />
3<br />
mu/<br />
l<br />
and <strong>the</strong> axial stiffness <strong>of</strong> a snug-tightened bolt row is equal to:<br />
EAs<br />
kebt<br />
. = 1.6<br />
(1.24)<br />
L b<br />
The stiffness coefficient ke.bt from Eq. (1.24) characterizes <strong>the</strong> deformation <strong>of</strong> a<br />
snug-tightened bolt row in tension and is determined assuming that <strong>the</strong> bolt<br />
force is increased from 0.5F to 0.63F due to <strong>the</strong> prying effect (cf. App<strong>end</strong>ix A).<br />
The initial stiffness <strong>of</strong> <strong>the</strong> overall connection is computed by means <strong>of</strong> <strong>the</strong> fol-<br />
(a) Actual behaviour.<br />
Q<br />
22<br />
B<br />
(1)<br />
F<br />
Q<br />
n m m n<br />
n = 1.25m<br />
B<br />
B = 0.63F<br />
F<br />
B<br />
n m m n<br />
Q = 0.13F<br />
(b) T-stub element alone. (c) Bolts alone.<br />
Fig. 1.16 Elastic deformation <strong>of</strong> <strong>the</strong> T-stub.<br />
B<br />
Q<br />
b<br />
F<br />
B = 0.63F
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
lowing relationship:<br />
keo<br />
. =<br />
1<br />
k<br />
1<br />
1<br />
+<br />
k<br />
1<br />
+<br />
k<br />
eT . . u eT . . l ebt .<br />
(1.25)<br />
Referring to Eqs. (1.16) and (1.23), <strong>the</strong> effective length b′ eff represents a<br />
new effective length for stiffness calculations, slightly different from <strong>the</strong> effective<br />
width beff for resistance calculations defined above. This new length may<br />
be taken as (cf. App<strong>end</strong>ix A):<br />
b′ eff = 0.9beff<br />
(1.26)<br />
Faella et al. [1.44] also adopt a procedure that neglects <strong>the</strong> compatibility<br />
requirements between <strong>the</strong> axial deformation <strong>of</strong> <strong>the</strong> bolts and <strong>the</strong> deformation <strong>of</strong><br />
<strong>the</strong> T-stub flanges and neglects <strong>the</strong> effect <strong>of</strong> prying action. The bolt deformability<br />
is again separated from that <strong>of</strong> <strong>the</strong> T-stub. They derive <strong>the</strong> initial stiffness<br />
<strong>of</strong> <strong>the</strong> single T-stub by means <strong>of</strong> a flexible beam model, i.e. <strong>the</strong> bolt restraining<br />
action is modelled as simple supports at <strong>the</strong> bolt axis (Fig. 1.17a). In<br />
this case (I: moment <strong>of</strong> inertia <strong>of</strong> <strong>the</strong> beam section):<br />
( )<br />
( )<br />
3<br />
3<br />
F 2m 2Fm<br />
δ = =<br />
(1.27)<br />
flex<br />
3<br />
48EI<br />
Eb′ t<br />
eff f<br />
δ(flex)<br />
B B<br />
(a) Flexible beam approach.<br />
B<br />
m m<br />
F<br />
F<br />
δ(rig)<br />
m m<br />
(b) Rigid beam approach.<br />
Fig. 1.17 Behavioural schemes <strong>of</strong> <strong>the</strong> equivalent T-stub modelled as a beam.<br />
B<br />
23
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
and so:<br />
F<br />
K(<br />
) = flex<br />
δ(<br />
flex)<br />
3<br />
0.5Eb′ eff t f<br />
= 3<br />
m<br />
(1.28)<br />
If <strong>the</strong> bolt acts as a fixed edge (Fig. 1.17b), <strong>the</strong>n <strong>the</strong> beam is fully restrained at<br />
<strong>the</strong> bolt line (rigid beam approach) and <strong>the</strong> displacement is evaluated as follows:<br />
3 3<br />
Fm 0.5Fm<br />
δ ( rig ) = =<br />
3<br />
24EI<br />
Eb′ t<br />
(1.29)<br />
24<br />
eff f<br />
3<br />
2Eb′ eff t f<br />
3<br />
F<br />
K(<br />
rig ) = = (1.30)<br />
δ(<br />
rig ) m<br />
In reality, <strong>the</strong> restraining action <strong>of</strong> <strong>the</strong> bolts lies in between <strong>the</strong>se two limit<br />
situations. In fact, <strong>the</strong> Eurocode 3 adopts an expression that yields results in between<br />
<strong>the</strong>se two boundaries (see Eq. (1.23)). By adopting a nomenclature similar<br />
to <strong>the</strong> above, according to Faella et al. [1.44], <strong>the</strong> axial stiffness <strong>of</strong> a single<br />
T-element is determined from <strong>the</strong> following relationship:<br />
3<br />
Eb′ eff t f<br />
ke. T . u / l = K(<br />
flex)<br />
= 0.5<br />
(1.31)<br />
3<br />
m<br />
For <strong>the</strong> bolt stiffness <strong>the</strong>y propose <strong>the</strong> expression from Eurocode 3 – Eqs.<br />
(1.18) and (1.24). The effective width b′ eff is now defined as follows:<br />
b′ eff = 2m+<br />
dh≤ b<br />
(1.32)<br />
where dh is <strong>the</strong> bolt head diameter and b is <strong>the</strong> actual width <strong>of</strong> <strong>the</strong> T-stub. This<br />
new width is obtained by consideration <strong>of</strong> a 45º spreading <strong>of</strong> <strong>the</strong> bolt action<br />
starting from <strong>the</strong> bolt head edge (Fig. 1.18) [1.47]. The accuracy <strong>of</strong> such an assumption<br />
is confirmed by experimental evidence in [1.44].<br />
b’eff<br />
45º<br />
Fig. 1.18 Effective width for stiffness calculations [1.47].<br />
1.4.2 Characterization <strong>of</strong> <strong>the</strong> several joint components<br />
The Eurocode 3 formulations for prediction <strong>of</strong> <strong>the</strong> full plastic resistance and<br />
initial stiffness for each component are summarized in Table 1.1. In this table,<br />
fy is <strong>the</strong> yield stress, fu is <strong>the</strong> ultimate stress, Avc is <strong>the</strong> shear area <strong>of</strong> <strong>the</strong> column<br />
dh<br />
m
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
pr<strong>of</strong>ile, dc is <strong>the</strong> clear depth <strong>of</strong> <strong>the</strong> column web, β is <strong>the</strong> transformation parameter<br />
defined in §1.2, γM are partial safety factors, Mb.Rd is <strong>the</strong> moment resistance<br />
<strong>of</strong> <strong>the</strong> beam cross-section and subscripts b, c, ep, f and w refer to <strong>the</strong> beam, <strong>the</strong><br />
column, <strong>the</strong> <strong>end</strong> <strong>plate</strong>, <strong>the</strong> flange and <strong>the</strong> web, respectively. The partial safety<br />
factors for design purposes are taken as γM0 = 1.1 = γM1 and γM2 = 1.25, for <strong>the</strong><br />
resistance <strong>of</strong> cross-sections and bolts, respectively [1.1]. The geometric parameters<br />
are defined in Table 1.2 and Fig. 1.19 for <strong>bolted</strong> joints.<br />
Regarding <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> plastic resistance <strong>of</strong> <strong>the</strong> components column<br />
web in compression and column web in tension, <strong>the</strong> reduction factors for<br />
<strong>plate</strong> buckling and interaction with shear in <strong>the</strong> column web panel, ρ and ω, respectively,<br />
are defined below:<br />
⎪<br />
⎧1.0 if λp<br />
≤ 0.72<br />
beff. cwcdcfy. wc<br />
ρ = ⎨<br />
with λ<br />
2<br />
p = 0.932<br />
(1.33)<br />
2<br />
⎪⎩ ( λp − 0.2 ) λp if λp<br />
> 0.72<br />
Etwc<br />
⎧1 if 0 ≤ β ≤0.5<br />
⎪<br />
ω = ⎨ω1+<br />
21 ( −β)( 1 − ω1) if 0.5< β ≤1<br />
(1.34)<br />
⎪<br />
⎩ω1−(<br />
1 −β)( ω2 − ω1) if 1< β ≤ 2<br />
with:<br />
1<br />
1<br />
ω 1 =<br />
and ω 2 =<br />
(1.35)<br />
2<br />
2<br />
⎛beff . cwctwc ⎞<br />
⎛beff . cwctwc⎞ 1+ 1.3⎜<br />
⎟<br />
1+ 5.2⎜<br />
⎟<br />
⎝ Avc<br />
⎠<br />
⎝ Avc<br />
⎠<br />
0.8rc<br />
mc<br />
ec<br />
rc<br />
aep.w<br />
mep<br />
eep<br />
( )<br />
n = min e , e ,1.25m<br />
c c ep c<br />
( )<br />
n = min e , e ,1.25m<br />
ep c ep ep<br />
0.8 2 aep.w<br />
(a) Column flange and <strong>end</strong> <strong>plate</strong> between beam<br />
flanges.<br />
aep.f<br />
LX<br />
bep<br />
eep eep w<br />
mep<br />
eX<br />
mX<br />
p1-2<br />
m2<br />
p2-3<br />
aep.w<br />
m = L −e − 0.8 2a<br />
X X X ep. f<br />
(b) End <strong>plate</strong> extension.<br />
Fig. 1.19 Definition <strong>of</strong> <strong>the</strong> geometric parameters m and n for <strong>the</strong> column<br />
flange and <strong>end</strong> <strong>plate</strong> (particular case <strong>of</strong> a hot rolled column section).<br />
25
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
κwc is a reduction factor to account for <strong>the</strong> effect <strong>of</strong> an axial force in <strong>the</strong> column.<br />
Generally, this reduction factor is unitary [1.1].<br />
As mentioned above, <strong>the</strong> components column flange in b<strong>end</strong>ing and <strong>end</strong><br />
<strong>plate</strong> in b<strong>end</strong>ing are modelled by means <strong>of</strong> equivalent T-stubs, provided that<br />
<strong>the</strong> effective width is properly defined (Table 1.2 – e1 is <strong>the</strong> <strong>end</strong> distance from<br />
<strong>the</strong> centre <strong>of</strong> <strong>the</strong> bolt hole to <strong>the</strong> adjacent edge, α is a coefficient obtained from<br />
an abacus provided by Eurocode 3). The design resistance associated to each <strong>of</strong><br />
<strong>the</strong> three possible failure modes is thus obtained from Eqs. (1.3-1.5) by introduction<br />
<strong>of</strong> <strong>the</strong> appropriate geometric and mechanic parameters and partial<br />
safety coefficients, γM0, γM1 or γM2.<br />
Table 1.1 Syn<strong>the</strong>sis <strong>of</strong> <strong>the</strong> code formulations for evaluation <strong>of</strong> <strong>the</strong> properties<br />
<strong>of</strong> basic <strong>bolted</strong> joint components.<br />
Component Plastic resistance Initial stiffness<br />
cws<br />
0.9 f ywc . Avc<br />
Fcws.<br />
Rd =<br />
3γ<br />
M 0<br />
0.38EAvc kecws<br />
. =<br />
β z<br />
cwc<br />
ωκwcbeff.<br />
cwctwcfy. wc<br />
Fcwc.<br />
Rd = , but:<br />
γ M 0<br />
ωκ wcρbeff. cwctwc fy.<br />
wc<br />
Fcwc.<br />
Rd ≤<br />
γ M 1<br />
0.7Ebeff . cwctwc kecwc<br />
. =<br />
dc<br />
cwt<br />
ωκ wcbeff. cwttwcfy. wc<br />
Fcwt.<br />
Rd =<br />
γ M 0<br />
0.7Ebeff . cwttwc kecwc<br />
. =<br />
dc<br />
cfb Fcfb. Rd = min ( Fcfb.1. Rd , Fcfb.2. Rd , Fcfb.3.<br />
Rd )<br />
3<br />
Eb′ eff . cfbt fc<br />
kecfb<br />
. = 3<br />
mc<br />
and:<br />
b′ eff . cfb= 0.9beff.<br />
cfb<br />
k<br />
Eb′ t<br />
26<br />
epb Fcfb. Rd = min ( Fepb.1. Rd , Fepb.2. Rd , Fepb.3.<br />
Rd )<br />
bfc<br />
bwt<br />
bt<br />
F<br />
F<br />
bfc. Rd<br />
M bRd .<br />
=<br />
hb − tfb<br />
b t f<br />
3<br />
eff . epb ep<br />
eepb . = 3<br />
mep<br />
and:<br />
b′ = 0.9b<br />
eff . epb eff . epb<br />
k ebfc . = ∞<br />
bwt. Rd =<br />
eff . bwt wb<br />
γ M 0<br />
y. wb<br />
ebwt .<br />
bt. Rd = Rd<br />
0.9 fub . As<br />
=<br />
γ M 2<br />
kebt<br />
.<br />
F B<br />
k = ∞<br />
1.6As =<br />
L<br />
b
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
With respect to <strong>the</strong> effective widths <strong>of</strong> <strong>the</strong> two above components, <strong>the</strong><br />
equivalence between <strong>the</strong> column flange in transverse b<strong>end</strong>ing and <strong>the</strong> T-stub<br />
model is quite straightforward. In fact, for an unstiffened column flange, <strong>the</strong> effective<br />
width is obtained directly from Fig. 1.13 by changing <strong>the</strong> geometry accordingly,<br />
except for <strong>the</strong> beam pattern that is unlikely to develop. In <strong>the</strong> case <strong>of</strong><br />
a stiffened column flange or <strong>the</strong> <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing, <strong>the</strong> groups <strong>of</strong> bolt rows at<br />
each side <strong>of</strong> a stiffener are treated as separate equivalent T-stubs. The exten-<br />
Table 1.2 Definition <strong>of</strong> geometric parameters that appear explicitly in <strong>the</strong><br />
above formulae.<br />
Component Geometric parameters<br />
beff . cwc = tfb + 2 2aep+ 5(<br />
tfc + s) + sp<br />
c hot-rolled pr<strong>of</strong>ile column section<br />
cwc<br />
⎧⎪ r ⇐<br />
s = ⎨<br />
⎪⎩ 2ac ⇐ welded pr<strong>of</strong>ile column section<br />
sp: length obtained by dispersion at 45º through <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
b = b<br />
cwt eff . cwt eff . cfb<br />
cfb<br />
⎧⎪ beff. nc but beff. cfb ≤beff . cp ⇐ type-1 mechanism<br />
beff<br />
. cfb = ⎨<br />
, whereby<br />
⎪⎩ beff<br />
. nc ⇐ type-2 mechanism<br />
subscript cp refers to circular yield line patterns and nc to noncircular<br />
yield line patterns.<br />
Bolt row<br />
location<br />
Inner bolt<br />
row<br />
End bolt<br />
row<br />
Bolt row<br />
adjacent to<br />
a stiffener<br />
Inner bolt<br />
row<br />
End bolt<br />
row<br />
Bolt row<br />
adjacent to<br />
a stiffener<br />
Circular pattern Non-circ. pattern<br />
Bolt row considered individually<br />
2 c m π 4mc + 1.25ec<br />
min ( 2 πmc, π mc + 2e1)<br />
min ( 4mc + 1.25 ec,<br />
2m + 0.625e<br />
+ e<br />
π c m α<br />
2 c m<br />
c c<br />
Bolt row as part <strong>of</strong> a group <strong>of</strong> bolt rows<br />
2 p p<br />
( π m + p e + p)<br />
min , 2<br />
π mc+ p<br />
c<br />
1<br />
(<br />
min e1+ 0.5 p,<br />
2mc + 0.625ec + 0.5p<br />
0.5p+<br />
αmc−<br />
− 2m +<br />
0.625e<br />
( )<br />
c c<br />
1<br />
)<br />
)<br />
27
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
Table 1.2 Definition <strong>of</strong> geometric parameters that appear explicitly in <strong>the</strong><br />
above formulae (cont.).<br />
Component Geometric parameters<br />
⎧⎪ beff. nc but beff . epb ≤beff . cp ⇐ type-1 mechanism<br />
beff<br />
. epb = ⎨<br />
⎪⎩ beff<br />
. nc ⇐ type-2 mechanism<br />
28<br />
epb<br />
Bolt row<br />
location<br />
Bolt row<br />
outside<br />
tension<br />
flange <strong>of</strong><br />
beam<br />
1 st row<br />
below<br />
tension<br />
flange <strong>of</strong><br />
beam<br />
O<strong>the</strong>r inner<br />
bolt<br />
row<br />
O<strong>the</strong>r <strong>end</strong><br />
bolt row<br />
Bolt row<br />
outside<br />
tension<br />
flange <strong>of</strong><br />
beam<br />
1 st row<br />
below<br />
tension<br />
flange <strong>of</strong><br />
beam<br />
O<strong>the</strong>r inner<br />
bolt<br />
row<br />
O<strong>the</strong>r <strong>end</strong><br />
bolt row<br />
b =<br />
b<br />
bwt eff . bwt eff . epb<br />
Circular pattern Non-circ. pattern<br />
Bolt row considered individually<br />
min ( 4mX+ 1.25 eX ,0.5 bep,<br />
min ( 2 πmX, πmX<br />
+ w,<br />
2mX + 0.625 eX + eep,<br />
π mX + 2eep)<br />
2m + 0.625e + 0.5w<br />
π ep m α<br />
2 ep m<br />
X c<br />
2 ep m π 4m + 1.25e<br />
ep ep<br />
2 ep m π 4m + 1.25e<br />
ep ep<br />
Bolt row as part <strong>of</strong> a group <strong>of</strong> bolt rows<br />
⎯⎯⎯ ⎯⎯⎯<br />
π mep + p<br />
2 p<br />
0.5p+<br />
αm<br />
−<br />
ep<br />
( 2mep 0.625eep)<br />
− +<br />
π mep + p<br />
2mep + 0.625eep + 0.5 p<br />
p<br />
)
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
sion <strong>of</strong> an <strong>end</strong> <strong>plate</strong> and <strong>the</strong> portion between <strong>the</strong> beam flanges are also modelled<br />
as two separate equivalent T-stubs [1.1] and <strong>the</strong> resistance and plastic<br />
failure modes are determined separately.<br />
1.5 CHARACTERIZATION OF THE POST-LIMIT BEHAVIOUR OF BASIC<br />
COMPONENTS OF BOLTED JOINTS<br />
Design codes as <strong>the</strong> Eurocode 3 do not give an accurate description <strong>of</strong> <strong>the</strong> postlimit<br />
response <strong>of</strong> <strong>the</strong> individual joint components and <strong>the</strong>ir deformation capacity,<br />
in particular. Within <strong>the</strong> framework <strong>of</strong> <strong>the</strong> component method, <strong>the</strong> overall<br />
joint behaviour is determined by <strong>the</strong> behaviour <strong>of</strong> its elementary parts. As a<br />
consequence, <strong>the</strong> rotation capacity <strong>of</strong> a joint is bound by <strong>the</strong> deformation capacity<br />
<strong>of</strong> <strong>the</strong> single components. In terms <strong>of</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> post-limit<br />
component behaviour with a bilinear approximation, two main properties have<br />
to be fully described: <strong>the</strong> post-limit stiffness, kp-l and deformation capacity, ∆u.<br />
Jaspart [1.20] and Jaspart and Maquoi [1.48] assume that this behaviour can<br />
be approximated by a linear relationship (Fig. 1.20) and propose a general,<br />
simple methodology for <strong>characterization</strong> <strong>of</strong> both properties for all components.<br />
kp-l is taken as <strong>the</strong> strain hardening stiffness since <strong>the</strong> effects <strong>of</strong> material strain<br />
hardening after yielding <strong>of</strong> <strong>the</strong> component are dominant. It is defined below:<br />
Eh<br />
kp− l = ke<br />
(1.36)<br />
E<br />
for components column web in compression, column web in tension, column<br />
flange in b<strong>end</strong>ing, <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing and:<br />
21 ( + υ ) Eh<br />
k = k<br />
(1.37)<br />
p−l e<br />
3<br />
E<br />
for component column web in shear. Eh is <strong>the</strong> strain hardening modulus <strong>of</strong> <strong>the</strong><br />
material and υ is <strong>the</strong> Poisson’s ratio. For <strong>the</strong> bolts in tension kp-l is taken as zero<br />
since this is a brittle component. Components beam web in tension and beam<br />
flange and web in compression are disregarded since <strong>the</strong>y only provide a limitation<br />
to <strong>the</strong> joint flexural resistance [1.1]. They also suggest expressions for<br />
computation <strong>of</strong> <strong>the</strong> ultimate resistance, Fu, and, consequently, ∆u. Fu is readily<br />
determined by formally equivalent expressions to those listed in Table 1.1, by<br />
replacing fy with fu, <strong>the</strong> ultimate stress <strong>of</strong> <strong>the</strong> structural steel. The deformation<br />
capacity is determined from <strong>the</strong> intersection <strong>of</strong> <strong>the</strong> post-limit behaviour with<br />
Fu:<br />
FRd Fu − FRd<br />
∆ u = + (1.38)<br />
ke kp−l From a qualitative point <strong>of</strong> view, <strong>the</strong> basic components can be grouped according<br />
to three <strong>ductility</strong> classes that reflect this post-limit behaviour [1.49].<br />
The component <strong>ductility</strong> reflects <strong>the</strong> “length” <strong>of</strong> <strong>the</strong> post-limit response and<br />
can be quantified by means <strong>of</strong> an index ϕ i for each component i. The author<br />
29
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
30<br />
F<br />
Fu<br />
FRd<br />
ke<br />
kp-l<br />
∆u<br />
Post-limit linear approximation<br />
Actual behaviour<br />
Elastic-plastic approximation<br />
(using <strong>the</strong> component initial<br />
stiffness)<br />
Fig. 1.20 Bilinear approximation <strong>of</strong> <strong>the</strong> component behaviour as proposed by<br />
Jaspart [1.20] and Jaspart and Maquoi [1.48].<br />
proposes <strong>the</strong> following expression for <strong>the</strong> definition <strong>of</strong> ϕ i:<br />
∆u<br />
ϕi<br />
= (1.39)<br />
∆ FRd<br />
whereby ∆u is <strong>the</strong> component deformation capacity and ∆ F = F<br />
Rd Rd ke<br />
is <strong>the</strong><br />
deformation value corresponding to <strong>the</strong> component plastic resistance, FRd.<br />
Kuhlmann et al. [1.49] propose three <strong>ductility</strong> classes: (i) components with<br />
high <strong>ductility</strong> (ϕi ≥ α) (e.g. cws, cfb, epb), (ii) components with limited <strong>ductility</strong><br />
(β ≤ ϕi < α) (e.g. cwc, cwt) and (iii) components with brittle failure (ϕi < β)<br />
(e.g. bt, welds). α and β represent <strong>ductility</strong> limits. Simões da Silva et al. [1.50]<br />
propose α = 20 and β = 3. The <strong>ductility</strong> behaviour <strong>of</strong> <strong>the</strong> several joint components<br />
is analysed in <strong>the</strong> following sub-sections according to alternative procedures<br />
from <strong>the</strong> literature.<br />
1.5.1 Column web in shear (component with high <strong>ductility</strong>)<br />
For <strong>the</strong> web panel subjected to shear, literature proposes an alternative model,<br />
<strong>the</strong> Krawinkler et al. model that can be used to predict <strong>the</strong> contribution <strong>of</strong> this<br />
component to <strong>the</strong> overall joint response [1.44]. This model was developed<br />
based on experimental observations regarding <strong>the</strong> significant post-yield resistance<br />
<strong>of</strong> <strong>the</strong> panel zone. Fig. 1.21 illustrates this model in terms <strong>of</strong> a global Vwpγ<br />
response. This curve is easily converted into a Vwp-∆wp response by means <strong>of</strong><br />
<strong>the</strong> following simplified relationship:<br />
∆ ∆ wp ( cws)<br />
γ = = (1.40)<br />
z z<br />
∆
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
Vwp<br />
Vwp.p<br />
Vwp.y<br />
γy<br />
4γy<br />
Fig. 1.21 Krawinkler et al. trilinear model.<br />
Kcws.h<br />
This relation was derived by idealizing <strong>the</strong> web panel as a short column stub <strong>of</strong><br />
height z, subjected to a shear force Vwp [1.5,1.20]. There are two swivel points<br />
in <strong>the</strong> model corresponding to: (i) first yielding <strong>of</strong> <strong>the</strong> panel zone, (Vwp.y,γy) and<br />
(ii) first yielding <strong>of</strong> <strong>the</strong> column flanges, (Vwp.p,4γy). According to Krawinkler et<br />
al., <strong>the</strong> rotational behaviour <strong>of</strong> <strong>the</strong> panel after yielding can be attributed to <strong>the</strong><br />
b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> column flanges [1.44]. The co-ordinates at <strong>the</strong> swivel points <strong>of</strong><br />
this curve are given by [1.44]:<br />
f y twchchb Vwp.<br />
y = (1.41)<br />
3 β z<br />
E f y<br />
γ y =<br />
(1.42)<br />
21+ υ 3<br />
( )<br />
⎛ bt ⎞<br />
2<br />
c fc<br />
Vwp. p = Vwp.<br />
y ⎜1+ 3.12 ⎟<br />
(1.43)<br />
⎜ hht ⎟<br />
⎝ c b wc ⎠<br />
whereby hc and hb are <strong>the</strong> column and beam depth, respectively. The residual<br />
stiffness Kcws.h is given by [1.44]:<br />
Eh twchchb Kcws.<br />
h =<br />
(1.44)<br />
21 ( + υ) βz<br />
This model imposes no limits on <strong>the</strong> deformation capacity <strong>of</strong> this component.<br />
Beg and co-workers [1.51-1.52] present some expressions to limit <strong>the</strong> ultimate<br />
shear panel rotation, γu:<br />
⎧⎛ d 1 ⎞ c<br />
⎪⎜28−0.38 ⎟κNif<br />
0 ≤ n ≤0.10<br />
⎪⎝<br />
twc<br />
ε ⎠<br />
γ u[%]<br />
= ⎨ (1.45)<br />
⎪⎡<br />
dc 1 ⎛ dc<br />
1⎞<br />
⎤<br />
⎪⎢28 −0.38−⎜55−0.81 ⎟(<br />
n− 0.1 ) ⎥κNif<br />
n><br />
0.10<br />
⎩⎢⎣<br />
twc ε ⎝ twc<br />
ε ⎠ ⎥⎦<br />
γ<br />
31
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
with:<br />
N<br />
n =<br />
N<br />
(1.46)<br />
32<br />
pl<br />
235<br />
ε = (1.47)<br />
f y<br />
The influence <strong>of</strong> <strong>the</strong> level <strong>of</strong> axial force in <strong>the</strong> column, N, can be assessed by<br />
means <strong>of</strong> <strong>the</strong> following parameter κN:<br />
2<br />
hc n<br />
κ N = 1 if hb hc<br />
≥ 1 and κ N = 1 − if hb hc<br />
< 1<br />
(1.48)<br />
h 0.4<br />
b<br />
1.5.2 Column flange in b<strong>end</strong>ing, <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing and bolts in tension<br />
(T-stub idealization)<br />
These three components can be idealized with <strong>the</strong> equivalent T-stub approach.<br />
The deformation capacity <strong>of</strong> a T-stub mainly dep<strong>end</strong>s on <strong>the</strong> <strong>plate</strong>/bolt resistance<br />
ratio. It has been well established that <strong>the</strong> best way to accomplish deformation<br />
capacity is a design in <strong>the</strong> type-1 situation [1.8], i.e. βRd < 2λ/(2λ+1). In<br />
type-1, <strong>the</strong> deformation can be regarded as “indefinitely large” because yielding<br />
occurs in <strong>the</strong> flange. The only limitations are <strong>the</strong> membrane stresses in <strong>the</strong><br />
<strong>plate</strong> that develop with large deformations. In a type-3 mechanism (βRd > 2), <strong>the</strong><br />
deformations are mainly determined by bolt tension elongation, which leads to<br />
a brittle failure. A thorough analysis <strong>of</strong> <strong>the</strong> post-limit behaviour <strong>of</strong> isolated Tstub<br />
connections is carried out later in <strong>the</strong> text (Chapter 6). Previous research<br />
work <strong>of</strong> several authors on this subject is also reviewed, namely <strong>the</strong> work <strong>of</strong><br />
Jaspart [1.20], Faella et al. [1.44], Beg et al. [1.51-1.52] and Swanson [1.53].<br />
1.5.3 Column web in compression (component with limited <strong>ductility</strong>)<br />
Aribert and co-workers opened up <strong>the</strong> elastoplastic studies <strong>of</strong> web pr<strong>of</strong>iles subjected<br />
to local compression forces [1.54-1.56]. Their studies mainly focused on<br />
resistance evaluation ra<strong>the</strong>r than a full description <strong>of</strong> <strong>the</strong> overall deformation<br />
behaviour.<br />
The component column web in compression in particular was extensively<br />
studied by Kuhlmann and Kühnemund [1.57-1.58] and Kühnemund [1.59].<br />
They performed numerous tests on this component and characterized its F-∆<br />
behaviour in detail (model depicted in Fig. 1.22). The elastic-plastic response is<br />
easily determined from <strong>the</strong> Eurocode 3 proposals (see Tables 1.1-1.2). The<br />
post-limit behaviour is described by two distinct branches. The first branch is<br />
defined between <strong>the</strong> plastic resistance and <strong>the</strong> maximum resistance, Fcwc.u. The<br />
second (s<strong>of</strong>tening) branch follows on until fracture. In this phase <strong>the</strong>y redefine
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
Fcwc<br />
Fcwc.u<br />
Fcwc.Rd<br />
2/3Fcwc.Rd<br />
ke.cwc<br />
∆e.cwc<br />
4.5∆e.cwc<br />
∆u.cwc<br />
Fig. 1.22 Kuhlmann and Kühnemund model.<br />
∆cwc<br />
<strong>the</strong> effective width beff.cwc. The procedure for assessment <strong>of</strong> <strong>the</strong> relevant ordinates<br />
<strong>of</strong> <strong>the</strong> curve in this post-limit regime can be found in [1.59].<br />
O<strong>the</strong>r authors also looked into <strong>the</strong> behaviour <strong>of</strong> this component. Huber and<br />
Tschemmernegg [1.60] suggested values for <strong>the</strong> deformation capacity for this<br />
component for different standard shapes <strong>of</strong> <strong>the</strong> column section (Table 1.3). Beg<br />
and co-workers [1.51-1.52] carried out a numerical analysis <strong>of</strong> <strong>the</strong> component<br />
and proposed expressions for evaluation <strong>of</strong> <strong>the</strong> deformation capacity that dep<strong>end</strong><br />
on <strong>the</strong> level <strong>of</strong> axial force in <strong>the</strong> column. These expressions are defined in<br />
a non-dimensional form below:<br />
⎧ dc 1 dc<br />
1<br />
⎪18.5<br />
− 0.75 if < 20<br />
⎪<br />
twc ε twcε<br />
⎪ dc 1 dc<br />
1<br />
δucwc<br />
. [%] = ⎨5.7<br />
−0.11 if 20 ≤ < 33 for n = 0 (1.49)<br />
⎪ twc ε twc<br />
ε<br />
⎪ dc<br />
1<br />
⎪2.07<br />
if ≥ 33<br />
⎪⎩ twc<br />
ε<br />
and:<br />
⎧ d 1 ⎛ c d 1⎞ c dc<br />
1<br />
⎪9.4 − 0.34 + ⎜15−0.75 ⎟(<br />
0.5 − n)<br />
if < 20<br />
⎪ twcε ⎝ twc ε ⎠ twc<br />
ε<br />
⎪ dc 1 dc<br />
1<br />
δucwc<br />
. [%] = ⎨4.8<br />
−0.11 if 20 ≤ < 33<br />
(1.50)<br />
⎪ twc ε twc<br />
ε<br />
⎪ dc<br />
1<br />
⎪1.17<br />
if ≥ 33<br />
⎪⎩<br />
twc<br />
ε<br />
33
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
for n ≥ 0.1. n and ε are defined in Eqs. (1.46) and (1.47), respectively. The deformation<br />
capacity <strong>of</strong> <strong>the</strong> column web in compression, ∆u.cwc, is computed<br />
from:<br />
d δ ∆ = (1.51)<br />
34<br />
ucwc . ucwc . c<br />
Table 1.3 Deformation capacity <strong>of</strong> <strong>the</strong> column web in compression according<br />
to Huber and Tschemmernegg [1.60].<br />
Column pr<strong>of</strong>ile IPE HEA HEB HEM<br />
∆u.cwc (mm) 1.5 3.0 5.0 7.5<br />
1.5.4 Column web in tension (component with limited <strong>ductility</strong>)<br />
Witteveen et al. [1.61] suggest a very simple expression for evaluation <strong>of</strong> <strong>the</strong><br />
deformation capacity <strong>of</strong> <strong>the</strong> column web subjected to tension:<br />
ucwt . 0.025 c h<br />
∆ = (1.52)<br />
This limit value is also adopted in Eurocode 3.<br />
Beg et al. [1.51-1.52] also studied this component and derived an analytical<br />
expression for evaluation <strong>of</strong> <strong>the</strong> ultimate deformation:<br />
u. cwt u. cwt c d δ ∆ = (1.53)<br />
with:<br />
2<br />
⎛ 2<br />
4−3sx −s⎞<br />
x<br />
δucwt . = ε ⎜ ⎟<br />
0<br />
(1.54)<br />
⎜ 2 ⎟<br />
⎝ ⎠<br />
sx is defined below (σx: transverse stress):<br />
σ x<br />
sx<br />
= (1.55)<br />
f ywc .<br />
and ε0 is <strong>the</strong> ultimate transverse strain in <strong>the</strong> case that <strong>the</strong> axial force in <strong>the</strong> column<br />
is absent. They suggest that this value should be set as equal to 0.1.<br />
1.6 EVALUATION OF THE MOMENT-ROTATION RESPONSE OF BOLTED<br />
JOINTS BY MEANS OF COMPONENT MODELS<br />
Mechanical (component) models use a set <strong>of</strong> rigid and flexible parts (springs)<br />
to simulate <strong>the</strong> interaction between <strong>the</strong> various sources <strong>of</strong> joint deformation.<br />
The springs are combined in series or in parallel dep<strong>end</strong>ing on <strong>the</strong> way <strong>the</strong>y interplay<br />
with each o<strong>the</strong>r. Springs in series are subjected to <strong>the</strong> same force whilst<br />
parallel springs undergo <strong>the</strong> same deformation.<br />
The active components <strong>of</strong> a joint are grouped according to <strong>the</strong>ir type <strong>of</strong><br />
loading (tension, compression or shear). They can also be distinguished be-
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
tween those linked to <strong>the</strong> web panel, <strong>the</strong> load-introduction into <strong>the</strong> column web<br />
panel and <strong>the</strong> connection. A sophisticated component interplay as assumed in<br />
<strong>the</strong> Innsbruck model [1.43,1.62] allows separate representation <strong>of</strong> <strong>the</strong> behaviour<br />
<strong>of</strong> <strong>the</strong> web panel in shear, <strong>the</strong> load-introduction and <strong>the</strong> connection elements<br />
(Fig. 1.23). Due to its complexity, this model is not easily implemented<br />
in a design code, as it requires successive iterations within <strong>the</strong> component assembly<br />
[1.62]. This is why a simplified component model, as depicted in Fig.<br />
1.7, is desirable.<br />
Huber highlights <strong>the</strong> two main differences between <strong>the</strong> two component<br />
models [1.62]. In <strong>the</strong> Eurocode 3 model <strong>the</strong>re is no separation between <strong>the</strong><br />
panel and connecting zone, which may lead to a non-straight deformation <strong>of</strong><br />
<strong>the</strong> column front in contradiction with experimental evidence. Also, <strong>the</strong> stiff<br />
separation bar between tension and compression components (see Fig. 1.7)<br />
prevents <strong>the</strong> interaction between <strong>the</strong>se components within <strong>the</strong> web panel that<br />
exists in reality. However, this simplified model yields analytical solutions<br />
ra<strong>the</strong>r than iterative, making it a simpler tool for daily design practice.<br />
Aribert et al. propose an alternative component model that is yet restricted<br />
to <strong>the</strong> case <strong>of</strong> internal flush <strong>end</strong> <strong>plate</strong> joints under balanced loading, i.e. <strong>the</strong><br />
web panel is not subjected to shear forces [1.63]. Basically <strong>the</strong>y assume <strong>the</strong><br />
same components but introduce an additional component at <strong>the</strong> level <strong>of</strong> <strong>the</strong> tension<br />
beam flange. This new component corresponds to <strong>the</strong> part <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
located between <strong>the</strong> tension beam flange and <strong>the</strong> first bolt row and is subjected<br />
to longitudinal b<strong>end</strong>ing. Also, <strong>the</strong>y assume a sophisticated component interplay<br />
since <strong>the</strong>y do not separate <strong>the</strong> components under compression from <strong>the</strong> tensile<br />
zone as in Eurocode 3 (see Fig. 1.7).<br />
Finally, reference is made to <strong>the</strong> component model commonly used at <strong>the</strong><br />
University <strong>of</strong> Coimbra (Fig. 1.24). This model assumes a sophisticated component<br />
interplay since it does not establish <strong>the</strong> equivalence <strong>of</strong> all tensile components<br />
into a single equivalent spring as in Eurocode 3. This equivalence is explained<br />
below. For fur<strong>the</strong>r reference, this model is designated by UC model.<br />
For illustration <strong>of</strong> <strong>the</strong> differences between <strong>the</strong> alternative spring models,<br />
consider <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> initial stiffness <strong>of</strong> a (single-sided) <strong>bolted</strong> ex-<br />
(cwt.1)<br />
(cwt.2)<br />
(cfb.2)<br />
(cws) (cwc) (bfc)<br />
(cfb.1) (epb.1) (bt.1)<br />
(epb.2)<br />
(bwt.2)<br />
(bt.2)<br />
Fig. 1.23 Innsbruck spring model (single-sided steel joint configuration).<br />
Φ<br />
M<br />
35
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
36<br />
(cwt.1) (cfb.1) (epb.1) (bt.1)<br />
(cwt.2)<br />
(cfb.2)<br />
(epb.2)<br />
(cws) (cwc) (bfc)<br />
(bwt.2)<br />
(bt.2)<br />
Fig. 1.24 UC spring model (single-sided steel joint configuration).<br />
30 100<br />
30<br />
Φ<br />
30<br />
35<br />
35<br />
165<br />
Fig. 1.25 Illustrative example: connection geometry.<br />
t<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connection. The column is made up <strong>of</strong> a HE240B pr<strong>of</strong>ile and<br />
<strong>the</strong> beam pr<strong>of</strong>ile is IPE240. Bolts M20 fasten <strong>the</strong> elements. The <strong>end</strong> <strong>plate</strong> dimensions<br />
are 315×160 mm 2 and 15 mm thickness. The continuous fillet welds<br />
between <strong>the</strong> beam and <strong>the</strong> <strong>end</strong> <strong>plate</strong> have a throat thickness aw = 8 mm. The geometry<br />
<strong>of</strong> <strong>the</strong> connection is depicted in Fig. 1.25. For all components, E = 210<br />
GPa. The <strong>characterization</strong> <strong>of</strong> <strong>the</strong> elastic stiffness <strong>of</strong> <strong>the</strong> single components is<br />
based on <strong>the</strong> Eurocode 3 proposals (Tables 1.1-1.2). The following results are<br />
obtained: (i) Eurocode 3 model: Sj.ini = 23.82 kNm/mrad, (ii) Innsbruck model:<br />
Sj.ini = 23.12 kNm/mrad (difference <strong>of</strong> -2.94% in comparison with <strong>the</strong> Eurocode<br />
3 model) and (iii) UC model: Sj.ini = 24.25 kNm/mrad (difference <strong>of</strong> 1.81% in<br />
comparison with <strong>the</strong> Eurocode 3 model).<br />
50<br />
M
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
1.6.1 Eurocode 3 component model<br />
Fig. 1.26a depicts <strong>the</strong> Eurocode 3 mechanical model for <strong>the</strong> particular case <strong>of</strong> a<br />
<strong>bolted</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connection, with two bolt rows in tension, which allows<br />
for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> rotational behaviour <strong>of</strong> such a joint type. The<br />
model assumes that <strong>the</strong> compressive springs are located at <strong>the</strong> centre <strong>of</strong> compression,<br />
which corresponds to <strong>the</strong> centre <strong>of</strong> <strong>the</strong> lower beam flange, and <strong>the</strong><br />
tensile springs are positioned at <strong>the</strong> corresponding bolt row level. The bolt row<br />
deformations are proportional to <strong>the</strong> distance to <strong>the</strong> centre <strong>of</strong> compression and<br />
<strong>the</strong> forces acting in each row dep<strong>end</strong> on <strong>the</strong> component stiffness [1.40].<br />
1.6.1.1 Model for stiffness evaluation<br />
For evaluation <strong>of</strong> <strong>the</strong> initial rotational stiffness, Sj.ini, <strong>the</strong> model (Fig. 1.26a) is<br />
simplified by replacing each assembly <strong>of</strong> springs in series with an equivalent<br />
spring, which retains all <strong>the</strong> relevant characteristics (Fig. 1.26b). Weynand fur<strong>the</strong>r<br />
simplifies this model by establishing <strong>the</strong> equivalence between <strong>the</strong> parallel<br />
spring assembly t.1 and t.2 and <strong>the</strong> spring t (Fig. 1.26c) [1.64]. By means <strong>of</strong><br />
simple equilibrium considerations and compatibility requirements, <strong>the</strong> follow-<br />
(cws)<br />
(cwc) (bfc)<br />
(cwt.1) (cfb.1) (epb.1) (bt.1)<br />
(cwt.2)<br />
(cfb.2)<br />
(epb.2)<br />
(bwt.2)<br />
(a) Eurocode 3 spring model: active components for a <strong>bolted</strong> ext<strong>end</strong>ed <strong>end</strong><br />
<strong>plate</strong> connection with two bolt rows in tension (see Fig. 1.7).<br />
(c)<br />
(t.2)<br />
(t.1)<br />
Φ M<br />
(c)<br />
(bt.2)<br />
Φ<br />
(t)<br />
M<br />
Φ M<br />
(b) Eurocode 3 equivalent model. (c) Eurocode 3 simplified model.<br />
Fig. 1.26 Eurocode 3 spring model and simplifications.<br />
37
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
ing expression for initial stiffness is derived in [1.64]:<br />
M 2 kecket Sjini . = = z<br />
Φ kec + ket<br />
(1.56)<br />
The elastic stiffness <strong>of</strong> each equivalent spring t.i and c, corresponding to a<br />
spring assembly in series, are readily obtained as follows (Fig. 1.26a):<br />
kec<br />
=<br />
1<br />
k<br />
1<br />
1<br />
+<br />
k<br />
1<br />
+<br />
k<br />
(1.57)<br />
38<br />
ecws . ecwc . ebfc .<br />
and:<br />
ket.<br />
i =<br />
1<br />
k<br />
1<br />
+<br />
k<br />
1<br />
1<br />
+<br />
k<br />
1<br />
+<br />
k<br />
1<br />
+<br />
k<br />
ecwti . . ecfbi . . eepbi . . ebwti . . ebti . .<br />
The lever arm z is given by [1.64]:<br />
z =<br />
n<br />
∑<br />
i=<br />
1<br />
n<br />
∑<br />
i=<br />
1<br />
k z<br />
2<br />
et. i i<br />
k z<br />
et. i i<br />
whereby zi is <strong>the</strong> distance from bolt row i to <strong>the</strong> centre <strong>of</strong> compression.<br />
1.6.1.2 Model for resistance evaluation<br />
(1.58)<br />
(1.59)<br />
For evaluation <strong>of</strong> <strong>the</strong> joint flexural resistance, Mj.Rd, simple equilibrium equations<br />
yield:<br />
n<br />
M = ∑ F z<br />
(1.60)<br />
j. Rd ti. Rd i<br />
i=<br />
1<br />
in <strong>the</strong> absence <strong>of</strong> an axial force. Fti.Rd is <strong>the</strong> potential resistance <strong>of</strong> bolt row i in<br />
<strong>the</strong> tension zone and zi is <strong>the</strong> distance <strong>of</strong> <strong>the</strong> i-th bolt row from <strong>the</strong> centre <strong>of</strong><br />
compression. Fti.Rd is taken as <strong>the</strong> least <strong>of</strong> <strong>the</strong> following values:<br />
Fti. Rd = min ( Fcwt. i. Rd , Fcfb. i. Rd , Fepb. i. Rd , Fbwt. i. Rd , Fbt.<br />
i. Rd )<br />
(1.61)<br />
The values <strong>of</strong> Fti.Rd are calculated starting at <strong>the</strong> top row and working down.<br />
Bolt rows below <strong>the</strong> current row are ignored. Each bolt row is analysed first in<br />
isolation and <strong>the</strong>n in combination with <strong>the</strong> successive rows above it. The procedure<br />
can be summarized as follows [1.1]:<br />
(i) Compute <strong>the</strong> plastic resistance <strong>of</strong> bolt row 1 omitting <strong>the</strong> bolt rows below:<br />
Ft1. Rd = min ( Fcws. Rd β , Fcwc. Rd , Fbfc. Rd , Fcwt.1. Rd , Fcfb.1. Rd , Fepb.1. Rd , Fbt.1.<br />
Rd ) (1.62)<br />
(ii) Compute <strong>the</strong> plastic resistance <strong>of</strong> bolt row 2 omitting <strong>the</strong> bolt rows below:<br />
min F β −F , F −F , F −<br />
F , F ,<br />
F t2. Rd = ( cws. Rd t1. Rd cwc. Rd t1. Rd bfc. Rd t1. Rd cwt.2. Rd
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
F , F , F , F , F F ,<br />
cfb.2. Rd epb.2. Rd bwt.2. Rd bt.2. Rd .1 ( 2. ) −<br />
cwt + Rd t1. Rd<br />
cfb.1 ( + 2. ) Rd t1. Rd, bt.1 ( + 2. ) Rd t1. Rd)<br />
F −F F − F<br />
(1.63)<br />
(iii) Compute <strong>the</strong> plastic resistance <strong>of</strong> bolt row 2 omitting <strong>the</strong> bolt rows below:<br />
F t3. Rd = min ( Fcws. Rd β −Ft1. Rd −Ft2. Rd , Fcwc. Rd −Ft1. Rd − Ft2.<br />
Rd ,<br />
F −F − F , F , F , F , F , F ,<br />
bfc. Rd t1. Rd t 2. Rd cwt.3. Rd cfb.3. Rd epb.3. Rd bwt.3. Rd bt.3. Rd<br />
F −F , F −F , F − F ,<br />
cwt.2 ( + 3. ) Rd t2. Rd cfb.2 ( + 3. ) Rd t2. Rd epb.2 ( + 3. ) Rd t2. Rd<br />
cwt.1 ( + 2+ 3. ) Rd t2. Rd t1. Rd, cfb.1 ( + 2+ 3. ) Rd t2. Rd t1. Rd)<br />
F<br />
and so forth.<br />
−F −F F −F − F (1.64)<br />
1.6.1.3 Idealization <strong>of</strong> <strong>the</strong> moment-rotation curve<br />
The conversion <strong>of</strong> <strong>the</strong> F-∆ curves <strong>of</strong> <strong>the</strong> individual active joint components into<br />
a global M-Φ curve is based on <strong>the</strong> spring model so that <strong>the</strong> compatibility and<br />
equilibrium requirements are met. Dep<strong>end</strong>ing on <strong>the</strong> desired level <strong>of</strong> accuracy<br />
and available s<strong>of</strong>tware, <strong>the</strong> rotational joint behaviour can be fully characterized<br />
(full nonlinear shape) or approximated by nonlinear or multilinear simplifications.<br />
The <strong>characterization</strong> <strong>of</strong> <strong>the</strong> actual nonlinear M-Φ curve is not easily open<br />
to simple analytical formulations and <strong>the</strong>refore <strong>the</strong> simplified approximations<br />
are preferred for hand calculations. Recently, <strong>the</strong> author proposed an energy<br />
approach for evaluation <strong>of</strong> <strong>the</strong> multilinear M-Φ response from component<br />
models in closed-form solutions. It also allows <strong>the</strong> identification <strong>of</strong> <strong>the</strong> yielding<br />
sequence <strong>of</strong> <strong>the</strong> individual components and <strong>the</strong> corresponding levels <strong>of</strong> deformation<br />
[1.65-1.70].<br />
The Eurocode 3 adopts two possible idealizations <strong>of</strong> <strong>the</strong> M-Φ curve, bilinear<br />
(elastic-plastic curve) and nonlinear, as depicted in Fig. 1.27. The stiffness<br />
modification factor, η, (Fig. 1.27a) dep<strong>end</strong>s on <strong>the</strong> joint type and configuration<br />
and is defined in <strong>the</strong> code [1.1]. For <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> beam-to-column joints<br />
this factor is taken as 2. The stiffness ratio µ that is used to define <strong>the</strong> nonlinear<br />
part <strong>of</strong> <strong>the</strong> idealized M-Φ curve in Fig. 1.27b is defined as follows:<br />
Ψ<br />
⎛1.5M ⎞ jEd . 2<br />
µ = ⎜ ⎟ for M jRd . < M jEd . ≤ M jRd .<br />
(1.65)<br />
⎜ M ⎟<br />
⎝ jRd . ⎠ 3<br />
and Ψ is a coefficient that dep<strong>end</strong>s on <strong>the</strong> type <strong>of</strong> connection. For <strong>bolted</strong> <strong>end</strong><br />
<strong>plate</strong> connections, this coefficient is taken as 2.7.<br />
1.6.2 Guidelines for evaluation <strong>of</strong> <strong>the</strong> <strong>ductility</strong> <strong>of</strong> <strong>bolted</strong> joints<br />
The <strong>ductility</strong> <strong>of</strong> a joint can be defined as <strong>the</strong> amount <strong>of</strong> a plastic rotation that<br />
39
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
40<br />
B<strong>end</strong>ing moment (kNm)<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Sj.ini/η<br />
0<br />
0 15 30 45 60 75 90 105 120<br />
Joint rotation (mrad)<br />
(a) Bilinear idealization <strong>of</strong> <strong>the</strong> moment-rotation curve.<br />
B<strong>end</strong>ing moment (kNm)<br />
180<br />
150<br />
120<br />
90<br />
2/3Mj.Rd<br />
60<br />
30<br />
Sj.ini<br />
Sj.ini/µ<br />
Actual response<br />
Mj.Rd<br />
0<br />
0 15 30 45 60 75 90 105 120<br />
Joint rotation (mrad)<br />
Actual response<br />
Mj.Rd<br />
(b) Nonlinear idealization <strong>of</strong> <strong>the</strong> moment-rotation curve.<br />
Fig. 1.27 Eurocode 3 idealizations <strong>of</strong> <strong>the</strong> actual rotational response.<br />
can be sustained while maintaining a certain percentage <strong>of</strong> its ultimate resistance<br />
[1.53]. It reflects <strong>the</strong> length <strong>of</strong> <strong>the</strong> yield <strong>plate</strong>au <strong>of</strong> <strong>the</strong> M-Φ response.<br />
This property can be quantified by means <strong>of</strong> an index ϑj that relates <strong>the</strong> rotation<br />
capacity <strong>of</strong> <strong>the</strong> joint, ΦCd to <strong>the</strong> rotation value corresponding to <strong>the</strong> joint plastic<br />
resistance [1.46,1.50]. In this work, <strong>the</strong> following relationship is proposed:<br />
ΦCd<br />
ϑ j = (1.66)<br />
ΦM<br />
Rd<br />
similarly to Eq. (1.39). This index allows a direct classification <strong>of</strong> a joint in<br />
terms <strong>of</strong> <strong>ductility</strong>, similarly to <strong>the</strong> basic joint components (§1.5). ΦM is <strong>the</strong><br />
Rd<br />
“analytical” rotation value corresponding to Mj.Rd and is given by <strong>the</strong> ratio<br />
M S (Fig. 1.28). Fig. 1.28 presents o<strong>the</strong>r distinctive rotation values. ΦXd<br />
j. Rd j. ini
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
M<br />
Mmax<br />
Mj.Rd<br />
Φ MRd<br />
ΦXd<br />
Fig. 1.28 Definitions <strong>of</strong> joint rotation.<br />
*<br />
Φ Cd ΦCd<br />
Mmax<br />
Φ<br />
Φ<br />
*<br />
Φ Cd is <strong>the</strong> rotation<br />
is <strong>the</strong> rotation at which <strong>the</strong> moment first reaches Mj.Rd and<br />
at which <strong>the</strong> moment deteriorates back to Mj.Rd after reaching a moment above<br />
Mj.Rd through deformation beyond ΦXd. Φ M is <strong>the</strong> rotation at which <strong>the</strong> mo-<br />
max<br />
ment resistance is maximum.<br />
Jaspart [1.46] classifies <strong>the</strong> joints in terms <strong>of</strong> available rotation capacity. He<br />
groups structural joints into three classes: (i) class 1 joints, which have a sufficiently<br />
good rotation capacity to allow a plastic frame analysis (high <strong>ductility</strong>),<br />
(ii) class 2 joints, with a limited rotation capacity (limited <strong>ductility</strong>) and (iii)<br />
class 3 joints, for which brittle failure or instability phenomena limits <strong>the</strong> rotation<br />
capacity.<br />
Literature reports several procedures for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> available<br />
rotation capacity. Zoetemeijer [1.10] proposes some criteria and simple empirical<br />
expressions for <strong>the</strong> estimation <strong>of</strong> a joint deformation capacity based on a series<br />
<strong>of</strong> experiments. He concluded that considerable rotation capacity was obtained<br />
from <strong>the</strong> tension side <strong>of</strong> a joint if βRd < 2λ/(2λ+1), within <strong>the</strong> T-stub idealization<br />
<strong>of</strong> <strong>the</strong> region. This means that <strong>the</strong> tension zone fails according to a<br />
type-1 mechanism, with complete yielding <strong>of</strong> one <strong>of</strong> <strong>the</strong> <strong>plate</strong> components<br />
(column flange or <strong>end</strong> <strong>plate</strong>). If βRd > 2, <strong>the</strong>n <strong>the</strong> joint behaves elastically up to<br />
failure <strong>of</strong> <strong>the</strong> bolts without deformation <strong>of</strong> <strong>the</strong> <strong>plate</strong>(s). In this case, <strong>the</strong> bolt<br />
elongation mainly supplies <strong>the</strong> joint deformation. To prevent this situation,<br />
Zoetemeijer suggests that <strong>the</strong> condition βRd < 1.75 should always be satisfied<br />
[1.10]. For <strong>the</strong> intermediate situations, i.e. 2λ/(2λ+1) < βRd ≤ 1.75, <strong>the</strong> joint rotational<br />
deformation remains limited since <strong>the</strong> bolt is also engaged in <strong>the</strong> collapse<br />
mode. In <strong>the</strong> latter situation, Zoetemeijer suggests an expression for<br />
evaluation <strong>of</strong> <strong>the</strong> rotation capacity, ΦCd [1.10]:<br />
41
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
10.6 − 4β<br />
Rd<br />
Φ Cd = (1.67)<br />
1.3z1<br />
whereby z1 is <strong>the</strong> distance in [mm] between <strong>the</strong> first bolt row from <strong>the</strong> tension<br />
flange and <strong>the</strong> centre <strong>of</strong> compression.<br />
Later, Jaspart [1.46] ext<strong>end</strong>ed <strong>the</strong> above criteria for inclusion in Eurocode<br />
3. The code states that a <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> joint may be assumed to have sufficient<br />
rotation capacity for plastic analysis, provided that both <strong>of</strong> <strong>the</strong> following<br />
conditions are satisfied: (i) <strong>the</strong> moment resistance <strong>of</strong> <strong>the</strong> joint is governed by<br />
<strong>the</strong> resistance <strong>of</strong> ei<strong>the</strong>r <strong>the</strong> column flange in b<strong>end</strong>ing or <strong>the</strong> <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing<br />
and (ii) <strong>the</strong> thickness t <strong>of</strong> ei<strong>the</strong>r <strong>the</strong> column flange or <strong>the</strong> <strong>end</strong> <strong>plate</strong> (not necessarily<br />
<strong>the</strong> same basic component as in (i)) satisfies:<br />
fub<br />
.<br />
t ≤ 0.36φ<br />
(1.68)<br />
f y<br />
where φ is <strong>the</strong> bolt diameter, fu.b is <strong>the</strong> tensile strength <strong>of</strong> <strong>the</strong> bolt and fy is <strong>the</strong><br />
yield strength <strong>of</strong> <strong>the</strong> relevant basic component. This expression is derived in<br />
[1.46]. These guidelines are yet insufficient to ensure adequate <strong>ductility</strong> in partial<br />
strength joints.<br />
More recently, Adegoke and Kemp [1.23] performed an experimental/analytical<br />
study on thin ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong>s and realized that most <strong>of</strong> <strong>the</strong><br />
connection rotation in <strong>the</strong>se cases came from <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformation. From<br />
<strong>the</strong>se observations, <strong>the</strong>y proposed a simple expression for evaluation <strong>of</strong> <strong>the</strong><br />
connection ultimate rotation:<br />
2<br />
mf fy. ep mfmX fy.<br />
ep<br />
φ Cd = 1.4 + 40<br />
(1.69)<br />
Etephyfn Etephmrn In this expression, <strong>the</strong> first part corresponds to <strong>the</strong> connection rotation when<br />
<strong>the</strong> yield lines in <strong>the</strong> ext<strong>end</strong>ed and flush zones <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> (above and below<br />
<strong>the</strong> top tension beam flange, respectively) are fully developed and is based<br />
on <strong>the</strong> elastic flexibility <strong>of</strong> <strong>the</strong> stronger flush bolt lines [1.23]. mf represents <strong>the</strong><br />
average distance from each bolt to <strong>the</strong> adjacent web and flange welds below<br />
<strong>the</strong> tension flange, i.e.:<br />
mep + m2<br />
m f = (1.70)<br />
2<br />
(see Fig. 1.19). hyfn is <strong>the</strong> height <strong>of</strong> <strong>the</strong> flush bolt row above <strong>the</strong> neutral axis at<br />
yield and hmrn is <strong>the</strong> height <strong>of</strong> <strong>the</strong> resultant tension force above <strong>the</strong> neutral axis<br />
at maximum strain. They assumed that <strong>the</strong> rotation capacity was attained when<br />
fracture <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> occurred. This would happen when <strong>the</strong> maximum<br />
strain would be thirty times <strong>the</strong> yield strain.<br />
In <strong>the</strong> context <strong>of</strong> <strong>the</strong> component method, several researchers have developed<br />
simplified approaches to quantify <strong>the</strong> overall rotation capacity. Since in<br />
many cases <strong>the</strong> most important sources <strong>of</strong> deformability in <strong>bolted</strong> joints can be<br />
idealized by means <strong>of</strong> <strong>the</strong> equivalent T-stub in tension, special attention has<br />
been devoted to <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> deformation capacity <strong>of</strong> this individual<br />
42
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
component. Swanson [1.53] developed a methodology for <strong>characterization</strong> <strong>of</strong><br />
<strong>the</strong> <strong>ductility</strong> <strong>of</strong> T-stub connections. Faella and co-workers [1.44,1.71-1.72] set<br />
up a procedure for computation <strong>of</strong> <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> isolated Tstub<br />
and <strong>the</strong> overall joint. O<strong>the</strong>r components have also been studied within this<br />
framework. Kuhlmann and Kuhnemund [1.57] performed tests on <strong>the</strong> component<br />
column web under transverse compression and proposed design rules for<br />
this component from <strong>the</strong> point <strong>of</strong> view <strong>of</strong> resistance and deformation capacity.<br />
The researchers also conducted a series <strong>of</strong> full-scale tests that are reported in<br />
[1.58-1.59]. The study was restricted to joints under balanced loading. The<br />
dominant component <strong>of</strong> all tests was <strong>the</strong> column web in compression. They<br />
also developed a procedure based on <strong>the</strong> component method to determine <strong>the</strong><br />
rotation capacity <strong>of</strong> <strong>the</strong> joint for those cases where <strong>the</strong> critical component was<br />
<strong>the</strong> column web under compression. Beg et al. [1.51] set up a methodology<br />
based on a simplified component model for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> rotational<br />
response to include <strong>the</strong> evaluation <strong>of</strong> rotation capacity. They analysed different<br />
components, <strong>the</strong> column web, <strong>the</strong> bolts in tension, <strong>the</strong> column flange and <strong>the</strong><br />
<strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing, and proposed simple expressions for evaluation <strong>of</strong> <strong>the</strong>ir<br />
deformation capacity based on numerical evidence, as already mentioned<br />
above (see §1.5). Yet, <strong>the</strong>re was no calibration <strong>of</strong> this work. For this reason,<br />
this methodology is questionable and should be used with special care. They<br />
<strong>the</strong>n established a simple mechanical model to mimic <strong>the</strong> joint rotational behaviour<br />
(Fig. 1.29). This model is composed <strong>of</strong> bilinear springs that represent<br />
<strong>the</strong> response <strong>of</strong> all relevant components. The overall joint rotation results from<br />
<strong>the</strong> contribution <strong>of</strong> all components and can be readily determined as follows<br />
(Fig. 1.29):<br />
∆ cwt +∆ 0 +∆cwc<br />
Φ= + γ<br />
(1.71)<br />
z<br />
The rotation capacity mainly dep<strong>end</strong>s on <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> weakest<br />
component, i.e. <strong>the</strong> component with lower resistance. In Fig. 1.29 <strong>the</strong> Tstub<br />
that represents <strong>the</strong> tension zone is <strong>the</strong> governing component. Consequently:<br />
∆ cwt. R +∆ 0. u +∆cwc.<br />
R<br />
Φ = + γ<br />
(1.72)<br />
Cd R<br />
z<br />
It is worth mentioning that this procedure is identical to <strong>the</strong> proposals <strong>of</strong> Faella<br />
and co-workers [1.44]. They also proposed a similar expression for evaluation<br />
<strong>of</strong> <strong>the</strong> rotation capacity, though <strong>the</strong>y mainly focused on <strong>the</strong> study <strong>of</strong> <strong>the</strong> tension<br />
zone idealized as a T-stub.<br />
Finally, and in <strong>the</strong> framework <strong>of</strong> <strong>the</strong> component method, <strong>the</strong> author’s proposals<br />
for evaluation <strong>of</strong> <strong>the</strong> rotation capacity are also referred [1.65-1.70]. By<br />
means <strong>of</strong> an elastic analogy <strong>of</strong> <strong>the</strong> nonlinear behaviour, <strong>the</strong> author proposed an<br />
elastic equivalence for <strong>the</strong> spring models mentioned above (Figs. 1.7 and 1.24).<br />
The basic building block <strong>of</strong> an equivalent model corresponds to replacing each<br />
nonlinear spring with an equivalent elastic spring consisting <strong>of</strong> a set <strong>of</strong> linear<br />
elastic springs with specific properties. Such equivalent models provide closed-<br />
43
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
form solutions to an o<strong>the</strong>rwise numerical problem and allow for <strong>the</strong> identification<br />
<strong>of</strong> <strong>the</strong> yielding sequence <strong>of</strong> <strong>the</strong> components and, ultimately, <strong>the</strong> computation<br />
<strong>of</strong> <strong>the</strong> joint rotation capacity.<br />
F<br />
FR<br />
44<br />
z<br />
∆cwt.R<br />
cwt<br />
cws<br />
∆<br />
F<br />
cwt T-stub<br />
idealization<br />
cwc<br />
T-stub<br />
M<br />
∆0.u<br />
∆<br />
F<br />
(cws)<br />
∆cwc.R<br />
cwc<br />
(cwt) (T-stub)<br />
(cwc)<br />
Fig. 1.29 Computation <strong>of</strong> <strong>the</strong> joint rotation capacity according to Beg et al.<br />
[1.51].<br />
1.7 REFERENCES<br />
[1.1] European Committee for Standardization (CEN). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[1.2] Zoetemeijer P. A design method for <strong>the</strong> tension side <strong>of</strong> statically loaded,<br />
<strong>bolted</strong> beam-to-column connections. Heron; 20(1):1-59, 1974.<br />
[1.3] Packer JA, Morris LJ. A limit state design method for <strong>the</strong> tension region<br />
<strong>of</strong> <strong>bolted</strong> beam-to-column connections. The Structural Engineer;<br />
55(10):446-458, 1977.<br />
[1.4] Aggerskov H. Analysis <strong>of</strong> <strong>bolted</strong> connections subjected to prying. Journal<br />
<strong>of</strong> <strong>the</strong> Structural Division ASCE; 103(ST11):2145-2163, 1977.<br />
[1.5] Yee YL, Melchers RE. Moment-rotation curves for <strong>bolted</strong> connections.<br />
Journal <strong>of</strong> Structural Engineering ASCE; 112(3):615-635, 1986.<br />
[1.6] Douty RT, McGuire W. High strength moment connections. Journal <strong>of</strong><br />
∆<br />
F<br />
cws<br />
Φ<br />
Φ<br />
M<br />
γR<br />
γ
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
Structural Division ASCE; 91(ST2):101-128, 1965.<br />
[1.7] Aggerskov H. High-strength <strong>bolted</strong> connections subjected to prying.<br />
Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE; 102(ST1):161-175, 1976.<br />
[1.8] Zoetemeijer P, Munter H. Ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong>s with disappointing rotation<br />
capacity – Test results and analysis. Stevin Laboratory Report 6-83-<br />
13. Faculty <strong>of</strong> Civil Engineering, Delft University <strong>of</strong> Technology, 1983.<br />
[1.9] Zoetemeijer P, Munter H. Proposal for <strong>the</strong> standardization <strong>of</strong> ext<strong>end</strong>ed<br />
<strong>end</strong> <strong>plate</strong> connections based on test results – Test and analysis. Stevin<br />
Laboratory Report 6-83-23. Faculty <strong>of</strong> Civil Engineering, Delft University<br />
<strong>of</strong> Technology, 1983.<br />
[1.10] Zoetemeijer P. Summary <strong>of</strong> <strong>the</strong> research on <strong>bolted</strong> beam-to-column<br />
connections. Report 25-6-90-2. Faculty <strong>of</strong> Civil Engineering, Stevin<br />
Laboratory – Steel Structures, Delft University <strong>of</strong> Technology. 1990.<br />
[1.11] Mann AP, Morris LJ. Limit design <strong>of</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections.<br />
Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE; 105(ST3):511-526, 1979.<br />
[1.12] Nair RS, Birkemoe PC, Munse WH. High strength bolts subject to tension<br />
and prying. Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE;<br />
100(ST2):351-372, 1974.<br />
[1.13] Kato B, McGuire W. Analysis <strong>of</strong> T-stub flange-to-column connections.<br />
Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE; 99(ST5):865-888, 1973.<br />
[1.14] Astaneh A. Procedure for design and analysis <strong>of</strong> hanger-type connections.<br />
Engineering Journal AISC; 22(2):63-66, 1985.<br />
[1.15] Thornton WA. Prying action – a general treatment. Engineering Journal<br />
AISC; 22(2):67-75, 1985.<br />
[1.16] Jenkins WM, Tong CS, Prescott AT. Moment-transmitting <strong>end</strong><strong>plate</strong><br />
connections in steel construction and a proposal basis for flush <strong>end</strong><strong>plate</strong><br />
design. The Structural Engineer; 64A(5):121-132, 1986.<br />
[1.17] Davison JB, Kirby PA, Ne<strong>the</strong>rcot DA. Rotational stiffness characteristics<br />
<strong>of</strong> steel beam-to-column connections. Journal <strong>of</strong> Constructional<br />
Steel Research; 8:17-54, 1987.<br />
[1.18] Davison JB, Kirby PA, Ne<strong>the</strong>rcot DA. Effect <strong>of</strong> lack <strong>of</strong> fit on connection<br />
restraint. Journal <strong>of</strong> Constructional Steel Research; 8:55-69, 1987.<br />
[1.19] Janss J, Jaspart JP, Maquoi R. Experimental study <strong>of</strong> <strong>the</strong> non-linear behaviour<br />
<strong>of</strong> beam-to-column <strong>bolted</strong> joints. In: Proceedings <strong>of</strong> <strong>the</strong> First International<br />
Workshop on Connections in Steel Structures, Behaviour,<br />
Strength and Design (Eds.: R. Bjorhovde, J. Brozzetti and A. Colson),<br />
Cachan, France; 26-32, 1988.<br />
[1.20] Jaspart JP. Study <strong>of</strong> <strong>the</strong> semi-rigid behaviour <strong>of</strong> beam-to-column joints<br />
and <strong>of</strong> its influence on <strong>the</strong> stability and strength <strong>of</strong> steel building frames.<br />
PhD <strong>the</strong>sis (in French). University <strong>of</strong> Liège, Liège, Belgium, 1991.<br />
[1.21] Aggarwal AK. Comparative tests on <strong>end</strong><strong>plate</strong> beam-to-column connections.<br />
Journal <strong>of</strong> Constructional Steel Research; 30:151-175, 1994.<br />
[1.22] Bose B, Sarkar S, Bahrami M. Ext<strong>end</strong>ed <strong>end</strong><strong>plate</strong> connections: comparison<br />
between three-dimensional nonlinear finite-element analysis and<br />
45
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
full-scale destructive tests. Structural Engineering Review; 8(4):315-<br />
328, 1996.<br />
[1.23] Adegoke IO, Kemp AR. Moment-rotation relationships <strong>of</strong> thin <strong>end</strong> <strong>plate</strong><br />
connections in steel beams. In: Proceedings <strong>of</strong> <strong>the</strong> International Conference<br />
on Advances in Structures (ASSCCA’03) (Eds.: G.J. Hancock,<br />
M.A. Bradford, T.J. Wilkinson, B. Uy and K.J.R. Rasmussen), Sydney,<br />
Australia; 119-124, 2003.<br />
[1.24] Zandonini R, Zanon P. Experimental analysis <strong>of</strong> <strong>end</strong> <strong>plate</strong> connections.<br />
In: Proceedings <strong>of</strong> <strong>the</strong> First International Workshop on Connections in<br />
Steel Structures, Behaviour, Strength and Design (Eds.: R. Bjorhovde, J.<br />
Brozzetti and A. Colson), Cachan, France; 40-51, 1988.<br />
[1.25] Bursi OS. An experimental-numerical method for <strong>the</strong> modelling <strong>of</strong> plastic<br />
failure mechanisms <strong>of</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> steel connections. Structural<br />
Engineering Review; 3:111-119, 1991.<br />
[1.26] Krishnamurthy N, Graddy DE. Correlation between 2- and 3dimensional<br />
finite element analysis <strong>of</strong> steel <strong>bolted</strong> <strong>end</strong>-<strong>plate</strong> connections.<br />
Computers and Structures; 6:381-389, 1976.<br />
[1.27] Krishnamurthy N, Huang HT, Jeffrey PK, Avery LK. Analytical M-θ<br />
curves for <strong>end</strong>-<strong>plate</strong> connections. Journal <strong>of</strong> <strong>the</strong> Structural Division<br />
ASCE; 105(ST1):133-145, 1979.<br />
[1.28] Krishnamurthy N. Modelling and prediction <strong>of</strong> steel <strong>bolted</strong> connection<br />
behaviour. Computers and Structures; 11:75-82, 1980.<br />
[1.29] Kukreti AR, Murray JM, Abolmaali A. End <strong>plate</strong> connection momentrotation<br />
relationship. Journal <strong>of</strong> Constructional Steel Research; 8:137-<br />
157, 1987.<br />
[1.30] Kukreti AR, Murray JM, Ghasseimieh M. Finite element modelling <strong>of</strong><br />
large capacity stiffened steel tee-hanger connections. Computers and<br />
Structures; 32(2):409-422, 1989.<br />
[1.31] Kukreti AR, Ghasseimieh M, Murray JM. Behaviour and design <strong>of</strong><br />
large capacity moment <strong>end</strong> <strong>plate</strong>s. Journal <strong>of</strong> Structural Engineering<br />
ASCE; 116(3): 809-828, 1990.<br />
[1.32] Bahaari MR, Sherbourne AN. Computer modelling <strong>of</strong> an ext<strong>end</strong>ed <strong>end</strong><strong>plate</strong><br />
<strong>bolted</strong> connection. Computers and Structures; 52(5):879-893,<br />
1994.<br />
[1.33] Sherbourne AN , Bahaari MR. 3D simulation <strong>of</strong> <strong>end</strong>-<strong>plate</strong> <strong>bolted</strong> connections.<br />
Journal <strong>of</strong> Structural Engineering ASCE; 120(11):3122-3136,<br />
1994.<br />
[1.34] Bahaari MR, Sherbourne AN. Structural behavior <strong>of</strong> <strong>end</strong>-<strong>plate</strong> <strong>bolted</strong><br />
connections to stiffened columns. Journal <strong>of</strong> Structural Engineering<br />
ASCE; 122(8):926-935, 1996.<br />
[1.35] Bahaari MR, Sherbourne AN. 3D simulation <strong>of</strong> <strong>bolted</strong> connections to<br />
unstiffened columns-II: Ext<strong>end</strong>ed <strong>end</strong><strong>plate</strong> connections. Journal <strong>of</strong> Constructional<br />
Steel Research; 40(3):189-223, 1996.<br />
[1.36] Bahaari MR, Sherbourne AN. Finite element prediction <strong>of</strong> <strong>end</strong> <strong>plate</strong><br />
<strong>bolted</strong> connection behaviour II: Analytic formulation. Journal <strong>of</strong> Struc-<br />
46
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
tural Engineering ASCE; 123(2):165-175, 1997.<br />
[1.37] Bursi OS, Jaspart JP. Calibration <strong>of</strong> a finite element model for <strong>bolted</strong><br />
<strong>end</strong> <strong>plate</strong> steel connections. Journal <strong>of</strong> Constructional Steel Research;<br />
44(3):225-262, 1997.<br />
[1.38] Bursi OS, Jaspart JP. Basic issues in <strong>the</strong> finite element simulation <strong>of</strong> ext<strong>end</strong>ed<br />
<strong>end</strong>-<strong>plate</strong> connections. Computers and Structures; 69:361-382,<br />
1998.<br />
[1.39] Choi CK, Chung GT. Refined three-dimensional finite element model<br />
for <strong>end</strong> <strong>plate</strong> connection. Journal <strong>of</strong> Structural Engineering ASCE;<br />
122(11):1307-1316, 1996.<br />
[1.40] Weynand K, Jaspart JP, Steenhuis M. The stiffness model <strong>of</strong> revised<br />
Annex J <strong>of</strong> Eurocode 3. In: Proceedings <strong>of</strong> <strong>the</strong> Third International<br />
Workshop on Connections in Steel Structures III, Behaviour, Strength<br />
and Design (Eds.: R. Bjorhovde, A. Colson and R. Zandonini), Trento,<br />
Italy; 441-452, 1995.<br />
[1.41] Borges LAC. Probabilistic evaluation <strong>of</strong> <strong>the</strong> rotation capacity <strong>of</strong> steel<br />
joints. MSc <strong>the</strong>sis. University <strong>of</strong> Coimbra, Coimbra, Portugal, 2003.<br />
[1.42] Ne<strong>the</strong>rcot DA, Zandonini R. Methods <strong>of</strong> prediction <strong>of</strong> joint behaviour:<br />
beam-to-column connections, Chapter 2 in Structural connections, stability<br />
and strength (Ed.: R. Narayanan). Elsevier Applied Science, London,<br />
UK; 23-62, 1989.<br />
[1.43] Huber G, Tschemmernegg F. Modelling <strong>of</strong> beam-to-column joints.<br />
Journal <strong>of</strong> Constructional Steel Research; 45:199-216, 1998.<br />
[1.44] Faella C, Piluso V, Rizzano G. Structural semi-rigid connections – <strong>the</strong>ory,<br />
design and s<strong>of</strong>tware. CRC Press, USA, 2000.<br />
[1.45] Holmes M, Martin LH. Analysis and design <strong>of</strong> structural connections:<br />
reinforced concrete and steel. Ellis Horwood Limited, Chichester, UK,<br />
1983.<br />
[1.46] Jaspart JP. Contributions to recent advances in <strong>the</strong> field <strong>of</strong> steel joints –<br />
column bases and fur<strong>the</strong>r configurations for beam-to-column joints and<br />
beam splices. Aggregation <strong>the</strong>sis. University <strong>of</strong> Liège, Liège, Belgium,<br />
1997.<br />
[1.47] Ballio G, Mazzolani FM. Theory and design <strong>of</strong> steel structures. Chapman<br />
and Hall, London, UK, 1983.<br />
[1.48] Jaspart JP, Maquoi R. Prediction <strong>of</strong> <strong>the</strong> semi-rigid and partial strength<br />
properties <strong>of</strong> structural joints. In: Proceedings <strong>of</strong> <strong>the</strong> Annual Technical<br />
Meeting on Structural Stability Research, Lehigh, USA; 177-191, 1994.<br />
[1.49] Kuhlmann U, Davison JB, Kattner M. Structural systems and rotation<br />
capacity. In: Proceedings <strong>of</strong> <strong>the</strong> International Conference on Control <strong>of</strong><br />
<strong>the</strong> Semi-Rigid Behaviour <strong>of</strong> Civil Engineering Structural Connections<br />
(Ed.: R. Maquoi), Liège, Belgium; 167-176, 1998.<br />
[1.50] Simões da Silva L, Santiago A, Vila Real P. Post-limit stiffness and<br />
<strong>ductility</strong> <strong>of</strong> <strong>end</strong> <strong>plate</strong> beam-to-column steel joints. Computers and Structures;<br />
80:515-531, 2002.<br />
[1.51] Beg D, Zupančič E, Vayas I. On <strong>the</strong> rotation capacity <strong>of</strong> moment con-<br />
47
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
nections. Journal <strong>of</strong> Constructional Steel Research; 60:601-620, 2004.<br />
[1.52] Zupančič E, Beg D, Vayas I. Deformation capacity <strong>of</strong> components <strong>of</strong><br />
moment resistant connections. European Convention for Constructional<br />
Steelwork – Technical Committee 10, Structural Connections (ECCS-<br />
TC10), Document ECCS-TWG 10.2-02-005, 2002.<br />
[1.53] Swanson JA. Characterization <strong>of</strong> <strong>the</strong> strength, stiffness and <strong>ductility</strong> behavior<br />
<strong>of</strong> T-stub connections. PhD dissertation, Georgia Institute <strong>of</strong><br />
Technology, Atlanta, USA, 1999.<br />
[1.54] Aribert JM, Lachal A. Étude élasto-plastique par analyse des contraintes<br />
de la compression locale sur l’âme d’un pr<strong>of</strong>ilé. Construction Métallique;<br />
4:51-66, 1977.<br />
[1.55] Aribert JM, Lachal A, El Nawawy O. Modélisation élasto-plastique de<br />
la résistance d’un pr<strong>of</strong>ilé en compression locale. Construction Métallique;<br />
2:3-26, 1981.<br />
[1.56] Aribert JM, Lachal A, Moheissen M. Interaction du voilement et de la<br />
résistance plastique de l’âme d’un pr<strong>of</strong>ilé laminé soumis à une double<br />
compression locale (nuance dácier allant jusqu’à FeE460. Construction<br />
Métallique; 2:3-23, 1990.<br />
[1.57] Kuhlmann U, Kühnemund F. Rotation capacity <strong>of</strong> steel joints: verification<br />
procedure and component tests. In: Proceedings <strong>of</strong> <strong>the</strong> NATO Advanced<br />
Research Workshop: The paramount role <strong>of</strong> joints into <strong>the</strong> reliable<br />
response <strong>of</strong> structures (Eds.: C.C. Baniotopoulos and F. Wald),<br />
Nato Science series, Kluwer Academic Publishers, Dordrecht, The Ne<strong>the</strong>rlands;<br />
363-372, 2000.<br />
[1.58] Kuhlmann U, Kühnemund F. Ductility <strong>of</strong> semi-rigid steel joints. In:<br />
Proceedings <strong>of</strong> <strong>the</strong> International Colloquium on Stability and Ductility<br />
<strong>of</strong> Steel Structures (SDSS 2002) (Ed.: M. Ivanyi), Budapest, Hungary;<br />
363-370, 2002.<br />
[1.59] Kühnemund F. On <strong>the</strong> verification <strong>of</strong> <strong>the</strong> rotation capacity <strong>of</strong> semi-rigid<br />
joints in steel structures. PhD Thesis (in German), University <strong>of</strong> Stuttgart,<br />
Stuttgart, Germany, 2003.<br />
[1.60] Huber G, Tschemmernegg F. Component characteristics, Chapter 4 in<br />
Composite steel-concrete joints in braced frames for buildings (Ed.: D.<br />
Anderson), COST C1, Brussels, Luxembourg; 4.1-4.49, 1996.<br />
[1.61] Witteveen J, Stark JWB, Bijlaard FSK, Zoetemeijer P. Welded and<br />
<strong>bolted</strong> beam-to-column connections. Journal <strong>of</strong> <strong>the</strong> Structural Division<br />
ASCE; 108(ST2):433-455, 1982.<br />
[1.62] Huber G. Nicht-lineare berechnungen von verbundquerschnitten und<br />
biegeweichen knoten. PhD Thesis (in English), University <strong>of</strong> Innsbruck,<br />
Innsbruck, Austria, 1999.<br />
[1.63] Aribert JM, Lachal A, Dinga ON. Modélisation du comportement<br />
d’assemblages métalliques semi-rigides de type pouter-pouteau boulonnés<br />
par platine d’extremité. Construction Métallique; 1:25-46, 1999.<br />
[1.64] Weynand K. Sicherheits-und Wirtsschaftlichkeitsuntersuchungen zur<br />
anw<strong>end</strong>ung nachgiebiger anschlüsse im stahlbau. PhD <strong>the</strong>sis (in Ger-<br />
48
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
man). University <strong>of</strong> Aachen, Aachen, Germany, 1996.<br />
[1.65] Girão Coelho AM. Equivalent elastic models for <strong>the</strong> analysis <strong>of</strong> steel<br />
joints. MSc <strong>the</strong>sis (in Portuguese). University <strong>of</strong> Coimbra, Coimbra,<br />
Portugal, 1999.<br />
[1.66] Simões da Silva LAP, Girão Coelho AM, Neto EL. Equivalent postbuckling<br />
models for <strong>the</strong> flexural behaviour <strong>of</strong> steel connections. Computers<br />
and Structures; 77:615-624, 2000.<br />
[1.67] Simões da Silva LAP, Girão Coelho AM. A <strong>ductility</strong> model for steel<br />
connections. Journal <strong>of</strong> Constructional Steel Research; 57:45-70, 2001.<br />
[1.68] Simões da Silva LAP, Girão Coelho AM, Simões RAD. Analytical<br />
Evaluation <strong>of</strong> <strong>the</strong> moment-rotation response <strong>of</strong> beam-to-column composite<br />
joints under static loading. Steel and Composite Structures;<br />
1(2):245-268, 2001.<br />
[1.69] Simões da Silva LAP, Girão Coelho AM. Mode interaction in nonlinear<br />
models for steel and steel-concrete composite structural connections.<br />
In: Proceedings <strong>of</strong> <strong>the</strong> Third International Conference on Coupled<br />
Instabilities in Metal Structures (CIMS’2000) (Eds.: D. Camotim, D.<br />
Dubina and J. Rondal), Lisbon, Portugal; 605-614, 2000.<br />
[1.70] Simões da Silva L, Calado L, Simões R, Girão Coelho A. Evaluation <strong>of</strong><br />
<strong>ductility</strong> in steel and composite beam-to-column joints: analytical<br />
evaluation. In: Proceedings <strong>of</strong> <strong>the</strong> Fourth International Workshop on<br />
Connections in Steel Structures IV: Steel Connections in <strong>the</strong> New Millennium<br />
(Ed.: R. Leon), Roanoke, USA; 2000 (available on CD).<br />
[1.71] Piluso V, Faella C, Rizzano G. Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. I:<br />
Theoretical model. Journal <strong>of</strong> Structural Engineering ASCE;<br />
127(6):686-693, 2001.<br />
[1.72] Piluso V, Faella C, Rizzano G. Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. II:<br />
Model validation. Journal <strong>of</strong> Structural Engineering ASCE; 127(6):694-<br />
704, 2001.<br />
49
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
APPENDIX A: DESIGN PROVISIONS FOR CHARACTERIZATION OF RESIS-<br />
TANCE AND STIFFNESS OF T-STUBS<br />
A.1 Basic formulations for prediction <strong>of</strong> plastic resistance <strong>of</strong> <strong>bolted</strong> Tstubs<br />
The equilibrium conditions <strong>of</strong> <strong>the</strong> mechanisms illustrated in Fig. 1.10 provide<br />
<strong>the</strong> equations for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> corresponding “plastic” resistance, FRd.0.<br />
A.1.1 Type-1 mechanism<br />
Regarding type-1 mechanism, <strong>the</strong> three equilibrium equations yield <strong>the</strong> following<br />
relationships:<br />
F1.<br />
Rd .0<br />
∑ Fv= 0 ⇔ Q = B−<br />
2<br />
(A.1)<br />
(1)<br />
∑ M = M f. Rd⇔ Bm− Q( n+ m) = M f. Rd<br />
(A.2)<br />
and:<br />
(2)<br />
∑ M = M f. Rd⇔ Qn= M f. Rd<br />
(A.3)<br />
Section (1) corresponds to <strong>the</strong> critical section at <strong>the</strong> flange-to-web connection<br />
and section (2) is <strong>the</strong> section at <strong>the</strong> bolt axis.<br />
By equating Eqs. (A.2-A.3), <strong>the</strong> bolt force is computed as:<br />
2n+<br />
m<br />
B = M f . Rd<br />
(A.4)<br />
mn<br />
Eqs. (A.1), (A.3) and (A.4) provide:<br />
⎛ M f . Rd⎞ 4M<br />
f . Rd<br />
F1. Rd .0 = 2⎜B−<br />
⎟=<br />
(A.5)<br />
⎝ n ⎠ m<br />
A.1.2 Type-2 mechanism<br />
For <strong>the</strong> second mechanism, <strong>the</strong> equilibrium equations are written as follows:<br />
F2.<br />
Rd .0<br />
∑ Fv= 0 ⇔ Q = B−<br />
2<br />
and:<br />
(A.6)<br />
(1)<br />
∑ M = M f. Rd⇔ Bm− Q( n+ m) = M f . Rd<br />
(A.7)<br />
This mechanism involves bolt fracture. Therefore, <strong>the</strong> bolt force at “plastic”<br />
conditions is equal to<br />
B = B<br />
(A.8)<br />
50<br />
Rd
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
The plastic resistance F2.Rd.0 is <strong>the</strong>n calculated by means <strong>of</strong> <strong>the</strong> following relationship:<br />
⎛ 2m⎞ 2 2Mf. Rd+ 2nBRd<br />
F2. Rd .0 = ⎜2− ⎟BRd+<br />
M f . Rd =<br />
(A.9)<br />
⎝ m+ n⎠ m+ n m+ n<br />
By inserting <strong>the</strong> parameter λ = nmin<br />
Eq. (A.9), F2.Rd.0 is re-written as follows:<br />
2λ<br />
1+<br />
2M f . Rd β 2M f. Rd⎡ ( 2 − β Rd<br />
Rd ) λ ⎤<br />
F2.<br />
Rd .0 = = ⎢1+ ⎥<br />
(A.10)<br />
m 1+ λ m ⎢⎣ βRd ( 1+<br />
λ)<br />
⎥⎦<br />
A.1.3 Type-3 mechanism<br />
Type-3 mechanism is characterized by bolt fracture only. The force equilibrium<br />
equation yields:<br />
∑ Fv = 0⇔ F3. Rd.0 = 2BRd<br />
(A.11)<br />
A.1.4 Supplementary mechanism<br />
Supplementary plastic mechanisms corresponding to <strong>the</strong> metal shear tearing<br />
around <strong>the</strong> bolt head or <strong>the</strong> washer should also be taken into account, though<br />
<strong>the</strong>y are not relevant in most cases. The yielding condition in this case provides<br />
<strong>the</strong> following relationship:<br />
*<br />
FRd = 2πdwτ<br />
y. ftf (A.12)<br />
whereby τy.f is <strong>the</strong> yield shear stress <strong>of</strong> <strong>the</strong> flanges.<br />
A.2 Influence <strong>of</strong> <strong>the</strong> moment-shear interaction on resistance formulations<br />
The moment-shear interaction can be approximately assessed by assuming that<br />
<strong>the</strong> external fibres take <strong>the</strong> b<strong>end</strong>ing moment stresses and <strong>the</strong> internal ones <strong>the</strong><br />
shear stresses, as illustrated in Fig. A.1 (see reference [1.44]). The reduced<br />
plastic flexural resistance <strong>of</strong> <strong>the</strong> flanges is given by:<br />
M f = ∫ σ dS= xt ( f<br />
S<br />
−xb<br />
) eff fy.<br />
f<br />
(A.13)<br />
and <strong>the</strong> reduced plastic shear resistance is defined as:<br />
Vf = ∫ τ dS = ( tf S<br />
−2x)<br />
beffτy.<br />
f<br />
(A.14)<br />
Under <strong>of</strong> <strong>the</strong> Von Mises criterion, <strong>the</strong> yield shear stress is computed as:<br />
51
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
52<br />
tf<br />
Fig. A.1 Distribution <strong>of</strong> internal stresses in <strong>the</strong> plastic condition under combined<br />
b<strong>end</strong>ing moment and shear force.<br />
τ y =<br />
f y<br />
3<br />
(A.15)<br />
From Eqs. (A.13-A.15), <strong>the</strong> distance x is derived:<br />
tf x = −<br />
2<br />
3 Vf<br />
2 beff fy.<br />
f<br />
(A.16)<br />
Substitution <strong>of</strong> x into Eq. (A.13) yields:<br />
M f<br />
2<br />
3 V f<br />
= M f. Rd −<br />
4 b f<br />
(A.17)<br />
eff y. f<br />
where Mf.Rd is given by Eq. (1.6). The pure plastic shear resistance, Vf.Rd, is ex-<br />
pressed by <strong>the</strong> following relationship:<br />
t f<br />
Vf . Rd= befffy. f<br />
(A.18)<br />
3<br />
Therefore, Mf.Rd and Vf.Rd are related by means <strong>of</strong>:<br />
M f . Rd<br />
2<br />
t f<br />
befffy. f<br />
3<br />
tfVf . Rd<br />
=<br />
4<br />
=<br />
4<br />
(A.19)<br />
Eqs. (A.17-A.19) provide <strong>the</strong> following yielding condition:<br />
2<br />
M ⎛ f V ⎞ f<br />
+ ⎜ = 1<br />
M ⎜<br />
⎟<br />
f . Rd V ⎟<br />
⎝ f. Rd⎠<br />
The shear force in <strong>the</strong> T-stub flange is given by:<br />
(A.20)<br />
Vf FRd.0<br />
= B− Q =<br />
2<br />
(A.21)<br />
A.2.1 Type-1 mechanism<br />
Regarding type-1 mechanism, <strong>the</strong> equilibrium condition provides (§A.1.1):<br />
4M f<br />
F1.<br />
Rd .0 =<br />
m<br />
(A.22)<br />
S<br />
x<br />
x<br />
fy<br />
Mf<br />
τy<br />
Vf
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
Mf and Vf are related by means <strong>of</strong> <strong>the</strong> following relationship that derives from<br />
Eqs. (A.21-A.22):<br />
Vf F1. Rd.0<br />
= ⇔ M f<br />
2<br />
m<br />
= Vf<br />
2<br />
(A.23)<br />
The yielding condition brings:<br />
2<br />
M ⎛ f V ⎞ f<br />
+ ⎜ = 1⇔1− M ⎜<br />
⎟<br />
f . Rd V ⎟<br />
⎝ f . Rd⎠ 2<br />
2 m V ⎛ f V ⎞ f<br />
− ⎜ = 0<br />
3 tfV ⎜<br />
⎟<br />
f . Rd V ⎟<br />
⎝ f. Rd⎠<br />
(A.24)<br />
that has <strong>the</strong> positive solution:<br />
( ) 2<br />
V f<br />
Vf . Rd<br />
=<br />
⎡<br />
1 m ⎢<br />
3 t ⎢<br />
f ⎢⎣ ⎤<br />
3<br />
1+ −1⎥<br />
mt<br />
⎥<br />
f ⎥⎦<br />
(A.25)<br />
By equating Eqs. (1.6), (A.18) and (A.23-A.25), <strong>the</strong> following relationship<br />
is obtained for <strong>the</strong> plastic resistance associated with type-1 mechanism:<br />
⎡<br />
2<br />
F1. Rd .0 = m⎢ 3 ⎢<br />
⎢⎣ ⎤<br />
3<br />
1+ − 1⎥b 2 eff fy.<br />
f<br />
( mt<br />
⎥<br />
f ) ⎥⎦ 2<br />
8⎛ m ⎞<br />
⎡<br />
= ⎜ ⎢<br />
3⎜<br />
⎟<br />
t ⎟ ⎢<br />
⎝ f ⎠ ⎢⎣ ⎤<br />
3 M f . Rd<br />
1+ −1⎥<br />
(A.26)<br />
2<br />
( mt<br />
⎥<br />
f ) m<br />
⎥⎦<br />
A.2.2 Type-2 mechanism<br />
With reference to type-2 mechanism (§A.1.2), <strong>the</strong> plastic resistance F2.Rd.0 is<br />
given by Eqs. (A.9-A.10). From Eq. (A.21), Mf and Vf are correlated by means<br />
<strong>of</strong> <strong>the</strong> following relationship:<br />
Vf F2.<br />
Rd.0<br />
= ⇔ M f<br />
2<br />
= ( m+ n) Vf − nBRd<br />
(A.27)<br />
The yielding condition provides:<br />
M f<br />
M f . Rd<br />
2<br />
⎛ V ⎞ f<br />
+ ⎜ = 1⇔ 1+ ⎜<br />
⎟<br />
V ⎟<br />
⎝ f . Rd⎠ 4 BRd 3Vf. Rd<br />
−<br />
4 m+ n Vf 3 tfVf. Rd<br />
2<br />
⎛ V ⎞ f<br />
− ⎜ ⎟ = 0<br />
⎜V ⎟<br />
⎝ f. Rd⎠<br />
(A.28)<br />
The ratio BRd Vf. Rd can be written:<br />
BRd<br />
Vf . Rd<br />
=<br />
3 t f 1<br />
2 m βRd<br />
(A.29)<br />
by taking Eqs. (1.6-1.7) and (A.18) into account. Thus, from Eq. (A.28), <strong>the</strong><br />
following condition is obtained:<br />
2λ 1+ −<br />
β Rd<br />
2<br />
4 m(<br />
1+<br />
λ ) V ⎛ f V ⎞ f<br />
− ⎜ = 0<br />
3 tfV ⎜<br />
⎟<br />
f . Rd V ⎟<br />
⎝ f . Rd ⎠<br />
(A.30)<br />
The positive solution for Eq. (A.30) is:<br />
53
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
V f<br />
Vf . Rd<br />
=<br />
⎡<br />
⎢<br />
2 m ⎢<br />
( 1+ λ ) ⎢<br />
3 tf⎢<br />
⎢<br />
⎢⎣ ⎤<br />
2λ<br />
+ 1<br />
⎥<br />
3 β ⎥<br />
Rd<br />
1+ −1<br />
2 ⎥<br />
4 ⎛m⎞ 2 ⎥<br />
⎜ ⎟ ( 1+<br />
λ<br />
⎜ ) ⎥<br />
t ⎟<br />
⎝ f ⎠<br />
⎥⎦<br />
(A.31)<br />
and, <strong>the</strong>refore, by means <strong>of</strong> Eqs. (A.7), (A.18) and (A.21), F2.Rd.0 is given by:<br />
⎡<br />
⎢ 2<br />
16 ⎛m⎞ ⎢<br />
F2.<br />
Rd .0 = ⎜ ( 1+ λ )<br />
3 ⎜<br />
⎟<br />
t ⎟<br />
⎢<br />
⎝ f ⎠ ⎢<br />
⎢<br />
⎢⎣ ⎤<br />
2λ<br />
+ 1<br />
⎥<br />
3 β ⎥ M<br />
Rd<br />
f . Rd<br />
1+ −1<br />
2 ⎥<br />
4⎛m⎞<br />
m<br />
2 ⎥<br />
⎜ ⎟ ( 1+<br />
λ<br />
⎜ ) ⎥<br />
t ⎟<br />
⎝ f ⎠ ⎥⎦<br />
(A.32)<br />
A.3 Influence <strong>of</strong> <strong>the</strong> bolt dimensions on resistance formulations<br />
To cater for <strong>the</strong> influence <strong>of</strong> <strong>the</strong> bolt finite size on <strong>the</strong> plastic resistance for<br />
type-1 mechanism, Jaspart (see reference [1.20]) provides an alternative formulation<br />
that assumes that <strong>the</strong> bolt action is uniformly distributed under <strong>the</strong><br />
washer, <strong>the</strong> bolt head or <strong>the</strong> nut, as appropriate. Consider <strong>the</strong> half T-stub represented<br />
in Fig. A.2, whereby qb is <strong>the</strong> uniformly distributed bolt action, which is<br />
statically equivalent to B, and dw is <strong>the</strong> diameter <strong>of</strong> <strong>the</strong> washer, <strong>the</strong> bolt head or<br />
nut, as suitable. Equilibrium conditions provide <strong>the</strong> following relationships:<br />
F1.<br />
Rd.0<br />
∑ Fv = 0 ⇔ Q = qdw<br />
−<br />
(A.33)<br />
2<br />
(1)<br />
∑ M = M f. Rd⇔ qbdwm− Q( n+ m) = M f . Rd<br />
(A.34)<br />
and:<br />
2<br />
(2)<br />
dw<br />
∑ M = M f. Rd⇔Qn− qb= M f. Rd<br />
(A.35)<br />
8<br />
By solving this system <strong>of</strong> equations, <strong>the</strong> prying force, <strong>the</strong> bolt force and <strong>the</strong><br />
plastic resistance are obtained:<br />
( 8m+<br />
dw) M f. Rd<br />
Q =<br />
(A.36)<br />
8mn<br />
− m + n d<br />
B = qd =<br />
and:<br />
54<br />
w<br />
( ) w<br />
8( m+ 2n)<br />
M<br />
8 − ( + )<br />
f . Rd<br />
mn m n d<br />
( )<br />
w<br />
( 32n−2d ) M<br />
8 − ( + )<br />
w f . Rd<br />
(A.37)<br />
F1. Rd .0 = 2 B− Q =<br />
mn m n dw<br />
(A.38)<br />
The previous relationships do not allow for moment-shear interaction. By
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
Q<br />
e m<br />
n<br />
(2)<br />
B<br />
(1)<br />
F1.Rd.0<br />
qb<br />
B<br />
Q<br />
tf<br />
Mf.Rd<br />
Mf.Rd<br />
Fig. A.2 Influence <strong>of</strong> <strong>the</strong> bolt-finite size on <strong>the</strong> T-stub resistance.<br />
applying a similar procedure to §A.2.1, <strong>the</strong> following relationships are derived<br />
for <strong>the</strong> plastic resistance. The plastic T-stub resistance is now given by:<br />
( 32n−2dw) M f<br />
F1.<br />
Rd .0 =<br />
8mn<br />
− ( m + n) dw<br />
From Eq. (A.21),<br />
(A.39)<br />
Vf F1.<br />
Rd.0<br />
= ⇔ M f<br />
2<br />
8mn<br />
− ( m + n) dw<br />
=<br />
Vf<br />
16n−dw<br />
(A.40)<br />
By re-arranging <strong>the</strong> equations, <strong>the</strong> positive solution for <strong>the</strong> yielding condition<br />
is written as follows:<br />
( ) 2<br />
V f<br />
Vf . Rd<br />
with:<br />
=<br />
⎡<br />
2 m<br />
Γ ⎢<br />
3 t ⎢<br />
f ⎢⎣ ⎤<br />
3<br />
1+ −1⎥<br />
2<br />
4Γ<br />
mt<br />
⎥<br />
f ⎥⎦<br />
(A.41)<br />
55
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
56<br />
( )<br />
8λm− 1+<br />
λ dw<br />
Γ=<br />
16λm−dw<br />
Thus:<br />
2<br />
16 ⎛ m ⎞<br />
⎡ ⎤<br />
3 M .<br />
F1.<br />
Rd .0 = ⎜ Γ ⎢ 1+ −1⎥<br />
2<br />
2<br />
3 ⎜<br />
⎟<br />
t ⎟ ⎢<br />
f 4 ( mt<br />
⎥<br />
⎝ ⎠ Γ f ) m<br />
⎢⎣ ⎥⎦<br />
A.4 Formulations for prediction <strong>of</strong> elastic stiffness <strong>of</strong> <strong>bolted</strong> T-stubs<br />
A.4.1 Elastic <strong>the</strong>ory for evaluation <strong>of</strong> <strong>the</strong> elastic stiffness <strong>of</strong> a <strong>bolted</strong> T-stub<br />
f Rd<br />
(A.42)<br />
(A.43)<br />
The elastic stiffness <strong>of</strong> a <strong>bolted</strong> T-stub can be computed by means <strong>of</strong> <strong>the</strong> <strong>the</strong>oretical<br />
model shown in Fig. 1.14. From simple elastic b<strong>end</strong>ing <strong>the</strong>ory, applying<br />
<strong>the</strong> double integration method, <strong>the</strong> deflection <strong>of</strong> <strong>the</strong> T-stub web, ∆e.0.u/l is computed<br />
as:<br />
2<br />
1 ⎧⎪ n⎡ 2 n ⎤<br />
∆ = y ( x = m)<br />
( )<br />
e.0. u/ l<br />
2<br />
=− ⎨ ⎢ m+ n − ⎥B−<br />
EI f ⎪⎩ 2⎣ 3 ⎦<br />
2 3<br />
⎡n⎡ 2<br />
2 n ⎤ m ⎤ F⎫⎪<br />
−⎢ ⎢m − ( m+ n)<br />
− ⎥−<br />
⎥ ⎬<br />
⎣2⎣ 3 ⎦ 3 ⎦ 2⎭⎪<br />
If represents <strong>the</strong> flange inertia and is defined as follows:<br />
(A.44)<br />
I f<br />
3<br />
b′ eff tf<br />
=<br />
12<br />
(A.45)<br />
where b′ eff is <strong>the</strong> effective width for stiffness calculations, computed per bolt<br />
row. Eq. (A.44) can be re-written in a simpler form by bringing in <strong>the</strong> parame-<br />
ters Zf and αf defined in Eqs. (1.16-1.17):<br />
Z f ⎡F⎛33⎞⎤ ∆ e.0. u/ l = 2B<br />
α f α f<br />
E<br />
⎢ − ⎜ − ⎟<br />
4 4<br />
⎥<br />
⎣ ⎝ ⎠⎦<br />
(A.46)<br />
The bolt-elastic deformation, ∆e.bt, is given by:<br />
BLb<br />
∆ ebt . = (A.47)<br />
EAs<br />
The compatibility requirement between <strong>the</strong> bolt and <strong>the</strong> flange deformation at<br />
<strong>the</strong> bolt centreline yields:<br />
∆ Z eb .<br />
f ⎡F⎛33⎞ 2 3 ⎤ BLb<br />
= y( x1= n) = α f 2α f B(<br />
6α f 8α<br />
f )<br />
2 E<br />
⎢ ⎜ − ⎟−<br />
− =<br />
2 2<br />
⎥<br />
(A.48)<br />
⎣ ⎝ ⎠<br />
⎦ 2EAs<br />
Thus:
Modelling <strong>of</strong> <strong>the</strong> M-Φ characteristics <strong>of</strong> <strong>bolted</strong> joints: background review<br />
⎛33⎞ ⎜ α f − 2α<br />
f ⎟Z<br />
fF<br />
2<br />
B =<br />
⎝ ⎠<br />
2 3 Lb<br />
26 ( α f − 8α<br />
f ) +<br />
2As<br />
q<br />
= F<br />
2<br />
(A.49)<br />
where q is defined by Eq. (1.15). From Eqs. (A.46-A.49) <strong>the</strong> deformation <strong>of</strong><br />
<strong>the</strong> upper (u) or <strong>the</strong> lower (l) T-stub element is given by:<br />
Z f ⎡1 q ⎛3 3 ⎞⎤<br />
∆ e.0. u/ l = α f 2α<br />
f F<br />
E<br />
⎢ − ⎜ − ⎟<br />
4 2 2<br />
⎥<br />
⎣ ⎝ ⎠⎦<br />
and <strong>the</strong> total deformation:<br />
(A.50)<br />
Z f ⎡1 ⎛3 3 ⎞⎤<br />
∆ e.0 = ∆ e.0. u +∆ e.0. l = q α f 2α<br />
f F<br />
E<br />
⎢ − ⎜ − ⎟<br />
2 2<br />
⎥<br />
⎣ ⎝ ⎠⎦<br />
The elastic axial stiffness <strong>of</strong> <strong>the</strong> <strong>bolted</strong> T-stub is <strong>the</strong>n defined as:<br />
(A.51)<br />
F<br />
ke.0<br />
= =<br />
∆e.0 Z f<br />
E<br />
⎡1 ⎛3 3 ⎞⎤<br />
⎢ −q⎜ α f −2α<br />
f ⎟<br />
2 2<br />
⎥<br />
⎣ ⎝ ⎠⎦<br />
(A.52)<br />
A.4.2 Simplification <strong>of</strong> <strong>the</strong> stiffness coefficients for inclusion in design codes<br />
As already mentioned above, Jaspart (see reference [1.46]) simplifies <strong>the</strong> complex<br />
above formulae for inclusion in Eurocode 3 – cf. §1.4.1.2. By supposing<br />
that <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> T-elements and <strong>the</strong> bolts is carried out separately, ke.T is<br />
derived by means <strong>of</strong> Eq. (A.52) adopting A s ≈ ∞ in Eq. (1.16). Fur<strong>the</strong>r, if n is<br />
taken as equal to 1.25m, as explained in <strong>the</strong> text, <strong>the</strong>n:<br />
n 1.25m<br />
α f = = = 0.278<br />
(A.53)<br />
2 m+ n 2 m+ 1.25m<br />
( ) ( )<br />
( + ) ( + )<br />
3 3<br />
⎡2m n ⎤<br />
Z f =<br />
⎣ ⎦<br />
3<br />
b′ eff tf and:<br />
⎡2 =<br />
⎣ m 1.25m<br />
⎤⎦<br />
3<br />
b′ eff tf 3<br />
m<br />
= 91.125 3<br />
b′ eff tf<br />
(A.54)<br />
⎛3 3⎞ 2⎜ α f − 2α f ⎟Z f<br />
2<br />
q =<br />
⎝ ⎠<br />
2 3 Lb 26 ( α f − 8α f ) Z f +<br />
2As<br />
⎛3 3⎞<br />
2⎜ × 0.278 − 2× 0.278 ⎟Z<br />
f<br />
2<br />
=<br />
⎝ ⎠<br />
2 3 Lb<br />
26 ( × 0.278− 8× 0.278)<br />
Z f +<br />
∞<br />
= 1.28(A.55)<br />
By replacing <strong>the</strong> above results in Eq. (A.52):<br />
k<br />
eT .<br />
3 3<br />
E ⎛m⎞ 0.5E⎛<br />
m⎞<br />
= ⎜ ⎟ ≈ ⎜ ⎟<br />
1.936b′<br />
⎜<br />
eff t ⎟<br />
f b ⎜<br />
eff t ⎟<br />
⎝ ⎠<br />
′<br />
⎝ f ⎠<br />
(A.56)<br />
57
State-<strong>of</strong>-<strong>the</strong>-art and literature review<br />
The effective width for stiffness calculations, b′ eff , is related to <strong>the</strong> effective<br />
width for strength calculations, beff as explained below. With reference to Fig.<br />
1.16, <strong>the</strong> elastic b<strong>end</strong>ing moment at <strong>the</strong> T-stub flange (section (1)) is evaluated<br />
as follows:<br />
( 1)<br />
M = (0.63 − 0.13× 2.25) Fm = 0.3375Fm<br />
(A.57)<br />
If <strong>the</strong> maximum elastic load corresponds to <strong>the</strong> formation <strong>of</strong> a plastic hinge at<br />
section (1), <strong>the</strong>n from internal equilibrium conditions and Eq. (A.57) <strong>the</strong> following<br />
relationship is obtained:<br />
2<br />
1 (1) 1 b′ eff tf<br />
F = M = f<br />
(A.58)<br />
58<br />
el max y. f<br />
0.3375m 0.3375m 4<br />
The maximum elastic load, Fel, corresponds to 2/3 <strong>of</strong> <strong>the</strong> plastic resistance, FRd,<br />
as in Eurocode 3 (see reference [1.1]), being FRd given by Eq. (A.5). As <strong>the</strong> Tstub<br />
flange is fixed at <strong>the</strong> bolt centreline, <strong>the</strong> only possible collapse mode <strong>of</strong><br />
<strong>the</strong> T-stub is that <strong>of</strong> <strong>the</strong> complete yielding <strong>of</strong> <strong>the</strong> flange (type-1 mechanism).<br />
Then, <strong>the</strong> maximum elastic moment is given by:<br />
2<br />
2 2beff<br />
tf<br />
Fel = F1. Rd .0 = fy.<br />
f<br />
(A.59)<br />
3 3 m<br />
by taking Eqs. (A.5) and (1.6) into consideration. By equating Eqs. (A.57-<br />
A.58), b′ eff is computed as follows:<br />
2 2<br />
1 b′ eff tf 2beff tf<br />
2<br />
f ′<br />
y. f = fy. f ⇔ beff= × 0.3375× 4beff= 0.9beff(A.60)<br />
0.3375m 4 3 m<br />
3<br />
The bolt elastic deformation can be computed by means <strong>of</strong> Eq. (A.47) and<br />
assuming that <strong>the</strong> prying effect increases <strong>the</strong> bolt force from 0.5F to 0.63F. The<br />
bolt elastic stiffness is expressed as <strong>the</strong> ratio between <strong>the</strong> tensile force F and<br />
∆e.bt:<br />
F F EAs<br />
kebt<br />
. = = ≈1.6<br />
(A.61)<br />
∆ 0.63FL<br />
L<br />
ebt .<br />
b<br />
b<br />
EAs
PART II: FURTHER DEVELOPMENTS ON THE T-STUB MODEL<br />
59
2 IMPROVEMENTS ON THE T-STUB MODEL: INTRODUCTION<br />
2.1 INTRODUCTION<br />
The T-stub model is widely accepted as a simplified model for <strong>the</strong> <strong>characterization</strong><br />
<strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> <strong>the</strong> tension zone <strong>of</strong> a <strong>bolted</strong> joint, which is <strong>of</strong>ten <strong>the</strong><br />
most important source <strong>of</strong> deformability <strong>of</strong> <strong>the</strong> whole joint. Within <strong>the</strong> framework<br />
<strong>of</strong> <strong>the</strong> component method, this connection behaviour is modelled by<br />
means <strong>of</strong> a F-∆ response that is intrinsically nonlinear, due to mechanical and<br />
geometrical nonlinearities and contact phenomena. Current design specifications<br />
based on <strong>the</strong> T-stub model rely on pure plastic yield line mechanisms and<br />
do not allow for a complete <strong>characterization</strong> <strong>of</strong> <strong>the</strong> deformation capacity at ultimate<br />
conditions.<br />
Modern design codes, as <strong>the</strong> Eurocode 3 [2.1], approximate <strong>the</strong> nonlinear<br />
component behaviour by means <strong>of</strong> a linearized response, characterized by a full<br />
“plastic” resistance, FRd.0 and initial stiffness, ke.0. The design rules for <strong>the</strong> prediction<br />
<strong>of</strong> both parameters are given in §1.4.1. Fig. 2.1 illustrates <strong>the</strong> bilinear<br />
approximation <strong>of</strong> <strong>the</strong> actual behaviour <strong>of</strong> an isolated T-stub connection tested<br />
by Bursi and Jaspart [2.2]. In this particular case, <strong>the</strong> plastic mechanism <strong>of</strong> <strong>the</strong><br />
connection is <strong>of</strong> type-1, which corresponds to double curvature <strong>of</strong> <strong>the</strong> flange,<br />
owing to <strong>the</strong> formation <strong>of</strong> plastic hinges at <strong>the</strong> bolt axes and at <strong>the</strong> flange-toweb<br />
connection. Therefore, <strong>the</strong> connection has considerable deformation capacity<br />
[2.3]. No quantitative guidance is given in <strong>the</strong> code to evaluate this<br />
property though, and <strong>the</strong>refore no limits are imposed to <strong>the</strong> extension <strong>of</strong> <strong>the</strong><br />
plastic <strong>plate</strong>au.<br />
This part <strong>of</strong> <strong>the</strong> research work is devoted to <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> full<br />
nonlinear (monotonic) behaviour <strong>of</strong> isolated T-stub connections, in order to<br />
provide insight into <strong>the</strong> actual component behaviour, failure modes and deformation<br />
capacity. Tests, both experimental and numerical, were hence carried<br />
out at <strong>the</strong> Delft University <strong>of</strong> Technology and at <strong>the</strong> University <strong>of</strong> Coimbra to<br />
fulfil those objectives. Additionally, this test programme clarified some aspects<br />
related to <strong>the</strong> differences between <strong>the</strong> assembly types. The T-stub assemblage<br />
may comprise hot rolled pr<strong>of</strong>iles or welded <strong>plate</strong>s as T-stub elements, denominated<br />
HR-T-stubs and WP-T-stubs, respectively. The current approach to account<br />
for <strong>the</strong> behaviour <strong>of</strong> T-stubs made up <strong>of</strong> welded <strong>plate</strong>s consists in a mere<br />
extrapolation <strong>of</strong> <strong>the</strong> existing rules for <strong>the</strong> o<strong>the</strong>r assembly type. This assumption<br />
can be erroneous and can lead to unsafe estimations <strong>of</strong> <strong>the</strong> characteristic properties,<br />
as reported earlier by <strong>the</strong> author in technical literature [2.4-2.5].<br />
The following sections present and discuss <strong>the</strong> results <strong>of</strong> thirty-two experimental<br />
tests and three numerical tests on WP-T-stubs, and twenty-six numerical<br />
tests on HR-T-stubs. The experimental programme is described in Chapter<br />
61
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
62<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Actual response<br />
Eurocode 3 bilinear approximation<br />
0<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Total deformation (mm)<br />
Fig. 2.1 Simplified approximations <strong>of</strong> <strong>the</strong> response <strong>of</strong> a <strong>bolted</strong> T-stub connection.<br />
3. Detailed results for <strong>the</strong> benchmark specimen are also provided. The numerical<br />
model is fully described in Chapter 4 where <strong>the</strong> calibration procedure is<br />
also explained. Chapter 5 is completely devoted to <strong>the</strong> parametric study that<br />
highlights <strong>the</strong> main parameters affecting <strong>the</strong> deformation capacity <strong>of</strong> <strong>bolted</strong> Tstubs<br />
and assesses, both qualitatively and quantitatively, <strong>the</strong>ir influence on <strong>the</strong><br />
overall behaviour <strong>of</strong> <strong>the</strong> connection. Moreover, this study adds fur<strong>the</strong>r examples<br />
to a database for future validation <strong>of</strong> a simplified analytical (beam) model<br />
that is addressed in Chapter 6. This model attempts at filling in some code gaps<br />
on <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> component behaviour, namely post-limit stiffness,<br />
kp-l.0 and deformation capacity, ∆u.0.<br />
2.2 FAILURE MODES<br />
In this work, two categories <strong>of</strong> failure modes are considered: plastic mechanisms,<br />
that rely on pure “plastic” conditions and ultimate conditions, which<br />
correspond to cracking <strong>of</strong> <strong>the</strong> material. “Plastic” failure mechanisms indicate<br />
<strong>the</strong> strength <strong>of</strong> connections for design purposes whereas ultimate conditions indicate<br />
failure <strong>of</strong> <strong>the</strong> connection after certain deformation. The three possible<br />
“plastic” failure mechanisms have been briefly described in §1.4.1 and correspond<br />
to: (i) type-1: complete yielding <strong>of</strong> <strong>the</strong> flange, with <strong>the</strong> development <strong>of</strong><br />
four plastic hinges (double curvature b<strong>end</strong>ing), (ii) type-2: partial yielding <strong>of</strong><br />
<strong>the</strong> flange with bolt “plastic” failure, with <strong>the</strong> development <strong>of</strong> two plastic<br />
hinges at <strong>the</strong> flange-to-web connection (single curvature b<strong>end</strong>ing) and (iii)<br />
type-3: bolt “plastic” failure without yielding <strong>of</strong> <strong>the</strong> flanges (<strong>the</strong> flange remains<br />
virtually undeformed). With respect to ultimate conditions, four different typologies<br />
for <strong>the</strong> failure modes <strong>of</strong> a <strong>bolted</strong> T-stub connection are defined: (i)<br />
type-11, characterized by a plastic type-1 mode and cracking <strong>of</strong> <strong>the</strong> flange ma-
Improvements on <strong>the</strong> T-stub model: introduction<br />
terial at ultimate conditions, (ii) type-13, also a type-1 plastic mechanism but<br />
with fracture <strong>of</strong> <strong>the</strong> bolt at limit conditions, (iii) type-23, where <strong>the</strong> plastic<br />
mode involves both flange and bolt and <strong>the</strong> deformation capacity is governed<br />
by <strong>the</strong> bolt itself and (iv) type-33, a type-3 “plastic” mode and deformation capacity<br />
determined by bolt fracture.<br />
The failure mechanism typology, in both cases, is governed by <strong>the</strong> β-ratio,<br />
which is a resistance-based parameter that expresses <strong>the</strong> ratio between <strong>the</strong> flexural<br />
resistance <strong>of</strong> <strong>the</strong> flanges and <strong>the</strong> axial strength <strong>of</strong> <strong>the</strong> bolts. It dep<strong>end</strong>s exclusively<br />
on geometric and mechanic characteristics <strong>of</strong> <strong>the</strong> connection. In plastic<br />
conditions, this parameter is defined by Eq. (1.7). With reference to ultimate<br />
conditions, <strong>the</strong> β-ratio, βu, is given by:<br />
2M fu .<br />
β u = (2.1)<br />
Bm u<br />
whereby Mf.u is <strong>the</strong> ultimate flexural resistance <strong>of</strong> <strong>the</strong> T-stub flanges and Bu is<br />
<strong>the</strong> tensile strength <strong>of</strong> <strong>the</strong> bolts. According to Piluso et al. [2.6], <strong>the</strong> limit value<br />
for this parameter to have a collapse failure mode governed by cracking <strong>of</strong> <strong>the</strong><br />
flange material is:<br />
2λ<br />
⎡ dw<br />
⎤<br />
βu.lim = 1− ( 1+<br />
λ)<br />
2λ+ 1<br />
⎢<br />
8n<br />
⎥<br />
⎣ ⎦ (2.2)<br />
O<strong>the</strong>rwise (βu > βu.lim), bolt fracture is likely to determine <strong>the</strong> ultimate conditions.<br />
Table 2.1 summarizes <strong>the</strong> various failure mechanism types.<br />
The ultimate tensile strength <strong>of</strong> a bolt subjected to an axial loading is evaluated<br />
by assuming that tension fracture <strong>of</strong> <strong>the</strong> bolt occurs before stripping <strong>of</strong> <strong>the</strong><br />
threads. Therefore, <strong>the</strong> axial ultimate strength is calculated as <strong>the</strong> ultimate<br />
strength <strong>of</strong> <strong>the</strong> bolt material multiplied by <strong>the</strong> effective tensile area:<br />
Bu = fu. bAs (2.3)<br />
For computation <strong>of</strong> <strong>the</strong> ultimate flexural resistance <strong>of</strong> <strong>the</strong> flange, two alternative<br />
expressions are suggested. Gioncu et al. [2.7] propose <strong>the</strong> following relationship:<br />
M f . Rd<br />
M fu . = (2.4)<br />
ρ<br />
y<br />
Table 2.1 Failure mechanisms typologies.<br />
“Plastic” conditions Ultimate conditions<br />
Typology βRd Typology βu<br />
2λ<br />
Type-11 ≤ βu.lim<br />
Type-1 ≤<br />
2λ+ 1<br />
Type-13<br />
Type-2<br />
2λ<br />
><br />
2λ+ 1<br />
but 2 ≤ Type-23<br />
Type-3 > 2<br />
Type-33<br />
><br />
βu.lim<br />
63
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
being ρ y = fy fu<br />
<strong>the</strong> yield ratio and Mf.Rd <strong>the</strong> full plastic flexural resistance <strong>of</strong><br />
<strong>the</strong> T-stub flanges, defined in Eq. (1.6). Faella and co-authors [2.6,2.8] present<br />
an alternative expression that derives from simple equilibrium equations between<br />
internal stresses and <strong>the</strong> external moment. They assumed that <strong>the</strong> flange<br />
behaves as a rectangular compact section <strong>of</strong> width beff and thickness tf. The<br />
flange constitutive law is approximated by means <strong>of</strong> a quadrilinear model (Fig.<br />
2.2). The following relationship is <strong>the</strong>refore obtained:<br />
M f . y⎡ 1 E ⎛ h µ ⎞⎛ h µ ⎞ h<br />
M fu . = ⎢3− + 2 ( µ u−µ h)<br />
⎜1− ⎟⎜2+ ⎟−<br />
2 ⎢⎣ µ u E ⎝ µ u ⎠⎝ µ u ⎠<br />
Eh − E ⎛ u µ ⎞⎛ m µ ⎞⎤<br />
m<br />
− ( µ u −µ m)<br />
⎜1− ⎟⎜2+ ⎟⎥ E<br />
⎝ µ u ⎠⎝ µ u ⎠⎦⎥<br />
(2.5)<br />
whereby Mf.y is <strong>the</strong> b<strong>end</strong>ing moment corresponding to first yielding <strong>of</strong> <strong>the</strong><br />
flange:<br />
2<br />
t f<br />
M f . y= 6<br />
and:<br />
fy. fbeff 2<br />
= M f . Rd<br />
3<br />
(2.6)<br />
ε h µ h =<br />
ε y<br />
εm µ m =<br />
εy εu<br />
µ u =<br />
εy<br />
(2.7)<br />
The ratio M fu . M fy . is a parameter that also dep<strong>end</strong>s exclusively on <strong>the</strong> mechanical<br />
properties <strong>of</strong> <strong>the</strong> flange material and can be written in a formally<br />
equivalent expression to Eq. (2.4):<br />
M f . Rd<br />
M fu . = *<br />
ρ<br />
(2.8)<br />
64<br />
y<br />
σ<br />
fy<br />
E<br />
Eh<br />
εy εh εm εu<br />
Fig. 2.2 Flange piecewise material constitutive law (qaudrilinear approximation<br />
proposed by Faella and co-authors [2.8]).<br />
Eu<br />
ε
with (cf. Eqs. (2.5-2.6)):<br />
* 4⎡1 E ⎛ h µ ⎞⎛ h µ ⎞ h<br />
ρ y = ⎢3− + 2 ( µ u −µ h)<br />
⎜1− ⎟⎜2+ ⎟−<br />
3 ⎢⎣ µ u E ⎝ µ u ⎠⎝ µ u ⎠<br />
−1<br />
Eh − E ⎛ u µ ⎞⎛ m µ ⎞⎤<br />
m<br />
− ( µ u −µ m)<br />
⎜1− ⎟⎜2+ ⎟⎥ E<br />
⎝ µ u ⎠⎝ µ u ⎠⎦⎥<br />
2.3 REFERENCES<br />
Improvements on <strong>the</strong> T-stub model: introduction<br />
(2.9)<br />
[2.1] European Committee for Standardization (CEN). PrEN 1993-1-8:2003,<br />
Eurocode 3: Design <strong>of</strong> steel structures, Part 1.8: Design <strong>of</strong> joints, Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[2.2] Bursi OS, Jaspart JP. Benchmarks for finite element modelling <strong>of</strong> <strong>bolted</strong><br />
steel connections. Journal <strong>of</strong> Constructional Steel Research; 43(1):17-<br />
42, 1997.<br />
[2.3] Zoetemeijer P. Summary <strong>of</strong> <strong>the</strong> research on <strong>bolted</strong> beam-to-column<br />
connections. Report 25-6-90-2. Faculty <strong>of</strong> Civil Engineering, Stevin<br />
Laboratory – Steel Structures, Delft University <strong>of</strong> Technology. 1990.<br />
[2.4] Girão Coelho AM, Bijlaard F, Simões da Silva L. On <strong>the</strong> deformation<br />
capacity <strong>of</strong> beam-to-column <strong>bolted</strong> connections. Document ECCS-<br />
TWG 10.2-02-003, European Convention for Constructional Steelwork<br />
– Technical Committee 10, Structural connections (ECCS-TC10), 2002.<br />
[2.5] Girão Coelho AM, Bijlaard F, Simões da Silva L. On <strong>the</strong> behaviour <strong>of</strong><br />
<strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections modelled by welded T-stubs. In: Proceedings<br />
<strong>of</strong> <strong>the</strong> Third European Conference on Steel Structures (Eurosteel)<br />
(Eds.: A. Lamas and L. Simões da Silva), Coimbra, Portugal, 907-918,<br />
2002.<br />
[2.6] Piluso V, Faella C, Rizzano G. Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs – I:<br />
Theoretical model. Journal <strong>of</strong> Structural Engineering ASCE;<br />
127(6):686-693, 2001.<br />
[2.7] Gioncu V, Mateescu G, Petcu D, Anastasiadis A. Prediction <strong>of</strong> available<br />
<strong>ductility</strong> by means <strong>of</strong> local plastic mechanism method: DUCTROT<br />
computer program, Chapter 2.1 in Moment resistant connections <strong>of</strong> steel<br />
frames in seismic areas (Ed.: F. Mazzolani). E&FN Spon, London, UK;<br />
95-146, 2000.<br />
[2.8] Faella C, Piluso V, Rizzano G. Structural semi-rigid connections – <strong>the</strong>ory,<br />
design and s<strong>of</strong>tware. CRC Press, USA, 2000.<br />
65
3 EXPERIMENTAL ASSESSMENT OF THE BEHAVIOUR OF T-<br />
STUB CONNECTIONS<br />
3.1 INTRODUCTION<br />
A series <strong>of</strong> thirty-two tests on <strong>bolted</strong> T-stub connections made up <strong>of</strong> welded<br />
<strong>plate</strong>s is presented in this chapter. Although T-stubs have been used for many<br />
years to model <strong>the</strong> tension zone <strong>of</strong> <strong>bolted</strong> joints, <strong>the</strong> research was mainly concentrated<br />
on rolled pr<strong>of</strong>iles as T-stub elements. To ext<strong>end</strong> this model to <strong>the</strong><br />
case <strong>of</strong> welded <strong>plate</strong>s as T-stub elements, a test programme was undertaken at<br />
<strong>the</strong> Delft University <strong>of</strong> Technology. It provides insight into <strong>the</strong> behaviour <strong>of</strong><br />
this different type <strong>of</strong> assembly, in terms <strong>of</strong> resistance, stiffness, deformation<br />
capacity and failure modes, in particular. The key variables tested include <strong>the</strong><br />
weld throat thickness, <strong>the</strong> size <strong>of</strong> <strong>the</strong> T-stub, <strong>the</strong> type and diameter <strong>of</strong> <strong>the</strong> bolts,<br />
<strong>the</strong> steel grade, <strong>the</strong> presence <strong>of</strong> transverse stiffeners and <strong>the</strong> T-stub orientation.<br />
The results show that <strong>the</strong> welding procedure is particularly important to ensure<br />
a ductile behaviour <strong>of</strong> <strong>the</strong> connection. Most <strong>of</strong> <strong>the</strong> T-stubs failed by tension<br />
fracture <strong>of</strong> <strong>the</strong> bolts after significant yielding <strong>of</strong> <strong>the</strong> flanges. However,<br />
some <strong>of</strong> <strong>the</strong> specimens have shown early damage <strong>of</strong> <strong>the</strong> <strong>plate</strong> material near <strong>the</strong><br />
weld toe due to <strong>the</strong> effect <strong>of</strong> <strong>the</strong> welding consumable that induced premature<br />
cracking and reduced <strong>the</strong> overall deformation capacity. A solution to this problem<br />
was given by setting requirements to <strong>the</strong> weld metal to be used.<br />
This chapter describes <strong>the</strong> main collapse modes observed and gives detailed<br />
information on <strong>the</strong> benchmark specimen WT1 (eight tests). The remaining results<br />
are discussed in Chapter 5, as part <strong>of</strong> <strong>the</strong> parametric study presented.<br />
3.2 DESCRIPTION OF THE EXPERIMENTAL PROGRAMME<br />
3.2.1 Geometrical properties <strong>of</strong> <strong>the</strong> specimens<br />
The basic configuration <strong>of</strong> <strong>the</strong> test specimens comprised two <strong>plate</strong>s <strong>of</strong> 10 mm<br />
thickness. The <strong>plate</strong>s were welded toge<strong>the</strong>r by means <strong>of</strong> a continuous 45º-fillet<br />
weld with similar <strong>plate</strong> characteristics. Snug-tightened high-strength bolts fastened<br />
<strong>the</strong> T-stub elements. The unstiffened specimens were designed to fail according<br />
to a plastic collapse mode 1 that ensures a good <strong>ductility</strong> <strong>of</strong> <strong>the</strong> connection<br />
[3.1].<br />
The general characteristics <strong>of</strong> <strong>the</strong> specimens are given in Table 3.1. For notation<br />
<strong>the</strong> reader should refer to Fig. 3.1. Both nominal and actual properties<br />
are reported. The actual geometry was measured before testing <strong>the</strong> specimens<br />
and is listed in Table 3.1 as an average value <strong>of</strong> <strong>the</strong> several T-elements from<br />
67
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
each series. In series WT7 and WT57, <strong>the</strong> T-stub elements were fastened by<br />
means <strong>of</strong> one bolt row only due to equipment limitation. The two T-stubs for<br />
most series were symmetrical. For series WT64A and WT64B, <strong>the</strong> T-stub elements<br />
included a stiffener only on one side <strong>of</strong> <strong>the</strong> connection.<br />
Table 3.1 Tests description [dimensions (nominal and averaged actual values<br />
– in bold) in mm; //: T-stub elements parallel, ⊥: T-stub elements<br />
orientated at right angles].<br />
Test ID #<br />
Geometry<br />
WT1 8<br />
tf<br />
10<br />
10.32<br />
tw<br />
10<br />
10.93<br />
b<br />
45<br />
45.05<br />
p<br />
50<br />
49.8<br />
e1<br />
20<br />
20.1<br />
w<br />
90<br />
89.7<br />
e<br />
30<br />
30.0<br />
aw<br />
5<br />
WT2A<br />
WT2B<br />
WT4A<br />
2<br />
2<br />
2<br />
10<br />
10.27<br />
10<br />
10.29<br />
10<br />
10.39<br />
10<br />
10.54<br />
10<br />
10.69<br />
10<br />
11.00<br />
45<br />
45.0<br />
45<br />
44.95<br />
75<br />
74.85<br />
50<br />
49.9<br />
50<br />
49.9<br />
90<br />
89.6<br />
20<br />
20.0<br />
20<br />
20.0<br />
30<br />
30.1<br />
90<br />
89.9<br />
90<br />
89.9<br />
90<br />
89.7<br />
30<br />
29.9<br />
30<br />
29.9<br />
30<br />
30.0<br />
3<br />
7<br />
5<br />
WT4B 1<br />
10<br />
10.37<br />
10<br />
10.92<br />
75<br />
74.8<br />
90<br />
89.8<br />
30<br />
29.9<br />
90<br />
89.8<br />
30<br />
29.9<br />
5<br />
WT51<br />
WT53C<br />
2<br />
1<br />
10<br />
9.98<br />
10<br />
10.09<br />
10<br />
10.01<br />
10<br />
10.10<br />
45<br />
45.0<br />
45<br />
45.05<br />
50<br />
50.7<br />
50<br />
50.0<br />
20<br />
19.6<br />
20<br />
20.0<br />
90<br />
90.1<br />
90<br />
90.1<br />
30<br />
30.2<br />
30<br />
30.0<br />
5<br />
5<br />
WT53D 1<br />
10<br />
10.14<br />
10<br />
10.22<br />
45<br />
45.0<br />
50<br />
49.9<br />
20<br />
20.0<br />
90<br />
90.0<br />
30<br />
30.0<br />
5<br />
WT53E<br />
WT61<br />
1<br />
2<br />
10<br />
10.09<br />
10<br />
10.31<br />
10<br />
10.17<br />
10<br />
10.93<br />
45<br />
44.65<br />
45<br />
45.1<br />
50<br />
49.2<br />
50<br />
49.9<br />
20<br />
20.0<br />
20<br />
20.1<br />
90<br />
90.0<br />
90<br />
89.8<br />
30<br />
30.1<br />
30<br />
29.4<br />
5<br />
5<br />
WT64A 1<br />
10<br />
10.28<br />
10<br />
10.94<br />
75<br />
74.95<br />
90<br />
90.0<br />
30<br />
29.9<br />
90<br />
89.7<br />
30<br />
29.8<br />
5<br />
WT64B<br />
WT64C<br />
2<br />
1<br />
10<br />
10.42<br />
10<br />
10.30<br />
10<br />
10.82<br />
10<br />
10.84<br />
75<br />
74.9<br />
75<br />
75.1<br />
90<br />
89.9<br />
90<br />
89.7<br />
30<br />
29.9<br />
30<br />
30.2<br />
90<br />
89.7<br />
90<br />
89.8<br />
30<br />
30.0<br />
30<br />
29.9<br />
5<br />
5<br />
WT7_M12 1<br />
10<br />
10.33<br />
10<br />
10.84<br />
75<br />
75.6<br />
30 90 30<br />
⎯<br />
30.0 89.9 29.9<br />
5<br />
WT7_M16 1<br />
10<br />
10.33<br />
10<br />
10.81<br />
75<br />
74.9<br />
30 90 30<br />
⎯<br />
WT7_M20 1<br />
10<br />
10.33<br />
10<br />
10.87<br />
75<br />
75.2<br />
30.0<br />
30<br />
89.9<br />
90<br />
29.8<br />
30<br />
5<br />
⎯<br />
WT57_M12 1<br />
10<br />
10.09<br />
10<br />
10.18<br />
75<br />
75.0<br />
29.9<br />
30<br />
89.8<br />
90<br />
29.7<br />
30<br />
5<br />
⎯<br />
30.0 89.7 30.2<br />
5<br />
WT57_M16 1<br />
10<br />
10.16<br />
10<br />
10.18<br />
75<br />
75.3<br />
30 90 30<br />
⎯<br />
WT57_M20 1<br />
10<br />
10.15<br />
10<br />
10.15<br />
75<br />
75.1<br />
30.0<br />
30<br />
90.0<br />
90<br />
30.1<br />
30<br />
5<br />
⎯<br />
30.0 90.0 30.2<br />
5<br />
68
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
3.2.2 Mechanical properties <strong>of</strong> <strong>the</strong> specimens<br />
3.2.2.1 Tension tests on <strong>the</strong> bolts<br />
In order to characterize <strong>the</strong> mechanical properties <strong>of</strong> <strong>the</strong> M12, grade 8.8 and<br />
10.9 bolts, two series <strong>of</strong> experiments were performed. In <strong>the</strong> first series, <strong>the</strong> ac-<br />
Table 3.1 Tests description (cont.).<br />
Test ID # Bolt Materials Stiff. Orient.<br />
φ # Type Plate Bolt<br />
WT1 8<br />
12<br />
11.8<br />
4 ST S355 8.8 No //<br />
WT2A<br />
WT2B<br />
WT4A<br />
2<br />
2<br />
2<br />
12<br />
11.8<br />
12<br />
11.8<br />
12<br />
11.8<br />
4<br />
4<br />
4<br />
ST<br />
ST<br />
ST<br />
S355<br />
S355<br />
S355<br />
8.8<br />
8.8<br />
8.8<br />
No<br />
No<br />
No<br />
//<br />
//<br />
//<br />
WT4B 1<br />
12<br />
11.8<br />
4 ST S355 8.8 No ⊥<br />
WT51<br />
WT53C<br />
2<br />
1<br />
12<br />
11.8<br />
12<br />
⎯<br />
4<br />
4<br />
ST<br />
FT<br />
S690<br />
S690<br />
8.8<br />
8.8<br />
No<br />
No<br />
//<br />
//<br />
WT53D 1<br />
12<br />
11.9<br />
4 ST S690 10.9 No //<br />
WT53E<br />
WT61<br />
1<br />
2<br />
12<br />
⎯<br />
12<br />
11.9<br />
4<br />
4<br />
FT<br />
ST<br />
S690<br />
S355<br />
10.9<br />
8.8<br />
No<br />
Yes<br />
//<br />
//<br />
WT64A 1<br />
12<br />
11.8<br />
4 ST S355 8.8 Yes //<br />
WT64B<br />
WT64C<br />
2<br />
1<br />
12<br />
11.8<br />
12<br />
11.8<br />
4<br />
4<br />
ST<br />
ST<br />
S355<br />
S355<br />
8.8<br />
8.8<br />
Yes<br />
Yes<br />
⊥<br />
//<br />
WT7_M12 1<br />
12<br />
11.9<br />
2 ST S355 8.8 No //<br />
WT7_M16<br />
WT7_M20<br />
WT57_M12<br />
1<br />
1<br />
1<br />
16<br />
⎯<br />
20<br />
⎯<br />
12<br />
2<br />
2<br />
2<br />
FT<br />
FT<br />
FT<br />
S355<br />
S355<br />
S690<br />
8.8<br />
8.8<br />
8.8<br />
No<br />
No<br />
No<br />
//<br />
//<br />
//<br />
WT57_M16<br />
WT57_M20<br />
1<br />
1<br />
16<br />
⎯<br />
20<br />
⎯<br />
2<br />
2<br />
FT<br />
FT<br />
S690<br />
S690<br />
8.8<br />
8.8<br />
No<br />
No<br />
//<br />
//<br />
69
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
70<br />
Section xx<br />
x x<br />
e w<br />
bf<br />
aw<br />
Fig. 3.1 T-stub geometry: notation.<br />
e<br />
e1<br />
p<br />
e1<br />
b<br />
b<br />
Plan<br />
tual bolt (short-threaded, ST or full-threaded, FT) was tested under tension<br />
(Fig. 3.2a). Failure always occurred in <strong>the</strong> threaded region. This type <strong>of</strong> test did<br />
not provide enough data to determine <strong>the</strong> Young modulus and <strong>the</strong> pro<strong>of</strong><br />
strength <strong>of</strong> <strong>the</strong> bolt. Then, in a second test series, <strong>the</strong> bolts were machined so<br />
that <strong>the</strong> threads within <strong>the</strong> bolt grip were removed and a constant diameter was<br />
obtained (Fig. 3.2b). This procedure was not expected to introduce major influences<br />
on <strong>the</strong> bolt behaviour since <strong>the</strong> removal <strong>of</strong> <strong>the</strong> material was limited to <strong>the</strong><br />
threads, even though <strong>the</strong> bolt mechanical properties were not uniform.<br />
Both specimen types were tested in tension under displacement control in a<br />
special test rig as shown in Fig. 3.3. The elongation behaviour <strong>of</strong> <strong>the</strong> bolt was<br />
measured by means <strong>of</strong> a measuring bracket (or horseshoe device, also illustrated<br />
in Fig. 3.3) in <strong>the</strong> first series <strong>of</strong> tests and by means <strong>of</strong> internal strain<br />
gauges in <strong>the</strong> second. The strain gauges (TML-BTM-6C) could measure strains<br />
up to 6000 µm/m. The graphs from Fig. 3.4a plot <strong>the</strong> bolt elongation curve for
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
one <strong>of</strong> <strong>the</strong> short-threaded M12 grade 8.8 (group 2) tested bolts. The graph includes<br />
<strong>the</strong> load cell displacement results and <strong>the</strong> measuring bracket data up to<br />
its removal in <strong>the</strong> elastic range. Clearly, <strong>the</strong> results obtained from <strong>the</strong> measuring<br />
bracket are stiffer since <strong>the</strong> displacement <strong>of</strong> <strong>the</strong> actuator also includes <strong>the</strong><br />
slippery <strong>of</strong> <strong>the</strong> clamps. Fig. 3.4b traces <strong>the</strong> force-strain results obtained for an<br />
identical bolt type (now chosen from <strong>the</strong> second test series). Naturally, <strong>the</strong><br />
Group 1 (8.8) Group 3 (10.9) Group 2 (8.8) Group 4 (10.9)<br />
(i) Full-threaded specimens.<br />
(a) First series: “actual” bolts.<br />
(ii) Short-threaded specimens.<br />
Group 1 (8.8) Group 3 (10.9) Group 2 (8.8) Group 4 (10.9)<br />
(i) Full-threaded specimens.<br />
(b) Second series: machined bolts.<br />
(ii) Short-threaded specimens.<br />
Fig. 3.2 Bolt specimens: two series <strong>of</strong> tests.<br />
Fig. 3.3 Test rig for <strong>the</strong> bolt tensile testing and horseshoe device.<br />
71
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
72<br />
Applied load (kN)<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
(a) Bolt elongation behaviour.<br />
Applied load (kN)<br />
20<br />
Results from <strong>the</strong> load cell<br />
10<br />
0<br />
Results from <strong>the</strong> measuring bracket<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
(b) Bolt strain behaviour.<br />
Applied load (kN)<br />
Deformation (mm)<br />
20<br />
10<br />
0<br />
0 600 1200 1800 2400 3000 3600 4200 4800 5400 6000 6600<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
Strain (µm/m)<br />
Maximum load<br />
Results from <strong>the</strong> measuring bracket<br />
Results from <strong>the</strong> strain gauge<br />
0<br />
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10<br />
Bolt elongation (mm)<br />
(c) Comparison <strong>of</strong> <strong>the</strong> bolt experimental (elastic) results for both test series.<br />
Fig. 3.4 Bolt tensile response (e.g. bolt from group 2).
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
maximum load level in <strong>the</strong> latter case decreases, as <strong>the</strong> bolt tensile area is<br />
smaller. To verify <strong>the</strong> accuracy <strong>of</strong> <strong>the</strong> special bolt-measuring bracket, Fig. 3.4c<br />
compares <strong>the</strong> elastic deformation <strong>of</strong> two bolts from <strong>the</strong> same group 2, representing<br />
each bolt series. The bolt elongation is given by ∆b = εbLg, in <strong>the</strong> case<br />
<strong>of</strong> <strong>the</strong> strain measurements, being εb <strong>the</strong> bolt strain and Lg <strong>the</strong> grip length. The<br />
results are identical for both series, which means that <strong>the</strong> measuring bracket<br />
that is much simpler to attach can be used to assess <strong>the</strong> bolt elongation behaviour<br />
in future tests.<br />
Bolts M16 and M20 used in series WT7 and WT57 have not been tested.<br />
Table 3.2 summarizes <strong>the</strong> average relevant characteristics for <strong>the</strong> tested<br />
bolts. Usually, for <strong>the</strong> bolts, <strong>the</strong> following parameters are measured: Young<br />
modulus, E, ultimate or tensile stress, fu and ultimate strain, εu.<br />
Table 3.2 Average characteristic values for <strong>the</strong> bolts.<br />
Bolt grade Type Group E (MPa) fu (MPa) εu<br />
8.8<br />
FT<br />
ST<br />
1<br />
2<br />
216942<br />
221886<br />
968.36<br />
919.91<br />
0.20<br />
0.13<br />
10.9<br />
FT<br />
ST<br />
3<br />
4<br />
217060<br />
217824<br />
1196.37<br />
1165.97<br />
0.14<br />
0.11<br />
3.2.2.2 Tension tests on <strong>the</strong> structural steel<br />
The test programme included two different steel grades: S355 and S690. According<br />
to <strong>the</strong> European Standards EN 10025 [3.2] and EN 10204 [3.3], <strong>the</strong><br />
steel qualities were S355J0 (ordinary steel) and N-A-XTRA M70 (highstrength<br />
steel for <strong>plate</strong>s), respectively. Table 3.3 summarizes <strong>the</strong> chemical<br />
composition for <strong>the</strong> two steel grades.<br />
The coupon tension testing <strong>of</strong> <strong>the</strong> structural steel material was performed<br />
according to <strong>the</strong> RILEM procedures [3.4]. The <strong>plate</strong> coupon specimens were <strong>of</strong><br />
a standard type for flat materials and were <strong>of</strong> full thickness <strong>of</strong> <strong>the</strong> product [3.4].<br />
Fig. 3.5 depicts <strong>the</strong> test arrangements for <strong>the</strong> standard tensile test. The experiments<br />
were driven under displacement control. The engineering stress-strain<br />
relation for <strong>the</strong> web and flange strips is represented in Fig. 3.6, for one <strong>of</strong> <strong>the</strong><br />
tested strip-coupons. The four typical regions <strong>of</strong> <strong>the</strong> stress-strain curve <strong>of</strong> a low<br />
carbon structural steel are very clear: linear elastic region, yield <strong>plate</strong>au, strain<br />
hardening region and strain s<strong>of</strong>tening or necking portion, after reaching <strong>the</strong><br />
maximum load.<br />
The average characteristics are set out in Table 3.4. In this table <strong>the</strong> values<br />
for <strong>the</strong> Young modulus, E, <strong>the</strong> strain hardening modulus, Eh, <strong>the</strong> static yield<br />
and tensile stresses, fy and fu, <strong>the</strong> yield ratio, ρy <strong>the</strong> strain at <strong>the</strong> strain hardening<br />
point, εh, <strong>the</strong> uniform strain, εuni, and <strong>the</strong> ultimate strain, εu, are given. The<br />
stress values indicated in <strong>the</strong> table correspond to <strong>the</strong> static stresses, which are<br />
<strong>the</strong> stress values obtained at zero strain rate, i.e. during a hold on <strong>of</strong> <strong>the</strong> defor-<br />
73
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(a) Test set up. (b) Detail <strong>of</strong> <strong>the</strong><br />
extensometer.<br />
Fig. 3.5 Tensile coupon tests.<br />
74<br />
Stress (MPa)<br />
900<br />
750<br />
600<br />
450<br />
(a) Steel grade S355.<br />
Stress (MPa)<br />
(c) Coupon<br />
necking.<br />
300<br />
Web strip coupon<br />
150<br />
Flange strip coupon<br />
0<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35<br />
900<br />
750<br />
600<br />
450<br />
Strain (m/m)<br />
300<br />
Web strip coupon<br />
150<br />
Flange strip coupon<br />
0<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35<br />
Strain (m/m)<br />
(b) Steel grade S690.<br />
Fig. 3.6 Engineering stress-strain relation.<br />
(d) S355 coupons after<br />
failure.
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
mation driven experiment. It has been observed that <strong>the</strong> static stresses were<br />
reached after a hold on <strong>of</strong> circa one minute. The total hold on lasted for three<br />
minutes. The yield ratio gives an idea on <strong>the</strong> material <strong>ductility</strong>. Gioncu and<br />
Mazzolani suggest that a good <strong>ductility</strong> is ensured if 0.5 ≤ ρy ≤ 0.7 [3.5]. High<br />
strength steel grades with ρy > 0.9 show a ra<strong>the</strong>r poor structural <strong>ductility</strong> [3.5].<br />
That is <strong>the</strong> case <strong>of</strong> <strong>the</strong> steel grade S690 (Table 3.4). In <strong>the</strong> author’s opinion,<br />
<strong>the</strong>se values are ra<strong>the</strong>r conservative. Eurocode 3 indicates that a good material<br />
<strong>ductility</strong> is guaranteed if ρy ≤ 0.83 (recomm<strong>end</strong>ed value for steel grades up to<br />
S460). The assurance <strong>of</strong> a good material <strong>ductility</strong> does not necessarily imply<br />
that <strong>the</strong> whole structure is ductile. The structural <strong>ductility</strong> dep<strong>end</strong>s on <strong>the</strong> yield<br />
ratio but especially on <strong>the</strong> structural discontinuities.<br />
Table 3.3 Chemical composition <strong>of</strong> <strong>the</strong> structural steels according to <strong>the</strong><br />
European standards.<br />
%C %Mn %Si %P %S %N %CEV<br />
max. max. max. max. max. max. max.<br />
S355J0 0.20 1.60 0.55 0.040 0.045 0.009 0.40<br />
0.20 1.60 0.80 0.020 0.010 ⎯ 0.48<br />
N-A-XTRA<br />
M70<br />
Table 3.4 Average characteristic values for <strong>the</strong> structural steels.<br />
Steel<br />
grade<br />
Strip # E<br />
(MPa)<br />
Eh<br />
(MPa)<br />
fy<br />
(MPa)<br />
fu<br />
(MPa)<br />
ρy<br />
S355<br />
Web<br />
Flange<br />
2<br />
2<br />
209211<br />
209856<br />
2145<br />
2264<br />
391.54<br />
340.12<br />
493.80<br />
480.49<br />
0.793<br />
0.708<br />
S690<br />
Web<br />
Flange<br />
2<br />
2<br />
208895<br />
204462<br />
2201<br />
2495<br />
706.31<br />
698.55<br />
742.96<br />
741.28<br />
0.950<br />
0.940<br />
Steel<br />
grade<br />
Strip # εh εuni εu<br />
S355<br />
Web<br />
Flange<br />
2<br />
2<br />
0.019<br />
0.015<br />
0.163<br />
0.224<br />
0.300<br />
0.361<br />
S690<br />
Web<br />
Flange<br />
2<br />
2<br />
0.018<br />
0.014<br />
0.082<br />
0.075<br />
0.160<br />
0.174<br />
3.2.3 Testing procedure<br />
The specimens were subjected to monotonic tensile force, which was applied to<br />
<strong>the</strong> webs that were clamped to <strong>the</strong> testing machine (Schenck, maximum test<br />
load 600 kN, maximum piston stroke ±125 mm) as shown in Fig. 3.7. The tests<br />
were carried out under displacement control with a speed <strong>of</strong> 0.01 mm/s up to<br />
collapse <strong>of</strong> <strong>the</strong> specimens. Two different ultimate failure modes were observed,<br />
75
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
as explained below in <strong>the</strong> text: (i) fracture <strong>of</strong> <strong>the</strong> bolts and (ii) cracking <strong>of</strong> <strong>the</strong><br />
flange near <strong>the</strong> weld toe.<br />
The gap <strong>of</strong> <strong>the</strong> flanges was measured at opposite sides <strong>of</strong> <strong>the</strong> specimen, in<br />
<strong>the</strong> centreline <strong>of</strong> <strong>the</strong> webs by means <strong>of</strong> Linear Variable Displacement Transducers<br />
(LVDTs). The bolt elongation was measured with a measuring bracket<br />
that was removed prior to collapse, as before, so that it was not damaged. In<br />
some <strong>of</strong> <strong>the</strong> specimens, internal strain gauges similar to those used in <strong>the</strong> tension<br />
tests were attached to <strong>the</strong> bolts. Strain gauges TML (maximum strain<br />
30000 µm/m) were used to monitor strains in <strong>the</strong> flange. Due to cost restrictions<br />
not all specimens have been instrumented with strain gauges. For illustration,<br />
Fig. 3.8 shows <strong>the</strong> instrumentation <strong>of</strong> some <strong>of</strong> <strong>the</strong> specimens.<br />
Before installation <strong>of</strong> <strong>the</strong> specimens into <strong>the</strong> testing machine, <strong>the</strong> dimensions<br />
<strong>of</strong> <strong>the</strong> <strong>plate</strong>s were recorded and <strong>the</strong> bolts were hand-tightened and measured.<br />
The specimen was next placed into <strong>the</strong> machine and aligned, so that <strong>the</strong><br />
clamping devices were centred with respect to <strong>the</strong> webs. The bolts were subsequently<br />
fastened by using an ordinary spanner (45º turn) and measured. After<br />
that, <strong>the</strong> measurement devices and strain gauges, if any, were connected. The<br />
test itself <strong>the</strong>n started with loading <strong>of</strong> <strong>the</strong> specimen up to 2/3FRd.0, which corresponded<br />
to <strong>the</strong> <strong>the</strong>oretical elastic limit. FRd.0 was determined according to<br />
Eurocode 3. Complete unloading followed on and <strong>the</strong> specimen was <strong>the</strong>n reloaded<br />
up to collapse. In this third phase <strong>the</strong> test was interrupted at <strong>the</strong> load<br />
(a) Unstiffened specimen. (b) Stif. spec. (T-stubs orientated at 90º).<br />
(c) Detail <strong>of</strong> <strong>the</strong> measuring devices.<br />
Fig. 3.7 Test set up for testing WP-T-stubs.<br />
76
30.0 mm<br />
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Left side Right side<br />
5.0 mm<br />
SG5<br />
SG4<br />
SG3<br />
SG2<br />
SG1<br />
90.0 mm<br />
SG6<br />
SG7<br />
30.0 mm<br />
20.0 mm<br />
50.0 mm<br />
20.0 mm<br />
Left side Right side<br />
30.0 mm<br />
SG4<br />
SG6z<br />
SG6x<br />
SG3<br />
SG2<br />
90.0 mm<br />
(a) Specimens WT1b/c/d/h, WT51a. (b) Specimen WT51b.<br />
Left side Right side<br />
LB RB<br />
30.0 mm<br />
SG3<br />
SG2<br />
LF RF<br />
SG1<br />
90.0 mm<br />
30.0 mm<br />
30.0 mm<br />
90.0 mm<br />
30.0 mm<br />
HP3<br />
SG7z<br />
SG7x<br />
30.0 mm<br />
Left side Right side<br />
HP1<br />
30.0 mm<br />
LB<br />
SG6 SG5<br />
LF<br />
SG4<br />
HP2<br />
90.0 mm<br />
(i) Upper pr<strong>of</strong>ile. (ii) Lower pr<strong>of</strong>ile.<br />
(c) Specimen WT64Bb.<br />
HP3<br />
14.0<br />
Left side Right side<br />
SG4<br />
SG2<br />
SG3<br />
40.0<br />
30.0 mm<br />
SG5<br />
16.0<br />
SG7<br />
SG6<br />
SG1<br />
48.0<br />
90.0 mm<br />
HP1<br />
HP2<br />
30.0 mm<br />
30.0 mm<br />
90.0 mm<br />
30.0 mm<br />
RB<br />
RF<br />
30.0 mm<br />
20.0 mm<br />
50.0 mm<br />
20.0 mm<br />
30.0 mm<br />
HP4 90.0 mm<br />
30.0 mm<br />
(d) Specimen WT64C (upper pr<strong>of</strong>ile; lower pr<strong>of</strong>ile not instrumented).<br />
Fig. 3.8 Instrumentation <strong>of</strong> some <strong>of</strong> <strong>the</strong> tested specimens (SG: strain gauge;<br />
L: left; R: right; B: back; F: front; HP: LVDT).<br />
77
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
levels <strong>of</strong> 2/3FRd.0, FRd.0, at <strong>the</strong> knee-range and after this level each six minutes,<br />
equivalent to an actuator displacement <strong>of</strong> 3.6 mm. The knee-range <strong>of</strong> <strong>the</strong> F-∆<br />
curve (K-R) corresponds to <strong>the</strong> transition from <strong>the</strong> stiff to <strong>the</strong> s<strong>of</strong>t part. The<br />
hold on <strong>of</strong> <strong>the</strong> test lasted for three minutes and int<strong>end</strong>ed to record <strong>the</strong> quasistatic<br />
forces.<br />
Regarding <strong>the</strong> stiffened specimens, <strong>the</strong> former load levels were taken as<br />
equal to <strong>the</strong> parent unstiffened cases. For <strong>the</strong> rotated configurations, <strong>the</strong> lower<br />
hydraulic actuator was rotated 90º so that <strong>the</strong> T-stub element was orientated at<br />
a right angle to <strong>the</strong> upper element (Fig. 3.7b).<br />
3.2.4 Aspects related to <strong>the</strong> welding procedure<br />
In this type <strong>of</strong> T-stub assembly, two <strong>plate</strong>s, web and flange, are welded toge<strong>the</strong>r<br />
by means <strong>of</strong> a continuous 45º-fillet weld. The fillet welds were done in<br />
<strong>the</strong> shop in a down-hand position. The procedure involved manual metal arc<br />
welding in which a consumable electrode was used. Three main zones could be<br />
identified after <strong>the</strong> welding process [3.6]: <strong>the</strong> weld metal (WM), <strong>the</strong> heat affected<br />
zone (HAZ) and <strong>the</strong> base metal (BM), which is <strong>the</strong> part <strong>of</strong> <strong>the</strong> parent<br />
<strong>plate</strong> that is not influenced by <strong>the</strong> heat input. The HAZ is <strong>the</strong> portion <strong>of</strong> <strong>the</strong><br />
<strong>plate</strong> on ei<strong>the</strong>r side <strong>of</strong> <strong>the</strong> weld affected by <strong>the</strong> heat in which metal suffers<br />
<strong>the</strong>rmal disturbances and <strong>the</strong>refore structural modifications that may include<br />
re-crystallization, refining and grain growth [3.7]. The hot WM causes <strong>the</strong> <strong>plate</strong><br />
to b<strong>end</strong> up due to shrinkage during cooling down and so considerable force is<br />
exerted to do this [3.7]. Residual stresses can <strong>the</strong>n be expected in <strong>the</strong> HAZ.<br />
Obviously, this will influence <strong>the</strong> overall behaviour <strong>of</strong> <strong>the</strong> connection.<br />
The composition <strong>of</strong> <strong>the</strong> WM deposited with <strong>the</strong> electrode compared to that<br />
<strong>of</strong> <strong>the</strong> BM is <strong>of</strong> great importance, since this will naturally alter <strong>the</strong> properties <strong>of</strong><br />
<strong>the</strong> steel at and near <strong>the</strong> weld toe [3.7]. For each steel quality <strong>the</strong>re are <strong>of</strong>ten a<br />
large number <strong>of</strong> electrode types to choose from. In this test programme two different<br />
types <strong>of</strong> carbon steel covered electrodes were used: rutile and basic (Table<br />
3.5). The distinction between <strong>the</strong>m lies in <strong>the</strong> type <strong>of</strong> covering that result in<br />
different performances. Rutile electrodes have high titanium oxide content and<br />
produce easy striking with a stable arc and low spatter. They are commonly<br />
known as general-purpose electrodes. The mechanical strength is generally<br />
classed as moderate. This type <strong>of</strong> consumable normally has high hydrogen content<br />
(higher than 10 ml/100 g all-WM). Basic electrodes <strong>of</strong>fer improved mechanical<br />
properties and superior weld penetration. The mechanical strength is<br />
generally classed as good to high and <strong>the</strong> resistance to cracking is enhanced.<br />
They have a high proportion <strong>of</strong> calcium carbonate and calcium fluoride in <strong>the</strong><br />
coating, which makes it more fluid than rutile coatings and also fast freezing.<br />
The hydrogen content is generally lower, which reduces <strong>the</strong> cracking problem.<br />
Table 3.5 summarizes <strong>the</strong> main characteristics <strong>of</strong> <strong>the</strong> various electrodes.<br />
The classification indicated in <strong>the</strong> table complies with <strong>the</strong> European standard<br />
EN 499 [3.8]. Regarding this classification standard, <strong>the</strong> first two digits desig-<br />
78
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
nate <strong>the</strong> minimum yield strength <strong>of</strong> <strong>the</strong> deposited WM and also refer to <strong>the</strong><br />
limit boundaries <strong>of</strong> <strong>the</strong> tensile strength and <strong>the</strong> minimum elongation <strong>of</strong> <strong>the</strong><br />
WM. For instance, <strong>the</strong> Kardo electrode (E35) has a minimum yield stress <strong>of</strong><br />
350 MPa (measured value: 396 MPa), <strong>the</strong> tensile strength varies between 440<br />
and 570 MPa (measured value: 453 MPa) and a minimum elongation <strong>of</strong> 22%.<br />
The latter value decreases as <strong>the</strong> strength <strong>of</strong> <strong>the</strong> WM increases [3.8], thus reducing<br />
<strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> weld.<br />
Table 3.6 lists <strong>the</strong> electrodes used in <strong>the</strong> welding <strong>of</strong> each specimen. Clearly,<br />
<strong>the</strong> Kardo and <strong>the</strong> Conarc 70G were <strong>the</strong> most utilized electrode types. These<br />
are s<strong>of</strong>t, low hydrogen electrodes. The experiences on <strong>the</strong> consumable performance<br />
were carried out in test series WT1. Fig. 3.9 shows <strong>the</strong> influence <strong>of</strong><br />
<strong>the</strong> deposited WM on <strong>the</strong> global behaviour <strong>of</strong> <strong>the</strong> eight specimens from series<br />
WT1. Essentially, such behaviour mainly dep<strong>end</strong>s on <strong>the</strong> mismatch in mechanical<br />
properties between <strong>the</strong> three different zones and <strong>the</strong> hydrogen content<br />
[3.6-3.7]. In <strong>the</strong> elastic range, <strong>the</strong> deformation behaviour is not too much dep<strong>end</strong>ent<br />
on <strong>the</strong> WM properties. However, when <strong>the</strong> connection is plastically<br />
deformed, <strong>the</strong> choice <strong>of</strong> <strong>the</strong> electrode type becomes crucial. The graphs show<br />
that <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> joint was greatly influenced by <strong>the</strong> depos-<br />
Table 3.5 Characteristics <strong>of</strong> <strong>the</strong> electrodes and mechanical properties <strong>of</strong> <strong>the</strong><br />
deposited weld metal.<br />
Brand Type Classif. Actual mech. prop.<br />
name (EN 499) fy (MPa) fu (MPa)<br />
Cumulo Rutile E38 O R12 Not provided.<br />
Conarc 51 Basic E42 4 B12 H5 Not provided.<br />
Kardo Basic E35 4 B32 H5 396 453<br />
Conarc 70G Basic E55 4 B32 H5 600 655<br />
Brand Chemical composition<br />
name %C %Mn %Si %P %S %Ni<br />
Cumulo 0.06 0.50 0.30 ⎯ ⎯ ⎯<br />
Conarc 51 Not provided.<br />
Kardo 0.016 0.30 0.21 0.010 0.008 0.03<br />
Conarc 70G 0.06 1.2 0.4 0.014 0.009 1.0<br />
Table 3.6 Types <strong>of</strong> electrode used in <strong>the</strong> tests.<br />
Test ID Electrode Test ID Electrode<br />
WT1a/b/c Cumulo Series WT51 Conarc 70G<br />
WT1d Conarc 51 Series WT53 Conarc 70G<br />
WT1e/f Cumulo Series WT61 Kardo<br />
WTg/h Kardo Series WT64 Kardo<br />
Series WT2 Kardo Series WT7 Kardo<br />
Series WT4 Kardo Series WT57 Conarc 70G<br />
79
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
80<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
(a) Cumulo electrode (rutile).<br />
Total applied load (kN)<br />
WT1f<br />
WT1e WT1a<br />
WT1b<br />
WT1c<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Deformation (mm)<br />
WT1d<br />
WT1c<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
(b) Conarc 51 and Cumulo electrodes (basic and rutile, respectively).<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
WT1d<br />
WT1c<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
WT1g<br />
WT1h<br />
(c) Kardo electrode (basic).<br />
Fig. 3.9 Performance <strong>of</strong> <strong>the</strong> different electrode types for steel grade S355.
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
(i) Deformation at failure (WT1e). (ii) Detail: typical crack (WT1a).<br />
(a) Cumulo electrode (rutile).<br />
(i) Deformation at failure. (ii) Detail: crack.<br />
(iii) Detail: bolts (no b<strong>end</strong>ing deformations).<br />
(b) Cumulo electrode (rutile) and aw = 8.0 mm (WT1f).<br />
(c) Conarc 51 electrode (basic) (WT1d). (d) Kardo electrode (basic) (WT1h).<br />
Fig. 3.10 Illustration: specimens (series WT1) after failure for comparison <strong>of</strong> <strong>the</strong><br />
effect <strong>of</strong> <strong>the</strong> deposited WM with different electrode-types.<br />
81
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
ited WM mechanical properties. Both Cumulo and Conarc 51 electrodes induced<br />
an early cracking <strong>of</strong> <strong>the</strong> <strong>plate</strong>s at <strong>the</strong> HAZ, limiting <strong>the</strong> deformation capacity<br />
<strong>of</strong> <strong>the</strong> T-stub and did not allow for <strong>the</strong> effective use <strong>of</strong> <strong>the</strong> bolts (Figs.<br />
3.9 and 3.10a-c). Also, <strong>the</strong> scatter <strong>of</strong> <strong>the</strong> responses was not acceptable (Fig.<br />
3.9a). Note that <strong>the</strong> load-carrying behaviour <strong>of</strong> WT1f deviates even more from<br />
<strong>the</strong> remaining tests because, by mistake, <strong>the</strong> actual weld throat thickness was<br />
8.0 mm instead <strong>of</strong> <strong>the</strong> specified value <strong>of</strong> 5.0 mm. The electrode that provided<br />
<strong>the</strong> best <strong>ductility</strong> to <strong>the</strong> overall connection (steel grade S355) is <strong>the</strong> Kardo<br />
(Figs. 3.9-3.10). Therefore, it was <strong>the</strong> most suitable consumable and it was<br />
used in <strong>the</strong> rest <strong>of</strong> <strong>the</strong> specimens to weld <strong>the</strong> <strong>plate</strong>s. This electrode is classified<br />
as an evenmatch electrode as <strong>the</strong> nominal properties <strong>of</strong> <strong>the</strong> WM and <strong>the</strong> BM<br />
are identical.<br />
Finally, regarding <strong>the</strong> welding <strong>of</strong> <strong>the</strong> <strong>plate</strong>s made up <strong>of</strong> S690, <strong>the</strong> electrode<br />
Conarc 70G, specified by <strong>the</strong> distributor as <strong>the</strong> proper electrode type for that<br />
steel quality, guaranteed a performance identical to <strong>the</strong> Kardo for S355.<br />
3.3 EXPERIMENTAL RESULTS<br />
3.3.1 Reference test series WT1<br />
Test series WT1 includes eight specimens that differ in <strong>the</strong> electrode type used<br />
in <strong>the</strong> welding procedure, as explained above. It has been shown previously<br />
that <strong>the</strong> Kardo electrode seemed to be <strong>the</strong> most suitable in terms <strong>of</strong> overall<br />
connection performance (Fig. 3.9). For fur<strong>the</strong>r analysis consider specimens<br />
WT1g/h whose collapse was determined by bolt fracture with some damage <strong>of</strong><br />
<strong>the</strong> <strong>plate</strong> in <strong>the</strong> HAZ in <strong>the</strong> first case, as well (Figs. 3.10d and 3.11).<br />
The load-carrying behaviour <strong>of</strong> <strong>the</strong> above specimens is compared with <strong>the</strong><br />
Eurocode 3 [3.1] predictions for elastic stiffness and plastic resistance in Fig.<br />
3.12 (results in Tables 3.7-3.8). Notice that <strong>the</strong> experimental F-∆ curves depicted<br />
throughout <strong>the</strong> text correspond to <strong>the</strong> third part <strong>of</strong> <strong>the</strong> test – reloading up<br />
to collapse (cf. §3.2.3). Eurocode 3 <strong>of</strong>ten underestimates both properties – see<br />
also Table 3.8. The experimental global elastic stiffness is computed by means<br />
<strong>of</strong> a regression analysis <strong>of</strong> <strong>the</strong> unloading portion <strong>of</strong> <strong>the</strong> F-∆ curve (which is not<br />
traced in <strong>the</strong> graphs). By comparing <strong>the</strong> results, <strong>the</strong>re is a ratio between <strong>the</strong> experiments<br />
and <strong>the</strong> code predictions <strong>of</strong> 1.58 and 1.48 for WT1g and WT1h, respectively.<br />
In addition, if <strong>the</strong> lower bound <strong>of</strong> <strong>the</strong> knee-range <strong>of</strong> <strong>the</strong> F-∆ curve is<br />
compared with FRd.0 predicted by Eurocode 3, deviations <strong>of</strong> 0.84 (WT1g) and<br />
0.81 (WT1h) are observed.<br />
The remaining characteristics <strong>of</strong> <strong>the</strong> F-∆ response (post-limit stiffness and<br />
deformation capacity) cannot be compared with any code provisions since it<br />
does not cover <strong>the</strong> post-limit behaviour. Table 3.8 sets out <strong>the</strong> values <strong>of</strong> maximum<br />
load, Fmax, post-limit stiffness (also determined by means <strong>of</strong> a regression<br />
analysis <strong>of</strong> <strong>the</strong> post-limit response) and deformation capacity, taken as <strong>the</strong> de-<br />
82
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
formation level corresponding to Fmax.<br />
Table 3.8 summarizes <strong>the</strong> results for <strong>the</strong> eight tests, corroborating <strong>the</strong><br />
above-mentioned scatter <strong>of</strong> results between <strong>the</strong> six initial specimens. Figs.<br />
3.13-3.14 show o<strong>the</strong>r results also obtained in this test series. The results <strong>of</strong> <strong>the</strong><br />
(a) Detail <strong>of</strong> WT1h.<br />
(b) WT1g: front view. (c) WT1g: top view.<br />
Fig. 3.11 Specimens WT1g/h after failure.<br />
Table 3.7 Eurocode 3 predictions <strong>of</strong> (global) initial stiffness and plastic resistance<br />
(evaluated using <strong>the</strong> average real dimensions <strong>of</strong> <strong>the</strong> specimens).<br />
Test ID ke.0<br />
(kN/mm)<br />
FRd.0<br />
(kN)<br />
Test ID ke.0<br />
(kN/mm)<br />
FRd.0<br />
(kN)<br />
WT1 217.28 96.66 WT53D 194.16 190.66<br />
WT2A 175.78 88.88 WT53E 189.84 187.23<br />
WT2B 254.24 102.18 WT57_M12 151.69 107.49<br />
WT4A 343.86 163.47 WT57_M16 163.02 159.18<br />
WT7_M12 168.64 81.00 WT57_M20 166.89 158.41<br />
WT7_M16 179.58 80.22 WT61 380.92 153.19<br />
WT7_M20 186.44 80.73 WT64A 388.02 172.85<br />
WT51 184.16 178.90 WT64C 425.46 182.45<br />
WT53C 190.10 187.35<br />
83
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
bolt elongation behaviour for specimen WT1h are given in Fig. 3.13. The<br />
graph does not apply up to collapse since <strong>the</strong> bolt deformation was measured<br />
by means <strong>of</strong> <strong>the</strong> horseshoe device that was removed before <strong>the</strong> maximum load<br />
was reached. The graph shows that <strong>the</strong> results are alike for <strong>the</strong> four bolts and,<br />
consequently, <strong>the</strong> four curves are nearly indistinguishable. Specimen WT1h<br />
was also instrumented with strain gauges (see Fig. 3.8a). These were attached<br />
close to <strong>the</strong> fillet weld, near <strong>the</strong> <strong>the</strong>oretical location <strong>of</strong> <strong>the</strong> expected yield line.<br />
Table 3.8 Main characteristics <strong>of</strong> <strong>the</strong> force-deformation curves for <strong>the</strong> unstiffened<br />
specimens [The elastic stiffness is quantified by using <strong>the</strong><br />
average deformation values. The deformation capacity here is<br />
taken as <strong>the</strong> average deformation, from <strong>the</strong> two opposite LVDTs,<br />
corresponding to <strong>the</strong> maximum load. Italic values for def. capacity<br />
refer to <strong>the</strong> readings <strong>of</strong> HP1].<br />
Test ID Resistance (kN)<br />
K-R Fmax ke.0<br />
Stiffness (kN/mm)<br />
kp-l.0 ke.0/ kp-l.0<br />
∆u.0<br />
(mm)<br />
WT1a 125-140 157.65 96.28 5.97 16.13 6.24<br />
WT1b 140-155 182.08 109.88 4.63 23.73 10.37<br />
WT1c 135-145 166.37 128.63 4.89 26.30 8.12<br />
WT1d 137-145 150.08 120.42 7.91 15.22 4.46<br />
WT1e 140-150 168.12 134.25 6.80 19.74 4.97<br />
WT1f 168-180 184.99 118.46 2.37 50.00 4.90<br />
WT1g 115-135 182.66 137.16 4.22 32.50 14.10<br />
WT1h 119-139 184.99 147.17 4.14 35.55 14.55<br />
WT2Aa 103-124 162.01 128.63 6.47 19.88 ⎯<br />
WT2Ab 106-130 173.64 123.65 3.80 32.54 17.98<br />
WT2Ba 118-156 191.97 127.15 6.47 19.65 10.09<br />
WT2Bb 123-160 195.75 159.49 4.43 36.00 13.09<br />
WT4Aa 118-209 216.40 150.15 5.50 27.30 5.35<br />
WT4Ab 140-196 206.51 173.91 8.74 19.90 4.33<br />
WT7_M12 60-96 100.64 91.18 3.78 24.12 4.60<br />
WT7_M16 80-104 132.34 116.09 5.08 22.85 11.47<br />
WT7_M20 88-118 145.72 137.70 5.61 24.55 9.12<br />
WT51a 155-188 193.71 119.24 3.47 34.36 4.10<br />
WT51b 158-189 194.59 123.67 3.98 31.07 3.82<br />
WT53C 166-192 197.79 128.46 4.75 27.04 4.24<br />
WT53D 185-218 234.72 105.79 9.52 11.11 5.54<br />
WT53E 178-215 230.07 129.63 8.25 15.71 5.26<br />
WT57_M12 75-119 121.87 85.78 1.14 75.25 4.33<br />
WT57_M16 104-165 173.64 110.43 6.99 15.80 5.88<br />
WT57_M20 126-204 241.71 150.96 6.32 23.89 15.98<br />
WT61a 128-180 203.89 164.65 9.75 16.89 6.18<br />
WT61b 119-177 213.20 152.05 11.09 13.71 7.96<br />
WT64A 121-200 220.47 164.04 9.39 17.47 4.60<br />
WT64C 118-214 236.47 172.45 8.84 19.51 4.59<br />
84
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
EC3: Initial<br />
stiffness<br />
30<br />
0<br />
WT1g WT1h<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
EC3: Plastic resistance<br />
Fig. 3.12 Experimental load-carrying behaviour <strong>of</strong> specimens WT1g/h and<br />
comparison with Eurocode 3 (EC3) predictions.<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
Bolt RB Bolt LB<br />
30<br />
0<br />
Bolt LF Bolt RF<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50<br />
Bolt elongation (mm)<br />
Fig. 3.13 Experimental results for <strong>the</strong> bolt elongation behaviour (WT1h).<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
SG1 SG3 SG5<br />
0 4000 8000 12000 16000 20000 24000 28000 32000<br />
Strain (µm/m)<br />
(a) Strain gauges SG1, SG3 and SG5.<br />
Fig. 3.14 Experimental results for <strong>the</strong> flange strain behaviour (WT1h).<br />
85
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
86<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
(b) Strain gauges SG4 and SG6.<br />
Total applied load (kN)<br />
30<br />
0<br />
SG4 SG6<br />
0 4000 8000 12000 16000 20000 24000 28000 32000<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
(c) Strain gauges SG2 and SG7.<br />
Total applied load (kN)<br />
Strain (µm/m)<br />
30<br />
0<br />
SG2 SG7<br />
0 4000 8000 12000 16000 20000 24000 28000 32000<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
Strain (µm/m)<br />
30<br />
0<br />
SG2 SG3 SG4<br />
0 4000 8000 12000 16000 20000 24000 28000 32000<br />
Strain (µm/m)<br />
(d) Strain gauges SG2, SG3 and SG4.<br />
Fig. 3.14 Experimental results for <strong>the</strong> flange strain behaviour (WT1h) (cont.).
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
The force-strain results are shown in Fig. 3.14. In particular, Fig.3.14a compares<br />
<strong>the</strong> results for <strong>the</strong> edge strain gauges (SG1, SG5) and <strong>the</strong> one attached at<br />
mid length <strong>of</strong> <strong>the</strong> specimen (SG3). The results are very similar. Figs. 3.14b-c<br />
depict symmetry. The results are analogous but not exactly <strong>the</strong> same for symmetric<br />
strain gauges, since <strong>the</strong>y may not be placed exactly on <strong>the</strong> same spot.<br />
Finally, Fig. 3.14d compares <strong>the</strong> strain results for <strong>the</strong> remaining strain gauges<br />
that are also alike.<br />
3.3.2 Failure modes and general characteristics <strong>of</strong> <strong>the</strong> overall behaviour<br />
<strong>of</strong> <strong>the</strong> test specimens<br />
The deformation capacity <strong>of</strong> a <strong>bolted</strong> T-stub connection made up <strong>of</strong> welded<br />
<strong>plate</strong>s primarily dep<strong>end</strong>s on <strong>the</strong> <strong>plate</strong>/bolt strength ratio and <strong>the</strong> weld resistance<br />
that is associated to <strong>the</strong> consumable type and properties. Collapse is eventually<br />
governed by brittle fracture <strong>of</strong> <strong>the</strong> bolts or <strong>the</strong> welds, or cracking <strong>of</strong> <strong>the</strong> <strong>plate</strong><br />
material near <strong>the</strong> weld toe. Most <strong>of</strong> <strong>the</strong> tested specimens failed by tension rupture<br />
<strong>of</strong> <strong>the</strong> bolts after b<strong>end</strong>ing deformation <strong>of</strong> <strong>the</strong> flange. The degree <strong>of</strong> plastic<br />
deformation <strong>of</strong> <strong>the</strong> flange dep<strong>end</strong>s first and foremost on <strong>the</strong> geometric characteristics<br />
<strong>of</strong> <strong>the</strong> connection and <strong>the</strong> mechanical properties <strong>of</strong> <strong>the</strong> elements. However,<br />
<strong>the</strong> collapse <strong>of</strong> some specimens was due to cracking <strong>of</strong> <strong>the</strong> <strong>plate</strong> material<br />
in <strong>the</strong> HAZ.<br />
In this T-stub assembly type, <strong>the</strong> collapse mode involving rupture <strong>of</strong> <strong>the</strong><br />
<strong>plate</strong> was also affected by residual stresses and modified microstructure in <strong>the</strong><br />
HAZ. This could lead to a reduction <strong>of</strong> <strong>the</strong> ultimate material strain with respect<br />
to <strong>the</strong> unaffected material and thus to an earlier failure <strong>of</strong> <strong>the</strong> whole connection.<br />
It was also observed that <strong>the</strong> extent <strong>of</strong> <strong>the</strong> properties variations in <strong>the</strong> HAZ,<br />
which were inherent to <strong>the</strong> welding procedure, was highly dep<strong>end</strong>ent on <strong>the</strong><br />
electrode type and <strong>the</strong> hydrogen content, in particular.<br />
The observed failure modes involved combined b<strong>end</strong>ing and tension bolt<br />
fracture (type-13 or -23) in nineteen specimens, stripping <strong>of</strong> <strong>the</strong> nut threads<br />
bolt fracture (type-23B) in one specimen (WT57_M16), cracking <strong>of</strong> <strong>the</strong> <strong>plate</strong><br />
material in <strong>the</strong> HAZ (type-11) in ten specimens and combined collapse modes<br />
11 and 13 (type-1(1+3)) in <strong>the</strong> remaining cases. Notice that <strong>the</strong> stiffened specimens<br />
failed in a combined b<strong>end</strong>ing and tension bolt fracture mode. Table 3.9<br />
summarizes <strong>the</strong> collapse modes <strong>of</strong> <strong>the</strong> several tests.<br />
Dep<strong>end</strong>ing on <strong>the</strong> failure mode and naturally on <strong>the</strong> connection configuration,<br />
a similar behaviour was observed between related specimens. The most<br />
significant characteristic describing <strong>the</strong> overall behaviour <strong>of</strong> <strong>the</strong> connection is<br />
<strong>the</strong> F-∆ response. Fig. 3.15 plots <strong>the</strong> load-carrying behaviour <strong>of</strong> six selected<br />
examples that illustrate <strong>the</strong> five above-mentioned collapse modes. For <strong>the</strong> parallel<br />
T-stub elements specimens, <strong>the</strong> deformation corresponds to <strong>the</strong> average<br />
value measured by <strong>the</strong> two opposite LVDTs at each specimen. For specimen<br />
WT64B that includes a stiffener and where <strong>the</strong> two T-stubs are orientated at<br />
right angles, <strong>the</strong> results for LVDTs HP1 and HP2 (see Fig. 3.8c) are shown.<br />
87
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 3.9 Observed failure modes.<br />
Test ID Failure Test ID Failure Test ID Failure<br />
mode<br />
mode<br />
mode<br />
Type # Type # Type #<br />
13 1 WT4A 13 2 WT64B “23” 2<br />
WT1 11 6 WT4B “13” 1 WT64C 23 1<br />
1(1+3) 1 WT51 23 2 WT7_M12 13 1<br />
WT2A<br />
13<br />
1(1+3)<br />
1<br />
1<br />
WT53C<br />
WT53D<br />
23<br />
13<br />
1<br />
1<br />
WT7_M16<br />
WT7_M20<br />
11<br />
11<br />
1<br />
1<br />
WT2B<br />
13<br />
1(1+3)<br />
1<br />
1<br />
WT53E<br />
WT61<br />
13<br />
23<br />
1<br />
2<br />
WT57_M12<br />
WT57_M16<br />
23<br />
23B<br />
1<br />
1<br />
WT64A 23 1 WT57_M20 1(1+3) 1<br />
88<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
WT64Bb (HP1)<br />
WT64Bb (HP2)<br />
WT61b<br />
WT57_M16<br />
WT2Ba<br />
0<br />
-2 0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
WT1g<br />
WT7_M16<br />
Fig. 3.15 Force-deformation characteristics <strong>of</strong> some <strong>of</strong> <strong>the</strong> tested specimens.<br />
Figs. 3.11b-c and 3.16 depict <strong>the</strong> six above specimens at collapse conditions<br />
[WT1g: collapse type-1(1+3)].<br />
First, consider specimens WT1g, WT7_M16, WT2Ba and WT61b, which<br />
exhibit failure modes type-1(1+3), type-11, type-13 and type-23, respectively<br />
(Figs. 3.11b-c and 3.16a-c). An elastic branch, with slope ke.0, that develops until<br />
yielding <strong>of</strong> <strong>the</strong> flange begins, characterizes <strong>the</strong> F-∆ curves. A loss <strong>of</strong> stiffness<br />
<strong>the</strong>n follows on and at a certain load level a quasi-linear branch with slope<br />
kp-l.0 arises. This post-limit region is longer for specimens WT1g and<br />
WT7_M16 that develop large b<strong>end</strong>ing deformations <strong>of</strong> <strong>the</strong> flange, when compared<br />
to tests WT2Ba and WT61b. For <strong>the</strong>se latter specimens, fracture <strong>of</strong> <strong>the</strong><br />
bolts determined collapse. In <strong>the</strong> specific case <strong>of</strong> WT61b, which was stiffened<br />
on one side, <strong>the</strong> bolts at <strong>the</strong> stiffened side fractured. Therefore, at failure, <strong>the</strong>re<br />
was a sudden drop <strong>of</strong> load with constant deformation, which characterizes a<br />
brittle failure type. Regarding specimen WT7_M16, <strong>the</strong> failure mechanism was
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
(a) Specimen WT7_M16 (collapse type-<br />
11).<br />
(b) Specimen WT2Ba (collapse<br />
type-13).<br />
(c) Specimen WT61b (collapse type-23). (d) Specimen WT64Bb (collapse<br />
type-“23”).<br />
(e) Specimen WT57_M16 (collapse type-23B).<br />
Fig. 3.16 Specimens at failure.<br />
very ductile and after <strong>the</strong> maximum load was reached, at a deformation <strong>of</strong><br />
about 12 mm, <strong>the</strong> drop <strong>of</strong> load was very smooth and proceeded with increasing<br />
deformation between <strong>the</strong> flanges. This test was stopped at ∆ ≈ 16 mm because<br />
<strong>the</strong> webs started to b<strong>end</strong> and twist excessively and that would damage <strong>the</strong><br />
equipment. If <strong>the</strong> test had continued, <strong>the</strong> behavioural t<strong>end</strong>ency would have<br />
been <strong>the</strong> same. Finally, with respect to specimen WT1g that exhibits a combined<br />
failure mechanism, <strong>the</strong> maximum load was reached for a deformation <strong>of</strong><br />
14 mm, after which it started decreasing. This decrease was smooth and corresponded<br />
to <strong>the</strong> beginning <strong>of</strong> cracking <strong>of</strong> <strong>the</strong> flange <strong>plate</strong> close to <strong>the</strong> weld toe.<br />
89
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Eventually, at ∆≈20 mm, <strong>the</strong>re was a sudden drop <strong>of</strong> load that coincided with<br />
<strong>the</strong> bolt fracture. Notice that bolt rupture took place at opposite side <strong>of</strong> <strong>plate</strong><br />
cracking. Apparently, a larger deformation capacity would be expected for<br />
specimen WT7_M16, when compared to WT1g, because <strong>the</strong> bolt did not govern<br />
<strong>the</strong> collapse. However, since <strong>the</strong> T-stub width tributary to a bolt row was<br />
higher in specimen WT7_M16, smaller deformation capacity was expected<br />
[3.9].<br />
Specimen WT64Bb tried to mimic <strong>the</strong> actual configuration <strong>of</strong> <strong>the</strong> tension<br />
side <strong>of</strong> a <strong>bolted</strong> connection: elements orientated at right angles, one T-element<br />
stiffened and <strong>the</strong> o<strong>the</strong>r unstiffened. When this assembly was subjected to a tensile<br />
force, <strong>the</strong> <strong>plate</strong>s became in contact except at <strong>the</strong> stiffener-web contact, as<br />
clearly shown in Fig. 3.16d. Therefore, <strong>the</strong> F-∆ response depicted in Fig. 3.15<br />
shows that <strong>the</strong> two flanges are opening at <strong>the</strong> stiffener side (HP2) and closing<br />
at <strong>the</strong> opposite side (HP1). The characteristics <strong>of</strong> <strong>the</strong> curve for LVDT HP2 are<br />
very similar to those described for WT61b, where bolt fracture at <strong>the</strong> stiffener<br />
side also governs <strong>the</strong> ultimate condition.<br />
Finally, type-23B failure that occurs in specimen WT57_M16 (and is not<br />
common) is a brittle rupture mode. The specimen at collapse is illustrated in<br />
Fig. 3.16e and <strong>the</strong> corresponding load-carrying behaviour is shown in <strong>the</strong> graph<br />
from Fig. 3.15.<br />
3.4 CONCLUDING REMARKS<br />
The experiments presented above can be regarded as a reliable database for <strong>the</strong><br />
<strong>characterization</strong> <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> <strong>the</strong> T-stub assembly made up <strong>of</strong> welded<br />
<strong>plate</strong>s. The test procedure and <strong>the</strong> instrumentation set up adopted for <strong>the</strong> programme<br />
were satisfactory, as evidenced by <strong>the</strong> identical results obtained from<br />
<strong>the</strong> various sets <strong>of</strong> tests from one series (Figs. 3.17-3.18, for illustration). Detailed<br />
information on <strong>the</strong> experimental results is given in Table 3.8, which set<br />
out <strong>the</strong> main characteristics <strong>of</strong> <strong>the</strong> load-carrying behaviour <strong>of</strong> <strong>the</strong> various specimens,<br />
and later in Chapter 5. Reference [3.10] also provides a thorough description<br />
<strong>of</strong> this experimental programme.<br />
The programme provides insight into <strong>the</strong> assessment <strong>of</strong> failure modes and<br />
available deformation capacity <strong>of</strong> <strong>bolted</strong> T-stub connections. The major contributions<br />
<strong>of</strong> <strong>the</strong> overall T-stub deformation are <strong>the</strong> flange deformation and <strong>the</strong><br />
tension bolt elongation. Usually, a higher deformation capacity <strong>of</strong> <strong>the</strong> T-stub is<br />
expected if <strong>the</strong> flange cracking governs <strong>the</strong> collapse instead <strong>of</strong> bolt fracture.<br />
However, in this type <strong>of</strong> assembly this statement is not so straightforward. The<br />
cracking associated to <strong>the</strong> flange mechanism, in this case, dep<strong>end</strong>s on structural<br />
constraint conditions and modifications in <strong>the</strong> mechanical properties in <strong>the</strong><br />
HAZ, particularly those linked to <strong>the</strong> presence <strong>of</strong> residual stresses. Therefore,<br />
<strong>the</strong> selection <strong>of</strong> <strong>the</strong> electrodes and welding procedures is <strong>of</strong> <strong>the</strong> utmost importance<br />
in this connection type to ensure a ductile behaviour. It has been found<br />
out that s<strong>of</strong>t, low hydrogen, mismatch (or evenmatch) electrodes prevent an<br />
90
Total applied load (kN)<br />
Experimental assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
WT2Aa WT2Ab<br />
30<br />
0<br />
WT2Ba WT2Bb<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
Fig. 3.17 (Experimental) load-carrying behaviour for test series WT2.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
Bolt RB<br />
150<br />
Bolt LB<br />
120<br />
90<br />
Bolt LF<br />
60<br />
Bolt RF<br />
30<br />
0<br />
Removal <strong>of</strong> measuring<br />
brackets<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Bolt elongation (mm)<br />
Fig. 3.18 Experimental results for <strong>the</strong> bolt elongation behaviour (specimen<br />
WT4Ab).<br />
early cracking <strong>of</strong> <strong>the</strong> flange thus enhancing <strong>the</strong> overall deformation capacity.<br />
Regarding <strong>the</strong> definition <strong>of</strong> “deformation capacity”, some clarification<br />
seems appropriate: “Which criterion should be considered to define <strong>the</strong> deformation<br />
capacity?”. This question has been addressed previously by <strong>the</strong> author<br />
[3.11] since <strong>the</strong> designation adopted so far (deformation capacity taken as <strong>the</strong><br />
deformation level at maximum load) seems very conservative (e.g.: WT1g,<br />
WT7_M16, among o<strong>the</strong>rs). In many examples, <strong>the</strong>re is a long <strong>plate</strong>au in <strong>the</strong> F-<br />
∆ response after <strong>the</strong> maximum load level is reached that cannot be disregarded.<br />
Then, some guidelines on this specific issue are desirable.<br />
91
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
3.5 REFERENCES<br />
[3.1] European Committee for Standardization (CEN). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[3.2] European Committee for Standardization (CEN). prEN 10025:2000E:<br />
Hot rolled products <strong>of</strong> structural steels, September 2000, Brussels, 2000.<br />
[3.3] European Committee for Standardization (CEN). EN 10204:1995E: Metallic<br />
products, October 1995, Brussels, 1995.<br />
[3.4] RILEM draft recomm<strong>end</strong>ation. Tension testing <strong>of</strong> metallic structural<br />
materials for determining stress-strain relations under monotonic and<br />
uniaxial tensile loading. Materials and Structures; 23:35-46, 1990.<br />
[3.5] Gioncu V, Mazzolani FM. Ductility <strong>of</strong> seismic resistant steel structures.<br />
Spon Press, London, UK, 2002.<br />
[3.6] Rodrigues DM, Menezes LF, Loureiro A, Fernandes JV. Numerical<br />
study <strong>of</strong> <strong>the</strong> plastic behaviour in tension <strong>of</strong> welds in high strength steels.<br />
International Journal <strong>of</strong> Plasticity; 20:1-18, 2004.<br />
[3.7] Davies AC. The science and practice <strong>of</strong> welding – welding science and<br />
technology – Vol. I. Cambridge University Press, Cambridge, UK,<br />
1992.<br />
[3.8] European Committee for Standardization (CEN). EN 499:1994E: Welding<br />
consumables – Covered electrodes for manual metal arc welding <strong>of</strong><br />
non alloy and fine grain steels - Classification, December 1994, Brussels,<br />
1994.<br />
[3.9] Girão Coelho AM, Simões da Silva L. Numerical evaluation <strong>of</strong> <strong>the</strong> <strong>ductility</strong><br />
<strong>of</strong> a <strong>bolted</strong> T-stub connection. In: Proceedings <strong>of</strong> <strong>the</strong> third international<br />
conference on advances in steel structures (ICASS’02) (Eds.: S.L.<br />
Chan, F.G. Teng and K.F.Chung), Hong Kong, China, 277-284, 2002.<br />
[3.10] Girão Coelho AM, Bijlaard F, Gresnigt N, Simões da Silva L. Experimental<br />
assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> <strong>bolted</strong> T-stub connections made<br />
up <strong>of</strong> welded <strong>plate</strong>s. Journal <strong>of</strong> Constructional Steel Research; 60:269-<br />
311, 2004.<br />
[3.11] Girão Coelho AM, Bijlaard F, Simões da Silva L. Experimental research<br />
work on T-stub connections made up <strong>of</strong> welded <strong>plate</strong>s. Document<br />
ECCS-TWG 10.2-217, European Convention for Constructional<br />
Steelwork – Technical Committee 10, Structural Connections (ECCS-<br />
TC10), 2002.<br />
92
4 NUMERICAL ASSESSMENT OF THE BEHAVIOUR OF T-STUB<br />
CONNECTIONS<br />
4.1 INTRODUCTION<br />
The behaviour <strong>of</strong> a <strong>bolted</strong> T-stub connection can be predicted with numerical<br />
simulations. The numerical modelling <strong>of</strong> this type <strong>of</strong> problem is complex since<br />
it requires an adequate representation <strong>of</strong> <strong>the</strong> connection geometry, <strong>the</strong> materials<br />
constitutive laws, boundary and load conditions [4.1]. Today, <strong>the</strong> FE method is<br />
widely accepted as <strong>the</strong> most expedite technique for obtaining numerical solutions<br />
for structural mechanical problems [4.2]. The basic steps <strong>of</strong> this method<br />
are [4.3]: (i) <strong>the</strong> continuum is divided into non-overlapping discrete elements,<br />
over which <strong>the</strong> main variables are interpolated, (ii) <strong>the</strong>se elements are interconnected<br />
at a number <strong>of</strong> points along <strong>the</strong>ir periphery (<strong>the</strong> nodal points), (iii) <strong>the</strong><br />
solution strategy can be obtained using implicit or explicit solvers and (iv) subsidiary<br />
quantities, such as stresses and strains, are evaluated for each element.<br />
Regarding <strong>the</strong> solution technique, <strong>the</strong> implicit method is based on static equilibrium<br />
and is characterized by <strong>the</strong> assembly <strong>of</strong> a global stiffness matrix, followed<br />
by simultaneous solution <strong>of</strong> <strong>the</strong> set <strong>of</strong> linear equations [4.4]. The resulting<br />
system <strong>of</strong> equations is solved for <strong>the</strong> nodal variables and so <strong>the</strong> nodal displacements<br />
are computed directly, i.e. implicitly. The explicit method is based<br />
on dynamic equilibrium.<br />
The FE model allows complex geometry to be modelled fairly accurately.<br />
Material and geometrical nonlinearities are also adequately simulated, as well<br />
as <strong>the</strong> boundary and load conditions. In terms <strong>of</strong> geometry modelling, <strong>the</strong> numerical<br />
model must reproduce <strong>the</strong> global behaviour <strong>of</strong> <strong>the</strong> connection. Such<br />
behaviour is three-dimensional in nature. The choice <strong>of</strong> elements must <strong>the</strong>n be<br />
made among three-dimensional elements: solid or shell elements. Several attempts<br />
<strong>of</strong> a two-dimensional approach were made in <strong>the</strong> past but proved to be<br />
unsatisfactory. Shell elements behave in a three-dimensional fashion and are<br />
able to reproduce <strong>the</strong> collapse mechanisms but are not suitable for element interfacing,<br />
in particular for bolt/<strong>plate</strong> contact simulation. For that purpose, solid<br />
elements are accurate and <strong>the</strong>refore this type <strong>of</strong> elements was used in <strong>the</strong> numeric<br />
simulations. Regarding <strong>the</strong> material properties, for steel components <strong>the</strong><br />
modelling <strong>of</strong> elastoplasticity is fundamental. In elasticity type problems, no<br />
permanent deformations occur. The plastic behaviour is characterized by a<br />
time-indep<strong>end</strong>ent irreversible straining that can only be sustained once a certain<br />
stress level has been reached [4.5]. The elastoplastic material response is<br />
taken into account through dissociation <strong>of</strong> <strong>the</strong> elastic and plastic deformations<br />
(ε e and ε p , respectively). The total strain ε is thus defined as <strong>the</strong> sum<br />
93
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
ε ε ε<br />
<strong>the</strong> yield stress is reached, <strong>the</strong> stress continues to increase with strain but with a<br />
reduced modulus <strong>of</strong> elasticity. The plasticity formulations are based on three<br />
fundamental concepts [4.6]: (i) a yield condition to specify <strong>the</strong> onset <strong>of</strong> plastic<br />
deformation, (ii) a flow rule to define <strong>the</strong> plastic straining and (iii) a hardening<br />
rule to define <strong>the</strong> evolution <strong>of</strong> <strong>the</strong> yield surface with plastic straining. For steel<br />
components <strong>the</strong> yield condition is usually defined under <strong>the</strong> Von Mises yield<br />
criterion. The flow rule defines <strong>the</strong> direction <strong>of</strong> <strong>the</strong> plastic straining. In most<br />
cases, <strong>the</strong> direction <strong>of</strong> <strong>the</strong> plastic strain vector is orthogonal to <strong>the</strong> yield surface<br />
(associated flow).<br />
With respect to <strong>the</strong> element-interfacing phenomenon, in FE analysis <strong>the</strong><br />
element penetration in contact zones is avoided by adding special interface or<br />
contact elements. Generally, it is not possible to define a priori <strong>the</strong> zones that<br />
come into contact because <strong>of</strong> <strong>the</strong> different load stages and corresponding deformations.<br />
This means that contact may not be attained for <strong>the</strong> same element<br />
under different loading conditions. As a result, <strong>the</strong> simulation <strong>of</strong> contact behaviour<br />
between <strong>the</strong> connection components is ra<strong>the</strong>r complex. Contact phenomenon<br />
is intrinsically nonlinear: <strong>the</strong> contact zones are very stiff (compression)<br />
whilst non-contacting zones are very flexible (tension). The interfacing forces<br />
that are developed when two parts come into contact transmit <strong>the</strong> applied<br />
forces. These contact forces are normal to <strong>the</strong> interface direction and <strong>the</strong> frictional<br />
forces are developed along <strong>the</strong> tangential direction <strong>of</strong> <strong>the</strong> interface. The<br />
distribution <strong>of</strong> <strong>the</strong> interface stresses and <strong>the</strong> contact conditions (sticking or sliding)<br />
are also unknown. Most FE packages <strong>of</strong>fer some facilities for dealing with<br />
<strong>the</strong> unilateral contact problem with friction.<br />
The modelling <strong>of</strong> a <strong>bolted</strong> T-stub connection is <strong>the</strong>refore highly nonlinear,<br />
involving complex phenomena such as material plasticity, second-order effects<br />
and unilateral contact boundary conditions. In <strong>the</strong> following sections, <strong>the</strong> procedures<br />
for <strong>the</strong> implementation <strong>of</strong> a FE model using <strong>the</strong> commercial FE package<br />
LUSAS [4.7-4.8] for <strong>the</strong> analysis <strong>of</strong> this type <strong>of</strong> problem are described.<br />
This numerical model is validated through comparison with experimental evidence.<br />
94<br />
e p<br />
= + . In general, plasticity is modelled with strain hardening, i.e. once<br />
4.2 PREVIOUS RESEARCH<br />
The FE modelling <strong>of</strong> an individual T-stub connection has been performed by a<br />
number <strong>of</strong> authors from different research centres. In <strong>the</strong> framework <strong>of</strong> <strong>the</strong><br />
Numerical Simulation Working Group <strong>of</strong> <strong>the</strong> European Research Project<br />
COST C1 “Civil Engineering Structural Connections”, this task was proposed<br />
as a benchmark for FE modelling <strong>of</strong> <strong>bolted</strong> steel connections. Jaspart provided<br />
<strong>the</strong> necessary experimental data for those simulations (T-stub T1) [4.9]. Bursi<br />
[4.10] and Bursi and Jaspart [4.11-4.12] developed and calibrated a threedimensional<br />
nonlinear model to mimic <strong>the</strong> experimental load-carrying response
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
<strong>of</strong> <strong>the</strong> given example (T1). Later, <strong>the</strong>y ext<strong>end</strong>ed <strong>the</strong> method to ano<strong>the</strong>r T-stub<br />
configuration to investigate ano<strong>the</strong>r connection representative <strong>of</strong> a different<br />
collapse mode [4.11-4.12]. The models are proposed as benchmarks in <strong>the</strong><br />
validation process <strong>of</strong> FE s<strong>of</strong>tware packages. The simulations <strong>of</strong> <strong>the</strong> individual<br />
connections were performed by means <strong>of</strong> solids and contact elements. The F-∆<br />
response as well as <strong>the</strong> bolt behaviour (elongation, preloading effects) and prying<br />
effects were addressed. The proposed model was satisfactory, in general.<br />
Gomes et al. [4.13] implemented a three-dimensional model as well for <strong>the</strong><br />
simulation <strong>of</strong> <strong>the</strong> test T1 but used shell elements instead <strong>of</strong> solids. Their model<br />
allowed for <strong>the</strong> assessment <strong>of</strong> second order effects and nonlinear material behaviour<br />
with strain hardening. The agreement between results was ra<strong>the</strong>r poor.<br />
Mistakidis et al. proposed a two-dimensional FE model capable <strong>of</strong> describing<br />
plasticity, large displacements and unilateral contact effects [4.14-4.15]. Although<br />
<strong>the</strong> model encompasses all <strong>the</strong> essential characteristics and dominant<br />
plastification mechanisms, <strong>the</strong> numerical results are much stiffer than <strong>the</strong> actual<br />
response. In general, <strong>the</strong> FE results do not compare well to <strong>the</strong> experiments.<br />
Zajdel also carried out a three-dimensional FE analysis <strong>of</strong> <strong>the</strong> benchmark<br />
problem and proposed a reliable model that accounted for most <strong>of</strong> <strong>the</strong> Tstub<br />
features [4.16].<br />
Wanzek and Gebbeken [4.17] validated a three-dimensional numerical<br />
model against experimental results performed in Munich [4.18]. They used<br />
o<strong>the</strong>r experimental results (e.g. strain results, bolts measurements) for calibration<br />
<strong>of</strong> <strong>the</strong> model. The agreement between responses was very good.<br />
More recently, Swanson [4.19] and Swanson et al. [4.20] performed tests<br />
on individual T-stubs and proposed a robust FE model to supplement <strong>the</strong>ir research.<br />
This sophisticated model provided insight into <strong>the</strong> characteristics <strong>of</strong> <strong>the</strong><br />
T-stub behaviour and stress distributions (namely, contact stresses). The results<br />
<strong>of</strong> this robust model were used to validate a simpler two-dimensional model.<br />
The main criticism to <strong>the</strong>ir approach lies in <strong>the</strong> input <strong>of</strong> <strong>the</strong> material properties.<br />
They used nominal properties instead <strong>of</strong> actual properties. This procedure is<br />
questionable. Naturally, this validation process was only applicable within <strong>the</strong><br />
range analysis, which was limited to a single example. The authors explored<br />
many features <strong>of</strong> <strong>the</strong> T-stub model, as <strong>the</strong> bolt response and <strong>the</strong> prying effect.<br />
They discussed <strong>the</strong> conclusions drawn from <strong>the</strong> FE analyses but <strong>the</strong>y did not<br />
broaden <strong>the</strong> scope <strong>of</strong> <strong>the</strong>ir analysis to conclude about <strong>the</strong> mechanisms and parameters<br />
that influence (and how) <strong>the</strong> T-stub behaviour.<br />
The main concern <strong>of</strong> all above models was <strong>the</strong> accomplishment <strong>of</strong> a reliable<br />
FE model that was calibrated against experiments to obtain <strong>the</strong> F-∆ response.<br />
Fur<strong>the</strong>rmore, only <strong>the</strong> case <strong>of</strong> HR-T-stubs was addressed. These models afforded<br />
some basis for <strong>the</strong> implementation <strong>of</strong> <strong>the</strong> FE models described below.<br />
Several model features have already been highlighted by <strong>the</strong>se authors. However,<br />
some aspects still have to be looked into. Additionally, this research also<br />
proposes a FE model for WP-T-stubs that necessarily includes specific aspects,<br />
namely <strong>the</strong> influence <strong>of</strong> <strong>the</strong> welding <strong>of</strong> <strong>the</strong> <strong>plate</strong>s.<br />
95
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
4.3 DESCRIPTION OF THE MODEL<br />
The T-stub connection was generated with three-dimensional elements, solid<br />
and joint elements. In particular, <strong>the</strong> solid elements were hexahedral bricks and<br />
were used to model <strong>the</strong> continuum. The joint elements were employed in <strong>the</strong><br />
simulation <strong>of</strong> element contact.<br />
In <strong>the</strong> FE library <strong>of</strong> <strong>the</strong> commercial package LUSAS [4.21] <strong>the</strong>re are three<br />
types <strong>of</strong> hexahedral solid elements: <strong>the</strong> eight-node brick (HX8(M)), <strong>the</strong> sixteen-node<br />
brick (HX16) and <strong>the</strong> twenty-node brick (HX20). These elements<br />
belong to a family <strong>of</strong> ser<strong>end</strong>ipity isoparametric elements, i.e. <strong>the</strong>y have no inside<br />
nodes and <strong>the</strong> geometry and displacement interpolation are carried out by<br />
means <strong>of</strong> <strong>the</strong> same shape functions. The elements have three degrees-<strong>of</strong>freedom<br />
per node (u, v and w) and are numerically integrated. The complete<br />
formulation <strong>of</strong> this element type is detailed in <strong>the</strong> literature [4.2-4.3]. The<br />
choice <strong>of</strong> one or ano<strong>the</strong>r type <strong>of</strong> brick dep<strong>end</strong>s on <strong>the</strong>ir application. In linear<br />
elastic problems, <strong>the</strong> higher order elements (sixteen and twenty nodes) are<br />
more accurate than <strong>the</strong> eight-node brick. For nonlinear problems, involving<br />
plasticity and contact phenomenon, in particular, <strong>the</strong> eight-node element, which<br />
has no mid-side nodes, leads to improved numerical solutions since <strong>the</strong>y allow<br />
for a better representation <strong>of</strong> <strong>the</strong> discontinuities at element edges and <strong>of</strong> <strong>the</strong><br />
strain field. The FE code LUSAS implements two eight-node bricks, HX8 e<br />
HX8M, as already pointed out. The element HX8M exhibits improved accuracy<br />
in coarse meshes when compared with <strong>the</strong> parent element HX8, particularly<br />
in b<strong>end</strong>ing dominated problems [4.22]. In addition, <strong>the</strong> element does not<br />
suffer from shear locking in <strong>the</strong> nearly incompressible limit. The element formulation<br />
is based on <strong>the</strong> works <strong>of</strong> Simo and Rifai [4.23]. It includes an assumed<br />
“enhanced” strain field related to <strong>the</strong> internal degrees-<strong>of</strong>-freedom that<br />
are eliminated at <strong>the</strong> element level before assembly <strong>of</strong> <strong>the</strong> structure stiffness<br />
matrix. Thereby, <strong>the</strong> eight-node brick with enhanced strains, HX8M, and full<br />
integration (2×2×2 Gauss points) was chosen for <strong>the</strong> numerical analysis.<br />
The kinematic description <strong>of</strong> solid elements in nonlinear geometrical analysis<br />
is based on three different formulations: (i) <strong>the</strong> total Lagrangian formulation,<br />
that accounts for large displacements and small strains; in this formulation<br />
all variables are referred to <strong>the</strong> undeformed configuration, (ii) <strong>the</strong> updated Lagrangian<br />
formulation that accounts for large displacements and moderately<br />
large strains; all variables are referred to <strong>the</strong> last converged solution configuration<br />
and (iii) <strong>the</strong> Eulerian formulation catering for large displacements and<br />
large strains; in each iteration at <strong>the</strong> same load increment, <strong>the</strong> deformed configuration<br />
is updated and it corresponds to <strong>the</strong> reference configuration for <strong>the</strong><br />
subsequent iteration. In <strong>the</strong> total Lagrangian formulation stresses and strains<br />
are output in terms <strong>of</strong> <strong>the</strong> “second Piola-Kirch<strong>of</strong>f stresses” and “Green-<br />
Lagrange strains”, with reference to <strong>the</strong> undeformed configuration [4.22]. The<br />
stresses and strains output for <strong>the</strong> updated Lagrangian and Eulerian formulations<br />
are <strong>the</strong> “Cauchy (or true) stresses” and <strong>the</strong> “natural (or logarithmic)<br />
strains” [4.22]. For elements with no rotational degrees-<strong>of</strong>-freedom <strong>of</strong> <strong>the</strong><br />
96
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
nodes, in which <strong>the</strong> internal displacement field is defined in terms <strong>of</strong> nodal<br />
displacements only, <strong>the</strong> three types <strong>of</strong> formulations yield identical results for<br />
arbitrarily large displacements, provided that <strong>the</strong> strains are small. In <strong>the</strong> FE<br />
code LUSAS, <strong>the</strong> Lagrangian approach is preferred in structural problems<br />
[4.22]. Consequently, <strong>the</strong> updated Lagrangian nonlinear formulation was employed.<br />
For <strong>the</strong> material nonlinearity, an elastoplastic constitutive law based on <strong>the</strong><br />
Von Mises yield criterion was adopted. A plastic potential defined <strong>the</strong> flow<br />
rule [4.5]. The constitutive model was integrated by means <strong>of</strong> <strong>the</strong> explicit forward<br />
Euler algorithm [4.5]. For this algorithm <strong>the</strong> hardening data and direction<br />
<strong>of</strong> <strong>the</strong> plastic flow are evaluated at <strong>the</strong> point at which <strong>the</strong> elastic stress increment<br />
crosses <strong>the</strong> yield surface.<br />
In order for an element formulation to be applicable to a specific response<br />
prediction, both kinematic and constitutive descriptions must be appropriate<br />
[4.2]. In a materially nonlinear only analysis, <strong>the</strong> configuration and volume <strong>of</strong><br />
<strong>the</strong> body under consideration are constant. In this type <strong>of</strong> analysis, both displacements<br />
and rotations are assumed infinitesimally small. Then, <strong>the</strong> engineering<br />
stress-strain constitutive law describes <strong>the</strong> material behaviour in a<br />
proper way. In a large displacement and strain elastoplastic analysis, <strong>the</strong> configuration<br />
and volume <strong>of</strong> <strong>the</strong> body do not remain constant. The Lagrangian<br />
formulation includes <strong>the</strong> kinematic nonlinear effects due to large displacements<br />
and strains, but whe<strong>the</strong>r <strong>the</strong> large strain behaviour is modelled accurately dep<strong>end</strong>s<br />
on <strong>the</strong> constitutive relations specified. This requires <strong>the</strong> use <strong>of</strong> a true<br />
stress-logarithmic strain measure (σn – εn) for <strong>the</strong> definition <strong>of</strong> <strong>the</strong> uniaxial material<br />
response, instead <strong>of</strong> <strong>the</strong> classic engineering constitutive law (σ – ε). These<br />
quantities are defined with respect to <strong>the</strong> current length and cross-sectional area<br />
<strong>of</strong> <strong>the</strong> coupon and are related to <strong>the</strong> engineering values by means <strong>of</strong> <strong>the</strong> following<br />
relationships:<br />
σ n = σ ( 1+<br />
ε)<br />
and ε n = ln ( 1+<br />
ε )<br />
(4.1)<br />
Node-to-node nonlinear contact friction elements simulated <strong>the</strong> interface<br />
boundary conditions. The contact between two bodies was modelled with a<br />
joint mesh interface, which used a “master” and “slave” connection to tie <strong>the</strong><br />
two surfaces toge<strong>the</strong>r at <strong>the</strong>ir nodes. The sliding and sticking conditions are<br />
reproduced with <strong>the</strong> classic isotropic Coulomb friction law. The selected element<br />
from LUSAS FE package for <strong>the</strong> contact analysis was <strong>the</strong> threedimensional<br />
joint element JNT4 [4.21] that connected two adjacent nodes by<br />
means <strong>of</strong> springs with adequate properties. This element is compatible with <strong>the</strong><br />
brick HX8M, comprising three nodal degrees-<strong>of</strong>-freedom (u, v and w). The<br />
element has four nodes: two active nodes, a third and fourth auxiliary nodes for<br />
definition <strong>of</strong> <strong>the</strong> local xy-plane. The two active nodes are connected with extensional<br />
springs in <strong>the</strong> three local directions x, y and z.<br />
The friction model was able to represent frictional and gap connections<br />
between adjacent nodes whereby on <strong>the</strong> closure <strong>of</strong> a specified initial gap, frictional<br />
forces were allowed to develop. In <strong>the</strong> proposed numerical model, this<br />
97
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
initial gap was set as equal to zero. In order to model <strong>the</strong> nonlinear relation<br />
between stresses F and relative displacements δ, <strong>the</strong> linear stiffness moduli<br />
have to be specified. The local element stiffness matrix is formulated directly<br />
from user input stiffness coefficients and is <strong>the</strong>n transformed to <strong>the</strong> global Cartesian<br />
system. The normal stiffness modulus, k1, should be set as equal to infinity,<br />
i.e. its value should be <strong>the</strong> biggest possible. However, a stiffness value too<br />
large could induce poor conditioning <strong>of</strong> <strong>the</strong> stiffness matrix. The optimum<br />
value was found when <strong>the</strong> change in <strong>the</strong> results for an additional increase in <strong>the</strong><br />
stiffness value was negligible or when <strong>the</strong> penetration between <strong>the</strong> bodies in<br />
contact reached a certain limit [4.11]. Concerning <strong>the</strong> tangential stiffness<br />
moduli, k2 and k3, <strong>the</strong>ir value must be non-zero, o<strong>the</strong>rwise <strong>the</strong> bodies in contact<br />
would have an unrealistic infinite movement in <strong>the</strong>se directions at <strong>the</strong> commencement<br />
<strong>of</strong> loading. The location and magnitude <strong>of</strong> <strong>the</strong> contact forces can be<br />
ascertained by <strong>the</strong> joint elements arrangement, since a zero force means separation<br />
<strong>of</strong> <strong>the</strong> flanges whilst a compressive force implies contact between <strong>the</strong><br />
<strong>plate</strong>s at that location.<br />
The joint element possesses no geometrically nonlinear terms in its formulation.<br />
However, it may be used in geometrically nonlinear analysis but it remains<br />
geometrically linear.<br />
To determine <strong>the</strong> structural response <strong>of</strong> <strong>the</strong> nonlinear problem an implicit<br />
solution strategy was used, which is suitable for problems involving smooth<br />
nonlinear analyses. A load stepping routine was hence used. There was no restriction<br />
on <strong>the</strong> magnitude <strong>of</strong> <strong>the</strong> load step as <strong>the</strong> procedure was unconditionally<br />
stable. The increment size followed from accuracy and convergence criteria.<br />
Within each increment, <strong>the</strong> equilibrium equations were solved by means <strong>of</strong><br />
<strong>the</strong> Newton-Raphson iterations, which is stable and converges quadratically. In<br />
<strong>the</strong> Newton-Raphson method, for each load step, <strong>the</strong> residuals are eliminated<br />
by an iterative scheme. In each iteration, <strong>the</strong> load level remains constant and<br />
<strong>the</strong> structure is analysed with a redefined tangent stiffness matrix. The accuracy<br />
<strong>of</strong> <strong>the</strong> solution is measured by means <strong>of</strong> appropriate convergence criteria. Their<br />
selection is <strong>of</strong> <strong>the</strong> utmost importance: too tight convergence criteria may lead<br />
to an unnecessary number <strong>of</strong> iterations and a consequent waste <strong>of</strong> computer<br />
resources, whilst a loose tolerance may result in incorrect solutions. Generally<br />
speaking, in nonlinear geometrical analysis relatively tight tolerances are required,<br />
while in nonlinear material problems slack tolerances are admitted,<br />
since high local residuals are not easy to eliminate. The FE code LUSAS disposes<br />
<strong>of</strong> six different convergence criteria [4.22]: (i) Euclidean residual norm,<br />
γψ, defined by <strong>the</strong> norm <strong>of</strong> <strong>the</strong> residuals, ψ , as a percentage <strong>of</strong> <strong>the</strong> norm <strong>of</strong> <strong>the</strong><br />
external forces, R : γψψ R<br />
2 2<br />
98<br />
= × 100 , (ii) Euclidean displacement norm,<br />
γd, defined by <strong>the</strong> norm <strong>of</strong> <strong>the</strong> iterative displacements, δ a , as a percentage <strong>of</strong><br />
<strong>the</strong> total displacements, a : 2 2 100<br />
γ d = δa<br />
a × , (iii) Euclidean iterative displacement<br />
norm, γdt, defined by <strong>the</strong> norm <strong>of</strong> <strong>the</strong> iterative displacements, δ a , as
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
a percentage <strong>of</strong> <strong>the</strong> total displacements ∆ a for a certain increment: γ dt =<br />
2 2 100<br />
a a δ = ∆ × , (iv) work norm, γw, corresponding to <strong>the</strong> work done by<br />
<strong>the</strong> residuals forces on <strong>the</strong> current iteration as <strong>the</strong> percentage <strong>of</strong> <strong>the</strong> work done<br />
() i<br />
T<br />
() i<br />
γ w = ⎡ψ ⎤<br />
⎣ ⎦<br />
δa (0)<br />
T<br />
(0)<br />
⎡R ⎤<br />
⎣ ⎦<br />
δa<br />
×<br />
by <strong>the</strong> external forces on iteration zero, ( )<br />
× 100 , (v) root mean square <strong>of</strong> residuals and (vi) maximum absolute residual.<br />
Based on nonlinear numerical analysis from literature [4.24-4.26], it was<br />
concluded that establishing displacement-based convergence criteria was<br />
enough. Never<strong>the</strong>less, Crisfield [4.25] suggests that any displacement constraint<br />
must be coupled with a force limitation. The following convergence<br />
criteria were hence used. LUSAS [4.22] suggests <strong>the</strong> following values as limit<br />
tolerances:<br />
⎧slack<br />
: 5.<br />
0 −1.<br />
0<br />
⎪<br />
(i) Euclidean displacement norm ⎨tight<br />
: 0.<br />
1−<br />
0.<br />
001<br />
⎪<br />
⎩reasonable<br />
: 0.<br />
1−<br />
1.<br />
0<br />
⎧slack<br />
: 5.<br />
0 −1.<br />
0<br />
⎪<br />
(ii) Euclidean incremental norm ⎨tight<br />
: 0.<br />
1−<br />
0.<br />
001<br />
⎪<br />
⎩reasonable<br />
: 0.<br />
1−<br />
1.<br />
0<br />
⎧slack<br />
: 0.<br />
1−<br />
0.<br />
001<br />
⎪<br />
−6<br />
−9<br />
(iii) Work norm ⎨tight<br />
: 10 −10<br />
⎪<br />
−6<br />
⎩reasonable<br />
: 0.<br />
001−<br />
10<br />
For predominantly materially nonlinear problems, where high local residuals<br />
have to be tolerated, slack convergence criteria are usually more effective<br />
[4.22]. As a consequence, <strong>the</strong> following slack tolerance values were used: γd =<br />
3.0, γdt = 3.0 and γw= 0.05.<br />
With respect to <strong>the</strong> incremental method, a load curve was defined. Loads<br />
were applied to <strong>the</strong> specimen in a displacement-control fashion that enforced a<br />
better conditioning <strong>of</strong> <strong>the</strong> tangent stiffness matrix when compared to <strong>the</strong> classical<br />
load-control procedure.<br />
4.4 CALIBRATION OF THE FINITE ELEMENT MODEL<br />
The FE model for both T-stub assembly-types was identical. The only difference<br />
lied in <strong>the</strong> representation <strong>of</strong> <strong>the</strong> flange-to-web connection. For <strong>the</strong> HR-Tstub,<br />
flange and web were connected by means <strong>of</strong> a fillet radius, r, that ensured<br />
<strong>the</strong> continuity between both <strong>plate</strong>s. In <strong>the</strong> case <strong>of</strong> WP-T-stubs, a continuous<br />
45º-fillet weld (throat thickness aw) linked <strong>the</strong> flange and <strong>the</strong> web, though <strong>the</strong><br />
two <strong>plate</strong>s were not necessarily in contact.<br />
99
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
The calibration <strong>of</strong> <strong>the</strong> FE model for <strong>the</strong> HR-T-sub was based on <strong>the</strong> experimental<br />
test programme carried out by Bursi and Jaspart [4.11-4.12]. The<br />
specimen T1, which was obtained from an IPE300 beam pr<strong>of</strong>ile, with snugtightened<br />
bolts was selected for <strong>the</strong> following study. Regarding <strong>the</strong> WP-T-stub,<br />
<strong>the</strong> approach was validated with experimental evidence from <strong>the</strong> series <strong>of</strong> tests<br />
WT1 reported in Chapter 3.<br />
4.4.1 Geometry<br />
The general geometrical characteristics <strong>of</strong> <strong>the</strong> specimens are specified in Table<br />
4.1 for <strong>the</strong> two specimens reported herein. By adopting <strong>the</strong> adequate boundary<br />
conditions only one eighth <strong>of</strong> <strong>the</strong> T-stub was modelled, owing to symmetry<br />
considerations (Fig. 4.1). The xy and yz planes are geometrical planes <strong>of</strong> symmetry.<br />
Although <strong>the</strong> xz plane does not meet such criterion, since <strong>the</strong> bolt elongation<br />
behaviour is not symmetrical along <strong>the</strong> y direction, some authors propose<br />
numerical models that account for a symmetric behaviour <strong>of</strong> an “equivalent<br />
bolt” complying with <strong>the</strong> requirements for symmetry in <strong>the</strong> xz plane<br />
[4.11,4.17]. If <strong>the</strong> “equivalent bolt” is defined in such a way that its geometrical<br />
stiffness is identical to that <strong>of</strong> <strong>the</strong> actual bolt, i.e. <strong>the</strong> elongation <strong>of</strong> <strong>the</strong><br />
“equivalent bolt” represents half <strong>of</strong> <strong>the</strong> elongation behaviour <strong>of</strong> <strong>the</strong> actual bolt,<br />
only one eight <strong>of</strong> <strong>the</strong> T-stub has to be considered. This approach can be very<br />
useful in terms <strong>of</strong> FE analysis, since <strong>the</strong> number <strong>of</strong> elements is significantly<br />
reduced. In this case, <strong>the</strong> xz symmetry plane between <strong>the</strong> two flanges was modelled<br />
by contact elements on a rigid foundation (Fig. 4.1). The interface<br />
boundaries between flanges and washer or bolt head and between web and<br />
flange <strong>plate</strong>s in <strong>the</strong> case <strong>of</strong> WP-T-stubs were also modelled by means <strong>of</strong> contact<br />
elements. In order to reduce <strong>the</strong> number <strong>of</strong> contact planes <strong>the</strong> bolt head or<br />
nut and <strong>the</strong> washer, if any, were assumed fully connected. This simplification<br />
led to slightly stiffer deformation behaviour, but <strong>the</strong> overall response was not<br />
greatly influenced, as already shown in <strong>the</strong> literature [4.12,4.16-4.17].<br />
The bolt modelling in this type <strong>of</strong> connection is very important since <strong>the</strong><br />
overall response <strong>of</strong> <strong>the</strong> T-stub is greatly influenced by <strong>the</strong> bolt behaviour. The<br />
bolt is composed <strong>of</strong> head, nut and shank (threaded and non-threaded part).<br />
Each <strong>of</strong> <strong>the</strong>se components constitutes a source <strong>of</strong> flexibility that must be taken<br />
into account when modelling <strong>the</strong> bolt. Bursi and Jaspart [4.12-4.16] defined <strong>the</strong><br />
above-mentioned “equivalent bolt” by means <strong>of</strong> <strong>the</strong> Aggerskov model [4.27]<br />
Table 4.1 Nominal geometrical properties <strong>of</strong> <strong>the</strong> various specimens (dimensions<br />
in [mm]; ST: short-threaded, FT: full-threaded).<br />
Test ID T-elements geometry Bolt characteristics<br />
tf tw w e p/2 e1 r/aw φ Washer Type #<br />
T1 10.7 7.1 90 30 20 20 15 12 Yes ST 4<br />
WT1 10.0 10.0 90 30 25 20 5 12 No ST 4<br />
100
y<br />
z<br />
x<br />
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
External load<br />
External load<br />
‘Equivalent’ bolt<br />
M12, short-threaded<br />
Plan<br />
10.7 mm<br />
x<br />
y<br />
Rigid foundation<br />
x<br />
d0 = 14.0 mm<br />
30.0 mm 26.45 mm<br />
Section xx<br />
Contact plane<br />
15.0 mm<br />
3.55 mm<br />
¼ external load<br />
Boundary geometric<br />
conditions<br />
x<br />
20.0 mm<br />
20.0 mm<br />
150.0 mm<br />
Fig. 4.1 Finite element geometry model assuming symmetry in <strong>the</strong> xy, xz and<br />
yz planes: particular specimen T1.<br />
and reproduced <strong>the</strong> bolt shank with a cylinder <strong>of</strong> cross-sectional area As (tensile<br />
stress area <strong>of</strong> <strong>the</strong> bolt). In <strong>the</strong> proposed numerical model, a different approach<br />
was implemented. The “equivalent bolt” had half <strong>of</strong> <strong>the</strong> conventional bolt<br />
length, as defined in Eurocode 3 [4.28] and <strong>the</strong> “equivalent shank” has a<br />
threaded part (cross-sectional area As) and a non-threaded portion (actual bolt<br />
diameter). The length <strong>of</strong> <strong>the</strong>se parts was proportional to that <strong>of</strong> <strong>the</strong> real bolt.<br />
4.4.2 Boundary and load conditions<br />
The nodes in <strong>the</strong> symmetry planes xy and yz were fixed with symmetric geo-<br />
101
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
metrical boundary conditions (Fig. 4.1): in plane xy, <strong>the</strong> nodes were fixed in <strong>the</strong><br />
z direction on one side and in plane yz in <strong>the</strong> x direction, along <strong>the</strong> back <strong>of</strong> <strong>the</strong><br />
half <strong>of</strong> <strong>the</strong> web. The nodes in plane xz between <strong>the</strong> two flanges and between <strong>the</strong><br />
washers and <strong>the</strong> flanges were restrained with contact elements. The boundary<br />
conditions for <strong>the</strong> second model were identical except for <strong>the</strong> symmetry plane<br />
xz between <strong>the</strong> two flanges. This plane was modelled by contact elements on a<br />
rigid foundation (Fig. 4.1). The nodes on this rigid base were fully restrained.<br />
Complying with geometrical symmetry, <strong>the</strong> bottom bolt nodes were also fixed<br />
in <strong>the</strong> y direction.<br />
No friction was assumed between <strong>the</strong> flanges interface because <strong>of</strong> <strong>the</strong> Telements<br />
symmetric behaviour. For <strong>the</strong> flange-washer and flange-web (in <strong>the</strong><br />
case <strong>of</strong> welded pr<strong>of</strong>iles) interfaces a non-zero friction coefficient, µ, was assumed.<br />
A value <strong>of</strong> 0.25 for this type <strong>of</strong> contact surface was suggested by Vasarhelyi<br />
and Chiang [4.29], who carried out an experimental study for supplying<br />
reliable values for this parameter. This value was adopted in <strong>the</strong> model.<br />
A uniform total prescribed displacement <strong>of</strong> 0.1 mm was applied at <strong>the</strong> top<br />
<strong>of</strong> <strong>the</strong> upper T-element in positive y direction (Fig. 4.1). In <strong>the</strong> nonlinear analysis,<br />
<strong>the</strong> total load factor was increased from 1.0 to <strong>the</strong> collapse, as explained<br />
below. A final remark concerning <strong>the</strong> nodal restraints must be made: in LU-<br />
SAS FE package, when applying total prescribed displacements in a certain<br />
direction, <strong>the</strong> corresponding nodes must be fixed in <strong>the</strong> same direction.<br />
4.4.3 Mechanical properties <strong>of</strong> steel components<br />
For a good correlation with experimental results, <strong>the</strong> full actual stress-strain<br />
relationship <strong>of</strong> <strong>the</strong> materials must be adopted in <strong>the</strong> numerical simulation. For<br />
both models a rate and temperature indep<strong>end</strong>ent plasticity law with hardening<br />
was used for <strong>the</strong> T-stub pr<strong>of</strong>ile and <strong>the</strong> high strength bolt. The constitutive laws<br />
were reproduced with a piecewise linear model [4.11-4.12,4.30]. As already<br />
pointed out, to perform realistic simulations, <strong>the</strong> conventional constitutive law<br />
had to be converted into a constitutive true law (Fig. 4.2). The material properties<br />
for <strong>the</strong> rigid foundation were also defined. Since it is a rigid element, a<br />
linear elastic material was assumed, with E = 10 15 MPa and υ = 0.45.<br />
4.4.4 Specimen discretization<br />
A FE mesh must be sufficiently refined to produce accurate results but <strong>the</strong><br />
number <strong>of</strong> elements and nodes should be kept as small as possible in order to<br />
limit <strong>the</strong> processing time needed for <strong>the</strong> analysis.<br />
The behaviour <strong>of</strong> a <strong>bolted</strong> T-stub connection is dominated by <strong>the</strong> flexural<br />
deformation <strong>of</strong> <strong>the</strong> flange. Particular attention must <strong>the</strong>n be devoted to <strong>the</strong> discretization<br />
<strong>of</strong> this part. Based on <strong>the</strong> study performed by Wanzek and Gebbeken<br />
[4.17], <strong>the</strong> flange discretization with HX8M elements was analysed with<br />
102
True stress (MPa)<br />
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
1200<br />
1000<br />
800<br />
600<br />
(a) HR-T-stub specimen T1 [4.11].<br />
True stress (MPa)<br />
400<br />
Bolt (fy=893MPa)<br />
200<br />
0<br />
T-flange (fy=431MPa)<br />
T-web (fy=496MPa)<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35<br />
1200<br />
1000<br />
800<br />
600<br />
Logarithmic strain<br />
400<br />
Bolt (fy=803MPa)<br />
200<br />
0<br />
T-flange (fy=340MPa)<br />
T-web (fy=391MPa)<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35<br />
Logarithmic strain<br />
(b) WP-T-stub specimen WT1.<br />
Fig. 4.2 Stress-strain true laws for specimens T1 and WT1.<br />
respect to two parameters: (i) degree <strong>of</strong> discretization in order to represent <strong>the</strong><br />
b<strong>end</strong>ing dominated problem and (ii) number <strong>of</strong> elements through thickness to<br />
check <strong>the</strong> capability <strong>of</strong> representing <strong>the</strong> yielding lines.<br />
The FE mesh depicted in Fig. 4.3a complies with <strong>the</strong> requirements for a<br />
reliable simulation and satisfies <strong>the</strong> mesh convergence study that was performed<br />
within <strong>the</strong> framework <strong>of</strong> this research work (cf. App<strong>end</strong>ix B).<br />
For <strong>the</strong> bolt discretization, in order to simulate <strong>the</strong> complex state <strong>of</strong> stress in<br />
<strong>the</strong> bolt, a reasonably refined mesh was essential. In a T-stub connection <strong>the</strong><br />
bolt works in tension and b<strong>end</strong>ing due to <strong>the</strong> deformation <strong>of</strong> <strong>the</strong> T-stub flanges.<br />
The overall response <strong>of</strong> <strong>the</strong> T-stub is greatly influenced by <strong>the</strong> bolt behaviour.<br />
As a result, <strong>the</strong> bolt modelling is crucial. The bolt is composed <strong>of</strong> head, nut and<br />
shank (threaded and non-threaded part). Each <strong>of</strong> <strong>the</strong>se components constitutes<br />
a source <strong>of</strong> flexibility that has to be taken into account when modelling <strong>the</strong><br />
bolt. The number <strong>of</strong> elements was determined decisively by <strong>the</strong> discretization<br />
103
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(a) Flange discretization. (b) Bolt discretization. (c) Global mesh.<br />
Fig. 4.3 Flange discretization adopted for fur<strong>the</strong>r analysis (e.g. T1).<br />
<strong>of</strong> <strong>the</strong> circumference <strong>of</strong> <strong>the</strong> bolt. Technical literature suggests a minimum <strong>of</strong> 12<br />
to 16 nodes around <strong>the</strong> circular hole [4.1]. The bolt mesh represented in Fig.<br />
4.3b also complied with <strong>the</strong> requests for an accurate modelling. In <strong>the</strong> proposed<br />
numerical model, a different approach was implemented. As already explained,<br />
<strong>the</strong> “equivalent bolt” has half <strong>of</strong> <strong>the</strong> conventional bolt length, Lb, defined in Eq.<br />
(1.18) and <strong>the</strong> “equivalent shank” a threaded part (cross-sectional As) and a<br />
non-threaded part (actual bolt diameter), whose lengths were proportional to<br />
those <strong>of</strong> <strong>the</strong> real bolt.<br />
Fig. 4.3c shows specimen T1 global mesh that comprises 3588 elements<br />
and 5680 nodes. For <strong>the</strong> welded specimen, similar discretization was adopted.<br />
The global mesh in this case included 4164 elements and 6618 nodes.<br />
4.4.5 Contact analysis<br />
App<strong>end</strong>ix B describes <strong>the</strong> models used for <strong>the</strong> calibration <strong>of</strong> <strong>the</strong> joint elements<br />
stiffness coefficients ki in <strong>the</strong> interface behaviour.<br />
4.5 FAILURE CRITERIA<br />
The deformation capacity <strong>of</strong> a T-stub is related to <strong>the</strong> <strong>plate</strong>/bolt resistance ratio<br />
and is eventually determined by bolt fracture or cracking <strong>of</strong> <strong>the</strong> <strong>plate</strong> material,<br />
as already mentioned. In both situations, <strong>the</strong> modelling <strong>of</strong> <strong>the</strong> failure condition<br />
104
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
can be ascertained by assuming that cracking occurs when <strong>the</strong> ultimate strain εu<br />
is attained, ei<strong>the</strong>r at <strong>the</strong> bolt or at <strong>the</strong> T-stub critical sections [4.31-4.32]. Due<br />
to <strong>the</strong> nature <strong>of</strong> <strong>the</strong> materials, <strong>the</strong> bolt deformation supply is substantially less<br />
than <strong>the</strong> <strong>plate</strong>. Whilst for high strength bolts <strong>the</strong> ultimate strain is circa 5%-6%,<br />
for constructional steels, ultimate strains <strong>of</strong> 25%-30%, at least, can be expected<br />
[4.33]. As a result, bolt fracture is likely to govern most ultimate conditions<br />
and its assessment is <strong>of</strong> primary importance.<br />
The potential failure mechanisms <strong>of</strong> a bolt under axial loading are: (i) tension<br />
failure, (ii) stripping <strong>of</strong> <strong>the</strong> bolt threads and (iii) stripping <strong>of</strong> <strong>the</strong> nut<br />
threads. Swanson [4.19] points out that high-strength fasteners are designed so<br />
that tension failure <strong>of</strong> <strong>the</strong> bolt occurs before stripping <strong>of</strong> <strong>the</strong> threads. The stripping<br />
phenomena should not be expected in most cases. Additionally, such a<br />
failure type is not easily opened to a numerical or analytical implementation.<br />
Therefore, <strong>the</strong> ultimate deformation <strong>of</strong> <strong>the</strong> bolt is frequently governed by tension<br />
failure. A comprehensive numerical study on <strong>the</strong> behaviour <strong>of</strong> a single<br />
bolt in tension was hence carried out to evaluate its maximum deformation<br />
capacity and has been recently reported by <strong>the</strong> author [4.34].<br />
Based on <strong>the</strong> study <strong>of</strong> a single bolt in tension, a failure criterion for <strong>the</strong> assessment<br />
<strong>of</strong> <strong>the</strong> T-stub collapse is now proposed. As a component <strong>of</strong> <strong>the</strong> T-stub<br />
connection, <strong>the</strong> bolt is subjected to combined tension and b<strong>end</strong>ing. In this case,<br />
<strong>the</strong> strain distribution at <strong>the</strong> bolt critical section changes from <strong>the</strong> symmetric<br />
case depicted in Fig. 4.4a to <strong>the</strong> case illustrated in Fig. 4.4b. The bolt axis direction<br />
is no longer a principal direction. However, if a similar failure criterion<br />
εmax<br />
εmin<br />
ε11.av = εy.av<br />
Bolt cross-section<br />
Bolt cross-section<br />
(a) Bolt under pure axial tension. (b) Bolt under combined tension and<br />
b<strong>end</strong>ing.<br />
Fig. 4.4 Sketch <strong>of</strong> <strong>the</strong> strain distribution within a bolt cross-section.<br />
εmin<br />
εmax<br />
105
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
is adopted to <strong>the</strong> single bolt in tension respecting to <strong>the</strong> maximum average<br />
principal strain, i.e. ε11. av = εu.<br />
b , <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> bolt in com-<br />
bined tension and b<strong>end</strong>ing can be determined. It has been concluded that since<br />
<strong>the</strong> bolt, as a T-stub element, is subjected to combined tension and b<strong>end</strong>ing<br />
deformations, failure should be assessed by comparison <strong>of</strong> <strong>the</strong> maximum average<br />
principal strain, ε11.av.b with εu.b.<br />
Should <strong>the</strong> flange section be critical, a similar criterion based on <strong>the</strong> maximum<br />
principal strain, i.e. ε11.av.f = εu.f, seems appropriate.<br />
4.6 NUMERICAL RESULTS FOR HR T-STUB T1<br />
The most significant characteristic describing <strong>the</strong> overall behaviour <strong>of</strong> <strong>the</strong> model<br />
is <strong>the</strong> F-∆ curve. The implementation <strong>of</strong> <strong>the</strong> above FE model yields <strong>the</strong><br />
results shown in Fig. 4.5. The numerical results are compliant with <strong>the</strong> experimental<br />
response [4.12] showing that <strong>the</strong> proposed model is ra<strong>the</strong>r accurate. The<br />
<strong>end</strong> <strong>of</strong> <strong>the</strong> numerical curve, i.e. <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> connection is<br />
established by application <strong>of</strong> <strong>the</strong> above failure criteria.<br />
For <strong>the</strong> T-stub specimen T1, experimental observations indicated that <strong>the</strong><br />
collapse is due to inelastic phenomena in <strong>the</strong> bolts and significant flange yielding<br />
[4.11]. Under <strong>the</strong> above failure criterion, <strong>the</strong> ultimate conditions are governed<br />
by bolt fracture. The maximum average bolt strain ε11.av.b equals εu.b for a<br />
global deformation <strong>of</strong> 9.20 mm. This value is very close to <strong>the</strong> experiments<br />
(9.49 mm; ratio = 0.97).<br />
Fig. 4.5 compares <strong>the</strong> actual T-stub behaviour with <strong>the</strong> FE model. The<br />
curves in this case include <strong>the</strong> web deformation. However, <strong>the</strong> real gap between<br />
<strong>the</strong> two flanges does not account for <strong>the</strong> web deformation. This response<br />
is depicted in Fig. 4.6a. The “real” flange deformation, ∆ is smaller than <strong>the</strong><br />
total deformation due to <strong>the</strong> contribution <strong>of</strong> <strong>the</strong> web. Therefore, <strong>the</strong> “real” F-∆<br />
106<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Experimental results<br />
Numerical results LUSAS<br />
0<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Total deformation (mm)<br />
Fig. 4.5 Global response <strong>of</strong> specimen T1: numerical and experimental results.
Total applied load (kN)<br />
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
210<br />
180<br />
150<br />
120<br />
(a) Load-deformation behaviour.<br />
Ratio<br />
90<br />
Num. - Total def.<br />
60<br />
Num. - Real def.<br />
Def. capacity (bolt)<br />
30<br />
0<br />
Bolt elongation<br />
0 1 2 3 4 5 6 7 8 9 10<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
Deformation (mm)<br />
0.2<br />
Prying force, Q Bolt force, B<br />
0.0<br />
0 15 30 45 60 75 90 105<br />
Applied load F, per bolt row (kN)<br />
(b) Bolt and prying force.<br />
Fig. 4.6 Numerical results for specimen T1.<br />
curve is stiffer, showing a deviation <strong>of</strong> nearly 25% in <strong>the</strong> elastic regime and 8%<br />
in <strong>the</strong> post-limit range. Fig. 4.6a also plots <strong>the</strong> bolt elongation behaviour<br />
against <strong>the</strong> applied load until collapse. The evolution <strong>of</strong> <strong>the</strong> ratios <strong>of</strong> prying and<br />
bolt forces with <strong>the</strong> applied load per bolt row, F, Q/F and B/F, respectively, is<br />
illustrated in Fig. 4.6b showing an increase <strong>of</strong> such ratios with plastic straining<br />
in <strong>the</strong> flange. The yielding <strong>of</strong> <strong>the</strong> flange starts at a load level <strong>of</strong> 96.29 kN (Fig.<br />
4.7a). The ratio Q/F for this load level is 0.22; at collapse (2F = 207.98 kN) it<br />
increases to 0.34, which means an enlargement <strong>of</strong> 1.5 times. This information<br />
on <strong>the</strong> contact pressures provided by <strong>the</strong> FE model is very useful and cannot be<br />
obtained from experiments. Fur<strong>the</strong>rmore, <strong>the</strong> model gives detailed results for<br />
<strong>the</strong> bolt behaviour, particularly in terms <strong>of</strong> bolt elongation behaviour (curve Bδb)<br />
– Fig. 4.8.<br />
The evolution <strong>of</strong> <strong>the</strong> flange yielding is represented in Fig. 4.7. The beam<br />
pattern governs <strong>the</strong> kinematic mechanism: two yield lines develop in <strong>the</strong><br />
flange, one near <strong>the</strong> bolt hole and ano<strong>the</strong>r close to <strong>the</strong> flange-to-web connections<br />
(see also App<strong>end</strong>ix C). This means that <strong>the</strong> flange is in double curvature,<br />
107
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(a) 2F=96.29kN;<br />
∆=0.61mm.<br />
(d) 2F=146.45kN;<br />
∆=1.10mm.<br />
(g) 2F=179.08kN;<br />
∆=3.04mm.<br />
108<br />
(b) 2F=117.44kN;<br />
∆=0.76mm.<br />
(e) 2F=159.55 kN;<br />
∆=1.43mm.<br />
(h) 2F=190.85kN;<br />
∆=4.82mm.<br />
Fig. 4.7 Flange yielding evolution with <strong>the</strong> applied load.<br />
(c) 2F=134.20kN;<br />
∆=0.93 mm.<br />
(f) 2F=166.25kN;<br />
∆=1.69mm.<br />
(i) 2F=207.98kN;<br />
∆=8.70mm.<br />
as Fig. 4.7c clearly shows.<br />
The location <strong>of</strong> <strong>the</strong> prying forces changes during <strong>the</strong> course <strong>of</strong> loading. Fig.<br />
4.9 shows <strong>the</strong> evolution <strong>of</strong> <strong>the</strong> contact area with <strong>the</strong> applied load. Clearly, as<br />
<strong>the</strong> load increases, <strong>the</strong> contact area spreads to <strong>the</strong> bolt axis. Let n be <strong>the</strong> distance<br />
between <strong>the</strong> prying forces and <strong>the</strong> bolt axis. The ratio n/e is plotted<br />
against <strong>the</strong> external load in Fig. 4.10. Two cases are taken into account: (i) <strong>the</strong><br />
overall contact area and (ii) <strong>the</strong> flange cross-section at <strong>the</strong> horizontal bolt axis<br />
x. In both cases, n is computed as follows:<br />
∑ FQi Linfluence. i × number <strong>of</strong> active joints×<br />
xi<br />
i<br />
n = e−<br />
(4.2)<br />
F L × number <strong>of</strong> active joints<br />
∑<br />
i<br />
Qi influence. i
Bolt internal force (kN)<br />
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
90<br />
75<br />
60<br />
45<br />
30<br />
15<br />
0<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Bolt deformation (mm)<br />
Fig. 4.8 Bolt elongation behaviour.<br />
(a) 2F = 96.29kN;<br />
∆ = 0.61mm.<br />
(d) 2F = 174.46kN;<br />
∆ = 2.45mm.<br />
(b) 2F = 159.55kN;<br />
∆ = 1.43mm.<br />
(e) 2F = 179.08kN;<br />
∆ = 3.04mm.<br />
Fig. 4.9 Evolution <strong>of</strong> <strong>the</strong> contact area with <strong>the</strong> applied load.<br />
Ratio n/e<br />
1.00<br />
0.96<br />
0.92<br />
0.88<br />
0.84<br />
0.80<br />
0.76<br />
Whole contact area<br />
Joint elements at <strong>the</strong> bolt x axis<br />
(c) 2F = 171.00kN;<br />
∆ = 2.06mm.<br />
(f) 2F ≥ 183.40kN;<br />
∆ ≥ 3.63mm.<br />
0.72<br />
0 15 30 45 60 75 90 105<br />
Applied load F, per bolt row (kN)<br />
Fig. 4.10 Evolution <strong>of</strong> <strong>the</strong> ratio n/e with <strong>the</strong> applied load per bolt row.<br />
109
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
where FQi is <strong>the</strong> force associated to a joint row (in <strong>the</strong> y direction), Linfluence.i is<br />
<strong>the</strong> influence length <strong>of</strong> each <strong>of</strong> those joint rows and xi is <strong>the</strong> distance <strong>of</strong> <strong>the</strong><br />
joint row to <strong>the</strong> tip <strong>of</strong> <strong>the</strong> flanges. Clearly, as <strong>the</strong> load increases, Q is shifted<br />
inside, from <strong>the</strong> tip <strong>of</strong> <strong>the</strong> flanges. Such situation is even more evident in <strong>the</strong><br />
second case.<br />
4.7 NUMERICAL RESULTS FOR WP T-STUB WT1<br />
The first series <strong>of</strong> tests WT1 included eight different specimens for analysis <strong>of</strong><br />
<strong>the</strong> adequate electrode for <strong>the</strong> welding procedure (cf. §3.3.1). The criterion for<br />
<strong>the</strong> choice <strong>of</strong> one or ano<strong>the</strong>r electrode type was based on <strong>the</strong> <strong>ductility</strong> provided<br />
to <strong>the</strong> connection. It was seen experimentally that basic electrodes with low<br />
hydrogen content ensured enhanced deformation capacity <strong>of</strong> <strong>the</strong> T-stub connection.<br />
Two tests (WT1g/h) from this series were performed with this electrode<br />
type and are used for fur<strong>the</strong>r comparisons. Fig. 4.11a compares <strong>the</strong> loadcarrying<br />
behaviour from <strong>the</strong> numerical model with <strong>the</strong> experiments. In <strong>the</strong> FE<br />
modelling, <strong>the</strong> average real dimensions are used. Exception is made for <strong>the</strong><br />
fillet weld throat thickness, as it was not measured. Therefore, <strong>the</strong> nominal<br />
value (aw = 5 mm) was used in <strong>the</strong> model. This can lead to some differences<br />
since <strong>the</strong> F-∆ response is sensitive to <strong>the</strong> value <strong>of</strong> m.<br />
The measurement <strong>of</strong> <strong>the</strong> gap between <strong>the</strong> two flanges in <strong>the</strong> test was performed<br />
by means <strong>of</strong> two LVDTs at opposite sides <strong>of</strong> <strong>the</strong> web. The numerical<br />
results that appear in <strong>the</strong> graph correspond to <strong>the</strong> location <strong>of</strong> those LVDTs. Fig.<br />
4.11b shows <strong>the</strong> bolt elongation response for specimen WT1h, for <strong>the</strong> broken<br />
bolts (LB: left back and LF: left front) – see also Figs. 3.10d and 3.11. The<br />
graph <strong>of</strong> Fig. 4.11b does not display <strong>the</strong> experimental results <strong>of</strong> <strong>the</strong> bolt elongation<br />
behaviour up to collapse since <strong>the</strong> measuring device was removed before<br />
<strong>the</strong> collapse.<br />
The FE model yields stiffer results than <strong>the</strong> experiments, though <strong>the</strong> agreement<br />
is good. The differences may derive from <strong>the</strong> insufficient geometrical and<br />
mechanical <strong>characterization</strong> <strong>of</strong> <strong>the</strong> fillet weld and also because <strong>of</strong> <strong>the</strong> modelling<br />
<strong>of</strong> <strong>the</strong> HAZ, near <strong>the</strong> weld toe. In fact, some authors [4.35] have already<br />
highlighted <strong>the</strong> fact that due to <strong>the</strong> welding process, <strong>the</strong> connection behaviour<br />
and <strong>the</strong> cracking <strong>of</strong> material, in particular, are influenced by <strong>the</strong> presence <strong>of</strong><br />
residual stresses and modified microstructures in <strong>the</strong> HAZ. It is very difficult to<br />
quantify <strong>the</strong>se effects and <strong>the</strong>refore <strong>the</strong>y were not included in <strong>the</strong> simulations.<br />
However, it should be borne in mind that if cracking <strong>of</strong> material governs <strong>the</strong><br />
collapse model, a reduction <strong>of</strong> <strong>the</strong> ultimate strain with respect to <strong>the</strong> unaffected<br />
material is advised. For both specimens WT1g/h, bolt fracture determines <strong>the</strong><br />
failure mode. Yet, for specimen WT1g <strong>the</strong>re was a combined failure type involving<br />
cracking <strong>of</strong> <strong>the</strong> flange in <strong>the</strong> HAZ and bolt fracture. Figs. 3.11b-c illustrate<br />
<strong>the</strong> specimen at failure. The graph from Fig. 4.11a also shows this type <strong>of</strong><br />
fracture: at a deformation level <strong>of</strong> circa 14 mm <strong>the</strong>re is a smooth drop <strong>of</strong> load<br />
that follows on until fracture <strong>of</strong> <strong>the</strong> bolt at 20.5 mm. Numerically and under <strong>the</strong><br />
110
Total applied load (kN)<br />
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
210<br />
180<br />
150<br />
(a) Load-deformation behaviour.<br />
Total applied load (kN)<br />
120<br />
90<br />
Experimental results: WT1g<br />
Experimental results: WT1h<br />
60<br />
Numerical results LUSAS<br />
30<br />
0<br />
Def. capacity (bolt - num. assessment)<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
210<br />
180<br />
150<br />
120<br />
Deformation (mm)<br />
90<br />
60<br />
Experimental results: bolt LB<br />
Experimental results: bolt LF<br />
30<br />
0<br />
Numerical results LUSAS<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1<br />
Bolt elongation (mm)<br />
(b) Bolt elongation behaviour.<br />
Fig. 4.11 Global response <strong>of</strong> specimen WT1: numerical and experimental results.<br />
above proposed failure criterion, it was established that bolt determines collapse.<br />
This was in line with experimental observations and <strong>the</strong> numerical prediction<br />
(13.98 mm) matches <strong>the</strong> experimental results for WT1h (15.11 mm at<br />
maximum load). The average maximum principal strain level in <strong>the</strong> HAZ is<br />
6.8% with a local maximum <strong>of</strong> 14% (FE results). For <strong>the</strong> flange <strong>plate</strong>, <strong>the</strong><br />
maximum (natural) strain measured in standard material tensile testing was<br />
30.8%.<br />
Finally, Fig. 4.12 plots <strong>the</strong> strains in <strong>the</strong> flange, close to <strong>the</strong> fillet weld.<br />
Specimen WT1h was instrumented with five strain gauges on one side <strong>of</strong> <strong>the</strong><br />
connection near <strong>the</strong> weld toe (Figs. 3.8a and 3.10d): (i) SG1 and SG5 are located<br />
near <strong>the</strong> flange edge, (ii) SG2 and SG4 are placed at <strong>the</strong> bolt x axis crosssection<br />
and (iii) SG3 is attached at <strong>the</strong> T-stub half width. The good correspondence<br />
between results is a valid statement <strong>of</strong> <strong>the</strong> reliability <strong>of</strong> <strong>the</strong> procedure.<br />
Regarding <strong>the</strong> ratios Q/F and n/e for <strong>the</strong> specimen WT1, Fig. 4.13 shows<br />
111
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
112<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
(a) Strain gauges SG1 and SG5.<br />
Total applied load (kN)<br />
0<br />
0.0% 0.6% 1.2% 1.8% 2.4% 3.0% 3.6% 4.2% 4.8% 5.4% 6.0%<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
(b) Strain gauges SG2 and SG4.<br />
Total applied load (kN)<br />
Strain<br />
Experimental results SG1<br />
Experimental results SG5<br />
Numerical results LUSAS<br />
30<br />
Numerical results LUSAS<br />
0<br />
0.0% 0.6% 1.2% 1.8% 2.4% 3.0% 3.6% 4.2% 4.8% 5.4% 6.0%<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Strain<br />
Experimental results SG2<br />
Experimental results SG4<br />
0<br />
0.0% 0.6% 1.2% 1.8% 2.4% 3.0% 3.6% 4.2% 4.8% 5.4% 6.0%<br />
Strain<br />
Experimental results SG3<br />
Numerical results LUSAS<br />
(c) Strain gauge SG3.<br />
Fig. 4.12 Force-strain in <strong>the</strong> x direction, εxx, response for specimen WT1.
Ratio<br />
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
(a) Ratio Q/F and B/F.<br />
Ratio n/e<br />
0.2<br />
Prying force, Q Bolt force, B<br />
0.0<br />
0 15 30 45 60 75 90 105<br />
1.00<br />
0.94<br />
0.88<br />
0.82<br />
0.76<br />
0.70<br />
0.64<br />
0.58<br />
Applied load F, per bolt row (kN)<br />
Whole contact area<br />
Joint elements at <strong>the</strong> bolt axis<br />
0.52<br />
0 15 30 45 60 75 90 105<br />
Applied load F, per bolt row (kN)<br />
(b) Evolution <strong>of</strong> <strong>the</strong> ratio n/e with <strong>the</strong> applied load per bolt row.<br />
Fig. 4.13 Numerical results for <strong>the</strong> prying forces (specimen WT1).<br />
<strong>the</strong>ir evolution with <strong>the</strong> external load. In <strong>the</strong> elastic regime Q/F = 0.26 and n/e =<br />
0.73 at <strong>the</strong> bolt horizontal axis; at failure, Q/F = 0.37 and n/e = 0.58 at <strong>the</strong> same<br />
section. There is an amplification in Q/F <strong>of</strong> 1.42 and <strong>the</strong> prying force shifts to<br />
<strong>the</strong> bolt vertical axis as <strong>the</strong> load increases.<br />
4.8 CONSIDERATIONS ON THE NUMERICAL MODELLING OF THE HEAT<br />
AFFECTED ZONE IN WP T-STUBS<br />
The numerical results presented above for WP-T-stub specimen WT1 do not<br />
account for <strong>the</strong> specific behaviour <strong>of</strong> <strong>the</strong> HAZ. The <strong>characterization</strong> <strong>of</strong> <strong>the</strong> mechanical<br />
properties <strong>of</strong> this zone is very complex and uncertain due to its heterogeneity<br />
and small size. Never<strong>the</strong>less, most <strong>of</strong> <strong>the</strong> high strength steels that<br />
113
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
have a high carbon content and various alloying elements as chromium, copper,<br />
nickel, etc. present a marked loss <strong>of</strong> hardness and strength in <strong>the</strong> HAZ that<br />
may affect <strong>the</strong> performance <strong>of</strong> <strong>the</strong> T-stub connection [4.36]. Moreover, localized<br />
heating from <strong>the</strong> welding process and subsequent rapid cooling induce a<br />
local triaxial residual tensile stress field in <strong>the</strong> HAZ and <strong>the</strong> constraint conditions<br />
affect <strong>the</strong> failure <strong>ductility</strong> <strong>of</strong> <strong>the</strong> metal in <strong>the</strong> zone. These effects are not<br />
easily modelled.<br />
However, <strong>the</strong>re is evidence that when s<strong>of</strong>t electrodes are used (cf. §3.2.4),<br />
<strong>the</strong> strength <strong>of</strong> <strong>the</strong> weld is slightly affected and <strong>the</strong> installed residual stress field<br />
is not significant [4.36]. As <strong>the</strong> fine microstructure is lost during <strong>the</strong> weld <strong>the</strong>rmal<br />
cycle, <strong>the</strong> HAZ strength and toughness are expected to decrease below<br />
those <strong>of</strong> <strong>the</strong> BM [4.37], i.e. <strong>the</strong> HAZ s<strong>of</strong>tens. This s<strong>of</strong>tening effect, which dep<strong>end</strong>s<br />
on <strong>the</strong> heat input, can be so severe that fracture can occur in <strong>the</strong> HAZ<br />
instead <strong>of</strong> <strong>the</strong> BM, as seen in <strong>the</strong> previous chapter 3. Bang and Kim [4.37] estimate<br />
<strong>the</strong> degree <strong>of</strong> HAZ s<strong>of</strong>tening in 20% at 6 kJ/mm. This means that <strong>the</strong><br />
strength properties in this zone should be reduced to a maximum <strong>of</strong> 80% in<br />
relation to those <strong>of</strong> <strong>the</strong> BM.<br />
Ano<strong>the</strong>r aspect that must be considered in <strong>the</strong> modelling <strong>of</strong> <strong>the</strong> HAZ is <strong>the</strong><br />
width itself, lHAZ. Rodrigues et al. show that <strong>the</strong> ratio lHAZ/tf is an important<br />
parameter in <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> change <strong>of</strong> strength in <strong>the</strong> zone [4.38].<br />
They studied <strong>the</strong> influence <strong>of</strong> <strong>the</strong> HAZ size in <strong>the</strong> geometrical constraint effect<br />
and consequent influence behaviour <strong>of</strong> <strong>the</strong> joint. The study covered <strong>the</strong> range<br />
<strong>of</strong> lHAZ/tf 1/6-1 and it demonstrated that this influence is negligible if <strong>the</strong> WM<br />
tensile strength evenmatches <strong>the</strong> BM. This is <strong>the</strong> case <strong>of</strong> <strong>the</strong> tested specimens.<br />
Taking <strong>the</strong>se considerations into account, a FE model was implemented for<br />
specimen WT1 in order to analyse <strong>the</strong> influence <strong>of</strong> <strong>the</strong> HAZ properties on <strong>the</strong><br />
overall behaviour. The width <strong>of</strong> <strong>the</strong> zone was taken as 5 mm, which corresponds<br />
to lHAZ/tf = 0.5. The model assumed a degree <strong>of</strong> s<strong>of</strong>tening <strong>of</strong> 15%,<br />
slightly below <strong>the</strong> maximum, as <strong>the</strong>re was no information on <strong>the</strong> heat input<br />
during <strong>the</strong> welding process. The results are illustrated in Fig. 4.14. Compari-<br />
114<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Experimental results: WT1h<br />
60<br />
Numerical results (mechanical properties <strong>of</strong> <strong>the</strong><br />
HAZ equal to those <strong>of</strong> <strong>the</strong> BM)<br />
30<br />
0<br />
Numerical results (strength mechanical<br />
properties <strong>of</strong> <strong>the</strong> HAZ reduced in 15%)<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
Fig. 4.14 Numerical results for specimen WT1 accounting for a reduction <strong>of</strong><br />
15% in <strong>the</strong> strength properties <strong>of</strong> <strong>the</strong> HAZ.
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
sons with <strong>the</strong> original numerical model and experimental results are also set.<br />
The correspondence between <strong>the</strong> FE results and <strong>the</strong> experiments improves<br />
when compared to <strong>the</strong> model described earlier. There is a slight drop in <strong>the</strong><br />
load-carrying behaviour in <strong>the</strong> post-limit domain (circa 7% in <strong>the</strong> load and 10%<br />
in <strong>the</strong> deformation capacity). These differences, however, can be considered<br />
insignificant. Therefore, for future analyses, <strong>the</strong> influence <strong>of</strong> <strong>the</strong> s<strong>of</strong>tening <strong>of</strong><br />
<strong>the</strong> HAZ is disregarded.<br />
4.9 CONCLUDING REMARKS<br />
The three-dimensional FE model presented above provides accurate deformation<br />
predictions (up to fracture) <strong>of</strong> <strong>the</strong> T-stub response. It allows for a complete<br />
<strong>characterization</strong> <strong>of</strong> <strong>the</strong> load-carrying behaviour <strong>of</strong> both types <strong>of</strong> T-stub assemblies.<br />
Table 4.2 compares <strong>the</strong> main characteristics <strong>of</strong> <strong>the</strong> F-∆ curve as ascertained<br />
numerically and experimentally. The results are very close, which means<br />
that <strong>the</strong> FE model is valid and reliable.<br />
The <strong>characterization</strong> <strong>of</strong> <strong>the</strong> T-stub collapse failure modes and corresponding<br />
<strong>ductility</strong> levels can be performed by means <strong>of</strong> this numerical procedure in<br />
order to clarify some code deficiencies. Additionally, <strong>the</strong> numerical model allows<br />
<strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> prying forces, thus opening <strong>the</strong> way to more reliable<br />
design rules. Fur<strong>the</strong>r, <strong>the</strong> parameters affecting <strong>the</strong> deformation capacity <strong>of</strong><br />
<strong>bolted</strong> T-stubs can be highlighted and <strong>the</strong>ir influence on <strong>the</strong> overall behaviour<br />
<strong>of</strong> <strong>the</strong> connection can be assessed both qualitatively and quantitatively. It is<br />
easy to recognize that <strong>the</strong> deformation capacity <strong>of</strong> isolated <strong>bolted</strong> T-stub connections<br />
mainly dep<strong>end</strong>s on <strong>the</strong> mechanical properties <strong>of</strong> <strong>the</strong> materials and on<br />
some geometrical parameters. The next logical step forward is <strong>the</strong> implementation<br />
<strong>of</strong> a parametric study based on <strong>the</strong> above procedures, in order to get insight<br />
on this particular aspect. The following chapter is devoted to such a<br />
study, presenting an experimental/numerical investigation that allows for a<br />
complete understanding <strong>of</strong> <strong>the</strong> main influences on <strong>the</strong> T-stub ultimate behaviour.<br />
Table 4.2 Results for <strong>the</strong> two specimens [values in bold correspond to averaged<br />
experimental results; underlined values include <strong>the</strong> web deformation;<br />
K-R refers to <strong>the</strong> knee-range <strong>of</strong> <strong>the</strong> curve].<br />
Spec. Stiffness (kN/mm) Strength (kN) ∆u Q/F<br />
T1<br />
WT1<br />
ke.0 kpl.0 ke.0/kpl.0 K-R Fu (mm) K-R Ult.<br />
83.54 2.68 29.60 65-85 103.99 8.70 0.24 0.34<br />
49.00 1.73 28.32 58-87 102.81 9.49 ⎯ ⎯<br />
69.29 1.57 44.24 55-76 94.98 14.20 0.27 0.37<br />
71.09 2.09 34.01 58-69 183.83 14.33 ⎯ ⎯<br />
115
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
4.10 REFERENCES<br />
[4.1] Virdi KS. Guidance on good practice in simulation <strong>of</strong> semi-rigid connections<br />
by <strong>the</strong> finite element method. In: Numerical simulation <strong>of</strong><br />
semi-rigid connections by <strong>the</strong> finite element method (Ed.: K.S. Virdi).<br />
COST C1, Report <strong>of</strong> working group 6 – Numerical simulation, Brussels;<br />
1-12, 1999.<br />
[4.2] Ba<strong>the</strong> KJ. Finite element procedures in engineering analysis. Prentice-<br />
Hall, Englewood Cliffs, New Jersey, USA, 1982.<br />
[4.3] Hinton E, Owen DR. An introduction to finite element computations.<br />
Pineridge Press Limited, Swansea, UK, 1979.<br />
[4.4] van der Vegte GJ, Makino Y, Sakimoto T. Numerical research on single-<strong>bolted</strong><br />
connections using implicit and explicit solution techniques.<br />
Memoirs <strong>of</strong> <strong>the</strong> Faculty <strong>of</strong> Engineering Kumamoto University;<br />
XXXXVII(1):19-44, 2002.<br />
[4.5] Owen DRJ, Hinton E. Finite elements in plasticity, <strong>the</strong>ory and practice.<br />
Pineridge Press Limited, Swansea, UK, 1980.<br />
[4.6] Ba<strong>the</strong> KJ, Wilson EL. Numerical methods in finite element analysis.<br />
Prentice-Hall, Englewood Cliffs, New Jersey, USA, 1976.<br />
[4.7] Lusas 13. Modeller reference manual. Finite element analysis Ltd, Version<br />
13.2. Surrey, UK, 2001.<br />
[4.8] Lusas 13. Solver reference manual. Finite element analysis Ltd, Version<br />
13.2. Surrey, UK, 2001.<br />
[4.9] Jaspart JP. Numerical simulation <strong>of</strong> a T-stub – experimental data. Cost<br />
C1, Numerical simulation group, Doc. C1WD6/94-09, 1994.<br />
[4.10] Bursi OS. A refined finite element model for T-stub steel connections.<br />
Cost C1, Numerical simulation group, Doc. C1WD6/95-07, 1995.<br />
[4.11] Bursi OS, Jaspart JP. Benchmarks for finite element modelling <strong>of</strong> <strong>bolted</strong><br />
steel connections. Journal <strong>of</strong> Constructional Steel Research; 43(1):17-<br />
42, 1997.<br />
[4.12] Bursi OS, Jaspart JP. Basic issues in <strong>the</strong> finite element simulation <strong>of</strong><br />
ext<strong>end</strong>ed <strong>end</strong>-<strong>plate</strong> connections. Computers and Structures; 69:361-382,<br />
1998.<br />
[4.13] Gomes FCT, Neves LFC, Silva LAPS, Simões RAD. Numerical simulation<br />
<strong>of</strong> a T-stub. Cost C1, Numerical simulation group, Doc.<br />
C1WG6/95-, 1995.<br />
[4.14] Mistakidis ES, Baniotopoulos CC, Bisbos CD, Panagiotopoulos PD. A<br />
2-D numerical model for <strong>the</strong> analysis <strong>of</strong> steel T-stub connections. Cost<br />
C1, Numerical simulation group, Doc. C1WD6/96-09, 1996.<br />
[4.15] Mistakidis ES, Baniotopoulos CC, Bisbos CD, Panagiotopoulos PD.<br />
Steel T-stub connections under static loading: an effective 2-D numerical<br />
model. Journal <strong>of</strong> Constructional Steel Research; 44(1-2):51-67,<br />
1997.<br />
[4.16] Zajdel M. Numerical analysis <strong>of</strong> <strong>bolted</strong> tee-stub connections. TNO-<br />
Report 97-CON-R-1123, 1997.<br />
116
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
[4.17] Wanzek T, Gebbeken N. Numerical aspects for <strong>the</strong> simulation <strong>of</strong> <strong>end</strong><br />
<strong>plate</strong> connections. In: Numerical simulation <strong>of</strong> semi-rigid connections<br />
by <strong>the</strong> finite element method (Ed.: K.S. Virdi). COST C1, Report <strong>of</strong><br />
working group 6 – Numerical simulation, Brussels; 13-31, 1999.<br />
[4.18] Gebbeken N, Wanzek T, Petersen, C. Semi-rigid connections, T-stub<br />
model – Report on experimental investigations. Report 97/2. Institut für<br />
Mechanik und Static, Universität des Bundeswehr München, Munich,<br />
Germany, 1997.<br />
[4.19] Swanson JA. Characterization <strong>of</strong> <strong>the</strong> strength, stiffness and <strong>ductility</strong><br />
behavior <strong>of</strong> T-stub connections. PhD dissertation, Georgia Institute <strong>of</strong><br />
Technology, Atlanta, USA, 1999.<br />
[4.20] Swanson JA, Kokan DS, Leon RT. Advanced finite element modelling<br />
<strong>of</strong> <strong>bolted</strong> T-stub connection components. Journal <strong>of</strong> Constructional<br />
Steel Research; 58:1015-1031, 2002.<br />
[4.21] Lusas 13. Element reference manual. Finite element analysis Ltd, Version<br />
13.2. Surrey, UK, 2001.<br />
[4.22] Lusas 13. Theory manual. Finite element analysis Ltd, Version 13.2.<br />
Surrey, UK, 2001.<br />
[4.23] Simo JC, Rifai MS. A class <strong>of</strong> mixed assumed strain methods and <strong>the</strong><br />
method <strong>of</strong> incompatible modes. International Journal for Numerical<br />
Methods in Engineering; 29:1595-1638, 1990.<br />
[4.24] Crisfield M. Large deflection elasto-plastic buckling analysis <strong>of</strong> <strong>plate</strong>s<br />
using finite elements. TRRL Report LR 593, Transport and Road Research<br />
Laboratory, Department <strong>of</strong> <strong>the</strong> Environment, Crowthorne, UK,<br />
1973.<br />
[4.25] Crisfield M. Non-linear finite element analysis <strong>of</strong> solids and structures,<br />
Volume 1 – Essentials. John Wiley & Sons Ltd., Chichester, UK, 1997.<br />
[4.26] Crisfield M. Non-linear finite element analysis <strong>of</strong> solids and structures,<br />
Volume 2 – Advanced topics. John Wiley & Sons Ltd., Chichester, UK,<br />
1997.<br />
[4.27] Aggerskov H. High-strength <strong>bolted</strong> connections subjected to prying.<br />
Journal <strong>of</strong> Structural Division ASCE; 102(ST1):161-175, 1976.<br />
[4.28] European Committee for Standardization (CEN). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[4.29] Vasarhelyi DD, Chiang KC. Coefficient <strong>of</strong> friction in joints <strong>of</strong> various<br />
steel. Journal <strong>of</strong> Structural Division ASCE; 93(ST4):227-243, 1967.<br />
[4.30] Girão Coelho AM. Material data <strong>of</strong> <strong>the</strong> <strong>plate</strong> sections <strong>of</strong> <strong>the</strong> welded Tstub<br />
specimens. Internal report, Steel and Timber Section, Faculty <strong>of</strong><br />
Civil Engineering, Delft University <strong>of</strong> Technology, 2002.<br />
[4.31] Faella C, Piluso V, Rizzano G. Structural semi-rigid connections – <strong>the</strong>ory,<br />
design and s<strong>of</strong>tware, CRC Press, USA, 2000.<br />
[4.32] Gioncu V, Mateescu G, Petcu D, Anastasiadis A. Prediction <strong>of</strong> available<br />
<strong>ductility</strong> by means <strong>of</strong> local plastic mechanism method: DUCTROT<br />
computer program, Chapter 2.1 in Moment resistant connections <strong>of</strong> steel<br />
117
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
frames in seismic areas (Ed.: F. Mazzolani). E&FN Spon, London, UK;<br />
95-146, 2000.<br />
[4.33] Hirt MA, Bez R. Construction métallique – Notions fondamentales et<br />
methods de dimensionnement. Traité de Génie Civil de l’École<br />
polytechnique fédérale de Lausanne, Volume 10. Presses Polytechniques et<br />
Universitaires Romandes, Lausanne, Switzerland, 1994.<br />
[4.34] Girão Coelho AM, Bijlaard F, Simões da Silva L. On <strong>the</strong> deformation<br />
capacity <strong>of</strong> beam-to-column <strong>bolted</strong> connections. Document ECCS-<br />
TWG 10.2-02-003, European Convention for Constructional Steelwork<br />
– Technical Committee 10, Structural connections (ECCS-TC10), 2002.<br />
[4.35] Piluso V, Faella C, Rizzano G. Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. II:<br />
model validation. Journal <strong>of</strong> Structural Engineering ASCE; 127(6):694-<br />
704, 2001.<br />
[4.36] Loureiro AJR. Effect <strong>of</strong> heat input on plastic deformation <strong>of</strong> underma<strong>the</strong>d<br />
welds. Journal <strong>of</strong> Materials Processing Technology; 128:240-<br />
249, 2002.<br />
[4.37] Bang KS, Kim WY. Estimation and prediction <strong>of</strong> <strong>the</strong> HAZ s<strong>of</strong>tening in<br />
<strong>the</strong>rmomechanically controlled-rolled and accelerated-cooled steel.<br />
Welding Journal; 81(8):174S-179S, 2002.<br />
[4.38] Rodrigues DM, Menezes LF, Loureiro A, Fernandes JV. Numerical<br />
study <strong>of</strong> <strong>the</strong> plastic behaviour in tension <strong>of</strong> welds in high strength steels.<br />
International Journal <strong>of</strong> Plasticity; 20:1-18, 2004.<br />
118
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
APPENDIX B: PRELIMINARY STUDY FOR CALIBRATION OF THE FINITE ELE-<br />
MENT MODEL (E.G. HR-T-STUB T1)<br />
B.1 Mesh convergence study<br />
As mentioned in §4.4.4, <strong>the</strong> flange mesh discretization with HX8M elements is<br />
analysed with respect to two parameters: (i) x: degree <strong>of</strong> discretization in order<br />
to represent <strong>the</strong> b<strong>end</strong>ing dominated problem and (ii) y: number <strong>of</strong> elements<br />
through thickness to check <strong>the</strong> capability <strong>of</strong> representing <strong>the</strong> yielding lines. The<br />
various discretizations are labelled ‘FxTy’, concerning <strong>the</strong> two above parameters,<br />
respectively (Fig. B1). The material properties adopted in <strong>the</strong>se simulations<br />
are those from Fig. 4.2a.<br />
Fig. B2 depicts <strong>the</strong> F-∆ response <strong>of</strong> discretization F0Ty, with y = 1, 2, 3, 4<br />
and 5. The deformation behaviour <strong>of</strong> F0T1 is stiffer than <strong>the</strong> o<strong>the</strong>r models because<br />
shear locking occurs. The remaining models yield identical solutions in<br />
<strong>the</strong> elastic domain but slightly different solutions in <strong>the</strong> plastic domain. Model<br />
F0T2 is more flexible than F0T3, F0T4 and F0T5, which show very small deviations.<br />
For future analysis, <strong>the</strong> model with three layers <strong>of</strong> HX8M is adopted.<br />
(a) F0T1. (b) F0T4. (c) F1T2.<br />
(d) F1T3. (e) F2T3.<br />
Fig. B1 Flange discretization: analysed models FxTy.<br />
119
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
To assess <strong>the</strong> influence <strong>of</strong> <strong>the</strong> degree <strong>of</strong> <strong>the</strong> flange discretization, <strong>the</strong> F-∆ behaviour<br />
<strong>of</strong> models F0T3, F1T3 and F2T3 is compared (Fig. B3). In <strong>the</strong> elastic<br />
domain, <strong>the</strong> three models yield identical stiffness. The model F1T3 is stiffer<br />
than model F2T3 in <strong>the</strong> plastic regime, but <strong>the</strong> post-limit stiffness is identical<br />
for both models. For model F0T3 <strong>the</strong> slope <strong>of</strong> <strong>the</strong> F-∆ curve is smaller than <strong>the</strong><br />
corresponding value for <strong>the</strong> finer meshes. Fig. B4 compares <strong>the</strong> curves for<br />
models F1Ty, y = 2, 3 and 4 and F2T3. Again, <strong>the</strong> model with two layers <strong>of</strong><br />
elements shows a weaker response than <strong>the</strong> remainders. This situation, again, is<br />
due to <strong>the</strong> shear locking effect, which is compensated in this particular case by<br />
a weaker plastic response. Comparison <strong>of</strong> models F1T3 and F1T4 shows that<br />
<strong>the</strong> lesser <strong>the</strong> number <strong>of</strong> elements across flange thickness, <strong>the</strong> stiffer <strong>the</strong> response.<br />
F1T4 and F2T3, however, yield similar results.<br />
Model F2T3 (11910 nodes and 7722 elements) satisfies convergence requirements<br />
but demands greater computation effort. Model F1T3 (5680 nodes<br />
and 3588 elements) shows small deviations from F2T3 and is not as timeconsuming.<br />
Therefore, it will be used extensively in future comparisons.<br />
120<br />
Total applied load (kN)<br />
210<br />
195<br />
180<br />
165<br />
150<br />
135<br />
120<br />
F0T1<br />
F0T4<br />
F0T2<br />
F0T5<br />
F0T3<br />
0 1 2 3 4 5 6 7 8<br />
Total deformation (mm)<br />
Fig. B2 Comparison <strong>of</strong> <strong>the</strong> deformation behaviour <strong>of</strong> models F0Ty.<br />
Total applied load (kN)<br />
200<br />
190<br />
180<br />
170<br />
160<br />
150<br />
140<br />
F0T3 F1T3 F2T3<br />
0 1 2 3 4 5 6 7 8<br />
Total deformation (mm)<br />
Fig. B3 Comparison <strong>of</strong> <strong>the</strong> deformation behaviour <strong>of</strong> models FxT3.
Total applied load (kN)<br />
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
200<br />
190<br />
180<br />
170<br />
160<br />
150<br />
140<br />
F1T2<br />
F1T4<br />
F1T3<br />
F2T3<br />
0 1 2 3 4 5 6 7 8<br />
Total deformation (mm)<br />
Fig. B4 Comparison <strong>of</strong> <strong>the</strong> deformation behaviour <strong>of</strong> models F1T2, F1T3,<br />
F1T4 and F2T3.<br />
B.2 Influence <strong>of</strong> <strong>the</strong> definition <strong>of</strong> <strong>the</strong> constitutive law and element formulation<br />
on <strong>the</strong> overall behaviour<br />
Fig. B5a compares <strong>the</strong> F-∆ curve for <strong>the</strong> different material stress-strain relationships.<br />
As expected, this relation does not influence <strong>the</strong> elastic behaviour,<br />
but <strong>the</strong> post-limit behaviour is ra<strong>the</strong>r stiffer in <strong>the</strong> case <strong>of</strong> true stresslogarithmic<br />
strain relation and closer to <strong>the</strong> experimental behaviour. Fig. B5b<br />
depicts <strong>the</strong> F-∆ characteristics for two different element formulations: total<br />
Lagrangian and updated Lagrangian formulations. Again, <strong>the</strong> elastic part <strong>of</strong> <strong>the</strong><br />
curve is not affected by <strong>the</strong> different element HX8M formulation. However, in<br />
<strong>the</strong> plastic range, <strong>the</strong> updated Lagrangian formulation yields closer results to<br />
<strong>the</strong> experimental curve. Therefore, to perform realistic simulations, a true<br />
stress-logarithmic strain relation must describe <strong>the</strong> constitutive material laws<br />
and <strong>the</strong> updated Lagrangian element formulation must be used.<br />
B.3 Calibration <strong>of</strong> <strong>the</strong> joint element stiffness<br />
Regarding <strong>the</strong> joint element stiffness, three cases are analysed in Fig. B6. The<br />
stiffer <strong>the</strong> elements, <strong>the</strong> stiffer <strong>the</strong> global T-stub response. From <strong>the</strong> stiffness<br />
values, it can be concluded that <strong>the</strong> results for <strong>the</strong> stiffness coefficient k1 =<br />
8000 N/mm/mm 2 are more realistic than <strong>the</strong> higher values. In terms <strong>of</strong> elastic<br />
stiffness and ultimate resistance, <strong>the</strong> three curves fit each o<strong>the</strong>r. However, in<br />
<strong>the</strong> knee-range <strong>of</strong> <strong>the</strong> global response, this model is more compliant and accurate<br />
than <strong>the</strong> remaining.<br />
The joint element stiffness k1 is hence taken as equal to 8000 N/mm/mm 2<br />
and <strong>the</strong> tangential stiffness coefficients are taken as k2 = k3 = 1000 N/mm/mm 2 .<br />
121
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
122<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
(a) Stress-strain relationships.<br />
Total applied load (kN)<br />
90<br />
60<br />
Experimental results<br />
Num. res.: Nominal law<br />
30<br />
0<br />
Num. res.: True law<br />
0 1 2 3 4 5 6 7 8 9 10<br />
210<br />
180<br />
150<br />
120<br />
Total deformation (mm)<br />
90<br />
60<br />
Experimental results<br />
Num. res.: total lagrangian formulation<br />
30<br />
0<br />
Num. res.: update lagrangian formulation<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Total deformation (mm)<br />
(b) Total and updated Lagrangian formulation.<br />
Fig. B5 Influence <strong>of</strong> constitutive laws and element formulation on <strong>the</strong> overall<br />
behaviour.<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Experimental results<br />
60<br />
Num. res.: k1=8000<br />
Num. res.: k1=20000<br />
30<br />
0<br />
Num. res.: k1=2000000<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Total deformation (mm)<br />
Fig. B6 Influence <strong>of</strong> contact element stiffness coefficients on <strong>the</strong> overall<br />
behaviour.
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
APPENDIX C: STRESS AND STRAIN NUMERICAL RESULTS FOR HR-T-STUB T1<br />
C.1 Load steps for stress and strain contours<br />
To illustrate <strong>the</strong> stress and strain contour results, four load steps are chosen<br />
(Fig. C1) as follows: (i) 2F = 96.29 kN (load case 4) for <strong>the</strong> elastic regime, (ii)<br />
2F = 159.55 kN (load case 9) for <strong>the</strong> knee-range, (iii) 2F = 190.85 kN (load<br />
case 25) for <strong>the</strong> subsequent linear part (in <strong>the</strong> post-limit regime) and (iv) 2F =<br />
207.98 kN (load case 46) for <strong>the</strong> collapse (maximum deformation).<br />
C.2 Von Mises equivalent stresses, σeq<br />
The Von Mises equivalent stress, σeq, combines <strong>the</strong> individual component<br />
stresses at a node according to <strong>the</strong> classical Von Mises failure criterion. The<br />
stress distribution within <strong>the</strong> T-stub flange is well reproduced with <strong>the</strong> generalized<br />
stress σeq. Fig. C2 illustrates <strong>the</strong> σeq contours in <strong>the</strong> three-dimensional<br />
view, for <strong>the</strong> four load levels. The b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> flange is well reproduced.<br />
The peak equivalent stress values are located at <strong>the</strong> bolt axis and at <strong>the</strong> flangeto-web<br />
connection, where <strong>the</strong> yield lines develop. Fig. C3 depicts <strong>the</strong> equivalent<br />
stresses in xy cross-section corresponding to <strong>the</strong> bolt axis.<br />
Regarding <strong>the</strong> bolt behaviour, Fig. C4 shows <strong>the</strong> equivalent stresses for <strong>the</strong><br />
chosen load levels. The b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> bolt is clearly present from <strong>the</strong> commencement<br />
<strong>of</strong> loading. For <strong>the</strong> first load stage, no yielding occurs. As <strong>the</strong> load<br />
increases, <strong>the</strong> bolt stresses and strains magnify and so do <strong>the</strong> yielded portions.<br />
The compression and tension zones <strong>of</strong> <strong>the</strong> bolt are also noticeable: <strong>the</strong> bolt area<br />
near <strong>the</strong> web is subjected to tension whilst <strong>the</strong> zone near <strong>the</strong> tips <strong>of</strong> <strong>the</strong> flange is<br />
in compression.<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
96.29 kN<br />
159.55 kN<br />
190.85 kN<br />
Experimental results<br />
Numerical results LUSAS<br />
0<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Total deformation (mm)<br />
207.98 kN<br />
Fig. C1 Selected load levels for stress and strain analyses.<br />
123
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C2 Von Mises equivalent stresses in <strong>the</strong> T-stub flange.<br />
(a) Elastic regime. (b) Knee-range.<br />
Fig. C3 Von Mises equivalent stresses in xy cross-section.<br />
C.3 Stresses σxx and strains εxx<br />
The stress component in <strong>the</strong> x direction, σxx, represents <strong>the</strong> b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> T-<br />
element flanges along this axis. The development <strong>of</strong> σxx is illustrated in Fig.<br />
124
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
C5. The double curvature b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> T-stub flange is clear. The high values<br />
<strong>of</strong> tension (positive values) occur in <strong>the</strong> upper part <strong>of</strong> <strong>the</strong> flange-to-web connection<br />
and on <strong>the</strong> lower part <strong>of</strong> <strong>the</strong> flange near <strong>the</strong> bolt axis. Conversely, <strong>the</strong><br />
high values <strong>of</strong> compression are located on <strong>the</strong> opposite parts. Fig. C6 presents<br />
<strong>the</strong> strain results for <strong>the</strong> same load levels.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C3 Von Mises equivalent stresses in xy cross-section (cont.).<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C4 Von Mises equivalent stresses in <strong>the</strong> bolt.<br />
125
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
C.4 Stresses σyy<br />
The stress component along <strong>the</strong> y direction, σyy, is depicted in Figs. C.7-C.8 for<br />
<strong>the</strong> T-stub flange, three-dimensional and bottom xz plane views, respectively.<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C5 Stresses σxx in <strong>the</strong> T-stub flange.<br />
(a) Elastic regime. (b) Knee-range.<br />
Fig. C6 Strains εxx in <strong>the</strong> T-stub flange.<br />
126
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C6 Strains εxx in <strong>the</strong> T-stub flange (cont.).<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C7 Stresses σyy in <strong>the</strong> T-stub flange.<br />
The positive stress σyy is quite uniform in <strong>the</strong> flange in <strong>the</strong> elastic areas. The<br />
stress uniform transfer between <strong>the</strong> flange and <strong>the</strong> web is also clear in Fig. C7<br />
(red contour). The concentration <strong>of</strong> negative stress σyy occurs at <strong>the</strong> contact<br />
127
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
areas: <strong>the</strong> washer/flange and <strong>the</strong> flange/rigid foundation contact planes (Figs.<br />
C7-C8, respectively). In <strong>the</strong> latter, <strong>the</strong> stress concentration is quite distinct in<br />
<strong>the</strong> middle <strong>of</strong> <strong>the</strong> flange-to-web connection, due to <strong>the</strong> b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> flanges<br />
and at <strong>the</strong> tips <strong>of</strong> <strong>the</strong> flanges, where <strong>the</strong> prying effect takes place.<br />
C.5 Stresses σzz<br />
The stress component on <strong>the</strong> z-axis, σzz, represents <strong>the</strong> deformation behaviour<br />
along <strong>the</strong> T-stub width. The distribution <strong>of</strong> <strong>the</strong> stress is not uniform and <strong>the</strong><br />
peak values occur at <strong>the</strong> washer/flange contact plane, due to sliding (Fig. C9).<br />
C.6 Principal stresses and strains, σ11 and ε11<br />
The principal stress σ11 and <strong>the</strong> principal strain ε11 represent <strong>the</strong> maximum<br />
stress and strain values, respectively. The maximum values <strong>of</strong> stress in <strong>the</strong> Tstub<br />
flange (Fig. C10) occur at <strong>the</strong> bolt axis and at <strong>the</strong> flange-to-web connection.<br />
Fig. C11 shows <strong>the</strong> corresponding strain contours.<br />
Figs. C12-C13 illustrate <strong>the</strong> principal stresses and strains in <strong>the</strong> bolt, respectively.<br />
The maximum strain ε11 in <strong>the</strong> bolt corresponds to <strong>the</strong> maximum allowed<br />
strain and <strong>the</strong>refore once it is attained, collapse occurs. The distribution<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C8 Stresses σyy in <strong>the</strong> T-stub flange/rigid foundation contact plane.<br />
128
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C9 Stresses σzz in <strong>the</strong> T-stub flange.<br />
(a) Elastic regime. (b) Knee-range.<br />
Fig. C10 Principal stresses σ11 in <strong>the</strong> T-stub flange.<br />
<strong>of</strong> principal stresses in <strong>the</strong> bolt shank is not uniform as <strong>the</strong> applied load increases.<br />
The maximum contour area enlarges with <strong>the</strong> increasing <strong>of</strong> load. The<br />
principal strain contours in <strong>the</strong> xy cross-section are illustrated in Fig. C14.<br />
129
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C10 Principal stresses σ11 in <strong>the</strong> T-stub flange (cont.).<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C11 Principal strains ε11 in <strong>the</strong> T-stub flange.<br />
130
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C12 Principal stresses σ11 in <strong>the</strong> bolt.<br />
(a) Elastic regime. (b) Knee-range.<br />
Fig. C13 Principal strains ε11 in <strong>the</strong> bolt.<br />
131
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C13 Principal strains ε11 in <strong>the</strong> bolt (cont.).<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C14 Principal strains ε11 in xy cross-section.<br />
C.7 Displacement results in xy cross-section<br />
Finally, <strong>the</strong> displacement contour in <strong>the</strong> x and y directions are illustrated in<br />
Figs. C15-C16 for <strong>the</strong> middle xy cross-section. The results are presented in <strong>the</strong><br />
deformed configuration (magnification factor = 1.0). Fig. C15 shows that no<br />
penetration occurs between <strong>the</strong> bolt and <strong>the</strong> flange at collapse conditions.<br />
132
Numerical assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
(a) Elastic regime. (b) Knee-range.<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C15 Horizontal displacement contours in xy cross-section.<br />
(a) Elastic regime. (b) Knee-range.<br />
Fig. C16 Vertical displacement contours in xy cross-section.<br />
133
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(c) Post-limit regime. (d) Collapse.<br />
Fig. C16 Vertical displacement contours in xy cross-section (cont.).<br />
134
5 PARAMETRIC STUDY<br />
5.1 DESCRIPTION OF THE SPECIMENS<br />
A parametric analysis was undertaken in order to identify <strong>the</strong> dep<strong>end</strong>ence <strong>of</strong><br />
<strong>the</strong> T-stub behaviour on <strong>the</strong> main geometrical and mechanical variables. The<br />
basic HR specimen is T1 from <strong>the</strong> previous section and <strong>the</strong> study on this assembly<br />
type was performed numerically. For <strong>the</strong> WP specimens, an experimental<br />
programme was devised along with a FE analysis. Three supplementary<br />
WP-T-stubs derived from HR-T-stub T1 were also considered to compare <strong>the</strong><br />
behaviour <strong>of</strong> <strong>the</strong> two assembly types. The main geometric parameters that were<br />
varied in <strong>the</strong> study are: (i) weld throat thickness, aw, for WP-T-stubs (ii) gauge<br />
<strong>of</strong> <strong>the</strong> bolts, w, (iii) pitch <strong>of</strong> <strong>the</strong> bolts, p, (iv) <strong>end</strong> distance, e1, (v) edge distance,<br />
e, and (vi) flange thickness, tf. The influence <strong>of</strong> <strong>the</strong> bolt is analysed by varying:<br />
(i) <strong>the</strong> diameter, φ, (ii) <strong>the</strong> thread length (spec. P24: FT bolt) and (iii) <strong>the</strong> preload,<br />
S0 (spec. P25). The steel constitutive law is <strong>the</strong> mechanical variable in <strong>the</strong><br />
study. Additionally, <strong>the</strong> question <strong>of</strong> <strong>the</strong> number <strong>of</strong> bolt rows is also tackled.<br />
Tables 3.1 and 5.1 sum up <strong>the</strong> main characteristics <strong>of</strong> <strong>the</strong> several specimens.<br />
In <strong>the</strong> following sections <strong>the</strong> load-carrying behaviour <strong>of</strong> <strong>the</strong> several specimens<br />
is compared to assess <strong>the</strong> influence <strong>of</strong> <strong>the</strong> main parameters. For some<br />
specimens o<strong>the</strong>r results are also included to get insight into o<strong>the</strong>r behaviour parameters.<br />
5.2 INFLUENCE OF THE ASSEMBLY TYPE AND THE WELD THROAT THICK-<br />
NESS<br />
Having validated <strong>the</strong> numerical procedure for both T-stub assemblies, in this<br />
section <strong>the</strong> influence <strong>of</strong> welding and <strong>of</strong> <strong>the</strong> size <strong>of</strong> <strong>the</strong> fillet weld on <strong>the</strong> overall<br />
behaviour are analysed. For that purpose, HR-T-stub specimen T1 is selected.<br />
The equivalent WP-T-stub (generally labelled Weld_T1 hereafter) is identical<br />
to T1 in terms <strong>of</strong> geometrical and mechanical properties. The flange-to-web<br />
connection radius is thus replaced with a continuous 45º-fillet weld <strong>of</strong> throat<br />
thickness (i) aw = 0.5tw = 3.55 mm [Weld_T1(i)], (ii) aw = tw<br />
= 7.1mm<br />
[Weld_T1(ii)] and (iii) a w = 10 mm [Weld_T1(iii)]. The values for aw are chosen<br />
to meet <strong>the</strong> Eurocode 3 requirements [5.1]. The first value, aw = 0.5tw,<br />
complies with <strong>the</strong> minimum dimension prescribed in <strong>the</strong> code (3.0 mm) and is<br />
commonly used in practice. The value aw = tw<br />
can be regarded, in design practice,<br />
as an upper value for <strong>the</strong> size <strong>of</strong> <strong>the</strong> fillet weld. Finally, <strong>the</strong> latter value<br />
135
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
yields a distance m similar to <strong>the</strong> one in <strong>the</strong> HR specimen T1. For T1, m =<br />
41.45 − 0.8× 15 = 29.45 mm and for Weld_T1(iii), m = 41.45 − 0.8 2 × 10 =<br />
30.14 mm.<br />
The numerical results for both specimens T1 and Weld_T1 are compared in<br />
Figs. 5.1-5.3. Concerning <strong>the</strong> overall behaviour, <strong>the</strong> connections clearly yield<br />
different responses. Bolt fracture determines collapse <strong>of</strong> all T-stubs. Comparison<br />
<strong>of</strong> <strong>the</strong> F-∆ responses <strong>of</strong> <strong>the</strong> welded specimens shows that as <strong>the</strong> weld throat<br />
thickness increases, <strong>the</strong> stiffness and resistance improve but <strong>the</strong> deformation<br />
capacity significantly decreases (Fig. 5.1). It should be noted that <strong>the</strong> increase<br />
Table 5.1 Details <strong>of</strong> <strong>the</strong> parametric numerical study [dimensions in mm;<br />
yield stress in MPa] (see Figs. 1.9 and 3.1 for notation).<br />
Type Test Pr<strong>of</strong>ile<br />
T-stub elements geometry<br />
HR specimens<br />
WP<br />
spec.<br />
136<br />
ID<br />
b p e1 w e r/aw tf tw<br />
P1 IPE300 40 40 20 100 25 15 10.7 7.1<br />
P2 IPE300 40 40 20 80 35 15 10.7 7.1<br />
P3 IPE300 35 30 20 90 30 15 10.7 7.1<br />
P4 IPE300 52.5 65 20 90 30 15 10.7 7.1<br />
P5 IPE300 60 80 20 90 30 15 10.7 7.1<br />
P6 IPE300 35 40 15 90 30 15 10.7 7.1<br />
P7 IPE300 45 40 25 90 30 15 10.7 7.1<br />
P8 HEA220 40 40 20 90 65 18 11.0 7.0<br />
P9 HEB180 40 40 20 90 45 15 14.0 8.5<br />
P10 HEAA160+ 40 40 20 90 30 15 7.0 4.5<br />
P11 HEB180 40 40 20 90 30 15 14.0 8.5<br />
P12 IPE300 40 40 20 90 30 15 10.7 7.1<br />
P13 IPE300 40 40 20 90 30 15 10.7 7.1<br />
P14 IPE300 40 40 20 90 30 15 10.7 7.1<br />
P15 IPE300 40 40 20 80 35 15 10.7 7.1<br />
P16 IPE300 70 70 35 90 30 15 10.7 7.1<br />
P17 IPE300 70 70 35 90 30 15 10.7 7.1<br />
P18 IPE300 70 70 35 90 30 15 10.7 7.1<br />
P19 IPE300 70 90 25 90 30 15 10.7 7.1<br />
P20 HEB180 70 70 35 90 30 15 14.0 8.5<br />
P21 IPE300 92.5 115 35 90 30 15 10.7 7.1<br />
P22 UB457×<br />
152×67<br />
P23 UB457×<br />
152×82<br />
70 70 35 90 30 10.2 15.0 9.0<br />
70 70 35 90 30 10.2 18.9 10.5<br />
P24 IPE300 40 40 20 90 30 15 10.7 7.1<br />
P25 IPE300 40 40 20 90 30 15 10.7 7.1<br />
Weld_T1(i) 40 40 20 90 30 3.55 10.7 7.1<br />
Weld_T1(ii) 40 40 20 90 30 7.1 10.7 7.1<br />
Weld_T1(iii) 40 40 20 90 30 10 10.7 7.1
HR specimens<br />
Parametric study<br />
Table 5.1 Details <strong>of</strong> <strong>the</strong> parametric numerical study (cont.).<br />
Type Test<br />
ID<br />
Pr<strong>of</strong>ile<br />
φ<br />
Bolt<br />
# Type<br />
Material (fy)<br />
Flange Bolt<br />
P1 IPE300 12 4 ST 431 893<br />
P2 IPE300 12 4 ST 431 893<br />
P3 IPE300 12 4 ST 431 893<br />
P4 IPE300 12 4 ST 431 893<br />
P5 IPE300 12 4 ST 431 893<br />
P6 IPE300 12 4 ST 431 893<br />
P7 IPE300 12 4 ST 431 893<br />
P8 HEA220 12 4 ST 431 893<br />
P9 HEB180 12 4 ST 431 893<br />
P10 HEAA160+ 12 4 ST 431 893<br />
P11 HEB180 12 4 ST 431 893<br />
P12 IPE300 16 4 ST 431 893<br />
P13 IPE300 12 4 ST 355 893<br />
P14 IPE300 12 4 ST 275 893<br />
P15 IPE300 16 4 ST 431 893<br />
P16 IPE300 12 4 ST 431 893<br />
P17 IPE300 16 4 ST 431 893<br />
P18 IPE300 20 4 ST 431 893<br />
P19 IPE300 16 4 ST 431 893<br />
P20 HEB180 16 4 ST 431 893<br />
P21 IPE300 20 4 ST 431 893<br />
P22<br />
UB457×<br />
152×67<br />
20<br />
4 ST<br />
431 893<br />
P23<br />
UB457×<br />
152×82<br />
20<br />
4 ST<br />
431 893<br />
P24 IPE300 12 4 FT 431 893<br />
P25 IPE300 12 4 ST 431 893<br />
Weld_T1(i) 12 4 ST 431 893<br />
Weld_T1(ii) 12 4 ST 431 893<br />
Weld_T1(iii) 12 4 ST 431 893<br />
WP<br />
spec.<br />
Table 5.2 Numerical results (per bolt row) for T1 and weld-equiv. Weld_T1.<br />
Test ID Stiffness (kN/mm) Resistance (kN) ∆u.0 Q/F<br />
ke.0 kp-l.0 ke.0/kp-l.0 K-R Fu.0 (mm) K-R Ult.<br />
T1 83.54 2.68 29.60 65-85 103.99 8.70 0.24 0.34<br />
Weld<br />
_T1(i)<br />
Weld<br />
_T1(ii)<br />
Weld<br />
_T1(iii)<br />
73.50 1.70 43.12 50-78 92.02 10.85 0.34 0.45<br />
88.04 2.51 35.07 60-87 102.75 8.01 0.27 0.36<br />
107.29 3.31 32.41 75-97 113.10 6.22 0.22 0.28<br />
137
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
138<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
T1 Weld_T1(i)<br />
Weld_T1(ii) Weld_T1(iii)<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation (mm)<br />
Fig 5.1 Overall response <strong>of</strong> specimens T1 and Weld_T1.<br />
(a) Ratio B/F.<br />
Ratio B/F<br />
Ratio Q/F<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
T1 Weld_T1(i)<br />
0.0<br />
Weld_T1(ii) Weld_T1(iii)<br />
0 15 30 45 60 75 90 105 120<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
Applied load F, per bolt row (kN)<br />
T1 Weld_T1(i)<br />
Weld_T1(ii) Weld_T1(iii)<br />
0.0<br />
0 15 30 45 60 75 90 105 120<br />
Applied load F, per bolt row (kN)<br />
b) Ratio Q/F.<br />
Fig. 5.2 Bolt and prying force ratios for specimens T1 and Weld_T1.
Ratio n/e<br />
1.00<br />
0.96<br />
0.92<br />
0.88<br />
0.84<br />
0.80<br />
0.76<br />
(a) Whole contact area.<br />
Ratio n/e<br />
T1 Weld_T1(i)<br />
Weld_T1(ii) Weld_T1(iii)<br />
0.72<br />
0 15 30 45 60 75 90 105 120<br />
1.00<br />
0.95<br />
0.90<br />
0.85<br />
0.80<br />
0.75<br />
0.70<br />
0.65<br />
Applied load F, per bolt row (kN)<br />
T1 Weld_T1(i)<br />
Weld_T1(ii) Weld_T1(iii)<br />
0.60<br />
0 15 30 45 60 75 90 105 120<br />
Applied load F, per bolt row (kN)<br />
(b) Joint elements at <strong>the</strong> bolt x axis.<br />
Fig. 5.3 Evolution <strong>of</strong> <strong>the</strong> ratios n/e for specimens T1 and Weld_T1.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
157.95 kN<br />
(Ld cs 6)<br />
140.11 kN<br />
(Ld cs 9)<br />
205.50 kN<br />
(Ld cs 29)<br />
184.05 kN<br />
(Ld cs 57)<br />
Weld_T1(i) Weld_T1(ii)<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation (mm)<br />
Parametric study<br />
Fig. 5.4 Selected load levels for stress and strain analyses <strong>of</strong> specimens<br />
Weld_T1.<br />
139
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range. (ii) Collapse.<br />
(c) Specimen Weld_T1(ii).<br />
Fig. 5.5 Von Mises equivalent stresses in <strong>the</strong> T-stub flange.<br />
140
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range. (ii) Collapse.<br />
(c) Specimen Weld_T1(ii).<br />
Fig. 5.6 Von Mises equivalent stresses in xy cross-section.<br />
Parametric study<br />
in aw leads to a decrease in m. To conclude about <strong>the</strong> influence <strong>of</strong> welding itself,<br />
<strong>the</strong> comparisons have to be made between specimen T1 and Weld_T1(iii)<br />
that have similar values <strong>of</strong> <strong>the</strong> distance m. Clearly, if <strong>the</strong> flange-to-web connection<br />
radius is replaced with a fillet weld, <strong>the</strong> stiffness and <strong>the</strong> resistance <strong>of</strong> <strong>the</strong><br />
connection improve, but <strong>the</strong> deformation capacity is greatly reduced: it drops<br />
141
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range.<br />
(c) Specimen Weld_T1(ii).<br />
(ii) Collapse.<br />
Fig. 5.7 Stresses σxx in <strong>the</strong> T-stub flange.<br />
142
(i) Knee-range.<br />
(a) Specimen T1.<br />
(ii) Collapse.<br />
Fig. 5.8 Strains εxx in <strong>the</strong> T-stub flange.<br />
Parametric study<br />
from a gap between flanges <strong>of</strong> 8.70 mm to 6.22 mm. For specimens T1 and<br />
Weld_T1(ii), <strong>the</strong> F-∆ curves are surprisingly coincident. However, in <strong>the</strong><br />
welded case, <strong>the</strong> <strong>ductility</strong> is smaller. Table 5.2 sets out <strong>the</strong> main characteristics<br />
<strong>of</strong> <strong>the</strong> four F-∆ curves.<br />
Regarding <strong>the</strong> bolt force and <strong>the</strong> prying forces, <strong>the</strong>ir magnitude in relation<br />
to <strong>the</strong> applied load is higher for smaller weld throat thickness, in <strong>the</strong> case <strong>of</strong> <strong>the</strong><br />
welded specimen for all course <strong>of</strong> loading (Fig. 5.2). The location <strong>of</strong> <strong>the</strong> contact<br />
forces also changes with <strong>the</strong> increasing <strong>of</strong> loading (Fig. 5.3). For <strong>the</strong> WP<br />
specimen Weld_T1(i), with smaller aw, Q is closer to <strong>the</strong> bolt axis than in <strong>the</strong><br />
remaining specimens. Concerning <strong>the</strong> influence <strong>of</strong> <strong>the</strong> assembly type, Fig. 5.2<br />
shows that in <strong>the</strong> elastic regime <strong>the</strong> ratios B/F and Q/F for specimens T1 and<br />
Weld_T1(iii) are coincident but as <strong>the</strong> load increases <strong>the</strong> same ratios decrease<br />
in <strong>the</strong> welded case. Worth mentioning is <strong>the</strong> fact that even if T1 and<br />
Weld_T1(ii) yield identical F-∆ behaviour, <strong>the</strong> evolution <strong>of</strong> B/F and Q/F with<br />
<strong>the</strong> course <strong>of</strong> loading is different (Fig. 5.2). With respect to <strong>the</strong> location <strong>of</strong> <strong>the</strong><br />
prying forces, Fig. 5.3 shows that for <strong>the</strong> welded specimen Weld_T1(iii) <strong>the</strong>re<br />
is a almost constant relationship <strong>of</strong> n/e from <strong>the</strong> commencement <strong>of</strong> loading to<br />
failure, with a slight increase near collapse. For specimen T1, <strong>the</strong> variation <strong>of</strong><br />
143
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) Knee-range.<br />
(b) Specimen Weld_T1(i).<br />
(ii) Collapse.<br />
Fig. 5.8 Strains εxx in <strong>the</strong> T-stub flange (cont.).<br />
Q with <strong>the</strong> applied load is more evident and Q is shifted to <strong>the</strong> bolt axis near<br />
collapse failure.<br />
The magnitude <strong>of</strong> <strong>the</strong> difference in performance <strong>of</strong> <strong>the</strong> HP-T-stub T1 and<br />
<strong>the</strong> welded equivalents Weld_T1 is ra<strong>the</strong>r surprising. In terms <strong>of</strong> <strong>the</strong> overall deformation<br />
behaviour, <strong>the</strong> differences can arise due to <strong>the</strong> redefinition <strong>of</strong> <strong>the</strong><br />
length m that slightly increases in <strong>the</strong> welded case. This accounts for <strong>the</strong> decrease<br />
in stiffness and resistance. Regarding <strong>the</strong> deformation capacity, as it will<br />
also be shown in <strong>the</strong> following section, <strong>the</strong> increase in <strong>the</strong> same distance m improves<br />
<strong>the</strong> ultimate deformation <strong>of</strong> <strong>the</strong> connection, ∆u. With respect to <strong>the</strong> prying<br />
effect, <strong>the</strong> disparity <strong>of</strong> results was not expected.<br />
To compare <strong>the</strong> stress and strain contour results for <strong>the</strong> above specimens,<br />
two load steps are chosen (Fig. 5.4), corresponding to <strong>the</strong> knee-range <strong>of</strong> <strong>the</strong><br />
curves and collapse (maximum deformation). For specimen T1 <strong>the</strong> reader<br />
should refer to Fig. C1 from App<strong>end</strong>ix C for indication <strong>of</strong> <strong>the</strong> analogous levels.<br />
As <strong>the</strong> contour results for <strong>the</strong> welded specimens are identical, only <strong>the</strong> results<br />
for specimens Weld_T1(i-ii) are shown. For <strong>the</strong> two selected load steps, Figs.<br />
5.5-5.6 show <strong>the</strong> Von Mises equivalent stress contours in <strong>the</strong> T-stub flange and<br />
in xy cross-section at <strong>the</strong> bolt axis. The figures show that <strong>the</strong> higher stress values<br />
in <strong>the</strong> flange concentrate at <strong>the</strong> bolt axis and near <strong>the</strong> flange-to-web con-<br />
144
(i) Knee-range.<br />
(c) Specimen Weld_T1(ii).<br />
(ii) Collapse.<br />
Fig. 5.8 Strains εxx in <strong>the</strong> T-stub flange (cont.).<br />
Parametric study<br />
nection. In particular, in <strong>the</strong> case <strong>of</strong> WP specimens, such concentration takes<br />
place at <strong>the</strong> “potential” HAZ ra<strong>the</strong>r than at <strong>the</strong> weld toe. The stress distribution<br />
in <strong>the</strong> bolt is identical in all three cases.<br />
Figs. 5.7-5.10 illustrate <strong>the</strong> stress and strain contours in <strong>the</strong> x direction.<br />
They clearly show <strong>the</strong> double curvature <strong>of</strong> <strong>the</strong> flange in <strong>the</strong> three cases and<br />
confirm <strong>the</strong> previous conclusions related to <strong>the</strong> location <strong>of</strong> yield lines and potential<br />
fracture lines. Finally, Figs. 5.11-5.14 display identical results with respect<br />
to <strong>the</strong> principal direction 1.<br />
The experimental programme also included <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> influence <strong>of</strong><br />
<strong>the</strong> fillet weld throat thickness, aw on <strong>the</strong> overall behaviour (series WT2 – cf.<br />
Table 3.1). Bolt fracture is still <strong>the</strong> determinant factor <strong>of</strong> collapse, though some<br />
damage in <strong>the</strong> HAZ has been observed in specimens WT2Aa and WT2Ba. In<br />
<strong>the</strong>se specimens <strong>the</strong> weld quality was inferior to <strong>the</strong> expected and this may explain<br />
such <strong>plate</strong> damage. Fig. 5.15a shows that if aw decreases, <strong>the</strong> resistance<br />
slightly decreases, whilst <strong>the</strong> deformation capacity improves, with little variation<br />
<strong>of</strong> stiffness. On <strong>the</strong> o<strong>the</strong>r hand, if aw increases, <strong>the</strong> deformation capacity is<br />
reduced, resistance increases and <strong>the</strong>re is still small change in <strong>the</strong> slope <strong>of</strong> <strong>the</strong><br />
two characteristic branches <strong>of</strong> <strong>the</strong> F-∆ curve (Fig. 5.15b; see also Table 3.8).<br />
The main (experimental) characteristics <strong>of</strong> <strong>the</strong> tests also confirm <strong>the</strong> above<br />
145
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range.<br />
(c) Specimen Weld_T1(ii).<br />
(ii) Collapse.<br />
Fig. 5.9 Stresses σxx in xy cross-section.<br />
statements related to <strong>the</strong> influence <strong>of</strong> <strong>the</strong> fillet weld throat thickness.<br />
In series WT2Aa <strong>the</strong>re was a malfunctioning <strong>of</strong> <strong>the</strong> LVDTs and <strong>the</strong>re is<br />
only a record <strong>of</strong> <strong>the</strong> deformation behaviour until ∆ ≈ 4.5 mm.<br />
146
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range. (ii) Collapse.<br />
(c) Specimen Weld_T1(ii).<br />
Fig. 5.10 Strains εxx in xy cross-section.<br />
5.3 INFLUENCE OF GEOMETRIC PARAMETERS<br />
Parametric study<br />
The main geometric connection parameters that were varied in this parametric<br />
study are indicated in §5.1.<br />
147
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) Knee-range. (ii) Collapse<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range. (ii) Collapse.<br />
(c) Specimen Weld_T1(ii).<br />
Fig. 5.11 Principal stresses σ11 in <strong>the</strong> T-stub flange.<br />
148
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
Fig. 5.12 Principal strains ε11 in <strong>the</strong> T-stub flange.<br />
5.3.1 Gauge <strong>of</strong> <strong>the</strong> bolts<br />
Parametric study<br />
To assess <strong>the</strong> influence <strong>of</strong> <strong>the</strong> variation <strong>of</strong> <strong>the</strong> distance w, two HR specimens,<br />
P1 and P2, were obtained from T1 by shifting <strong>the</strong> bolt axis centreline (Table<br />
5.1). Naturally, this will also alter <strong>the</strong> distance m between yield lines. Note that<br />
in all three cases <strong>the</strong> determinant plastic failure mechanism was <strong>of</strong> type-1. Yet,<br />
<strong>the</strong> collapse condition <strong>of</strong> <strong>the</strong> several specimens was determined by bolt fracture<br />
(black circles).<br />
Fig. 5.16 shows that if <strong>the</strong> gauge <strong>of</strong> <strong>the</strong> bolts increases, consequently increasing<br />
<strong>the</strong> distance m between plastic hinges, <strong>the</strong> connection strength and<br />
stiffness decrease but <strong>the</strong> deformation capacity improves. These results can be<br />
found in Table 5.3.<br />
5.3.2 Pitch <strong>of</strong> <strong>the</strong> bolts and <strong>end</strong> distance<br />
The enlargement <strong>of</strong> <strong>the</strong> pitch <strong>of</strong> <strong>the</strong> bolts and/or <strong>the</strong> <strong>end</strong> distance implies larger<br />
T-stub widths and <strong>the</strong>refore higher stiffness and resistance values but reduced<br />
deformation capacity (Figs. 5.17-5.18). As <strong>the</strong> T-stub width increases, <strong>the</strong> ef-<br />
149
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
Fig. 5.12 Principal strains ε11 in <strong>the</strong> T-stub flange (cont.).<br />
fective width also increases, since <strong>the</strong> beam pattern governs <strong>the</strong> plastic mechanisms.<br />
Therefore, <strong>the</strong> flexural resistance <strong>of</strong> <strong>the</strong> flanges is enhanced and so βRd<br />
assumes larger values. For a certain ratio λ, <strong>the</strong>n <strong>the</strong> starting governing plastic<br />
mechanism type-1 changes into type-2 and eventually into type-3, as beff grows.<br />
This transition <strong>of</strong> plastic modes is associated with <strong>the</strong> increase in resistance and<br />
initial stiffness and <strong>the</strong> reduction <strong>of</strong> deformation capacity. For all <strong>the</strong> analysed<br />
specimens, bolt determined collapse and <strong>the</strong> plastic mechanism was <strong>of</strong> type-1<br />
(flange yielding). For specimen WT4A, however, βRd was very close to <strong>the</strong><br />
boundary limit <strong>of</strong> type-2. This is ra<strong>the</strong>r evident in Fig. 5.19a that shows<br />
WT4Ab at collapse conditions. Apparently, <strong>the</strong> flange is in single b<strong>end</strong>ing curvature.<br />
Table 5.3 sets out <strong>the</strong> main characteristics <strong>of</strong> <strong>the</strong> above F-∆ responses and<br />
confirms <strong>the</strong> above statements concerning <strong>the</strong> major influences <strong>of</strong> <strong>the</strong> T-stub<br />
width on <strong>the</strong> overall behaviour <strong>of</strong> T-stubs. For better understanding, part <strong>of</strong> Table<br />
3.8 for <strong>the</strong> welded specimens is included here. The results in Table 5.3 are<br />
presented for a bolt row. This means that <strong>the</strong> previous experimental results for<br />
stiffness and resistance are divided by 2.<br />
With respect to <strong>the</strong> experiments, Fig. 5.18c compares <strong>the</strong> results for <strong>the</strong> two<br />
tests in this series with WT1h. The connection <strong>ductility</strong> clearly decreases.<br />
150
(i) Knee-range. (ii) Collapse.<br />
(c) Specimen Weld_T1(ii).<br />
Fig. 5.12 Principal strains ε11 in <strong>the</strong> T-stub flange (cont.).<br />
Parametric study<br />
When comparing specimens WT1g/h and WT4Aa/b, <strong>the</strong> reduction <strong>of</strong> ∆u.0 is, on<br />
average, 66%. Bolt determined collapse in all cases. In particular, for WT4Aa,<br />
only bolt RB did not fail and for specimen WT4Ab, for which <strong>the</strong>re is a record<br />
<strong>of</strong> <strong>the</strong> bolt elongation behaviour up to collapse (Fig. 5.20), <strong>the</strong> bolts on <strong>the</strong> left<br />
hand side were broken (Fig. 5.19 – <strong>the</strong> specimen is rotated in this figure). In<br />
fact, <strong>the</strong> graph from Fig. 5.20 shows that <strong>the</strong> bolts on <strong>the</strong> right hand side<br />
yielded smaller deformation than <strong>the</strong> o<strong>the</strong>rs. Fig. 5.19b shows that <strong>the</strong> bolts<br />
were highly deformed at collapse. In this figure, an unbroken bolt is shown and<br />
<strong>the</strong> combined b<strong>end</strong>ing and tension deformations are very clear. It should be<br />
stressed that <strong>the</strong> bolt measurement up to collapse has only been carried out in<br />
this specific specimen as an experiment. Unfortunately, it was observed that<br />
<strong>the</strong> measuring brackets were damaged in <strong>the</strong> <strong>end</strong> and <strong>the</strong>refore <strong>the</strong>y had to be<br />
replaced prior to collapse.<br />
5.3.3 Edge distance and flange thickness<br />
The variation <strong>of</strong> <strong>the</strong> edge distance e is analysed in Fig. 5.21 that depicts <strong>the</strong> F-<br />
∆ behaviour <strong>of</strong> specimens T1, P8, P9 and P11. For <strong>the</strong>se specimens not only<br />
151
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range. (ii) Collapse.<br />
(c) Specimen Weld_T1(ii).<br />
Fig. 5.13 Principal stresses σ11 in xy cross-section.<br />
<strong>the</strong> edge distance was varied, but also <strong>the</strong> flange thickness and <strong>the</strong> distance m<br />
were slightly different, as <strong>the</strong> beam pr<strong>of</strong>iles changed (P8: tf = 11.0 mm; P9-11: tf<br />
= 14.0 mm). To conclude about <strong>the</strong> single effect <strong>of</strong> e, only specimens P9 and<br />
P11 can be compared. Clearly, both stiffness and resistance are identical and<br />
<strong>the</strong> deformation capacity does not vary significantly ei<strong>the</strong>r. If e is bigger, ∆u.0 is<br />
152
(i) Knee-range. (ii) Collapse.<br />
(a) Specimen T1.<br />
(i) Knee-range. (ii) Collapse.<br />
(b) Specimen Weld_T1(i).<br />
(i) Knee-range. (ii) Collapse.<br />
(c) Specimen Weld_T1(ii).<br />
Fig. 5.14 Principal strains ε11 in xy cross-section.<br />
Parametric study<br />
somewhat improved. Fig. 5.22 depicts <strong>the</strong> influence <strong>of</strong> <strong>the</strong> flange thickness on<br />
<strong>the</strong> overall behaviour (<strong>the</strong> distance m also varies as <strong>the</strong> pr<strong>of</strong>ile changes). From<br />
<strong>the</strong> graphs it can be concluded that as <strong>the</strong> flange thickness decreases and all <strong>the</strong><br />
153
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
154<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
WT1h WT2Aa WT2Ab<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
(a) Series WT2A: smaller weld throat thickness.<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
WT1h WT2Ba WT2Bb<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
(b) Series WT2B: smaller weld throat thickness.<br />
Fig. 5.15 Experimental load-carrying behaviour <strong>of</strong> specimen series WT2 and<br />
comparison with WT1h.<br />
Total applied load (kN)<br />
280<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
0<br />
T1 P1 P2<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation (mm)<br />
Fig. 5.16 Influence <strong>of</strong> <strong>the</strong> gauge <strong>of</strong> <strong>the</strong> bolts on <strong>the</strong> overall behaviour.
Total applied load (kN)<br />
280<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
T1 P3 P4 P5<br />
0<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation (mm)<br />
(a) Pitch <strong>of</strong> <strong>the</strong> bolts, p.<br />
Total applied load (kN)<br />
280<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
0<br />
T1 P6 P7<br />
Parametric study<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation (mm)<br />
(b) End distance, e1.<br />
Fig. 5.17 Influence <strong>of</strong> <strong>the</strong> T-stub width on <strong>the</strong> overall behaviour: single effects<br />
<strong>of</strong> <strong>the</strong> pitch and <strong>end</strong> distance.<br />
Total applied load (kN)<br />
280<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
T1 P16<br />
0<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation (mm)<br />
(a) Numerical results for HR specimens.<br />
Fig. 5.18 Influence <strong>of</strong> <strong>the</strong> T-stub width on <strong>the</strong> overall behaviour: combined effects<br />
<strong>of</strong> <strong>the</strong> pitch and <strong>end</strong> distance.<br />
155
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
156<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Experimental results: WT4Aa<br />
60<br />
Experimental results: WT4Ab<br />
30<br />
0<br />
Numerical results LUSAS<br />
0.0 0.7 1.4 2.1 2.8 3.5 4.2 4.9 5.6 6.3 7.0<br />
Total deformation (mm)<br />
(b) Experimental and numerical load-carrying behaviour <strong>of</strong> WP specimen<br />
WT4A.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
WT1h WT4Aa WT4Ab<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
(c) Comparison <strong>of</strong> <strong>the</strong> responses <strong>of</strong> <strong>the</strong> original specimen WT1 and WT4A: experimental<br />
assessment.<br />
Total applied load (kN)<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
0<br />
WT1 WT4A<br />
0 2 4 6 8 10 12 14 16<br />
Deformation (mm)<br />
(d) Comparison <strong>of</strong> <strong>the</strong> responses <strong>of</strong> <strong>the</strong> original specimen WT1 and WT4A:<br />
numerical assessment.<br />
Fig. 5.18 Influence <strong>of</strong> <strong>the</strong> T-stub width on <strong>the</strong> overall behaviour: combined effects<br />
<strong>of</strong> <strong>the</strong> pitch and <strong>end</strong> distance (cont.).
Parametric study<br />
(a) Deformation at failure. (b) Detail <strong>of</strong> an unbroken bolt<br />
after failure <strong>of</strong> <strong>the</strong> connection.<br />
Fig. 5.19 Specimen WT4Ab at collapse conditions.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Experimental results: bolt LF<br />
60<br />
Experimental results: bolt LB<br />
30<br />
0<br />
Numerical results LUSAS<br />
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0<br />
Bolt deformation (mm)<br />
(a) Comparison <strong>of</strong> <strong>the</strong> numerical results with experimental evidence.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
Bolt RB Bolt LB<br />
30<br />
0<br />
Bolt LF Bolt RF<br />
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8<br />
Bolt elongation (mm)<br />
(b) Experimental results.<br />
Fig. 5.20 Bolt elongation behaviour for specimen WT4Ab.<br />
157
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 5.3 Syn<strong>the</strong>sis <strong>of</strong> <strong>the</strong> characteristic results (per bolt row) <strong>of</strong> <strong>the</strong> curves<br />
comparing <strong>the</strong> effect <strong>of</strong> <strong>the</strong> geometric parameters on <strong>the</strong> overall<br />
behaviour [underlined values correspond to experimental results].<br />
Test<br />
ID<br />
Stiffness (kN/mm)<br />
ke.0 kp-l.0 ke.0/kp-l.0<br />
Resistance (kN)<br />
K-R Fu.0<br />
∆u.0<br />
(mm)<br />
Q/F<br />
K-R Ult.<br />
T1 83.54 2.68 29.60 65-85 103.99 8.70 0.24 0.34<br />
P1 63.27 2.01 31.48 60-73 91.76 10.77 0.33 0.44<br />
P2 117.06 3.49 33.54 70-100 116.72 6.18 0.18 0.25<br />
P3 72.62 2.30 31.62 60-75 95.41 10.17 0.26 0.40<br />
P4 97.86 4.24 23.08 70-103 115.97 4.68 0.20 0.26<br />
P5 101.23 6.00 16.88 90-120 130.20 3.63 0.18 0.18<br />
P6 76.56 2.36 32.44 55-75 95.53 10.06 0.26 0.42<br />
P7 91.45 2.97 30.79 70-93 111.34 7.56 0.22 0.29<br />
P8 95.12 3.06 31.04 75-93 112.71 8.08 0.19 0.28<br />
P9 128.47 6.19 20.76 85-120 131.43 3.31 0.13 0.16<br />
P10 43.88 0.90 48.63 40-47 76.79 32.75 0.56 0.77<br />
P11 122.80 6.82 18.01 90-110 121.15 2.94 0.20 0.21<br />
P16 111.23 8.44 13.18 85-112 125.69 3.06 0.19 0.20<br />
WT1 69.29 1.57 44.24 55-76 94.98 14.20 0.27 0.37<br />
WT1g 68.58 2.11 32.50 58-68 91.33 14.10 ⎯ ⎯<br />
WT1h 73.58 2.07 35.55 60-70 92.50 14.55 ⎯ ⎯<br />
WT4A 85.95 3.84 22.38 67-99 107.95 5.18 0.22 0.27<br />
WT4Aa 75.08 2.75 27.30 59-105 108.20 5.35 ⎯ ⎯<br />
WT4Ab 86.96 4.37 19.90 70-98 103.26 4.33 ⎯ ⎯<br />
o<strong>the</strong>r parameters remain <strong>the</strong> same, <strong>the</strong> component <strong>ductility</strong> improves considerably,<br />
whilst stiffness and resistance decrease. In this case, <strong>the</strong> flange-bolt<br />
stiffness decreases and, consequently, <strong>the</strong> degree <strong>of</strong> plastic deformation in <strong>the</strong><br />
flange increases.<br />
5.4 INFLUENCE OF THE BOLT AND FLANGE STEEL GRADE<br />
Fig. 5.23 illustrates <strong>the</strong> influence <strong>of</strong> <strong>the</strong> bolt diameter on <strong>the</strong> overall behaviour<br />
<strong>of</strong> HR-T-stubs. Essentially, if <strong>the</strong> bolt diameter is bigger, <strong>the</strong> initial stiffness,<br />
<strong>the</strong> strength and <strong>the</strong> <strong>ductility</strong> improve greatly but <strong>the</strong> post-limit stiffness decreases.<br />
For a given geometry, <strong>the</strong> bolt ceases to be <strong>the</strong> determining factor <strong>of</strong><br />
collapse. The bolt-threaded length has an effect on <strong>the</strong> overall response if <strong>the</strong><br />
bolt governs <strong>the</strong> specimen collapse. In that case, if <strong>the</strong> threaded portion <strong>of</strong> <strong>the</strong><br />
bolt is longer, <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> whole connection increases. The<br />
remaining properties do not change much (Fig. 5.24). The effect <strong>of</strong> a bolt preloading<br />
is <strong>the</strong> enhancement <strong>of</strong> <strong>the</strong> initial stiffness (Fig. 5.25). The quantification<br />
<strong>of</strong> <strong>the</strong> observed behaviour variations is summarized in Table 5.4.<br />
158
Total applied load (kN)<br />
280<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
0<br />
T1 P8 P9 P11<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation (mm)<br />
Fig. 5.21 Influence <strong>of</strong> <strong>the</strong> edge distance on <strong>the</strong> overall behaviour.<br />
Total applied load (kN)<br />
280<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
0<br />
T1 P9 P10 P11<br />
0 3 6 9 12 15 18 21 24 27 30 33<br />
Deformation (mm)<br />
Fig. 5.22 Influence <strong>of</strong> <strong>the</strong> flange thickness on <strong>the</strong> overall behaviour.<br />
Parametric study<br />
Fig. 5.26 compares <strong>the</strong> behaviour <strong>of</strong> <strong>the</strong> specimens with larger bolts (M16<br />
and M20), confirming <strong>the</strong> previous considerations on <strong>the</strong> influence <strong>of</strong> <strong>the</strong> bolt<br />
diameter. This conclusion is also supported by experimental evidence (Fig.<br />
5.27). Surprisingly, if <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> connection is evaluated<br />
at <strong>the</strong> maximum load level, specimen WT7_M16 yields a higher value when<br />
compared to WT7_M20 (Fig. 5.27). However, <strong>the</strong> ductile branch after collapse<br />
starts is longer in <strong>the</strong> latter case. These conclusions can also be taken from Table<br />
5.4 that repeats part <strong>of</strong> <strong>the</strong> information contained in Table 3.8 for this test<br />
series. For illustration, Fig. 5.28 shows <strong>the</strong> specimens from <strong>the</strong> experiments at<br />
failure conditions.<br />
For specimen WT7_M20 whose failure mode is <strong>of</strong> type-11 (cracking <strong>of</strong> <strong>the</strong><br />
material at <strong>the</strong> HAZ – Fig. 5.28c), a numerical model was also implemented.<br />
159
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
160<br />
Total applied load (kN)<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
(a) Geometry from T1.<br />
Total applied load (kN)<br />
0<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
(b) Geometry from P2.<br />
Total applied load (kN)<br />
0<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
T1 P12<br />
0 3 6 9 12 15 18 21 24 27<br />
Deformation (mm)<br />
P2 P12 P15<br />
0 3 6 9 12 15 18 21 24 27<br />
Deformation (mm)<br />
P16 P17 P18<br />
0 3 6 9 12 15 18 21 24 27<br />
Deformation (mm)<br />
(c) Geometry from P16.<br />
Fig. 5.23 Influence <strong>of</strong> <strong>the</strong> bolt diameter on <strong>the</strong> overall behaviour: numerical<br />
results.
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
T1 P24<br />
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14<br />
Deformation (mm)<br />
Parametric study<br />
Fig. 5.24 Influence <strong>of</strong> <strong>the</strong> bolt threaded length on <strong>the</strong> overall behaviour.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
T1 P25<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation (mm)<br />
Fig. 5.25 Influence <strong>of</strong> a bolt preloading on <strong>the</strong> overall behaviour.<br />
Again, this model does not cater for <strong>the</strong> specific behaviour <strong>of</strong> <strong>the</strong> HAZ, as already<br />
mentioned in §4.8. The differences between <strong>the</strong> two F-∆ responses<br />
shown in Fig. 5.29 derive from this simplification. In particular, <strong>the</strong> failure<br />
<strong>ductility</strong> <strong>of</strong> <strong>the</strong> metal in this HAZ is clearly reduced. Experimentally, <strong>the</strong> deformation<br />
<strong>of</strong> <strong>the</strong> T-stub flange at maximum load is 9.12 mm, whilst numerically<br />
a total deformation <strong>of</strong> 25.37 mm is reached. However, since <strong>the</strong> s<strong>of</strong>tening<br />
branch is sufficiently large, this numerical value is comparable to <strong>the</strong> maximum<br />
deformation <strong>of</strong> 18.70 mm that was reached in <strong>the</strong> experiments. At this<br />
displacement level, <strong>the</strong> test was stopped to prevent damage <strong>of</strong> <strong>the</strong> equipment.<br />
The numerical deformation capacity was established by setting <strong>the</strong> maximum<br />
average principal strain, at <strong>the</strong> critical zone, ε11.av.f, as equal to <strong>the</strong> ultimate<br />
strain <strong>of</strong> <strong>the</strong> flange material, εu.f (cf. §4.5). In this specific case, since <strong>the</strong> critical<br />
section is located at <strong>the</strong> HAZ, <strong>the</strong> deformation <strong>of</strong> <strong>the</strong> flanges was also<br />
161
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
162<br />
Total applied load (kN)<br />
490<br />
420<br />
350<br />
280<br />
210<br />
140<br />
70<br />
(a) Specimens with bolt M16.<br />
Total applied load (kN)<br />
0<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
P17 P19 P20<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation (mm)<br />
P18 P21 P22 P23<br />
0 3 6 9 12 15 18 21 24 27<br />
Deformation (mm)<br />
(b) Specimens with bolt M20.<br />
Fig. 5.26 Influence <strong>of</strong> some geometric variations for bolts M16 and M20 and<br />
<strong>the</strong> corresponding geometries P17 and P18 (HR-T-stub series).<br />
Total applied load (kN)<br />
150<br />
125<br />
100<br />
75<br />
50<br />
25<br />
0<br />
WT7_M20 WT7_M16 WT7_M20<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Total deformation (mm)<br />
Fig. 5.27 Influence <strong>of</strong> <strong>the</strong> bolt diameter on <strong>the</strong> overall behaviour: experimental<br />
results (series WT7).
Parametric study<br />
(a) Spec. WT7_M12. (b) Spec. WT7_M16. (c) Spec. WT7_M20.<br />
Fig. 5.28 Deformation <strong>of</strong> specimens WT7 at failure.<br />
Table 5.4 Syn<strong>the</strong>sis <strong>of</strong> <strong>the</strong> characteristic results (per bolt row) <strong>of</strong> <strong>the</strong> curves<br />
comparing <strong>the</strong> effect <strong>of</strong> <strong>the</strong> bolt on <strong>the</strong> overall behaviour [underlined<br />
values correspond to experimental results].<br />
Test ID Stiffness (kN/mm)<br />
ke.0 kpl.0 ke.0/kpl.0<br />
Strength (kN)<br />
K-R Fu<br />
∆u<br />
(mm)<br />
Q/F<br />
K-R Ult.<br />
T1 83.54 2.68 29.60 65-85 103.99 8.70 0.24 0.34<br />
P12 102.05 2.19 46.68 80-103 154.06 24.22 0.29 0.57<br />
P24 80.46 1.89 42.51 65-87 108.14 13.80 0.24 0.31<br />
P25 127.66 2.59 49.29 65-85 104.26 8.72 0.32 0.34<br />
P2 117.06 3.49 33.54 70-100 116.72 6.18 0.18 0.25<br />
P12 102.05 2.19 46.68 80-103 154.06 24.22 0.29 0.57<br />
P15 128.41 2.71 47.47 80-120 171.08 18.02 0.21 0.48<br />
P16 111.23 8.44 13.18 85-112 125.69 3.06 0.19 0.20<br />
P17 138.31 3.86 35.81 115-165 192.01 9.29 0.22 0.36<br />
P18 171.57 2.56 66.96 150-200 266.57 26.07 0.27 0.44<br />
P19 127.27 3.93 32.36 115-160 186.52 9.29 0.22 0.38<br />
P20 181.68 9.73 18.67 160-200 225.94 4.06 0.18 0.23<br />
P21 168.61 3.22 52.38 155-230 281.33 17.67 0.23 0.42<br />
P22 250.69 7.08 32.07 190-270 305.21 6.40 0.21 0.34<br />
P23 322.09 9.61 33.53 275-305 346.01 5.22 0.20 0.24<br />
WT4Aa 75.08 2.75 27.30 59-105 108.20 5.35 ⎯ ⎯<br />
WT4Ab 86.96 4.37 19.90 70-98 103.26 4.33 ⎯ ⎯<br />
WT7_M12 91.18 3.78 24.12 60-96 100.64 3.86 ⎯ ⎯<br />
WT7_M16 116.09 5.08 17.54 80-104 132.34 5.88 ⎯ ⎯<br />
WT7_M20 142.80 2.86 49.93 90-131 177.53 25.37 0.30 0.45<br />
WT7_M20 137.70 5.61 16.33 88-118 145.72 15.98 ⎯ ⎯<br />
evaluated for o<strong>the</strong>r flange strain levels, as indicated in Fig. 5.29 (e.g.: for an<br />
ε11.av.f = 0.15, corresponding to 0.5εu.f, ∆ = 12.21 mm).<br />
Ano<strong>the</strong>r comparison that can be performed with <strong>the</strong> experimental test series<br />
163
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
164<br />
Total applied load (kN)<br />
180<br />
150<br />
120<br />
90<br />
Numerical results<br />
60<br />
Experimental results<br />
E11.p = 0.15 (def. = 12.21 mm)<br />
30<br />
0<br />
E11.p = 0.20 (def. = 16.60 mm)<br />
E11.p = 0.25 (def. = 20.98 mm)<br />
0 3 6 9 12 15 18 21 24 27<br />
Deformation (mm)<br />
Fig. 5.29 Global response <strong>of</strong> specimen WT7_M20: numerical and experimental<br />
results.<br />
Applied load per bolt row (kN)<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
WT4Aa WT4Ab WT7_M12<br />
0 1 2 3 4 5 6 7<br />
Deformation (mm)<br />
Fig. 5.30 Experimental load-carrying behaviour <strong>of</strong> specimen WT7_M12 and<br />
comparison with series WT4Aa (per bolt row): assessment <strong>of</strong> <strong>the</strong> influence<br />
<strong>of</strong> number <strong>of</strong> bolt rows for identical geometries.<br />
WT7 (specimen WT7_M12, more specifically) and series WT4A relates to <strong>the</strong><br />
influence <strong>of</strong> <strong>the</strong> number <strong>of</strong> bolts fastening <strong>the</strong> T-stub elements. Fig. 5.30 compares<br />
<strong>the</strong> F-∆ response, per bolt row, for <strong>the</strong> three specimens, and shows a<br />
good agreement. This means that <strong>the</strong> symmetric behaviour is valid. This graph<br />
also shows that for specimen WT7_M12 at a load level <strong>of</strong> 58 kN some slippage<br />
occurred, resulting in a sharp decrease <strong>of</strong> stiffness in <strong>the</strong> response. Identical<br />
situation is observed in WT4Aa.<br />
Regarding <strong>the</strong> effect <strong>of</strong> <strong>the</strong> flange steel grade, Fig. 5.31 shows that <strong>the</strong> initial<br />
stiffness is not affected by <strong>the</strong> steel properties (as long as <strong>the</strong> Young
Total applied load (kN)<br />
240<br />
200<br />
160<br />
120<br />
80<br />
40<br />
(a) Numerical results.<br />
Total applied load (kN)<br />
0<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
T1 P13 P14<br />
0 3 6 9 12 15 18 21 24 27<br />
Deformation (mm)<br />
30<br />
0<br />
WT1h WT51a WT51b<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
(b) Experimental results.<br />
Fig. 5.31 Influence <strong>of</strong> <strong>the</strong> flange steel grade on <strong>the</strong> overall behaviour.<br />
Parametric study<br />
modulus is constant) but as <strong>the</strong> yield stress <strong>of</strong> <strong>the</strong> flange, fy.f increases <strong>the</strong> resistance<br />
and <strong>the</strong> post-limit stiffness also increase and <strong>the</strong> deformation capacity<br />
decreases. Table 5.5 confirms <strong>the</strong>se conclusions.<br />
The FE models <strong>of</strong> P13 and P14 were obtained from <strong>the</strong> original specimen<br />
T1 by reducing <strong>the</strong> stress values <strong>of</strong> <strong>the</strong> flange mechanical properties and maintaining<br />
<strong>the</strong> strain ordinates. Both new specimens exhibit a type-1 plastic failure<br />
mechanism. The flexural resistance <strong>of</strong> <strong>the</strong> flanges increases with <strong>the</strong> flange yield<br />
stress and so βRd becomes greater. This explains <strong>the</strong> improvement in <strong>the</strong> resistance<br />
properties despite a reduction in <strong>the</strong> deformation capacity. In <strong>the</strong> above case,<br />
specimen P14, whose flange is steel grade S275, is typified by a failure type-<br />
11, i.e. cracking <strong>of</strong> <strong>the</strong> flange material is <strong>the</strong> determining factor <strong>of</strong> collapse. For<br />
<strong>the</strong> o<strong>the</strong>r two specimens, bolt failure governs <strong>the</strong> ultimate conditions.<br />
Test series WT51 comprises <strong>the</strong> testing <strong>of</strong> two specimens geometrically<br />
165
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 5.5 Syn<strong>the</strong>sis <strong>of</strong> <strong>the</strong> characteristic numerical results (per bolt row) <strong>of</strong><br />
<strong>the</strong> curves comparing <strong>the</strong> effect <strong>of</strong> <strong>the</strong> flange steel grade.<br />
Test ID<br />
166<br />
Stiffness (kN/mm) Strength (kN) ∆u Q/F<br />
ke.0 kpl.0 ke.0/kpl.0 K-R Fu (mm) K-R Ult.<br />
T1 83.54 2.68 29.60 65-85 103.99 8.70 0.24 0.34<br />
P13 81.31 2.19 37.17 55-72 93.71 11.38 0.24 0.42<br />
P14 81.97 1.07 76.69 45-65 86.57 24.15 0.24 0.53<br />
identical to <strong>the</strong> original test series WT1 and whose T-stub elements are made<br />
up <strong>of</strong> high-strength steel S690. According to Eurocode 3, <strong>the</strong>se specimens exhibit<br />
a type-2 plastic mechanism. Bolt governs collapse in <strong>the</strong> three cases and<br />
<strong>the</strong> deformation capacity is far reduced in series WT51 because <strong>the</strong> bolts are<br />
engaged in collapse <strong>of</strong> <strong>the</strong> specimen at an earlier stage. The knee-range <strong>of</strong> <strong>the</strong><br />
F-∆ response <strong>of</strong> specimens WT51 develops for higher loads in comparison to<br />
WT1h. The slope <strong>of</strong> <strong>the</strong> post-limit part <strong>of</strong> <strong>the</strong>se curves is lower than in <strong>the</strong><br />
original case. Here, <strong>the</strong> single curvature <strong>of</strong> <strong>the</strong> flange is evident (Fig. 5.32) and<br />
<strong>the</strong> deformation <strong>of</strong> <strong>the</strong> flanges is far less than in series WT1 (see Fig. 3.11, for<br />
instance). This is also clear in Fig. 5.33a where <strong>the</strong> strains for WT1h and<br />
WT51b are compared for SG3, on <strong>the</strong> same location in both specimens (Figs.<br />
3.8a-b). Fig. 5.33b plots <strong>the</strong> force-strain results for <strong>the</strong> two T-rosettes attached<br />
to specimen WT51b. It shows that <strong>the</strong> flange strain level <strong>the</strong>re at collapse is<br />
ra<strong>the</strong>r low. Symmetry <strong>of</strong> results is also ra<strong>the</strong>r obvious.<br />
Now consider test series WT53 to assess <strong>the</strong> influence <strong>of</strong> <strong>the</strong> bolt type on<br />
<strong>the</strong> overall response. Naturally, since <strong>the</strong> actual bolt properties also vary (Table<br />
3.2), <strong>the</strong> global results will include not only <strong>the</strong> effect <strong>of</strong> <strong>the</strong> bolt type (short- or<br />
full-threaded) but also <strong>the</strong>ir mechanical properties. Fig. 5.34 depicts <strong>the</strong> F-∆ response<br />
<strong>of</strong> identical T-stub elements connected by means <strong>of</strong> <strong>the</strong> four different<br />
M12 bolt types tested (cf. §3.2.2.1). The graphs show that if higher strength<br />
bolts are used (WT53D/E), since bolt determines failure in <strong>the</strong> four cases, <strong>the</strong><br />
maximum load reached is also higher (see Table 3.8). For <strong>the</strong> four specimens<br />
compared in this figure, <strong>the</strong> initial stiffness is identical because <strong>the</strong> Young<br />
modulus, which is one <strong>of</strong> <strong>the</strong> main parameters used in <strong>the</strong> computation <strong>of</strong> ke.0,<br />
is identical for <strong>the</strong> four bolt types (Table 3.2).<br />
If bolt governs <strong>the</strong> failure mode <strong>of</strong> <strong>the</strong> T-stub, <strong>the</strong> overall deformation capacity<br />
mainly dep<strong>end</strong>s on <strong>the</strong> maximum elongation <strong>of</strong> <strong>the</strong> bolt, or, in o<strong>the</strong>r<br />
words, on <strong>the</strong> ultimate strain values. Table 3.2 shows that full-threaded bolts<br />
exhibit higher values <strong>of</strong> εu (though for bolt grade 10.9 that difference is<br />
smaller) and higher bolt grades exhibit smaller deformations, i.e. <strong>the</strong> failure<br />
type is more brittle. When taking into account <strong>the</strong> T-stubs WT51b and<br />
WT53C/D/E, <strong>the</strong> above considerations are still valid. Specimen WT53C is<br />
more ductile than <strong>the</strong> remaining since <strong>the</strong> fasteners are full-threaded M12 grade<br />
8.8, even though <strong>the</strong> deformation level at Fmax is lower (Table 3.8). For this<br />
specimen, <strong>the</strong> <strong>plate</strong>au that follows Fmax is far longer than in <strong>the</strong> o<strong>the</strong>r cases
Parametric study<br />
(a) Deformation at failure (WT51b). (b) Detail <strong>of</strong> a broken bolt (WT51a).<br />
Fig. 5.32 Deformation <strong>of</strong> specimens WT51 at failure.<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
WT1h WT51b<br />
0 4000 8000 12000 16000 20000 24000 28000 32000<br />
Strain (µm/m)<br />
(a) Comparison <strong>of</strong> <strong>the</strong> results for SG3 in specimens WT1h and WT51b.<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
SG6x<br />
SG7x<br />
SG6z<br />
SG7z<br />
-4000 -3200 -2400 -1600 -800 0 800 1600 2400<br />
Strain (µm/m)<br />
(b) Results for <strong>the</strong> rosettes.<br />
Fig. 5.33 Experimental results for <strong>the</strong> flange strain behaviour (specimen<br />
WT51b).<br />
167
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
168<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
WT51b WT53C<br />
30<br />
0<br />
WT53D WT53E<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation (mm)<br />
Fig. 5.34 Experimental load-carrying behaviour <strong>of</strong> specimen series WT53 and<br />
comparison with WT51b.<br />
Fig. 5.35 Comparison <strong>of</strong> <strong>the</strong> deformation <strong>of</strong> specimens WT51, WT53C,<br />
WT53D and WT53E (from left to right) at failure.<br />
(Fig. 5.34). Surprisingly, <strong>the</strong> deformation capacity for both tests WT53D/E that<br />
use bolt grade 10.9 is identical. These conclusions are also indicated in Table<br />
3.8. Fig. 5.35 illustrates <strong>the</strong> four specimens after failure.<br />
Having analysed <strong>the</strong> influence <strong>of</strong> <strong>the</strong> steel grade on <strong>the</strong> overall T-stub behaviour<br />
(mainly: increase <strong>of</strong> strength and decrease <strong>of</strong> <strong>ductility</strong> for higher<br />
strength steel grades), <strong>the</strong> response obtained for series WT57 and WT7 can be<br />
compared. In series WT57, when using bolts M12 and M16, <strong>the</strong> plastic resistance<br />
<strong>of</strong> <strong>the</strong> specimens, as determined according to Eurocode 3 (Table 3.7) corresponds<br />
to that <strong>of</strong> a plastic mechanism type-2 whilst for M20 it corresponds to<br />
a type-1 plastic mechanism. This is evident in <strong>the</strong> graphs from Fig. 5.36 where<br />
<strong>the</strong> responses <strong>of</strong> <strong>the</strong> three specimens are shown. For comparison, WT7_M20 is<br />
also included. It is worth mentioning that <strong>the</strong> bolt is also engaged in collapse in<br />
<strong>the</strong> case <strong>of</strong> <strong>the</strong> high-strength steel (WT57_M20) since <strong>the</strong> specimen fails in a<br />
combined failure mode (type-13), whilst in WT7_M20 collapse is governed by<br />
<strong>plate</strong> cracking near <strong>the</strong> weld toe only. Basically, <strong>the</strong> conclusions drawn above<br />
are supported with this series <strong>of</strong> experiments (summary in Table 3.8).
Total applied load (kN)<br />
270<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
WT7_M20 WT57_M12<br />
30<br />
0<br />
WT57_M16 WT57_M20<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Parametric study<br />
Deformation (mm)<br />
Fig. 5.36 Experimental load-carrying behaviour <strong>of</strong> specimen series WT57 and<br />
comparison with WT7_M20.<br />
5.5 EXPERIMENTAL RESULTS FOR THE STIFFENED TEST SPECIMENS AND<br />
THE ROTATED CONFIGURATIONS<br />
The experimental programme included <strong>the</strong> test <strong>of</strong> some transversely stiffened<br />
specimens and T-stub connections with <strong>the</strong> elements orientated at right angles,<br />
in order to simulate <strong>the</strong> actual behaviour in tension <strong>of</strong> <strong>the</strong> components modelling<br />
<strong>the</strong> <strong>end</strong> <strong>plate</strong> side. The results obtained for those cases are discussed in this<br />
section. For complete description <strong>of</strong> <strong>the</strong> specimens and <strong>the</strong> characteristic results<br />
<strong>of</strong> <strong>the</strong> load-carrying behaviour, <strong>the</strong> reader should refer to Chapter 3.<br />
5.5.1 Influence <strong>of</strong> a transverse stiffener<br />
If a transverse stiffener is added to a T-stub connection, stiffness and resistance<br />
properties improve and deformation capacity decreases. To support this statement,<br />
first consider series WT61 that is obtained from <strong>the</strong> original WT1 by including<br />
a transverse stiffener in order to simulate <strong>the</strong> T-stub model for <strong>the</strong> <strong>end</strong><br />
<strong>plate</strong> side (Fig. 1.8c). The load-carrying behaviour <strong>of</strong> <strong>the</strong> two specimens included<br />
in this series is compared with specimen WT1h and <strong>the</strong> code predictions<br />
[5.1] in Fig. 5.37 and Tables 3.7-3.8. The collapse <strong>of</strong> <strong>the</strong> specimens is determined<br />
by bolt fracture at <strong>the</strong> stiffener side (Fig. 3.16c) – labelled “left side”.<br />
Fig. 5.38 plots <strong>the</strong> bolt elongation behaviour against <strong>the</strong> overall deformation<br />
for specimens WT61a and WT1h. Whilst for WT61a <strong>the</strong> record <strong>of</strong> <strong>the</strong> bolt<br />
elongation was carried out nearly until collapse, for WT1h, <strong>the</strong> measuring<br />
brackets were removed at an earlier stage. Therefore, <strong>the</strong> loss <strong>of</strong> stiffness in<br />
this response, which is evident for WT61a, is not plotted in <strong>the</strong> graph. This loss<br />
<strong>of</strong> stiffness does not occur for <strong>the</strong> unbroken bolt RB (unstiffened side), <strong>of</strong><br />
which <strong>the</strong> response is very close to <strong>the</strong> bolts from WT1h.<br />
169
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
170<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
EC3: Init.<br />
stiffness<br />
30<br />
0<br />
WT1h WT61a WT61b<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation (mm)<br />
EC3: Plastic resistance<br />
Fig. 5.37 Experimental load-carrying behaviour <strong>of</strong> specimen series WT61 and<br />
comparison with WT1h and Eurocode 3 predictions.<br />
Deformation (mm)<br />
24<br />
21<br />
18<br />
15<br />
12<br />
9<br />
6<br />
3<br />
Bolt RB (WT61a) Bolt LF (WT61a)<br />
Bolt RB (WT1h) Bolt LF (WT1h)<br />
WT1h: max. deformation<br />
WT61a: max. deformation<br />
0<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2<br />
Bolt elongation (mm)<br />
Fig. 5.38 Comparison <strong>of</strong> <strong>the</strong> overall deformation-bolt elongation response for<br />
bolts LF and RB in specimens WT1h and WT61a.<br />
Now consider <strong>the</strong> stiffened specimen WT64C that derives from series<br />
WT4A by inclusion <strong>of</strong> <strong>the</strong> stiffeners. The above conclusions are not so obvious<br />
in this case. Fig. 5.39 and Table 3.8 show that both <strong>the</strong> initial and <strong>the</strong> post-limit<br />
stiffness values are identical for <strong>the</strong> two series. Never<strong>the</strong>less, resistance is still<br />
higher in <strong>the</strong> stiffened case. With respect to <strong>the</strong> <strong>ductility</strong> properties, if <strong>the</strong> absolute<br />
maximum deformation is taken into account, <strong>the</strong>n WT64C shows improved<br />
<strong>ductility</strong>. If <strong>the</strong> deformation capacity is assumed as <strong>the</strong> level corresponding<br />
to Fmax instead, <strong>the</strong> same conclusion applies. For specimen WT64C<br />
some strain results are given in Fig. 5.40 (see Fig. 3.8d for an indication <strong>of</strong> <strong>the</strong><br />
strain gauges nomenclature) and <strong>the</strong>y prove that <strong>the</strong> flange is not engaged in<br />
collapse, as <strong>the</strong> strain level is low at failure conditions. Fig. 5.41 compares <strong>the</strong><br />
strains at equivalent strain gauges in WT4A and WT64C. At <strong>the</strong> bolt axis, <strong>the</strong>
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
WT4Ab WT64C<br />
0 1 2 3 4 5 6 7 8<br />
Parametric study<br />
Deformation (mm)<br />
Fig. 5.39 Experimental load-carrying behaviour <strong>of</strong> specimen series WT64C<br />
and comparison with WT4Ab.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
SG1 SG6 SG7<br />
0 4000 8000 12000 16000 20000 24000 28000 32000<br />
Strain (µm/m)<br />
(a) Strain gauges SG1, SG6 and SG7.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Limit <strong>of</strong> <strong>the</strong> strain gauges<br />
60<br />
30<br />
0<br />
SG2<br />
SG4<br />
SG3<br />
SG5<br />
0 4000 8000 12000 16000 20000 24000 28000 32000<br />
Strain (µm/m)<br />
(b) Strain gauges SG2, SG3, SG4 and SG5.<br />
Fig. 5.40 Experimental results for flange behaviour (specimen WT64C).<br />
Limit <strong>of</strong> <strong>the</strong> strain gauges<br />
171
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
172<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
SG1 SG2<br />
0 600 1200 1800 2400 3000 3600 4200 4800 5400 6000<br />
Strain (µm/m)<br />
(c) Strain gauges SG1 and SG2.<br />
Fig. 5.40 Experimental results for flange behaviour (specimen WT64C) (cont.).<br />
Deformation (mm)<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
SG2 (WT64C)<br />
1<br />
0<br />
SG4 (WT64C)<br />
SG6z (WT4Aa)<br />
0 1200 2400 3600 4800 6000 7200 8400 9600<br />
Strain (µm/m)<br />
(a) Strain gauges SG2 and SG2 from WT64C and SG6 from WT4Aa.<br />
Deformation (mm)<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
SG1 (WT64C)<br />
1<br />
0<br />
SG7 (WT64C)<br />
SG2 (WT4Ab)<br />
0 1500 3000 4500 6000 7500 9000 10500 12000 13500<br />
Strain (µm/m)<br />
(b) Strain gauges SG1 and SG7 from WT64C and SG2 from WT4Aa.<br />
Fig. 5.41 Experimental results for <strong>the</strong> flange behaviour: comparison between<br />
specimens WT64C and WT4Ab.
Parametric study<br />
strain level is higher in WT64C (Fig. 5.41a), whilst at <strong>the</strong> weld toe <strong>the</strong> strains<br />
are higher in WT4Aa (Fig. 5.41b).<br />
Finally, series WT64A is identical to WT64C but only one <strong>of</strong> <strong>the</strong> T-stub<br />
elements is stiffened. The results for both specimens are analogous (Fig. 5.42,<br />
Table 3.8). The deformation behaviour is illustrated at two different load stages<br />
in Fig. 5.43.<br />
The main effect <strong>of</strong> <strong>the</strong> transverse stiffness is in fact <strong>the</strong> increase <strong>of</strong> stiffness<br />
and resistance and decrease <strong>of</strong> <strong>ductility</strong> <strong>of</strong> <strong>the</strong> connection. The two stiffened<br />
specimen series also indicate that a trilinear curve best fits <strong>the</strong> experiments<br />
ra<strong>the</strong>r than a bilinear approximation as suggested for <strong>the</strong> o<strong>the</strong>r cases.<br />
A final remark concerns <strong>the</strong> evaluation <strong>of</strong> ke.0 and FRd for <strong>the</strong>se specimens,<br />
according to Eurocode 3. A simplification has been introduced: both properties<br />
are evaluated for full stiffened and unstiffened specimens and <strong>the</strong>n <strong>the</strong> average<br />
value is taken (Table 3.7).<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
WT64A WT64C<br />
0 1 2 3 4 5 6 7 8<br />
(a) Average gap (LVDTs HP1 and HP2).<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
Deformation (mm)<br />
30<br />
0<br />
WT64A WT64C<br />
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0<br />
Deformation measured by HP3 (mm)<br />
(b) LVDT HP3.<br />
Fig. 5.42 Experimental load-carrying behaviour <strong>of</strong> specimen series WT64A<br />
and comparison with WT64C.<br />
173
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
(i) WT64A (F = 214 kN; ∆ = 4.46<br />
mm).<br />
(a) After removal <strong>of</strong> <strong>the</strong> measuring brackets.<br />
174<br />
(ii) WT64C (F = 231 kN; ∆ = 4.49<br />
mm).<br />
(i) WT64A. (ii) WT64C.<br />
(b) After failure.<br />
Fig. 5.43 Deformation <strong>of</strong> <strong>the</strong> specimens WT64A and WT64C at two different<br />
load stages.<br />
5.5.2 Influence <strong>of</strong> <strong>the</strong> T-stub orientation<br />
To assess <strong>the</strong> influence <strong>of</strong> <strong>the</strong> T-stub orientation, consider series WT4B and<br />
WT64B. They are identical to specimens WT4A and WT64A, respectively, by<br />
rotating 90º one <strong>of</strong> <strong>the</strong> T-stubs. In <strong>the</strong>se tests <strong>the</strong> flanges are bent in two directions<br />
with double curvature in <strong>the</strong> plan (x and z directions) – Figs. 3.16d and<br />
5.44. Contrary to <strong>the</strong> previous tests, here <strong>the</strong>re is no gap between <strong>the</strong> flanges,<br />
except at <strong>the</strong> stiffener side in WT64B (Fig. 3.16d). For WT4B, both flanges are<br />
bent as a whole until <strong>the</strong> bolt starts deforming excessively. At this time, at <strong>the</strong><br />
bolt centrelines, <strong>the</strong> flanges start opening and <strong>the</strong> maximum deformation at <strong>the</strong><br />
web is nearly equal to <strong>the</strong> bolt deformation capacity. Fig. 5.45 illustrates <strong>the</strong> F-<br />
∆ response. It shows results for HP1, HP2 and HP3. HP1 (at <strong>the</strong> back, from <strong>the</strong><br />
eye position) shows that <strong>the</strong> two <strong>plate</strong>s on this side are compressed and <strong>the</strong>ir<br />
displacement is negative. HP2 and HP3, located at <strong>the</strong> front and left sides, respectively,<br />
show that <strong>the</strong> <strong>plate</strong>s are compressed until a certain load level is<br />
reached, but <strong>the</strong>n <strong>the</strong>y start “opening” and <strong>the</strong>re is an inversion <strong>of</strong> <strong>the</strong> deforma-
Parametric study<br />
tion. That inversion starts at a lower load level for HP3 and becomes positive<br />
closer to <strong>the</strong> maximum load. For comparison, Fig. 5.45 also plots <strong>the</strong> deformation<br />
<strong>of</strong> WT4B against <strong>the</strong> average gap <strong>of</strong> specimen WT4Ab. Clearly, no resemblance<br />
between results is observed. The maximum load for specimen<br />
WT4B (223.67 kN) is close to <strong>the</strong> maximum load <strong>of</strong> WT4Aa, but a bit higher<br />
though.<br />
For specimen WT64B similar conclusions are drawn (Figs. 5.46a-b) except<br />
at <strong>the</strong> stiffener side where <strong>the</strong> two flange <strong>plate</strong>s “open” from <strong>the</strong> commencement<br />
<strong>of</strong> loading.<br />
The results for LVDTs HP2 and HP3 for both specimens WT4B and<br />
WT64B are compared in Fig. 5.46c. They are identical apart from <strong>the</strong> influence<br />
<strong>of</strong> <strong>the</strong> stiffener. The F-∆ response, as given by HP2, for WT64Bb and WT64A<br />
is compared in Fig. 5.46d. A similar behaviour is observed.<br />
It should be noted though that some perturbations might have occurred in<br />
<strong>the</strong> measurement by means <strong>of</strong> <strong>the</strong> LVDTs in <strong>the</strong>se series since <strong>the</strong> devices are<br />
not so easily attached here.<br />
Fig. 5.44 Deformation <strong>of</strong> specimen WT4B at failure (two different views).<br />
5.6 SUMMARY OF THE PARAMETRIC STUDY AND CONCLUDING REMARKS<br />
The experimental/numerical investigation presented in this chapter provides<br />
accurate deformation predictions (up to failure) <strong>of</strong> <strong>the</strong> T-stub response. Particular<br />
emphasis on <strong>the</strong> identification <strong>of</strong> <strong>the</strong> main parameters affecting <strong>the</strong> deformation<br />
capacity <strong>of</strong> <strong>bolted</strong> T-stubs has been given. Their influence on <strong>the</strong> overall<br />
behaviour <strong>of</strong> <strong>the</strong> connection has been assessed both qualitatively and quantitatively.<br />
The main conclusions drawn from this study are listed below and summarized<br />
in Table 5.6:<br />
1. The enlargement <strong>of</strong> <strong>the</strong> weld throat thickness improves stiffness and resistance<br />
but decreases <strong>the</strong> deformation capacity;<br />
2. The effect <strong>of</strong> <strong>the</strong> width <strong>of</strong> <strong>the</strong> T-stub is identical to <strong>the</strong> above;<br />
3. The increase <strong>of</strong> <strong>the</strong> distance m leads to lower stiffness and resistance values<br />
and improves <strong>the</strong> deformation capacity;<br />
4. Long-threaded bolts increase <strong>the</strong> overall deformation capacity <strong>of</strong> a T-stub<br />
175
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
176<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
WT4B (HP1)<br />
60<br />
WT4B (HP2)<br />
30<br />
0<br />
WT4B (HP3)<br />
WTAb (av.def.)<br />
-2.1 -1.4 -0.7 0.0 0.7 1.4 2.1 2.8 3.5 4.2 4.9 5.6<br />
Deformation (mm)<br />
Fig. 5.45 Experimental load-carrying behaviour <strong>of</strong> specimen series WT4B and<br />
comparison with WT4A.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
WT64Ba (HP1)<br />
60<br />
WT64Ba (HP2)<br />
30<br />
0<br />
WT64Bb (HP1)<br />
WT64Bb (HP2)<br />
-2.1 -1.4 -0.7 0.0 0.7 1.4 2.1 2.8 3.5 4.2 4.9 5.6<br />
Deformation (mm)<br />
(a) Results measured by LVDTs HP1 and HP2 for <strong>the</strong> two tests WT64B.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
HP3 HP4<br />
-2.1 -1.4 -0.7 0.0 0.7 1.4 2.1 2.8 3.5 4.2 4.9 5.6<br />
Deformation (mm)<br />
HP1 HP2<br />
(b) Results measured by <strong>the</strong> four LVDTs for test WT64Bb.<br />
Fig. 5.46 Experimental load-carrying behaviour <strong>of</strong> specimen series WT64B<br />
and comparison with o<strong>the</strong>r test series.
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
WT64Bb (HP2)<br />
60<br />
WT64Bb (HP3)<br />
30<br />
0<br />
WT4B (HP2)<br />
WT4B (HP3)<br />
-2.1 -1.4 -0.7 0.0 0.7 1.4 2.1 2.8 3.5 4.2 4.9 5.6<br />
Deformation (mm)<br />
Parametric study<br />
(c) Comparison <strong>of</strong> <strong>the</strong> results from WT64Bb with WT4B (measured with<br />
LVDT HP2 and HP3).<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
0.0 0.6 1.2 1.8 2.4 3.0 3.6 4.2 4.8 5.4 6.0<br />
Deformation (mm)<br />
WT64Bb (HP2)<br />
WT64A (HP2)<br />
(d) Comparison <strong>of</strong> <strong>the</strong> results from WT64Bb with WT64A (measured with<br />
LVDT HP2).<br />
Fig. 5.46 Experimental load-carrying behaviour <strong>of</strong> specimen series WT64B<br />
and comparison with o<strong>the</strong>r test series (cont.).<br />
connection when compared to short-threaded equivalent bolts, if collapse is<br />
governed by bolt fracture;<br />
5. Higher bolt diameters increase <strong>the</strong> strength <strong>of</strong> <strong>the</strong> bolt and <strong>the</strong>refore enhance<br />
<strong>the</strong> three characteristic properties <strong>of</strong> <strong>the</strong> load-carrying behaviour <strong>of</strong> <strong>the</strong> connection:<br />
resistance, stiffness and <strong>ductility</strong>;<br />
6. Identical T-stubs yield higher resistance and lower deformation capacity for<br />
higher steel grades.<br />
Regarding <strong>the</strong> influence <strong>of</strong> <strong>the</strong> stiffener, its main effect is <strong>the</strong> decrease <strong>of</strong><br />
<strong>the</strong> deformation capacity (note that, for <strong>the</strong> stiffened specimens, a trilinear approximation<br />
for simplified calculations best fits <strong>the</strong> experimental results ra<strong>the</strong>r<br />
177
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
than <strong>the</strong> classical bilinear approximation. Moreover, for stiffened T-stubs, <strong>the</strong><br />
influence <strong>of</strong> <strong>the</strong> elements orientation is not relevant at <strong>the</strong> stiffener side; in <strong>the</strong><br />
case <strong>of</strong> unstiffened specimens it has been shown that <strong>the</strong> two <strong>plate</strong>s become in<br />
contact when <strong>the</strong> connection is subjected to tension.<br />
Table 5.6 Summary <strong>of</strong> <strong>the</strong> main conclusions drawn from <strong>the</strong> parametric<br />
study [Notation: x↑ ⇒ y↑ means that if x increases <strong>the</strong>n y also increases;<br />
similarly, x↑ ⇒ y↓ means that if x increases <strong>the</strong>n y decreases].<br />
Strength Stiffness Ductility<br />
FRd Fu ke.0<br />
Assembly type<br />
kpl.0 ∆u<br />
WP Rd F ⇒ ↑ WP u F ⇒ ↑ WP e.0<br />
k ⇒ ↑ WP ⇒kpl.0 ↑ WP ⇒∆u ↓<br />
Throat thickness (WP-T-stubs only)<br />
aw aw ↑⇒ FRd<br />
↑ aw ↑⇒ Fu<br />
↑ aw ↑⇒ke.0 ↑ aw ↑⇒kpl.0 ↑ aw ↑⇒∆u ↓<br />
Connection geometry<br />
w w FRd<br />
p p FRd<br />
e1<br />
tf<br />
↑⇒ ↓<br />
↑⇒ ↑<br />
e1↑⇒ FRd<br />
↑<br />
tf ↑⇒ FRd<br />
↑<br />
w↑⇒ Fu↓<br />
p ↑⇒ Fu<br />
↑<br />
e1↑⇒ Fu↑<br />
tf ↑⇒ Fu<br />
↑<br />
w↑⇒ ke.0<br />
↓<br />
p↑⇒ke.0 ↑<br />
e1 ↑⇒ke.0 ↑<br />
tf ↑⇒ke.0 ↑<br />
w↑⇒kpl.0 ↓<br />
p↑⇒kpl.0 ↑<br />
e1 ↑⇒kpl.0 ↑<br />
tf ↑⇒ kpl.0<br />
↑<br />
w ↑⇒∆u ↑<br />
p ↑⇒∆u ↓<br />
e1 ↑⇒∆u ↓<br />
t f ↑⇒∆u ↓<br />
Bolt characteristics<br />
φ Rd F φ ↑⇒ ↑ u F φ ↑⇒ ↑ φ ↑⇒ke.0 ↑ φ ↑⇒kpl.0 ↓ φ ↑⇒∆u ↑<br />
Ltg<br />
S0 No influence<br />
No influence<br />
S0 ↑⇒ke.0 ↑<br />
Plate material<br />
Lt ↑⇒∆u ↑<br />
No influence<br />
fyf . ↑⇒ FRd↑<br />
f y. f ↑⇒ Fu<br />
↑ No influence<br />
fyf . ↑⇒kpl.0 ↑ f y. f ↑⇒∆u↓ 5.7 REFERENCES<br />
[5.1] European Committee for Standardization (CEN). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
178
6 SIMPLIFIED METHODOLOGIES FOR ASSESSMENT OF THE<br />
BEHAVIOUR OF SINGLE T-STUB CONNECTIONS<br />
6.1 INTRODUCTION<br />
Previous chapters deal with <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> overall behaviour <strong>of</strong><br />
single T-stub connections by means <strong>of</strong> experimental tests or numerical threedimensional<br />
models. Both approaches provide a complete definition <strong>of</strong> <strong>the</strong> F-∆<br />
response up to collapse <strong>of</strong> <strong>the</strong> connection. From a practical point <strong>of</strong> view, nei<strong>the</strong>r<br />
<strong>of</strong> <strong>the</strong> above methods seems appropriate. Therefore, a simple methodology<br />
for prediction <strong>of</strong> <strong>the</strong> connection response up to collapse is desired. As already<br />
pointed out, <strong>the</strong> collapse is governed by fracture <strong>of</strong> <strong>the</strong> bolts and/or cracking <strong>of</strong><br />
<strong>the</strong> T-stub material. Because <strong>of</strong> <strong>the</strong> emphasis placed on connection <strong>ductility</strong>,<br />
<strong>the</strong> methodology must be able to predict <strong>the</strong> response <strong>of</strong> <strong>the</strong> T-stub well into its<br />
plastic and strain hardening range with a reasonable degree <strong>of</strong> accuracy.<br />
This chapter presents simplified methods for determining <strong>the</strong> monotonic deformation<br />
response <strong>of</strong> T-stubs. First, existing models from <strong>the</strong> literature are<br />
discussed. Next, a two-dimensional beam model is proposed and calibrated<br />
against test results from <strong>the</strong> database compiled previously. Some recomm<strong>end</strong>ations<br />
for modifications are also given. Finally, conclusions are drawn.<br />
6.2 PREVIOUS RESEARCH<br />
The analytical prediction <strong>of</strong> <strong>the</strong> overall response <strong>of</strong> <strong>bolted</strong> T-stub connections<br />
is very complex. The behaviour <strong>of</strong> this type <strong>of</strong> connections is intrinsically<br />
three-dimensional and involves both geometrical and material nonlinearities. It<br />
includes <strong>the</strong> b<strong>end</strong>ing deformations <strong>of</strong> <strong>the</strong> flange and <strong>the</strong> combined axial and<br />
b<strong>end</strong>ing deformations <strong>of</strong> <strong>the</strong> bolts.<br />
Several <strong>the</strong>oretical approaches for <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong><br />
T-stubs have already been proposed in <strong>the</strong> literature. Essentially, <strong>the</strong>y use <strong>the</strong><br />
same basic prying mechanism, which is also <strong>the</strong> model implemented in Eurocode<br />
3 [6.1] (Fig. 6.1). The model is two-dimensional, i.e. <strong>the</strong> threedimensional<br />
effects are not accounted for. The system is statically indeterminate<br />
to <strong>the</strong> first degree. It is loaded by applying a vertical force F/2 to <strong>the</strong> support<br />
(1), which corresponds to <strong>the</strong> critical section at <strong>the</strong> flange-to-web connection.<br />
Only one quarter-model is taken into account due to symmetry considerations.<br />
The contact points at <strong>the</strong> tips <strong>of</strong> <strong>the</strong> flange are modelled with a pinned<br />
support and reproduce <strong>the</strong> effect <strong>of</strong> <strong>the</strong> prying forces. The T-stub flange behaves<br />
as a rectangular cross-section <strong>of</strong> width beff and depth tf. Such width, beff,<br />
represents <strong>the</strong> flange <strong>plate</strong> width tributary to a bolt row that contributes to load<br />
transmission. This width varies with increasing loading but cannot exceed <strong>the</strong> ac-<br />
179
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
180<br />
F<br />
2<br />
F<br />
2<br />
(1)<br />
F ⎛∆⎞ ⎜ ⎟<br />
2⎝ 2⎠<br />
Fig. 6.1 Typical T-stub prying model.<br />
∆<br />
tual flange width, b. At pure plastic conditions and for evaluation <strong>of</strong> <strong>the</strong> plastic<br />
(design) resistance, it accounts for all possible yield line mechanisms <strong>of</strong> <strong>the</strong> T-stub<br />
flange. Despite <strong>the</strong>se major simplifications, <strong>the</strong> nonlinear analysis <strong>of</strong> this prying<br />
model is still very complex and requires an incremental procedure. Therefore,<br />
it is not int<strong>end</strong>ed for hand computation unless some simplifications that reduce<br />
<strong>the</strong> model complexity to a reasonable level are assumed.<br />
In this section three alternative simplified models developed by Jaspart<br />
[6.2], Faella and co-workers [6.3-6.5] and Swanson [6.6] are briefly addressed.<br />
These models yield a piecewise F-∆ relationship for <strong>characterization</strong> <strong>of</strong> <strong>the</strong><br />
connection response. In addition, <strong>the</strong> proposals <strong>of</strong> Beg et al. [6.7] for assessment<br />
<strong>of</strong> <strong>the</strong> deformation capacity are also reviewed.<br />
6.2.1 Jaspart proposal (1991)<br />
Jaspart approximates <strong>the</strong> nonlinear T-stub behaviour to a bilinear response<br />
[6.2]. The characteristics <strong>of</strong> this bilinear behaviour are summarized as follows:<br />
(i) The initial elastic region has a slope ke.0 that is evaluated by application <strong>of</strong><br />
Eqs. (1.21-1.22) (later, Jaspart simplified this expression for inclusion in<br />
Eurocode 3 in [6.8] – cf. Eqs. (1.23-1.25), in Chapter 1).<br />
B<br />
(2)<br />
Q
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
(ii) The swivel point in <strong>the</strong> bilinear relationship represents <strong>the</strong> full development<br />
<strong>of</strong> <strong>the</strong> yield lines and <strong>the</strong> correponding force is FRd.0. This “plastic” resistance<br />
is is determined from <strong>the</strong> above prying model as explained in Chapter 1 –<br />
Eqs. (1.3-1.5,1.10).<br />
(iii) In <strong>the</strong> plastic region, above <strong>the</strong> swivel point, <strong>the</strong> effects <strong>of</strong> material strain<br />
hardening are dominant. The slope <strong>of</strong> this second linear region is given by:<br />
Eh<br />
k = k<br />
(6.1)<br />
p− l.0 e.0<br />
E<br />
whereby Eh is <strong>the</strong> strain hardening modulus <strong>of</strong> <strong>the</strong> flange material.<br />
(iv) The point <strong>of</strong> maximum force, Fu.0, is determined by formally equivalent<br />
expressions to FRd.0, by replacing <strong>the</strong> plastic conditions (index Rd) with<br />
ultimate conditions (index u). This means that <strong>the</strong>se expressions are based on<br />
<strong>the</strong> same geometric characteristics but <strong>the</strong> plastic moment <strong>of</strong> <strong>the</strong> flange, Mf.Rd<br />
is replaced with:<br />
2<br />
M fu . = 0.25tffuf<br />
. beff<br />
(6.2)<br />
which is an identical expression to Eq. (1.6) and BRd is replaced with:<br />
Bu = fu. bAs (6.3)<br />
The following expressions are <strong>the</strong>n obtained for Fu.0:<br />
Fu.0 = min ( F1. u.0; F2. u.0; F3.<br />
u.0<br />
)<br />
(6.4)<br />
and:<br />
4M<br />
fu .<br />
F1.<br />
u.0<br />
= ( basic formulation)<br />
m<br />
(6.5)<br />
( 32n−2dw) M f . u<br />
F1.<br />
u.0<br />
=<br />
( formulation accounting for <strong>the</strong> bolt)<br />
8mn<br />
− d m + n<br />
w<br />
( )<br />
( − )<br />
β ( 1 λ )<br />
2M fu . + 2Bun 2M fu . ⎡<br />
F2.<br />
u.0<br />
= = ⎢1+ m+ n m ⎢⎣ 2<br />
u<br />
βu +<br />
λ ⎤<br />
⎥<br />
⎥⎦<br />
(6.6)<br />
F3. u.0 = 2Bu = 2fu.<br />
bAs (6.7)<br />
The deformation capacity is readily determined by intersection <strong>of</strong> <strong>the</strong> plastic<br />
region, with slope kp-l.0, with <strong>the</strong> maximum resistance, Fu.0, i.e.:<br />
FRd.0 Fu.0 − FRd.0<br />
∆ u.0<br />
= +<br />
(6.8)<br />
ke.0 kp−l.0<br />
This methodology can be easily ext<strong>end</strong>ed to a nonlinear idealization <strong>of</strong> <strong>the</strong> F-∆<br />
response, similar to that proposed in Eurocode 3 for <strong>the</strong> joint overall M-Φ response<br />
(cf. §1.6.1.3), provided that <strong>the</strong> transition portion <strong>of</strong> <strong>the</strong> two straight curves is well<br />
established.<br />
6.2.2 Faella and co-workers model (2000)<br />
Faella and co-workers [6.3-6.5] developed a procedure based on <strong>the</strong> resem-<br />
181
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
blance <strong>of</strong> <strong>the</strong> distribution <strong>of</strong> internal forces at plastic and ultimate conditions<br />
(Figs. 1.10 and 6.2). They assumed <strong>the</strong> following simplifications [6.4]: (i)<br />
geometrical nonlinearities are neglected, (ii) compatibility between bolt and<br />
flange deformation is not considered, (iii) <strong>the</strong> shear interaction is disregarded,<br />
(iv) prying forces are located at <strong>the</strong> tip <strong>of</strong> <strong>the</strong> flanges, (v) b<strong>end</strong>ing <strong>of</strong> <strong>the</strong> bolts is<br />
neglected and (vi) cracking <strong>of</strong> <strong>the</strong> material is modelled by assuming <strong>the</strong> cracking<br />
condition as <strong>the</strong> occurrence <strong>of</strong> <strong>the</strong> ultimate strain in <strong>the</strong> extreme fibres <strong>of</strong><br />
<strong>the</strong> T-stub flanges. The plastic deformation <strong>of</strong> <strong>the</strong> flange is computed from <strong>the</strong><br />
corresponding moment-curvature (M-χ) diagram. This is obtained from simple<br />
internal equilibrium conditions <strong>of</strong> <strong>the</strong> section and by assuming that <strong>the</strong> material<br />
constitutive law can be approximated by a quadrilinear relationship (see Fig.<br />
2.2). This stress-strain relationship is defined in natural coordinates. The basic<br />
formulations for computation <strong>of</strong> plastic deformations are derived from <strong>the</strong> integration<br />
<strong>of</strong> <strong>the</strong> M-χ diagram over a certain length, <strong>the</strong> cantilever length, L, that<br />
remains unchanged during <strong>the</strong> loading process and equal to that occurring at ultimate<br />
conditions [6.3].<br />
This simple model yields a multilinear F-∆ curve for <strong>the</strong> behaviour <strong>of</strong> <strong>the</strong><br />
T-stub. The characteristic coordinates <strong>of</strong> this curve are determined according to<br />
<strong>the</strong> potential failure mode (Fig. 6.2). In particular, for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong><br />
characteristic force coordinates, <strong>the</strong>y use <strong>the</strong> same expression as <strong>the</strong> Eurocode<br />
3 for plastic conditions (cf. Chapter 1 and [6.3-6.5]).<br />
Q<br />
182<br />
F1.u.0<br />
B B<br />
Q<br />
(=F1.u/2+Q)<br />
n m m n<br />
(a) Flange fracture<br />
mechanism:<br />
β ≤ β .<br />
( )<br />
u u.lim<br />
b<br />
Mf.u<br />
Mf.u<br />
Q<br />
Bu<br />
F2.u.0<br />
Bu<br />
n m m n<br />
Q<br />
b<br />
Mf.u<br />
ξMf.u<br />
(b) Combined<br />
bolt/flange mechanism:<br />
β < β ≤ 2 .<br />
( )<br />
u.lim u<br />
Bu<br />
F3.u.0<br />
Bu<br />
n m m n<br />
b<br />
ξMf.u<br />
(c) Bolt fracture mecha-<br />
β > 2 .<br />
nism: ( )<br />
Fig. 6.2 Collapse mechanism typologies <strong>of</strong> a single T-stub prying at ultimate<br />
conditions according to Faella et al. [6.3].<br />
6.2.3 Swanson model (1999)<br />
Swanson developed a prying model that uses <strong>the</strong> geometrical properties defined<br />
in Fig. 6.3a, which is consistent with <strong>the</strong> strength model proposed by Ku-<br />
u
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
lak et al. [6.6]. The author uses <strong>the</strong> following dimensions:<br />
b′ = d −0.5r− 0.5φ<br />
(6.9)<br />
( )<br />
a′ = min 1.25 b′ ; e+ 0.5φ<br />
(6.10)<br />
For comparison, Fig. 6.3b shows <strong>the</strong> dimensions used in Eurocode 3:<br />
m= d − 0.8r<br />
(6.11)<br />
n= min ( 1.25 m; e)<br />
(6.12)<br />
The model includes: (i) nonlinear material properties, (ii) a variable bolt<br />
stiffness that captures <strong>the</strong> changing behaviour <strong>of</strong> <strong>the</strong> bolts as a function <strong>of</strong> <strong>the</strong><br />
loads <strong>the</strong>y are subjected to, (iii) partially plastic hinges in <strong>the</strong> flange and (iv)<br />
second order membrane behaviour <strong>of</strong> thin flanges [6.6].<br />
The bolt behaviour is incorporated by means <strong>of</strong> an extensional spring located<br />
at <strong>the</strong> inside edge <strong>of</strong> <strong>the</strong> bolt shank. This spring is characterized by a<br />
piecewise linear force-deformation, B-δb, response. Swanson [6.6] proposes an<br />
analytical model for <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> bolt deformation behaviour<br />
similar to that depicted in Fig. 6.4. The multilinear curve refers to handtightened<br />
bolts and its characteristic coordinates are set out in Table 6.1. The<br />
bolt elastic stiffness, Kb is evaluated as follows [6.6]:<br />
Eb<br />
Kb<br />
=<br />
(6.13)<br />
L A + L A<br />
s b tg s<br />
whereby Ls and Ltg are <strong>the</strong> shank and threaded lengths <strong>of</strong> <strong>the</strong> bolt included in<br />
0.5r<br />
F<br />
2<br />
B-δb<br />
0.5φ<br />
B-δb<br />
b’ a’<br />
m n<br />
(a) Dimensions used by Swanson. (b) Dimensions used in Eurocode 3.<br />
Fig. 6.3 T-stub prying model proposed by Swanson [6.6].<br />
0.8r<br />
F<br />
2<br />
183
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
184<br />
B<br />
0.90Bu<br />
0.85Bu<br />
Yielding,<br />
0.1Kb<br />
Elastic, Kb<br />
δb.1 δb.2<br />
Plastic,<br />
0.02Kb<br />
Bolt fracture<br />
δb.fract<br />
δb<br />
Fig. 6.4 Bolt force-deformation model according to Swanson.<br />
<strong>the</strong> grip length, respectively, Ab is <strong>the</strong> nominal area <strong>of</strong> <strong>the</strong> bolt shank, Ab = πφ 2 /4<br />
and φ is <strong>the</strong> bolt nominal diameter.<br />
Based on mechanistic considerations, <strong>the</strong> deformation capacity <strong>of</strong> <strong>the</strong> single<br />
bolt in tension, δb.fract, is easily assessed as follows [6.6]:<br />
0.90BL ⎛ u s<br />
2 ⎞<br />
δbfract . = + εub<br />
. ⎜Ltg+ ⎟<br />
(6.14)<br />
AbEb ⎝ nth<br />
⎠<br />
being nth <strong>the</strong> number <strong>of</strong> threads per unit length <strong>of</strong> <strong>the</strong> bolt. These predictions are<br />
based on <strong>the</strong> assumption that <strong>the</strong> bolt shank remains elastic with inelastic deformation<br />
concentrated in <strong>the</strong> threads that are included in <strong>the</strong> grip length [6.6]. It is also<br />
recognized that a portion <strong>of</strong> <strong>the</strong> bolt inside <strong>the</strong> nut will deform inelastically. As a<br />
result, two <strong>of</strong> <strong>the</strong> threads within <strong>the</strong> nut are included in <strong>the</strong> predictions.<br />
The flange mechanistic model assumes an elastic-yielding-plastic constitutive<br />
relationship for <strong>the</strong> steel. It also accommodates <strong>the</strong> shear deformations as<br />
well as <strong>the</strong> membrane effect, which can be particularly relevant for flexible<br />
flanges. Plastic hinges will develop at <strong>the</strong> flange-to-web connection and at <strong>the</strong><br />
bolt axis and <strong>the</strong>ir length is taken as equal to <strong>the</strong> flange thickness. Strain hardening<br />
is assumed to start immediately following <strong>the</strong> formation <strong>of</strong> a plastic<br />
hinge and was modelled with rotational springs [6.6]. The partially plastic<br />
states were incorporated in <strong>the</strong> model in a simplified way, as reported in [6.6].<br />
Swanson derived <strong>the</strong> stiffness coefficients and corresponding prying gradients,<br />
qij,k = ∂Q/∂∆, by using <strong>the</strong> direct stiffness method. Both parameters are<br />
used in an incremental solution technique. First, <strong>the</strong> initial stiffness and <strong>the</strong> initial<br />
prying gradient, qee, are determined:<br />
12EI ( 3EI + Kbγ<br />
3 )<br />
ke.0<br />
= (6.15)<br />
γ ee<br />
2<br />
9EI ( Kba′′ b βb<br />
− 2EI)<br />
qee<br />
= (6.16)<br />
γ<br />
ee
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table 6.1 Characteristic coordinates <strong>of</strong> <strong>the</strong> bolt deformation response.<br />
Bolt force, B Bolt stiffness, Kb Bolt elongation, δb<br />
0≤ B< 0.85Bub K 0.85Bu δ b.1<br />
=<br />
Kb<br />
0.85Bu ≤ B < 0.90Bu<br />
0.10K b<br />
0.05Bu<br />
δb.2 = δb.1+<br />
0.10Kb<br />
0.90Bu ≤ B≤ Bfract<br />
0.02K b<br />
δ .<br />
0.90BL u s<br />
= + ε . L<br />
AE<br />
whereby:<br />
3<br />
beff tf<br />
bfract ub tg<br />
b b<br />
I =<br />
12<br />
(6.17)<br />
is <strong>the</strong> inertia <strong>of</strong> <strong>the</strong> flange cross-section and <strong>the</strong> remaining coefficients are defined<br />
below:<br />
γ ee = 12EIγ1+<br />
Kbγ2<br />
3 2 2 3<br />
γ1= a′ βa + ( 3a′ b′ + 3ab<br />
′ ′ + b′<br />
) βb<br />
(6.18)<br />
(6.19)<br />
3 3 2 4 2<br />
γ2= 4a′ b′ βaβb + 3a′<br />
b′<br />
βb<br />
(6.20)<br />
3 2<br />
γ3= a′ βa + 3a′<br />
b′<br />
βb<br />
(6.21)<br />
12EI βa = 1+ 2<br />
Gb′ t a′ 12EI<br />
βb<br />
= 1+<br />
2<br />
Gb′ t b′<br />
(6.22)<br />
eff f eff f<br />
The coefficients βa and βb account for shear deformations.<br />
Next, several checks are made to determine which limit is reached first<br />
(bolt force or flange internal stresses limits). Incremental deformations are <strong>the</strong>n<br />
calculated for each <strong>of</strong> <strong>the</strong> potential limits with <strong>the</strong> smallest value governing.<br />
The F-∆ curve can yield up to nine linear branches, with different stiffness, before<br />
failure. Swanson states that <strong>the</strong> strength and <strong>the</strong> deformation capacity <strong>of</strong><br />
<strong>the</strong> flange are not always predicted accurately because <strong>of</strong> sensitivities <strong>of</strong> <strong>the</strong><br />
model to strain hardening parameters and bolt <strong>ductility</strong> [6.6]. It should be<br />
stressed that this model is not int<strong>end</strong>ed for hand calculations and it will not be<br />
used for fur<strong>the</strong>r comparisons.<br />
6.2.4 Beg and co-workers proposals for evaluation <strong>of</strong> <strong>the</strong> deformation capacity<br />
(2002)<br />
Beg et al. developed a set <strong>of</strong> simple analytical expressions for evaluation <strong>of</strong> <strong>the</strong><br />
deformation capacity <strong>of</strong> single T-stub connections [6.7]. They also assumed<br />
two alternative cracking conditions: (i) attainment <strong>of</strong> <strong>the</strong> ultimate strain at <strong>the</strong><br />
185
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
outer fibre <strong>of</strong> <strong>the</strong> flange section and (ii) fracture <strong>of</strong> <strong>the</strong> bolt. The maximum<br />
strain allowed at <strong>the</strong> flange section is 0.20 and <strong>the</strong> fracture <strong>of</strong> <strong>the</strong> bolt is assessed<br />
as follows:<br />
*<br />
ub . 2 ub . b L δ = ε<br />
(6.23)<br />
whereby εu.b is taken as 0.10 for full-threaded bolts and 0.02-0.05 for small-<br />
threaded bolts, and<br />
186<br />
L is <strong>the</strong> clamping length <strong>of</strong> bolts, i.e. thickness <strong>of</strong> clamped<br />
*<br />
b<br />
<strong>plate</strong>s including thickness <strong>of</strong> washers [6.7]. Factor 2 results from symmetry.<br />
For each potential plastic failure mode (see Fig. 1.10) <strong>the</strong> authors propose <strong>the</strong><br />
following relationships (δu is <strong>the</strong> deformation capacity <strong>of</strong> a half-T-stub):<br />
(i) Mode 1:<br />
δu = 0.4m⇒∆ u.0 = 2δ u = 0.8m<br />
(6.24)<br />
(ii) Mode 2:<br />
δub<br />
. ⎛ m⎞ ⎛ m⎞<br />
δu = ⎜1+ k ⎟⇒ ∆ u.0 = 2δu = δu.<br />
b⎜1+<br />
k ⎟<br />
2 ⎝ n ⎠ ⎝ n ⎠ (6.25)<br />
whereby k is an empirical factor varying from 3.0 to 4.0 [6.7].<br />
(iii) Mode 3:<br />
ub .<br />
*<br />
u u.0 2 u. b b<br />
2<br />
L<br />
δ<br />
δ = ⇒∆ = ε<br />
(6.26)<br />
These expressions account for <strong>the</strong> dep<strong>end</strong>ence <strong>of</strong> <strong>the</strong> deformation capacity<br />
<strong>of</strong> a T-stub on <strong>the</strong> fracture elongation <strong>of</strong> <strong>the</strong> bolts, on <strong>the</strong> ultimate strain <strong>of</strong> <strong>the</strong><br />
steel and on <strong>the</strong> geometrical parameters m and n. However, <strong>the</strong> dep<strong>end</strong>ence on<br />
<strong>the</strong> flange thickness is neglected. In Chapter 5 it has been shown evidenced <strong>the</strong><br />
strong dep<strong>end</strong>ence <strong>of</strong> <strong>the</strong> deformation behaviour on this geometrical parameter.<br />
For identical geometry connections that fail according to a plastic mechanism<br />
<strong>of</strong> type-1, <strong>the</strong> T-stub with thicker flanges exhibits a lower deformation capacity<br />
than <strong>the</strong> thinner flange. Eq. (6.24) does not account for this effect.<br />
6.2.5 Examples<br />
To illustrate <strong>the</strong> alternative procedures, <strong>the</strong> T-stub (unstiffened) specimens referred<br />
in Chapters 3, 4 and 5 are used. These specimens constitute a database<br />
for exemplification <strong>of</strong> <strong>the</strong> procedures presented in this section. Later, <strong>the</strong> same<br />
specimens will be used for validation <strong>of</strong> an alternative model for <strong>characterization</strong><br />
<strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> isolated T-stub connections.<br />
6.2.5.1 Evaluation <strong>of</strong> initial stiffness<br />
Chapter 1 already presented some procedures for evaluation <strong>of</strong> <strong>the</strong> initial stiffness<br />
<strong>of</strong> single T-stub connections, namely, <strong>the</strong> Yee and Melchers standard proposals<br />
[6.9-6.10] and <strong>the</strong> subsequent modifications suggested by Jaspart in<br />
[6.2]. The Eurocode 3 simple expressions were also derived. These are <strong>the</strong>
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
same expressions adopted by Jaspart in his simple methodology [6.2] (§6.2.1).<br />
Faella et al. presented an alternative formulation for <strong>the</strong> definition <strong>of</strong> ke.0 in<br />
[6.3] (cf. Chapter 1).<br />
Table 6.2 compares <strong>the</strong> initial stiffness predictions (per bolt row) <strong>of</strong> some<br />
T-stub specimens by application <strong>of</strong> two <strong>of</strong> <strong>the</strong> above procedures: Faella et al.<br />
formulation and Eurocode 3. Identical tables for <strong>the</strong> o<strong>the</strong>r methodologies are<br />
shown in App<strong>end</strong>ix D. The specimens were grouped according to <strong>the</strong> assembly<br />
type (hot rolled pr<strong>of</strong>iles or welded <strong>plate</strong>s). The results do not differ substantially.<br />
The ratio value in <strong>the</strong> tables is given by:<br />
Ratio = Predicted value Actual value (6.27)<br />
As a general conclusion, it can be stated that <strong>the</strong> procedures proposed by<br />
Faella et al. provide <strong>the</strong> best prediction for evaluation <strong>of</strong> ke.0 (third column in<br />
Table 6.2). Eurocode 3 overestimates ke.0 (fifth column Table 6.2). Regarding<br />
<strong>the</strong> remaining methods, <strong>the</strong> following conclusions are also drawn (App<strong>end</strong>ix<br />
D): (i) <strong>the</strong> location <strong>of</strong> <strong>the</strong> prying forces for application <strong>of</strong> <strong>the</strong> Yee and Melchers<br />
procedures does not introduce major differences within <strong>the</strong> limits analysed and<br />
(ii) <strong>the</strong> Swanson model is more accurate in <strong>the</strong> elastic domain if <strong>the</strong> geometrical<br />
dimensions from Eurocode 3 are used. In both cases, however, <strong>the</strong> average<br />
error is systematically over 100%.<br />
6.2.5.2 Evaluation <strong>of</strong> plastic resistance<br />
The alternative methodologies analysed in this work use <strong>the</strong> same approach as<br />
<strong>the</strong> Eurocode 3 for evaluation <strong>of</strong> <strong>the</strong> “plastic” resistance, FRd.0 (see Chapter 1).<br />
Table 6.3 summarizes <strong>the</strong> predictions for FRd.0 using <strong>the</strong> expressions from<br />
Eurocode 3. The potential plastic mode is also indicated. For those specimens<br />
failing according to a plastic collapse mode 1, both results from application <strong>of</strong><br />
<strong>the</strong> basic formulation and <strong>the</strong> formulation accounting for <strong>the</strong> bolt action are<br />
given. In some cases, if <strong>the</strong> latter formulation is taken into account, <strong>the</strong> collapse<br />
type-2 may become critical (specimens P4, WT4, for instance). In those cases,<br />
<strong>the</strong> values for type-2 are shown in bold/italic.<br />
These predictions are compared with <strong>the</strong> knee-range <strong>of</strong> <strong>the</strong> actual F-∆ (numerical<br />
or experimental) response since <strong>the</strong> definition <strong>of</strong> FRd.0 in this case is not<br />
straightforward. The predictions <strong>of</strong> FRd.0 are within <strong>the</strong>se limits, which means<br />
that <strong>the</strong> Eurocode proposals are accurate.<br />
6.2.5.3 Piecewise multilinear approximation <strong>of</strong> <strong>the</strong> overall response and evaluation<br />
<strong>of</strong> <strong>the</strong> deformation capacity and ultimate resistance<br />
The global F-∆ response <strong>of</strong> a T-stub is characterized in this section by using<br />
<strong>the</strong> bilinear approximation suggested by Jaspart and <strong>the</strong> multilinear model proposed<br />
by Faella and co-workers. The same examples from above are used for<br />
187
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 6.2 Prediction <strong>of</strong> axial stiffness by application <strong>of</strong> <strong>the</strong> Faella, Piluso and<br />
Rizzano procedures and <strong>the</strong> Eurocode 3.<br />
Test ID Num./Exp.<br />
stiffness<br />
188<br />
Faella et al. predictions<br />
Eurocode 3 predictions<br />
ke.0 Ratio ke.0 Ratio<br />
T1 83.54 87.77 1.05 144.36 1.73<br />
P1 63.27 57.38 0.91 97.27 1.54<br />
P2 117.06 140.95 1.20 220.29 1.88<br />
P3 72.62 77.95 1.07 129.45 1.78<br />
P4 97.86 111.10 1.14 178.63 1.83<br />
P5 101.23 124.32 1.23 197.37 1.95<br />
P9 128.47 173.54 1.35 255.91 1.99<br />
P10 43.88 23.85 0.54 42.08 0.96<br />
P12 102.05 92.09 0.90 156.40 1.53<br />
P14 81.97 87.77 1.07 144.36 1.76<br />
P15 128.41 152.41 1.19 249.63 1.94<br />
P16 111.23 141.11 1.27 220.51 1.98<br />
P18 171.57 158.73 0.93 266.81 1.56<br />
P20 181.68 300.54 1.65 441.03 2.43<br />
P23 322.09 487.19 1.51 675.18 2.10<br />
Average 1.13 1.80<br />
Coefficient <strong>of</strong> variation<br />
0.24<br />
0.18<br />
Weld_T1(i) 73.50 45.50 0.62 78.07 1.06<br />
Weld_T1(ii) 88.04 62.40 0.71 105.25 1.20<br />
Weld_T1(iii) 107.29 82.56 0.77 136.49 1.27<br />
WT1g 68.58 63.38 0.92 108.64 1.58<br />
WT1h 73.58 63.38 0.86 108.64 1.48<br />
WT2Aa 64.32 50.79 0.79 87.89 1.37<br />
WT2Ab 61.75 50.79 0.82 87.89 1.42<br />
WT2Ba 63.58 74.80 1.18 127.12 2.00<br />
WT2Bb 79.75 74.80 0.94 127.12 1.59<br />
WT4Aa 75.08 103.38 1.38 171.93 2.29<br />
WT4Ab 86.96 103.38 1.19 171.93 1.98<br />
WT7_M12 91.18 101.21 1.11 168.64 1.85<br />
WT7_M16 116.09 104.59 0.90 179.58 1.55<br />
WT7_M20 137.70 107.38 0.78 186.44 1.35<br />
WT51a 59.62 53.27 0.89 92.08 1.54<br />
WT51b 61.84 53.27 0.86 92.08 1.49<br />
WT53C 64.23 55.13 0.86 95.05 1.48<br />
WT53D 52.90 56.36 1.07 97.08 1.84<br />
WT53E 64.82 55.05 0.85 94.92 1.46<br />
WT57_M12 42.89 90.34 2.11 151.69 3.54<br />
WT57_M16 55.22 94.71 1.72 163.02 2.95<br />
WT57_M20 75.48 95.72 1.27 166.89 2.21<br />
Average 1.03 1.75<br />
Coefficient <strong>of</strong> variation<br />
0.34 0.33
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table 6.3 Prediction <strong>of</strong> <strong>the</strong> plastic resistance Eurocode 3 (per bolt row).<br />
Test ID Num./Exp.<br />
Eurocode 3 predictions<br />
kneerange<br />
Potential plastic<br />
mode<br />
FRd.0 (kN)<br />
Basic for- Formul. accountmulationing<br />
for <strong>the</strong> bolt<br />
T1 65 - 85 1 67.02 79.78<br />
P1 60 - 73 1 57.29 67.92<br />
P2 70 - 100 1 80.73 98.53<br />
P3 60 - 75 1 58.64 69.80<br />
P4 70 - 103 1 or 2 87.97 96.37<br />
P5 90 - 120 2 99.48 ⎯<br />
P9 85 - 120 2 108.23 ⎯<br />
P10 40 - 47 1 27.47 32.52<br />
P12 80 - 103 1 67.02 84.04<br />
P14 45 - 65 1 42.76 50.90<br />
P15 80 - 120 1 80.73 104.67<br />
P16 85 - 112 2 103.63 ⎯<br />
P18 150 - 200 1 117.29 157.16<br />
P20 160 - 200 2 190.88 ⎯<br />
P23 275 - 305 2 296.71 ⎯<br />
Weld_T1(i) 50 - 78 1 57.23 61.10<br />
Weld_T1(ii) 60 - 87 1 59.07 69.26<br />
Weld_T1(iii) 75 - 97 1 65.50 77.73<br />
WT1g 58 - 68 1 48.33 55.77<br />
WT1h 60 - 70 1 48.33 44.44<br />
WT2Aa 52 - 62 1 44.44 50.94<br />
WT2Ab 53 - 65 1 44.44 50.94<br />
WT2Ba 59 - 78 1 51.09 59.35<br />
WT2Bb 62 - 80 1 51.09 59.35<br />
WT4Aa 89 - 105 1 or 2 81.62 87.34<br />
WT4Ab 70 - 98 1 or 2 81.62 87.34<br />
WT7_M12 60 - 96 1 or 2 81.00 86.96<br />
WT7_M16 80 - 104 1 80.22 96.14<br />
WT7_M20 88 - 118 1 80.73 100.79<br />
WT51a 78 - 94 2 89.48 ⎯<br />
WT51b 79 - 95 2 89.48 ⎯<br />
WT53C 79 - 94 1 or 2 93.19 93.38<br />
WT53D 83 - 96 1 or 2 94.31 107.76<br />
WT53E 93 - 109 1 92.59 106.67<br />
WT57_M12 75 - 119 2 110.48 ⎯<br />
WT57_M16 104 - 165 1 or 2 158.52 163.78<br />
WT57_M20 126 - 204 1 157.72 196.15<br />
189
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
comparison with <strong>the</strong> simple methodology <strong>of</strong> Jaspart. For application <strong>of</strong> <strong>the</strong><br />
Faella et al. procedures only six examples are considered. As already stated<br />
above, <strong>the</strong> Swanson proposals are not illustrated herein.<br />
a) Methodology recomm<strong>end</strong>ed by Jaspart<br />
Fig. 6.5 illustrates <strong>the</strong> bilinear approximation <strong>of</strong> <strong>the</strong> F-∆ response <strong>of</strong> some selected<br />
specimens, as proposed by Jaspart. The graphs compare <strong>the</strong> actual response<br />
with four alternative approaches <strong>of</strong> <strong>the</strong> methodology, regarding <strong>the</strong> mechanical<br />
properties <strong>of</strong> <strong>the</strong> T-stub flange, <strong>the</strong> resistance formulation and a combination<br />
<strong>of</strong> both. The two alternative resistance formulations are <strong>the</strong> basic formulations<br />
(BF) and <strong>the</strong> formulation accounting for <strong>the</strong> bolt action (FBA) when<br />
applicable. The complete <strong>characterization</strong> <strong>of</strong> <strong>the</strong> actual material properties <strong>of</strong><br />
<strong>the</strong> various specimens from <strong>the</strong> database was given in Chapters 3, 4 and 5. The<br />
190<br />
Load, F (kN)<br />
120<br />
105<br />
(a) HR-T-stub T1 (fy.f = 430 MPa).<br />
Load, F (kN)<br />
90<br />
75<br />
60<br />
Actual response<br />
45<br />
Bilinear response (actual Eh and BF)<br />
30<br />
Bilinear response (actual Eh and FBA)<br />
15<br />
0<br />
Bilinear response (nominal Eh and BF)<br />
Bilinear response (nominal Eh and FBA)<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
(b) HR-T-stub P14 (fy.f = 275 MPa).<br />
Deformation, ∆ (mm)<br />
Actual response<br />
Bilinear response (actual Eh and BF)<br />
Bilinear response (actual Eh and FBA)<br />
Bilinear response (nominal Eh and BF)<br />
Bilinear response (nominal Eh and FBA)<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20 22 24 26<br />
Deformation, ∆ (mm)<br />
Fig. 6.5 Illustration <strong>of</strong> <strong>the</strong> methodology proposed by Jaspart.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Bilinear response (actual Eh and BF)<br />
Bilinear response (actual Eh and FBA)<br />
15<br />
0<br />
Bilinear response (nominal Eh and BF)<br />
Bilinear response (nominal Eh and FBA)<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
(c) WP-T-stub WT1 (fy.f = 340 MPa).<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Deformation, ∆ (mm)<br />
30<br />
Actual response<br />
15<br />
0<br />
Bilinear response (actual Eh and FBA)<br />
0 1 2 3 4 5 6 7 8 9 10<br />
(d) WP-T-stub WT51 (fy.f = 698 MPa).<br />
Deformation, ∆ (mm)<br />
Fig. 6.5 Illustration <strong>of</strong> <strong>the</strong> methodology proposed by Jaspart (cont.).<br />
actual strain hardening modulus, Eh, for <strong>the</strong>se specimens however is always<br />
lower than <strong>the</strong> nominal properties [6.3,6.11]. For steel grade S355, Eh = E/48.2<br />
and for S275, Eh = E/42.8. No quantitative guidance is given in nei<strong>the</strong>r references<br />
for steel grade S690. Hence, both actual and nominal values for Eh are<br />
taken into account for those specimens where steel grade S355 and S275 was<br />
employed (S275 was used in specimen P14).<br />
For fur<strong>the</strong>r details on this methodology, <strong>the</strong> reader should refer to App<strong>end</strong>ix<br />
D. Generally speaking, <strong>the</strong> bilinear approximation proposed by this author reproduces<br />
well <strong>the</strong> actual behaviour for those specimens made up <strong>of</strong> S690, with<br />
an overestimation <strong>of</strong> <strong>the</strong> deformation capacity. For <strong>the</strong> remaining cases, <strong>the</strong><br />
predictions are fine provided that <strong>the</strong> nominal value <strong>of</strong> <strong>the</strong> strain hardening<br />
modulus is used. If <strong>the</strong> actual value <strong>of</strong> Eh is used instead, <strong>the</strong>n <strong>the</strong> predictions<br />
are not so good.<br />
191
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
b) Methodology recomm<strong>end</strong>ed by Faella, Piluso and Rizzano<br />
Fig. 6.6 shows <strong>the</strong> overall F-∆ response for some T-stub specimens that were<br />
chosen to illustrate <strong>the</strong> different failure modes. The graphs trace <strong>the</strong> response<br />
for actual and nominal flange mechanical properties and include <strong>the</strong> compatibility<br />
<strong>of</strong> <strong>the</strong> deformations <strong>of</strong> <strong>the</strong> flange and <strong>the</strong> bolt (App<strong>end</strong>ix D). In general,<br />
this model does not provide an accurate modelling <strong>of</strong> <strong>the</strong> deformation behaviour.<br />
c) Methodology recomm<strong>end</strong>ed by Beg, Zupančič and Vayas<br />
The procedures for a direct computation <strong>of</strong> <strong>the</strong> deformation capacity from Beg<br />
et al. [6.7] are illustrated in App<strong>end</strong>ix D for <strong>the</strong> various specimens. The predictions<br />
are not satisfactory though, particularly for those specimens that fail ac-<br />
192<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
Actual response<br />
45<br />
Quadrilinear approximation (BF: type-2<br />
30<br />
15<br />
0<br />
governs failure)<br />
Quadrilinear approximation(FBA: (BF: type-1<br />
governs failure)<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation, ∆ (mm)<br />
(a) HR-specimen T1 (actual material properties).<br />
Load, F (kN)<br />
270<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
Actual response<br />
Quadrilinear approximation (actual mat. prop.)<br />
Quadrilinear approximation (nominal mat. prop.)<br />
0 3 6 9 12 15 18 21 24 27 30<br />
Deformation, ∆ (mm)<br />
(b) HR-specimen P18.<br />
Fig. 6.6 Illustration <strong>of</strong> <strong>the</strong> methodology proposed by Faella, Piluso and Rizzano.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
120<br />
105<br />
(c) WP-T-stub specimen WT4A.<br />
Load, F (kN)<br />
90<br />
75<br />
60<br />
45<br />
30<br />
Actual response<br />
15<br />
0<br />
Quadrilinear approximation (actual mat. prop.)<br />
Quadrilinear approximation (nominal mat. prop.)<br />
0 1 2 3 4 5 6 7 8 9 10<br />
105<br />
(d) WP-T-stub specimen WT51.<br />
90<br />
75<br />
60<br />
45<br />
Deformation, ∆ (mm)<br />
30<br />
Actual response<br />
15<br />
Quadrilinear approximation (actual mat. prop.)<br />
0<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5<br />
Deformation, ∆ (mm)<br />
Fig. 6.6 Illustration <strong>of</strong> <strong>the</strong> methodology proposed by Faella, Piluso and Rizzano<br />
(cont.).<br />
cording to a plastic mode 1. The average ratios <strong>of</strong> <strong>the</strong> actual numerical or experimental<br />
predictions are 2.10 for HR-T-stubs and 3.55 for WP-T-stubs, with<br />
coefficients <strong>of</strong> variation <strong>of</strong> 0.50 and 0.48, respectively. Again, it is noted that<br />
this methodology only gives an estimation <strong>of</strong> <strong>the</strong> T-stub deformation capacity,<br />
ra<strong>the</strong>r than a description <strong>of</strong> <strong>the</strong> full nonlinear behaviour.<br />
6.2.5.4 Summary<br />
This section described and illustrated several methodologies for <strong>the</strong> assessment<br />
<strong>of</strong> <strong>the</strong> F-∆ response <strong>of</strong> T-stubs (or some <strong>of</strong> its characteristics). Table 6.4 compares<br />
<strong>the</strong> different methods from a statistical point <strong>of</strong> view, in terms <strong>of</strong> average<br />
ratios <strong>of</strong> <strong>the</strong> sample <strong>of</strong> examples and coefficient <strong>of</strong> variation. The examples are<br />
divided according to <strong>the</strong> assembly type (HR-T-stub or WP-T-stub). The pa-<br />
193
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 6.4 Summary <strong>of</strong> <strong>the</strong> different proposals from a statistical point <strong>of</strong> view<br />
(average ratios and coefficients <strong>of</strong> variation, <strong>the</strong> latter in italic) for<br />
evaluation <strong>of</strong> <strong>the</strong> force-deformation characteristics.<br />
Methodology T-stub<br />
assembly<br />
194<br />
Stiffness Ultimate resistance<br />
Deformation<br />
capacity<br />
ke.0 Fu.0 ∆u.0<br />
Beg et al.<br />
HR<br />
WP<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
2.10 (0.50)<br />
3.55 (0.48)<br />
Eurocode 3<br />
HR<br />
WP<br />
1.80 (0.18)<br />
1.75 (0.33)<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
Faella et al.<br />
HR<br />
WP<br />
1.13 (0.24)<br />
1.03 (0.34)<br />
0.80 (0.46)<br />
0.95 (0.23)<br />
0.73 (0.54)<br />
1.16 (0.36)<br />
Jaspart<br />
HR<br />
WP<br />
2.30 (0.21)<br />
2.30 (0.31)<br />
0.96 (0.11)<br />
0.93 (0.08)<br />
0.91 (0.35)<br />
1.13 (0.30)<br />
Swanson<br />
HR<br />
WP<br />
2.35 (0.26)<br />
3.08 (0.37)<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
Yee and<br />
HR 2.32 (0.18) ⎯ ⎯<br />
Melchers WP 2.23 (0.32) ⎯ ⎯<br />
rameters chosen for comparison are <strong>the</strong> initial stiffness, <strong>the</strong> ultimate resistance<br />
and <strong>the</strong> deformation capacity.<br />
The best approach for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> initial stiffness is that proposed<br />
by Faella et al. [6.3], though <strong>the</strong> coefficient <strong>of</strong> variation <strong>of</strong> <strong>the</strong> sample is<br />
slightly higher than for <strong>the</strong> Eurocode 3. Regarding <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> deformation<br />
capacity, Jaspart [6.2] gives accurate predictions. The scatter <strong>of</strong> results<br />
for <strong>the</strong> deformation capacity is ra<strong>the</strong>r high when compared to <strong>the</strong> o<strong>the</strong>r properties<br />
as shown by <strong>the</strong> coefficient <strong>of</strong> variation. The results shown for <strong>the</strong> methodology<br />
proposed by Faella and co-authors, in terms <strong>of</strong> ultimate resistance and<br />
deformation capacity, are merely illustrative since <strong>the</strong> sample is not big enough<br />
for a statistical analysis <strong>of</strong> this type.<br />
6.3 PROPOSAL AND VALIDATION OF A BEAM MODEL FOR CHARACTERI-<br />
ZATION OF THE FORCE-DEFORMATION RESPONSE OF T-STUBS<br />
6.3.1 Description <strong>of</strong> <strong>the</strong> model<br />
The models described above afford some basis for <strong>the</strong> development <strong>of</strong> an analytical<br />
method for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> deformation capacity and <strong>the</strong> loadcarrying<br />
capacity <strong>of</strong> single T-stub connections. The mechanical model is similar<br />
to that depicted in Fig. 6.3b. A model that uses geometrical and mechanical
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
properties consistent with <strong>the</strong> Eurocode 3 prying model is desired so as to<br />
make implementation easier.<br />
The prying model is analysed up to collapse. From static equilibrium (Fig.<br />
6.7):<br />
F<br />
B = + Q<br />
(6.28)<br />
2<br />
() 1 F<br />
M = Qn − m<br />
(6.29)<br />
2<br />
( 2)<br />
M = Qn<br />
(6.30)<br />
being (1) <strong>the</strong> section at <strong>the</strong> flange-to-web connection and (2) <strong>the</strong> flange section<br />
at <strong>the</strong> bolt line. Normally, M1 ≥ M2. However, Swanson points out that in some<br />
cases this inequality may not be observed due to <strong>the</strong> effect <strong>of</strong> <strong>the</strong> removal <strong>of</strong><br />
flange material at <strong>the</strong> bolt line when <strong>the</strong> holes are drilled for <strong>the</strong> bolts [6.6].<br />
The deflected shape and <strong>the</strong> moment diagram <strong>of</strong> one side <strong>of</strong> <strong>the</strong> flange in pure<br />
elastic conditions and after separation at <strong>the</strong> bolt axis are as shown in Fig. 6.8a<br />
[6.12]. If <strong>the</strong> bolt is strong enough, a stage <strong>of</strong> loading will be reached when <strong>the</strong><br />
plastic moment <strong>of</strong> <strong>the</strong> flange, Mf.p, is attained at <strong>the</strong> flange-to-web connection<br />
(Fig. 6.8b, [6.12]). Any additional load will cause fur<strong>the</strong>r flange deflection that<br />
results in strain hardening <strong>of</strong> <strong>the</strong> flange and an increase in <strong>the</strong> internal moment<br />
at that section. The zone <strong>of</strong> full plastification spreads into <strong>the</strong> flange. With continued<br />
loading, a similar condition may develop at <strong>the</strong> bolt axis. When <strong>the</strong> most<br />
(1)<br />
F<br />
2<br />
(1*)<br />
0.2r or 0.2 2a w<br />
Fig. 6.7 Internal forces.<br />
(2)<br />
m n<br />
M1 = Qn - 0.5Fm<br />
B<br />
M2 = Qn<br />
Q<br />
195
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
highly strained flange fibres are strained to <strong>the</strong> breaking point (εu) and fracture,<br />
<strong>the</strong> ultimate resisting moment <strong>of</strong> <strong>the</strong> flange, Mf.u, is also reached and <strong>the</strong> ultimate<br />
conditions (subscript u) are attained (Fig. 6.8c, [6.12]). In this figure, Bu<br />
is not necessarily <strong>the</strong> tensile strength <strong>of</strong> <strong>the</strong> bolt.<br />
For common steels, Mf.u is significantly higher than Mf.p [6.3,6.11,6.13]. If<br />
simple plastic <strong>the</strong>ory is applied, <strong>the</strong> limit resistance and <strong>the</strong> deformation would<br />
be determined by Mf.p. Therefore, consideration <strong>of</strong> strain hardening is crucial to<br />
carry out an ultimate analysis <strong>of</strong> <strong>the</strong> system up to a fracture condition.<br />
M (1)<br />
196<br />
B<br />
0.5F<br />
M (1) = Qn-0.5Fm<br />
Q<br />
M (2) = Qn<br />
m n<br />
(a) Pure elastic conditions.<br />
Mf.p<br />
B<br />
0.5F<br />
M (1) = Mf.p<br />
Mf.y<br />
Q<br />
M (2) = Qn<br />
m n<br />
(b) Full plastification <strong>of</strong><br />
section (1).<br />
Fig. 6.8 Effect <strong>of</strong> material strain hardening.<br />
6.3.1.1 Fracture conditions<br />
Mf.u<br />
Bu<br />
M (1) 0.5Fu<br />
= Mf.u<br />
Mf.p<br />
Mf.y<br />
Mf.y<br />
Mf.p<br />
Qu<br />
M (2) = Qun<br />
m n<br />
(c) Fracture <strong>of</strong> section<br />
(1).<br />
The two possible ultimate fracture conditions are: (i) fracture <strong>of</strong> <strong>the</strong> bolt and<br />
(ii) cracking <strong>of</strong> material <strong>of</strong> <strong>the</strong> flange near <strong>the</strong> web as already explained in<br />
Chapter 2.<br />
In <strong>the</strong> context <strong>of</strong> a two-dimensional model, where <strong>the</strong> flange is modelled as<br />
a rectangular cross-section, this latter condition may be too severe. Critical section<br />
(1) is defined at a distance m from <strong>the</strong> bolt axis, where <strong>the</strong> flange thickness<br />
is higher than tf owing to <strong>the</strong> fillet weld or radius that provide some extra material<br />
thickness. Therefore, <strong>the</strong> imposition <strong>of</strong> cracking <strong>of</strong> <strong>the</strong> material should also<br />
be checked at <strong>the</strong> <strong>end</strong> <strong>of</strong> this fillet, section (1*) – Fig. 6.7, i.e. at a distance m * =<br />
d – r or m * = d – 2 aw for HR- and WP-T-stubs, respectively.<br />
6.3.1.2 Bolt deformation behaviour<br />
The bolt elongation response is based on <strong>the</strong> Swanson’s proposals. Its influence
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
on <strong>the</strong> overall response is accounted for by means <strong>of</strong> an extensional spring with<br />
similar characteristics to <strong>the</strong> Swanson bolt model, as just explained (Fig. 6.4<br />
and Table 6.1). For <strong>the</strong> computation <strong>of</strong> <strong>the</strong> bolt deformation at fracture, however,<br />
<strong>the</strong> parameter nth that appears in Eq. (6.14) is disregarded.<br />
For design calculations, <strong>the</strong> nominal material characteristics <strong>of</strong> <strong>the</strong> bolts<br />
have to be defined. The parameters given in Table 6.5 are suggested by Hirt<br />
and Bez [6.14] for high-strength bolts.<br />
Table 6.5 Minimum mechanical properties <strong>of</strong> high-strength bolts.<br />
Bolt grade fy fu εu<br />
(MPa) (MPa)<br />
8.8 640 800 0.12<br />
10.9 900 1000 0.09<br />
6.3.1.3 Flange constitutive law<br />
The flange material constitutive law is modelled by means <strong>of</strong> a piecewise linear<br />
curve, that accounts for <strong>the</strong> strain hardening effects. This law is a true stresslogarithmic<br />
strain relationship, i.e. it is defined in natural coordinates in order<br />
to capture <strong>the</strong> actual material behaviour. Faella et al., in fact, adopted <strong>the</strong> same<br />
approach since <strong>the</strong> prediction <strong>of</strong> <strong>the</strong> plastic deformation capacity <strong>of</strong> compact<br />
sections is more accurate if natural stress-strain coordinates are used [6.3].<br />
The above model is not suitable for a hand calculation. Instead, a numerical<br />
FE method is used to determine <strong>the</strong> structural response. Consequently, <strong>the</strong><br />
piecewise constitutive law may contain numerous branches. It should be<br />
stressed that many FE codes do not allow for an elastoplastic analysis with<br />
strain hardening for beam elements. The FE code LUSAS [6.15] implements a<br />
beam element that belongs to <strong>the</strong> Kirch<strong>of</strong>f beams group (with quadrilateral<br />
cross-section) [6.15]. Shear deformations are excluded in this element formulation.<br />
It has a quadrilateral cross-section.<br />
From a design point <strong>of</strong> view, <strong>the</strong> constitutive law should be <strong>of</strong> a standard<br />
type though. The stress-strain curve can be idealized by means <strong>of</strong> a multilinear<br />
model with a straight line for hardening range, as suggested in [6.13] (Fig. 6.9).<br />
The maximum stress is reached for a strain value:<br />
fu − fy<br />
εhs = εh+<br />
(6.31)<br />
Eh<br />
With reference to Fig. 6.9, Gioncu and Mazzolani give no guidance on <strong>the</strong><br />
<strong>characterization</strong> <strong>of</strong> <strong>the</strong> s<strong>of</strong>tening branch <strong>of</strong> this curve [6.13]. For current structural<br />
steel grades, <strong>the</strong> characteristic coordinates are set out in Table 6.6, for<br />
<strong>plate</strong> thickness smaller than 40 mm [6.13]. These coordinates are transformed<br />
into a true stress-logarithmic strain curve by means <strong>of</strong> Eq. (4.1), which is reproduced<br />
below:<br />
197
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
198<br />
= ( 1+<br />
) and ε ln ( 1 ε )<br />
σ n σ ε<br />
n = + (6.32)<br />
The stress-strain characteristics in natural coordinates for <strong>the</strong> three above steel<br />
grades are depicted in Fig. 6.10. The fifth linear part <strong>of</strong> <strong>the</strong> curve in natural coordinates<br />
has a slope <strong>of</strong> Eu = fu, according to Faella et al. [6.3].<br />
σ<br />
fu<br />
fy<br />
E<br />
Eh<br />
εy εh εhs εuni εu<br />
Fig. 6.9 Idealization <strong>of</strong> <strong>the</strong> stress-strain diagram with a multi-linear model.<br />
True stress (MPa)<br />
700<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
0.00 0.03 0.06 0.09 0.12 0.15 0.18 0.21 0.24<br />
Logarithmic strain<br />
S355<br />
S275<br />
S235<br />
Fig. 6.10 True stress-logarithmic strain characteristics for steel grades S235,<br />
S275 and S355.<br />
Table 6.6 Characteristics <strong>of</strong> <strong>the</strong> stress-strain curve (stress values in [MPa]).<br />
Steel<br />
grade<br />
fy fu εy εh εhs εuni εu Eh Eu<br />
S235 235 360 0.001 0.014 0.037 0.140 0.250 5500 360<br />
S275 275 430 0.001 0.015 0.047 0.120 0.220 4800 430<br />
S355 355 510 0.002 0.017 0.053 0.110 0.200 4250 510<br />
ε
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
6.3.2 Analysis <strong>of</strong> <strong>the</strong> model in <strong>the</strong> elastic range<br />
First, <strong>the</strong> proposed model is analysed and validated in <strong>the</strong> elastic domain by using<br />
<strong>the</strong> specimens from <strong>the</strong> author’s database. The initial stiffness <strong>of</strong> a T-stub<br />
connection is evaluated and compared with <strong>the</strong> actual predictions, corresponding<br />
to experimental or (three-dimensional) numerical values.<br />
In order to assess <strong>the</strong> importance <strong>of</strong> <strong>the</strong> shear deformability on <strong>the</strong> flange<br />
rectangular cross-section, two beam elements are tested: thin beam and thick<br />
beam. The thick beam element includes <strong>the</strong> shear deformations in its formulation<br />
but does not allow for a material nonlinear analysis with strain hardening.<br />
The results are compared in Table 6.7 for fifteen selected examples that represent<br />
<strong>the</strong> three different failure types. Clearly, <strong>the</strong> thick beam model provides a<br />
better agreement with <strong>the</strong> actual results (second column <strong>of</strong> Table 6.7), which<br />
indicates that <strong>the</strong> shear deformation in <strong>the</strong> elastic domain may be significant.<br />
From <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> average ratio to <strong>the</strong> actual values, <strong>the</strong> model with a<br />
thin beam element was re-analysed with a reduced Young modulus for <strong>the</strong><br />
flange material. This reduction was taken as half <strong>of</strong> <strong>the</strong> actual E since <strong>the</strong> earlier<br />
results were nearly twice as much as <strong>the</strong> actual. This factor <strong>of</strong> reduction<br />
may be slightly increased in order to best fit <strong>the</strong> actual results (unitary average<br />
ratio). However, <strong>the</strong> calibration <strong>of</strong> this factor should be based on a larger sam-<br />
Table 6.7 Influence <strong>of</strong> shear deformations on <strong>the</strong> initial stiffness <strong>of</strong> some <strong>of</strong><br />
<strong>the</strong> tested T-stubs (stiffness values in [kN/mm]).<br />
Test<br />
ID<br />
Num.<br />
res.<br />
Thick beam model<br />
(Thin) Beam model predictions<br />
Actual Young Reduced Young<br />
modulus modulus (0.5E)<br />
ke.0 Ratio ke.0 Ratio ke.0 Ratio<br />
T1 83.54 149.35 1.79 175.06 2.10 99.68 1.19<br />
P1 63.27 108.03 1.71 124.27 1.96 69.86 1.10<br />
P2 117.06 215.72 1.84 259.43 2.22 153.84 1.31<br />
P3 72.62 133.92 1.84 157.91 2.17 88.80 1.22<br />
P4 97.86 184.95 1.89 214.04 2.19 125.31 1.28<br />
P5 101.23 204.49 2.02 235.17 2.32 139.71 1.38<br />
P9 128.47 262.74 2.05 292.60 2.28 185.87 1.45<br />
P10 43.88 49.35 1.12 54.57 1.24 28.14 0.64<br />
P12 102.05 162.51 1.59 194.79 1.91 106.02 1.04<br />
P14 81.97 149.35 1.82 175.06 2.14 99.68 1.22<br />
P15 128.41 240.45 1.87 298.29 2.32 167.26 1.30<br />
P16 111.23 228.71 2.06 261.17 2.35 157.91 1.42<br />
P18 171.57 279.74 1.63 333.71 1.95 183.33 1.07<br />
P20 181.68 422.05 2.32 509.64 2.81 328.23 1.81<br />
P23 322.09 605.32 1.88 775.43 2.41 522.49 1.62<br />
Average 1.83 2.16 1.27<br />
Coeff. variation<br />
0.15<br />
0.16<br />
0.21<br />
199
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
ple <strong>of</strong> examples, which also includes o<strong>the</strong>r connections beyond those from <strong>the</strong><br />
author. The results for this reduced modulus <strong>of</strong> elasticity are set out in Table<br />
6.7 as well. As expected, <strong>the</strong> initial stiffness values decrease. These values will<br />
improve fur<strong>the</strong>r if an additional correction <strong>of</strong> <strong>the</strong> Young modulus for shear is<br />
introduced. It is desirable to obtain this correction by means <strong>of</strong> a simple formula<br />
ra<strong>the</strong>r than an empirical correction. In App<strong>end</strong>ix A <strong>of</strong> Chapter 1, <strong>the</strong><br />
shear interaction was already taken into consideration for resistance purposes.<br />
In mode-1 plastic failure types <strong>the</strong> ratio between <strong>the</strong> design resistance <strong>of</strong><br />
mechanism type-1 accounting for shear and that corresponding to <strong>the</strong> basic<br />
formulation is given by (cf. App<strong>end</strong>ix A):<br />
2<br />
2⎛ m ⎞ ⎡ 3 ⎤4Mf.<br />
Rd<br />
⎜ 1+ −1<br />
2<br />
3 ⎜<br />
⎟<br />
t ⎟ ⎢ ⎥<br />
2<br />
⎝ f ⎠ ⎢ ( mtf ) ⎥ m<br />
2⎛m⎞ ⎡ 3 ⎤<br />
F1.<br />
Rd .0 =<br />
⎣ ⎦<br />
= ⎜ ⎟ 1 1 2<br />
4M ⎢ + − ⎥ (6.33)<br />
f . Rd 3 ⎜t⎟ ⎝ f ⎠ ⎢ ( mt<br />
⎣ f ) ⎥<br />
⎦<br />
m<br />
The above relationship dep<strong>end</strong>s exclusively on <strong>the</strong> ratio m/tf. Fig. 6.11 plots<br />
that relationship and shows that lim F1.<br />
Rd .0 = 1.<br />
The value <strong>of</strong> F 1. Rd .0 is signifi-<br />
cant for mtf 2.5( F1.<br />
Rd.0<br />
0.9)<br />
200<br />
mtf<br />
→∞<br />
≤ < . If <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> T-stub elements and <strong>the</strong><br />
bolts is carried out separately, <strong>the</strong> T-flange is fixed at <strong>the</strong> bolt centreline.<br />
Therefore, <strong>the</strong> only possible collapse mode is <strong>of</strong> type-1 and so <strong>the</strong> above relationship<br />
applies. The following expression is <strong>the</strong>n proposed for determining <strong>the</strong><br />
reduced modulus to employ in <strong>the</strong> beam model:<br />
2 2<br />
2⎛m⎞ ⎡ 3 ⎤ E⎛m⎞ ⎡ 3 ⎤<br />
Ered = 0.5E× ⎜ ⎟ ⎢ 1+ − 1 1 1<br />
2 ⎥ = ⎜ ⎟ ⎢ + − 2 ⎥ (6.34)<br />
3⎜t ⎟<br />
f ( mt ) 3 ⎜t ⎟<br />
⎝ ⎠ ⎢ f ⎥ f ⎢ ( mt<br />
⎣ ⎦<br />
⎝ ⎠<br />
⎣ f ) ⎥<br />
⎦<br />
This reduction does not have much influence at ultimate conditions. In fact, <strong>the</strong><br />
effect <strong>of</strong> shear on <strong>the</strong> moment resistance <strong>of</strong> <strong>the</strong> flange is apparently beneficial<br />
[6.12].<br />
1.0<br />
0.9<br />
0.8<br />
0.7<br />
0.6<br />
F 0.5<br />
1.Rd.0<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
Fig. 6.11 Interaction F 1. Rd.0<br />
vs. mt f .<br />
m/tf
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table 6.8 shows <strong>the</strong> results for initial stiffness obtained through application<br />
<strong>of</strong> <strong>the</strong> above expression, as well as <strong>the</strong> reduced Young modulus. In this table<br />
both <strong>the</strong> ratio to <strong>the</strong> actual stiffness values (eighth column – Act.) and to <strong>the</strong><br />
beam model with <strong>the</strong> simple reduction (ninth column – 0.5E) are calculated.<br />
The difference between <strong>the</strong> two approaches is 8% on average. The results obtained<br />
with Ered as in Eq. (6.34) show a good agreement with <strong>the</strong> actual predictions<br />
(error <strong>of</strong> 17%, on average).<br />
This value <strong>of</strong> Ered is <strong>the</strong>n used onwards. It is important to stress that this<br />
value has little influence at ultimate conditions, as already explained. The beam<br />
model referred hereafter is hence <strong>the</strong> model that employs <strong>the</strong> thin beam model<br />
and Ered for <strong>the</strong> flange material.<br />
Table 6.9 evaluates <strong>the</strong> initial stiffness for o<strong>the</strong>r T-stubs from <strong>the</strong> database.<br />
The results are in line with <strong>the</strong> previous predictions. Additionally and in <strong>the</strong><br />
elastic behaviour domain, a set <strong>of</strong> sixteen T-stub connections tested by Faella<br />
et al. [6.3] are considered. Unfortunately, <strong>the</strong>se specimens cannot be used for<br />
fur<strong>the</strong>r comparisons due to lack <strong>of</strong> data. The geometrical and mechanical characteristics<br />
<strong>of</strong> <strong>the</strong> latter specimens are set out in Table 6.10. The initial stiffness<br />
predictions by application <strong>of</strong> this model are summarized in Table 6.11 along<br />
with <strong>the</strong> reductions to be applied. In this table <strong>the</strong>se results are also compared<br />
with <strong>the</strong> experiments and <strong>the</strong> beam model with <strong>the</strong> simple reduction, as before.<br />
Table 6.8 Values <strong>of</strong> <strong>the</strong> reduced Young modulus accounting for shear.<br />
Test<br />
ID<br />
E<br />
(MPa)<br />
m/tf F 1. Rd .0 0.5F 1. Rd.0<br />
Ered<br />
(MPa)<br />
Beam model predictions<br />
Reduced Young modulus<br />
( Ered = 0.5EF1.<br />
Rd.0<br />
)<br />
ke.0 Ratio<br />
kN/mm Act. 0.5E<br />
T1 2.75 0.92 0.46 95416 92.48 1.11 0.93<br />
P1 3.22 0.94 0.47 97472 65.96 1.04 0.94<br />
P2 2.29 0.89 0.44 92315 139.46 1.19 0.91<br />
P3 2.75 0.92 0.46 95416 82.29 1.13 0.93<br />
P4 2.75 0.92 0.46 95416 116.60 1.19 0.93<br />
P5 2.75 0.92 0.46 95416 130.20 1.29 0.93<br />
P9 2.05 0.87 0.43 90180 167.39 1.30 0.90<br />
P10 4.39 0.96 0.48 100318 27.16 0.62 0.96<br />
P12 2.75 0.92 0.46 95416 97.94 0.96 0.92<br />
P14 2.75 0.92 0.46 95416 92.48 1.13 0.93<br />
P15 2.29 0.89 0.44 92315 154.94 1.21 0.93<br />
P16 2.75 0.92 0.46 95416 147.44 1.33 0.93<br />
P18 2.75 0.92 0.46 95416 169.58 0.99 0.93<br />
P20 2.05 0.87 0.43 90180 296.64 1.63 0.90<br />
P23 1.67 0.82 0.41 85306 461.22 1.43 0.88<br />
Average 1.17 0.92<br />
Coeff. variation 0.20 0.02<br />
208153<br />
201
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 6.9 Validation <strong>of</strong> <strong>the</strong> approach with fur<strong>the</strong>r examples from <strong>the</strong> database:<br />
comparison <strong>of</strong> initial stiffness predictions (Young modulus<br />
in [MPa] stiffness values in [kN/mm]).<br />
Test<br />
ID<br />
E m/tf F1.<br />
Rd .0<br />
2<br />
Beam model predictions<br />
0.5E Ered = 0.5EF1.<br />
Rd.0<br />
Ratio<br />
Actual<br />
ke.0<br />
ke.0 Ra- ke.0<br />
tio<br />
Act. 0.5E<br />
HR-T-stubs<br />
P6 2.75 0.46 76.68 88.80 1.16 82.29 1.07 0.93<br />
P7 2.75 0.46 91.72 110.19 1.20 102.35 1.12 0.93<br />
P8 2.46 0.45 95.19 127.28 1.34 116.44 1.22 0.91<br />
P11 2.05 0.43 122.93 188.83 1.54 170.56 1.39 0.90<br />
P13 2.75 0.46 83.46 99.68 1.19 92.48 1.11 0.93<br />
P17 2.75 0.46 138.60 172.19 1.24 159.89 1.15 0.93<br />
P19 2.75 0.46 130.47 172.19 1.32 159.89 1.23 0.93<br />
P21 2.75 0.46 174.25 228.11 1.31 211.79 1.22 0.93<br />
P22<br />
2.16 0.44 253.74 328.89 1.30 296.85 1.17 0.90<br />
202<br />
208153<br />
Average 1.29 1.19 0.92<br />
0.09<br />
0.08 0.01<br />
209856<br />
204462<br />
Coeff. variation<br />
WP-T-stubs<br />
Weld_<br />
T1(i)<br />
3.50 0.47 73.77 55.10 0.75 52.36 0.71 0.95<br />
Weld_<br />
T1(ii)<br />
3.12 0.47 89.12 73.19 0.82 68.82 0.77 0.94<br />
Weld_<br />
T1(iii)<br />
2.82 0.46 107.29 94.26 0.88 87.68 0.82 0.93<br />
WT1<br />
WT2A<br />
WT2B<br />
WT4A<br />
3.27<br />
3.53<br />
3.08<br />
3.24<br />
0.47<br />
0.47<br />
0.47<br />
0.47<br />
71.08<br />
61.83<br />
79.75<br />
86.96<br />
75.96<br />
62.09<br />
88.19<br />
122.66<br />
1.07<br />
1.00<br />
1.11<br />
1.41<br />
71.61<br />
58.97<br />
82.62<br />
115.86<br />
1.01<br />
0.95<br />
1.04<br />
1.33<br />
0.94<br />
0.95<br />
0.94<br />
0.94<br />
WT51<br />
WT53C<br />
WT53D<br />
WT53E<br />
3.45<br />
3.40<br />
3.38<br />
3.40<br />
0.47<br />
0.47<br />
0.47<br />
0.47<br />
60.73<br />
64.23<br />
52.90<br />
64.82<br />
64.11<br />
65.60<br />
67.77<br />
65.44<br />
1.06<br />
1.02<br />
1.28<br />
1.01<br />
60.73<br />
62.12<br />
64.08<br />
61.94<br />
1.00<br />
0.97<br />
1.21<br />
0.96<br />
0.95<br />
0.95<br />
0.95<br />
0.95<br />
WT7_<br />
M12<br />
3.28 0.47 91.18 120.34 1.32 113.79 1.25 0.95<br />
WT7_<br />
M16<br />
3.28 0.47 116.09 124.29 1.07 117.27 1.01 0.94<br />
WT7_<br />
M20<br />
3.27 0.47 137.70 128.08 0.93 120.60 0.88 0.94<br />
WT57_<br />
M12<br />
3.38 0.47 85.78 105.35 1.23 99.95 1.17 0.95<br />
WT57_<br />
M16<br />
3.37 0.47 110.43 113.07 1.02 106.94 0.97 0.95<br />
WT57_<br />
M20<br />
3.38 0.47 150.96 115.51 0.77 109.09 0.72 0.94<br />
208153<br />
209856<br />
204462<br />
Average 1.04 0.99 0.94<br />
0.18<br />
0.18 0.01<br />
Coeff. variation
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table 6.10 Geometric characteristics <strong>of</strong> <strong>the</strong> specimens tested by Faella et al.<br />
Test ID Geometric characteristics (dimensions in [mm])<br />
b beff tf m n λ φ<br />
Kb<br />
(kN/mm)<br />
TS1 189.0 90.70 11.40 28.35 35.44 1.25 20 1.79×10 6<br />
TS2 189.0 116.03 11.00 41.01 49.85 1.22 20 1.84×10 6<br />
TS3 189.0 64.05 9.10 30.03 34.10 1.14 20 2.13×10 6<br />
TS4 189.0 96.33 9.35 31.16 33.10 1.06 20 2.09×10 6<br />
TS5 189.0 98.13 13.75 32.06 32.21 1.00 20 1.54×10 6<br />
TS6 189.0 95.50 13.45 30.75 32.98 1.07 20 1.57×10 6<br />
TS7 188.0 80.28 14.85 29.99 37.48 1.25 12 5.10×10 5<br />
TS8 189.0 77.12 14.90 28.41 35.51 1.25 12 5.09×10 5<br />
TS9 189.0 78.65 16.50 29.18 36.47 1.25 12 4.66×10 5<br />
TS10 189.0 78.15 15.50 28.93 36.16 1.25 12 4.92×10 5<br />
TS11 189.5 101.55 11.05 40.63 50.25 1.24 12 6.53×10 5<br />
TS12 189.5 76.50 10.70 28.10 35.13 1.25 12 6.71×10 5<br />
TS13 189.0 79.45 12.65 29.58 36.97 1.25 12 5.84×10 5<br />
TS14 189.0 80.65 12.70 30.18 37.72 1.25 12 5.82×10 5<br />
TS15 189.0 84.20 13.80 31.95 39.94 1.25 12 5.43×10 5<br />
TS16 189.5 82.65 13.45 31.18 38.97 1.25 12 5.55×10 5<br />
Table 6.11 Validation <strong>of</strong> <strong>the</strong> approach with fur<strong>the</strong>r examples tested by Faella,<br />
Piluso and Rizzano [6.3]: comparison <strong>of</strong> initial stiffness predictions<br />
(Young modulus in [MPa]; stiffness values in [kN/mm]).<br />
Test<br />
ID<br />
E m/tf 0.5F 1. Rd.0<br />
Exp.<br />
ke.0<br />
Beam model predictions<br />
0.5E Ered = 0.5EF1.<br />
Rd.0<br />
ke.0 Ratio<br />
ke.0 Ratio<br />
Act. 0.5E<br />
TS1 2.49 0.45 167 290.59 1.74 265.65 1.59 0.91<br />
TS2 3.73 0.48 112 122.13 1.09 116.51 1.04 0.95<br />
TS3 3.30 0.47 99 145.17 1.47 136.88 1.38 0.94<br />
TS4 3.33 0.47 103 146.62 1.42 138.43 1.34 0.94<br />
TS5 2.33 0.45 229 364.25 1.59 347.20 1.52 0.95<br />
TS6 2.29 0.44 214 372.57 1.74 338.93 1.58 0.91<br />
TS7 2.02 0.43 237 295.17 1.25 270.92 1.14 0.92<br />
TS8 1.91 0.43 213 317.39 1.49 289.89 1.36 0.91<br />
TS9 1.77 0.42 214 346.13 1.62 315.21 1.47 0.91<br />
TS10 1.87 0.42 266 325.77 1.22 297.45 1.12 0.91<br />
TS11 3.68 0.47 82 101.57 1.24 97.15 1.18 0.96<br />
TS12 2.63 0.45 168 184.66 1.10 171.63 1.02 0.93<br />
TS13 2.34 0.45 163 234.90 1.44 217.04 1.33 0.92<br />
TS14 2.38 0.45 156 229.40 1.47 212.29 1.36 0.93<br />
TS15 2.32 0.44 192 242.82 1.26 224.84 1.17 0.93<br />
TS16 2.32 0.44 179 241.14 1.35 223.10 1.25 0.93<br />
Average 1.41 1.30 0.93<br />
Coeff. variation<br />
0.14 0.14 0.02<br />
210000<br />
203
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Regarding <strong>the</strong> latter specimens, <strong>the</strong> Young modulus <strong>of</strong> <strong>the</strong> flange crosssection<br />
was taken as equal to 210 GPa, since this particular parameter was not<br />
defined by <strong>the</strong> authors. And since <strong>the</strong> characteristics <strong>of</strong> <strong>the</strong> bolts were not provided<br />
as well, <strong>the</strong> following assumptions were made to define <strong>the</strong> bolt elastic<br />
elongation behaviour: (i) Eb = 210 GPa, (ii) bolts are full-threaded, (iii) two<br />
washers per bolt with twsh = 2.95 mm for M20 bolts and twsh = 2.50 mm for<br />
M12 bolts, (iv) th = 13.10 mm for M20 bolts and th = 8.90 mm for M12 bolts<br />
and (v) tn = 16.00 mm for M20 bolts and tn = 11.90 mm for M12 bolts.<br />
Finally, a remark concerning <strong>the</strong> two values <strong>of</strong> b and beff that appear in Table<br />
6.10 is required. The beam model employs beff for <strong>the</strong> definition <strong>of</strong> <strong>the</strong><br />
cross-section width, which is kept constant throughout <strong>the</strong> load history. As<br />
mentioned earlier, this width accounts for all possible yield line mechanisms <strong>of</strong><br />
<strong>the</strong> T-stub flange and cannot exceed <strong>the</strong> actual flange width, b. For <strong>the</strong> examples<br />
from <strong>the</strong> database, beff = b; however, for <strong>the</strong> set <strong>of</strong> examples tested by<br />
Faella and co-authors, b > beff and so <strong>the</strong> smallest value governs.<br />
Physically, it is quite clear that <strong>the</strong> section width that contributes to load<br />
transmission expands with <strong>the</strong> loading provided that <strong>the</strong> actual flange width is<br />
not exceeded. The assumption <strong>of</strong> a constant flange width <strong>of</strong> beff seems appropriate<br />
until strain hardening begins (<strong>the</strong> reduction at elastic conditions can be<br />
done indirectly with <strong>the</strong> reduction <strong>of</strong> <strong>the</strong> Young modulus, for instance), but it<br />
may be too conservative at ultimate conditions.<br />
6.3.3 Analysis <strong>of</strong> <strong>the</strong> model in <strong>the</strong> elastoplastic range<br />
Having carried out <strong>the</strong> full nonlinear analysis <strong>of</strong> <strong>the</strong> beam model, numerous results<br />
can be extracted, namely: (i) load-carrying behaviour, (ii) evolution <strong>of</strong> <strong>the</strong><br />
prying forces, (iii) flange moment diagram, (iv) flange plastic strain diagram,<br />
etc. Thorough results for specimens T1 and WT1 are given in App<strong>end</strong>ix D to<br />
illustrate <strong>the</strong> capacities <strong>of</strong> <strong>the</strong> model. In this section, <strong>the</strong> F-∆ response <strong>of</strong> <strong>the</strong><br />
specimens from <strong>the</strong> database is characterized up to collapse in <strong>the</strong> framework<br />
<strong>of</strong> <strong>the</strong> proposed methodology. This simplified load-carrying behaviour is compared<br />
with <strong>the</strong> actual response and <strong>the</strong> bilinear approximation proposed by Jaspart.<br />
Firstly, <strong>the</strong> collapse modes are defined. Table 6.12 sets out <strong>the</strong> actual determining<br />
fracture element: bolt or T-stub flange. The prediction <strong>of</strong> <strong>the</strong> failure<br />
modes as described in Chapter 2, which is based on a force criterion, is also<br />
given. These predictions are generally in line with <strong>the</strong> observed failure modes,<br />
except for specimens P13, WT53D and WT53E. Such situation may derive<br />
from <strong>the</strong> fact that <strong>the</strong> fracture criterion for <strong>the</strong> numerical three-dimensional<br />
model and for <strong>the</strong> beam model is based on a deformation condition. This may<br />
introduce some differences. This table also indicates <strong>the</strong> fracture element that is<br />
determinant in <strong>the</strong> two-dimensional model. Here, <strong>the</strong> differences are more frequent<br />
(underlined specimens). The critical section (1*) that is referred to in <strong>the</strong><br />
table is <strong>the</strong> critical section right at <strong>the</strong> <strong>end</strong> <strong>of</strong> <strong>the</strong> fillet weld or radius. After<br />
204
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table 6.12 Prediction <strong>of</strong> <strong>the</strong> failure modes.<br />
Test ID Actual deter- Predicted poten- Determining fracture<br />
mining fracture tial failure mode. element in <strong>the</strong> beam<br />
element Mf.u Eq.<br />
(2.4)<br />
Mf.u Eq.<br />
(2.5)<br />
model<br />
T1 Bolt 13 13 Bolt<br />
P1 Bolt 13 13 Bolt<br />
P2 Bolt 13 13 Bolt<br />
P3 Bolt 13 13 Flange, at (1*)<br />
P4 Bolt 13 13 Bolt<br />
P5 Bolt 23 23 Bolt<br />
P6 Bolt 13 13 Bolt<br />
P7 Bolt 13 13 Bolt<br />
P8 Bolt 13 13 Bolt<br />
P9 Bolt 23 23 Bolt<br />
P10 Flange 11 11 Flange, at (1*)<br />
P11 Bolt 23 23 Bolt<br />
P12 Flange 11 11 Flange, at (1*)<br />
P13 Bolt 11 11 Flange, at (1*)<br />
P14 Flange 11 11 Flange, at (1*)<br />
P15 Flange 11 11 Flange, at (1*)<br />
P16 Bolt 23 23 Bolt<br />
P17 Bolt 13 13<br />
Bolt and flange at<br />
(1*) (simultaneously)<br />
P18 Flange 11 11 Flange, at (1*)<br />
P19 Bolt 13 13<br />
Bolt and flange at<br />
(1*) (simultaneously)<br />
P20 Bolt 23 23 Bolt<br />
P21 Bolt 13 13 Flange at <strong>the</strong> bolt axis<br />
P22 Bolt 13 13 Bolt<br />
P23 Bolt 23 23 Bolt<br />
Weld_T1(i) Bolt 11 11 Flange, at (1*)<br />
Weld_T1(ii) Bolt 13 13 Flange, at (1*)<br />
Weld_T1(iii) Bolt 13 13 Bolt<br />
WT1 Bolt and flange 11 13 Flange, at (1*)<br />
WT2A Bolt and flange 11 13 Flange, at (1*)<br />
WT2B Bolt and flange 11 13 Flange, at (1*)<br />
WT4A Bolt 13 13 Bolt<br />
WT51 Bolt 23 23 Flange, at (1*)<br />
WT53C Bolt 13 13 Flange, at (1*)<br />
WT53D Bolt 11 13 Flange, at (1*)<br />
WT53E Bolt 11 13 Flange, at (1*)<br />
WT7_M12 Bolt 13 13 Bolt<br />
WT7_M16 Flange 11 13 Flange, at (1*)<br />
WT7_M20 Flange 11 11 Flange, at (1*)<br />
WT57_M12 Bolt 23 23 Flange, at (1*)<br />
WT57_M16 Bolt (stripping) 13 13 Flange, at (1*)<br />
WT57_M20 Bolt and flange 11 11 Flange, at (1*)<br />
205
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
application <strong>of</strong> <strong>the</strong> model and within <strong>the</strong> process <strong>of</strong> calibration <strong>of</strong> <strong>the</strong> model, it<br />
was observed that <strong>the</strong> imposition <strong>of</strong> cracking <strong>of</strong> <strong>the</strong> material at section (1), at<br />
<strong>the</strong> flange-to-web connection, as an ultimate condition was a too severe condition<br />
indeed (§6.3.1.1). Therefore, collapse occurs when ei<strong>the</strong>r or both <strong>of</strong> <strong>the</strong><br />
following conditions are verified: (i) fracture <strong>of</strong> <strong>the</strong> bolt and (ii) cracking <strong>of</strong><br />
material <strong>of</strong> <strong>the</strong> flange at <strong>the</strong> <strong>end</strong> <strong>of</strong> <strong>the</strong> fillet weld or radius, section (1*).<br />
Table 6.13 summarizes <strong>the</strong> predictions <strong>of</strong> deformation capacity and ultimate<br />
resistance <strong>of</strong> <strong>the</strong> proposed model. The predictions are compared with <strong>the</strong><br />
actual results. The ultimate resistance is well estimated and <strong>the</strong> scatter <strong>of</strong> results<br />
is lower since <strong>the</strong> variation is also low (coefficient <strong>of</strong> variation <strong>of</strong> 0.13 for<br />
HR-T-stubs and 0.15 for WP-T-stubs). Regarding <strong>the</strong> deformation capacity, <strong>the</strong><br />
differences are more relevant. However, this disparity has to be accepted<br />
within reasonable limits due to all <strong>the</strong> simplifications inherent to this twodimensional<br />
approach. In this table <strong>the</strong> specimens are separated according to<br />
<strong>the</strong>ir assembly type. Table 6.14 presents identical results but with specimens<br />
grouped according to <strong>the</strong> potential failure type (Mf.u from Eq. (2.4)). The specimens<br />
whose actual failure type was not well predicted were excluded from this<br />
table. If now <strong>the</strong> specimens are analysed in this context, <strong>the</strong> following conclusions<br />
may be drawn: (i) for specimens that fail according to type-23, both predictions<br />
<strong>of</strong> deformation and resistance at ultimate conditions are good, with average<br />
errors smaller than 10%, (ii) identical conclusions are valid for specimens<br />
<strong>of</strong> type-11 failure, regarding <strong>the</strong> evaluation <strong>of</strong> deformation capacity, (iii)<br />
for <strong>the</strong>se latter specimens <strong>the</strong> ultimate resistance prediction is conservative, (iv)<br />
for those specimens failing according to a type-13 mode, <strong>the</strong> predictions for resistance<br />
are good and (v) <strong>the</strong> evaluation <strong>of</strong> deformation capacity for <strong>the</strong>se<br />
specimens is ra<strong>the</strong>r weak. It should be noted that <strong>the</strong> deformation capacity, ∆u.0,<br />
that appears in <strong>the</strong> above tables corresponds to <strong>the</strong> deformation <strong>of</strong> <strong>the</strong> T-stub<br />
when <strong>the</strong> maximum load is reached. This definition may be quite conservative<br />
when <strong>the</strong> experimental results are taken for comparison since <strong>the</strong> s<strong>of</strong>tening<br />
branches sometimes can be quite ext<strong>end</strong>ed (specimen WT7_M20, for example).<br />
Figs. 6.12-6.14 illustrate <strong>the</strong> load-carrying behaviour for some specimens<br />
that represent <strong>the</strong> various collapse modes. The curves are compared with <strong>the</strong><br />
actual response and <strong>the</strong> bilinear approximation <strong>of</strong> Jaspart. This bilinear approximation<br />
was defined using <strong>the</strong> formulation accounting for <strong>the</strong> bolt action<br />
for specimens <strong>of</strong> failure type-11 and <strong>the</strong> actual material strain hardening<br />
modulus. It becomes clear from <strong>the</strong>se curves that <strong>the</strong> beam model provides a<br />
better agreement with <strong>the</strong> real F-∆ response and in general <strong>the</strong>se two curves fit<br />
well. Finally, Fig. 6.15 compares <strong>the</strong> responses for those specimens whose<br />
failure mode was not well predicted by <strong>the</strong> beam model. None<strong>the</strong>less, <strong>the</strong><br />
agreement is surprisingly good.<br />
Fig. 6.15e traces <strong>the</strong> F-∆ behaviour <strong>of</strong> specimen WT57_M12. For this<br />
specimen <strong>the</strong> flange <strong>plate</strong>s are fastened by means <strong>of</strong> two full-threaded bolts. At collapse,<br />
<strong>the</strong> bolt model estimates a fracture elongation <strong>of</strong> 4 mm – Eq. (6.14). Since<br />
206
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table 6.13 Prediction <strong>of</strong> deformation capacity and ultimate resistance.<br />
Test ID Potential<br />
failure<br />
Actual results Beam model predictions<br />
Fmax ∆u.0 Fu.0 Ratio ∆u.0 Ratio<br />
type (kN) (mm) (kN) (mm)<br />
T1 13 103.99 8.70 114.45 1.10 16.76 1.93<br />
P1 13 91.76 10.77 103.25 1.13 25.34 2.35<br />
P2 13 116.72 6.18 124.34 1.07 8.75 1.42<br />
P3 13 95.41 10.17 111.96 1.17 24.17 2.38<br />
P4 13 115.97 4.68 120.25 1.04 7.23 1.55<br />
P5 23 130.20 3.63 123.76 0.95 4.63 1.28<br />
P6 13 95.53 10.06 111.96 1.17 24.17 2.40<br />
P7 13 111.34 7.56 116.43 1.05 11.66 1.54<br />
P8 13 112.71 8.08 124.24 1.10 12.20 1.51<br />
P9 23 131.43 3.31 137.48 1.05 3.67 1.11<br />
P10 11 76.79 32.75 50.25 0.65 32.40 0.99<br />
P11 23 121.15 2.94 134.34 1.11 3.47 1.18<br />
P12 11 154.06 24.22 122.43 0.79 19.67 0.81<br />
P13 11 93.71 11.38 97.44 1.04 18.57 1.63<br />
P14 11 86.57 24.15 79.54 0.92 20.49 0.85<br />
P15 11 171.08 18.02 153.17 0.90 16.48 0.91<br />
P16 23 125.69 3.06 127.98 1.02 3.00 0.98<br />
P17 13 192.01 9.29 212.18 1.11 20.67 2.22<br />
P18 11 266.57 26.07 215.75 0.81 20.18 0.77<br />
P19 13 186.52 9.29 212.18 1.14 20.67 2.23<br />
P20 23 225.94 4.06 246.14 1.09 4.03 0.99<br />
P21 13 281.33 17.67 285.70 1.02 20.61 1.17<br />
P22 13 305.21 6.40 345.44 1.13 13.75 2.15<br />
P23 23 346.01 5.22 408.62 1.18 5.24 1.00<br />
Average 1.03 1.47<br />
Coefficient <strong>of</strong> variation 0.13 0.38<br />
Weld_T1(i) 11 92.02 10.85 84.80 0.92 18.27 1.68<br />
Weld_T1(ii) 13 102.75 8.01 101.03 0.98 20.00 2.50<br />
Weld_T1(iii) 13 113.10 6.22 114.21 1.01 18.82 3.02<br />
WT1 11 91.91 14.32 85.60 0.93 18.51 1.29<br />
WT2A 11 86.82 17.98 75.16 0.87 16.51 0.92<br />
WT2B 11 97.88 13.09 92.51 0.95 19.18 1.47<br />
WT4A 13 103.26 4.33 124.40 1.20 10.77 2.49<br />
WT51 23 97.08 3.96 112.86 1.16 9.43 2.38<br />
WT53C 13 98.90 4.24 114.45 1.16 8.50 2.00<br />
WT53D 11 117.36 5.54 117.02 1.00 8.23 1.49<br />
WT53E 11 115.04 5.26 116.04 1.01 9.03 1.72<br />
WT7_M12 13 100.34 4.60 123.83 1.23 11.00 2.39<br />
WT7_M16 11 132.34 11.47 140.79 1.06 17.78 1.55<br />
WT7_M20 11 145.72 9.12 141.17 0.97 17.87 1.96<br />
WT57_M12 23 121.87 4.33 174.51 1.43 8.07 1.86<br />
WT57_M16 13 173.64 5.88 196.62 1.13 9.13 1.55<br />
WT57_M20 11 241.71 15.98 196.77 0.81 8.36 0.52<br />
Average 1.05 1.81<br />
Coefficient <strong>of</strong> variation<br />
0.15<br />
0.34<br />
207
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 6.14 Prediction <strong>of</strong> deformation capacity and ultimate resistance: specimens<br />
organized by failure type group.<br />
208<br />
Test ID Potential<br />
failure<br />
Actual results Beam model predictions<br />
Fmax ∆u.0 Fu.0 Ratio ∆u.0 Ratio<br />
type (kN) (mm) (kN) (mm)<br />
T1 13 103.99 8.70 114.45 1.10 16.76 1.93<br />
P1 13 91.76 10.77 103.25 1.13 25.34 2.35<br />
P2 13 116.72 6.18 124.34 1.07 8.75 1.42<br />
P4 13 115.97 4.68 120.25 1.04 7.23 1.55<br />
P6 13 95.53 10.06 111.96 1.17 24.17 2.40<br />
P7 13 111.34 7.56 116.43 1.05 11.66 1.54<br />
P8 13 112.71 8.08 124.24 1.10 12.20 1.51<br />
P17 13 192.01 9.29 212.18 1.11 20.67 2.22<br />
P19 13 186.52 9.29 212.18 1.14 20.67 2.23<br />
P22 13 305.21 6.40 345.44 1.13 13.75 2.15<br />
Weld_T1(iii) 13 113.10 6.22 114.21 1.01 18.82 3.02<br />
WT4A 13 103.26 4.33 124.40 1.20 10.77 2.49<br />
WT7_M12 13 100.34 4.60 123.83 1.23 11.00 2.39<br />
Average 1.11 2.09<br />
Coefficient <strong>of</strong> variation 0.06 0.23<br />
P10 11 76.79 32.75 50.25 0.65 32.40 0.99<br />
P12 11 154.06 24.22 122.43 0.79 19.67 0.81<br />
P14 11 86.57 24.15 79.54 0.92 20.49 0.85<br />
P15 11 171.08 18.02 153.17 0.90 16.48 0.91<br />
P18 11 266.57 26.07 215.75 0.81 20.18 0.77<br />
WT7_M16 11 132.34 11.47 140.79 1.06 17.78 1.55<br />
WT7_M20 11 145.72 9.12 141.17 0.97 17.87 1.96<br />
WT57_M20 11 241.71 15.98 196.77 0.81 8.36 0.52<br />
Average 0.86 1.05<br />
Coefficient <strong>of</strong> variation 0.14 0.45<br />
P5 23 130.20 3.63 123.76 0.95 4.63 1.28<br />
P9 23 131.43 3.31 137.48 1.05 3.67 1.11<br />
P11 23 121.15 2.94 134.34 1.11 3.47 1.18<br />
P16 23 125.69 3.06 127.98 1.02 3.00 0.98<br />
P20 23 225.94 4.06 246.14 1.09 4.03 0.99<br />
P23 23 346.01 5.22 408.62 1.18 5.24 1.00<br />
Average 1.07 1.09<br />
Coefficient <strong>of</strong> variation<br />
0.07<br />
0.11<br />
bolt governs fracture <strong>of</strong> this specimen, <strong>the</strong> post-limit behaviour proceeds until this<br />
deformation <strong>of</strong> 4 mm is attained, leading to an overall deformation <strong>of</strong> 8.1 mm and<br />
ultimate resistance <strong>of</strong> 174.5 kN, corresponding to 1.43 times <strong>the</strong> maximum resis-
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
(a) Specimen P1.<br />
Load, F (kN)<br />
(b) Specimen P2.<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Beam model)<br />
15<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 3 6 9 12 15 18 21 24 27 30 33<br />
Deformation, ∆ (mm)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Beam model)<br />
15<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 1 2 3 4 5 6 7 8 9 10 11 12 13<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
15<br />
Deformation, ∆ (mm)<br />
Actual response<br />
Simplified response (Beam model)<br />
Bilinear approximation (Jaspart)<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20 22 24 26<br />
Deformation, ∆ (mm)<br />
(c) Specimen Weld_T1(iii).<br />
Fig. 6.12 Specimens that fail according to type-13: comparison <strong>of</strong> <strong>the</strong> actual<br />
response with <strong>the</strong> beam model predictions.<br />
209
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
210<br />
Load, F (kN)<br />
(a) Specimen P10.<br />
Load, F (kN)<br />
(b) Specimen P18.<br />
Load, F (kN)<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
Actual response<br />
20<br />
Simplified response (Beam model)<br />
10<br />
0<br />
Bilinear approximation<br />
0 3 6 9 12 15 18 21 24 27 30 33 36<br />
280<br />
240<br />
200<br />
160<br />
Deformation, ∆ (mm)<br />
120<br />
Actual response<br />
80<br />
Simplified response (Beam model)<br />
40<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 3 6 9 12 15 18 21 24 27 30<br />
160<br />
140<br />
120<br />
100<br />
80<br />
Deformation, ∆ (mm)<br />
60<br />
Actual response<br />
40<br />
Simplified response (Beam model)<br />
20<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Deformation, ∆ (mm)<br />
(c) Specimen WT7_M20.<br />
Fig. 6.13 Specimens that fail according to type-11: comparison <strong>of</strong> <strong>the</strong> actual<br />
response with <strong>the</strong> beam model predictions.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
(a) Specimen P5.<br />
Load, F (kN)<br />
(b) Specimen P16.<br />
Load, F (kN)<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
Actual response<br />
Simplified response (Beam model)<br />
20<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
140<br />
120<br />
100<br />
80<br />
Deformation, ∆ (mm)<br />
60<br />
Actual response<br />
40<br />
Simplified response (Beam model)<br />
20<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
450<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
Deformation, ∆ (mm)<br />
Actual response<br />
Simplified response (Beam model)<br />
50<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 1 2 3 4 5 6 7 8 9 10 11<br />
Deformation, ∆ (mm)<br />
(c) Specimen P23.<br />
Fig. 6.14 Specimens that fail according to type-23: comparison <strong>of</strong> <strong>the</strong> actual<br />
response with <strong>the</strong> beam model predictions.<br />
211
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
212<br />
Load, F (kN)<br />
(a) Specimen Weld_T1(i).<br />
Load, F (kN)<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
Actual response<br />
20<br />
Simplified response (Beam model)<br />
10<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 3 6 9 12 15 18 21 24 27 30 33 36 39<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
15<br />
(b) Specimen WT2B.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
Actual response<br />
Simplified response (Beam model)<br />
Bilinear approximation (Jaspart)<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
15<br />
Deformation, ∆ (mm)<br />
Actual response<br />
Simplified response (Beam model)<br />
Bilinear approximation (Jaspart)<br />
0<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
(c) Specimen WT51.<br />
Fig. 6.15 WP-T-stub specimens whose observed failure types are not coincident<br />
with <strong>the</strong> predictions.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
(d) Specimen WT53D.<br />
Load, F (kN)<br />
45<br />
Actual response<br />
30<br />
Simplified response (Beam model)<br />
15<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 1 2 3 4 5 6 7 8 9 10<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
(e) Specimen WT57_M12.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
Actual response<br />
Simplified response (Beam model)<br />
20<br />
0<br />
Bilinear approximation (Jaspart)<br />
0 1 2 3 4 5 6 7 8 9 10<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Deformation, ∆ (mm)<br />
Actual response<br />
Simplified response (Beam model)<br />
Bilinear approximation (Jaspart)<br />
0<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
(f) Specimen WT57_M16.<br />
Fig. 6.15 WP-T-stub specimens whose observed failure types are not coincident<br />
with <strong>the</strong> predictions (cont.).<br />
213
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
tance from <strong>the</strong> tests. With reference to Eq. (6.14) derived for δb.fract, it was developed<br />
for short-threaded bolts [6.6]. If a full-threaded bolt is considered instead,<br />
this expression seems to overestimate <strong>the</strong> bolt fracture deformation. Consequently,<br />
some guidelines concerning this matter are required.<br />
In terms <strong>of</strong> design calculations, <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> <strong>the</strong><br />
above specimens should be based in nominal properties, as already mentioned.<br />
The results obtained by this procedure are fully described and analysed in App<strong>end</strong>ix<br />
D.<br />
6.3.4 Sophistication <strong>of</strong> <strong>the</strong> proposed method: modelling <strong>of</strong> <strong>the</strong> bolt action<br />
as a distributed load<br />
Jaspart has shown that a significant increase in <strong>the</strong> resistance <strong>of</strong> single T-stubs<br />
that fail according to a plastic mechanism type-1 can be expected due to <strong>the</strong> influence<br />
<strong>of</strong> <strong>the</strong> bolt action on a finite contact area [6.2]. This effect is taken into<br />
account in <strong>the</strong> beam model in this section. The bolt is <strong>the</strong>n modelled as a spring<br />
assembly in parallel, as shown in Fig. 6.16. The length <strong>of</strong> this assembly is <strong>the</strong><br />
bolt diameter. The behaviour <strong>of</strong> this spring assembly is <strong>the</strong> same as <strong>the</strong> original<br />
single spring, i.e. <strong>the</strong> spring stiffness and force values are divided by <strong>the</strong> number<br />
<strong>of</strong> springs in <strong>the</strong> assembly.<br />
Eleven examples were chosen to illustrate this modification, five examples<br />
from type-11 group and three examples from each <strong>of</strong> <strong>the</strong> o<strong>the</strong>r two groups. The<br />
analysis <strong>of</strong> <strong>the</strong> first group is quite straightforward as <strong>the</strong> bolt is not engaged in<br />
<strong>the</strong> failure mode. For those specimens whose fracture is governed by <strong>the</strong> bolt<br />
itself, <strong>the</strong> fracture condition has to be redefined. Below, this condition is imposed<br />
at two different sections and <strong>the</strong> results are <strong>the</strong>n compared: (i) section (2)<br />
from above, at mid- bolt section and (ii) section (2*), located at ¼ <strong>of</strong> <strong>the</strong> inside<br />
bolt edge, from section (2). For those specimens that fail according to a type-23<br />
(1)<br />
mechanism, when <strong>the</strong> bolt fractures, ε p < ε pu . , specimens belonging to type-<br />
(1)<br />
13 failure mode may exhibit ε > ε when <strong>the</strong> bolt fracture but<br />
(1*) (2)<br />
ε p , ε p < ε p. u .<br />
214<br />
p pu .<br />
Table 6.15 summarizes <strong>the</strong> predicted resistance and deformation values at<br />
collapse when this modification is introduced. The examples are grouped according<br />
to <strong>the</strong> failure mode. The underlined connections, again, refer to those<br />
cases where <strong>the</strong> predicted failure mode does not match <strong>the</strong> observed type. For<br />
evaluation <strong>of</strong> ratios to actual values <strong>the</strong>se examples are not taken into consideration.<br />
The fracture condition here was identical to <strong>the</strong> above.<br />
The application <strong>of</strong> this modified model provides a significant enhancement <strong>of</strong><br />
results in terms <strong>of</strong> resistance, particularly for <strong>the</strong> evaluation <strong>of</strong> Fu.0. So are <strong>the</strong> predictions<br />
<strong>of</strong> deformation capacity for specimens from type-11 group. Specimens<br />
that fail according to a type-23 mechanism show worse predictions <strong>of</strong> deformation<br />
capacity. For specimens belonging to type-13 fracture mode, <strong>the</strong>se predict-
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
0.8r<br />
F<br />
2<br />
B-δb<br />
φ<br />
m n<br />
Fig. 6.16 Beam model accounting for <strong>the</strong> bolt action on a finite area.<br />
ions improve but are still distant from <strong>the</strong> actual deformation values. Fur<strong>the</strong>r<br />
comparisons are carried out in App<strong>end</strong>ix D.<br />
It is worth mentioning that this sophistication enforced <strong>the</strong> “correct” fracture<br />
element to be critical in specimens P3 and Weld_T1(ii), for instance. Never<strong>the</strong>less,<br />
<strong>the</strong> beam model is not yet able to simulate <strong>the</strong> fracture <strong>of</strong> <strong>the</strong> bolt simultaneously<br />
to cracking <strong>of</strong> <strong>the</strong> flange material in some WP-T-stubs.<br />
6.3.5 Influence <strong>of</strong> <strong>the</strong> distance m for <strong>the</strong> WP-T-stubs<br />
Distance m is one <strong>of</strong> <strong>the</strong> geometrical parameters that most influences <strong>the</strong> deformation<br />
behaviour <strong>of</strong> T-stub connections. For HR-T-stubs, this distance is<br />
well established and <strong>the</strong>re is experimental and numerical evidence for its definition.<br />
Common procedures for WP-T-stubs consisted in extrapolating <strong>the</strong> design<br />
rules defined for HR-T-stubs. Parameter m defines <strong>the</strong> location <strong>of</strong> <strong>the</strong> potential<br />
yield line at <strong>the</strong> flange-to-web connection with respect to <strong>the</strong> bolt line.<br />
In previous chapters, it has been shown that this procedure may not be accurate<br />
enough. In fact, <strong>the</strong>re is evidence that in some cases <strong>the</strong> yield line at <strong>the</strong> flangeto-web<br />
connection develops in <strong>the</strong> flange at <strong>the</strong> <strong>end</strong> <strong>of</strong> <strong>the</strong> fillet root, i.e. for a<br />
distance m defined as follows:<br />
m = d − 2aw (6.35)<br />
215
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table 6.15 Prediction <strong>of</strong> deformation capacity and ultimate resistance by applying<br />
<strong>the</strong> sophisticated beam model accounting for bolt action:<br />
specimens organized by failure type group.<br />
216<br />
Test ID Potential<br />
failure<br />
type<br />
Actual results Sophisticated beam model predictions<br />
Fmax ∆ u.0 Fu.0 Ratio ∆ u.0 Ratio<br />
(kN) (mm) (kN) (mm)<br />
T1 13 103.99 8.70 113.87 1.09 12.74 1.46<br />
P1 13 91.76 10.77 103.27 1.13 19.88 1.85<br />
P3 13 95.41 10.17 108.18 1.13 16.06 1.58<br />
Weld_T1(ii) 13 102.75 8.01 104.18 1.01 17.61 2.20<br />
Weld_T1(iii) 13 113.10 6.22 113.33 1.00 14.16 2.27<br />
WT57_M16 13 173.64 5.88 212.19 1.22 8.35 1.42<br />
Average 1.07 1.87<br />
Coefficient <strong>of</strong> variation 0.06 0.19<br />
P10 11 76.79 32.75 59.17 0.77 27.79 0.85<br />
P12 11 154.06 24.22 149.02 0.97 19.22 0.79<br />
P14 11 86.57 24.15 88.79 1.03 19.87 0.82<br />
P15 11 171.08 18.02 187.94 1.10 16.16 0.90<br />
P18 11 266.57 26.07 280.34 1.05 18.69 0.72<br />
Weld_T1(i) 11 92.02 10.85 88.57 0.96 17.17 1.58<br />
WT1 11 91.91 14.32 90.37 0.98 15.67 1.09<br />
WT2A 11 86.82 17.98 81.50 0.94 16.11 0.90<br />
WT2B 11 97.88 13.09 102.67 1.05 19.83 1.10<br />
WT7_M16 11 132.34 11.47 158.47 1.20 16.24 1.42<br />
WT7_M20 11 145.72 9.12 177.93 1.22 14.71 1.61<br />
WT57_M20 11 241.71 15.98 233.41 0.97 7.56 0.47<br />
Average 1.02 0.97<br />
Coefficient <strong>of</strong> variation 0.12 0.33<br />
P5 23 130.20 3.63 121.71 0.93 3.37 0.93<br />
P20 23 225.94 4.06 247.76 1.10 3.51 0.87<br />
P23 23 346.01 5.22 399.89 1.16 4.00 0.77<br />
WT51 23 97.08 3.96 119.08 1.23 9.00 2.27<br />
Average 1.06 0.85<br />
Coefficient <strong>of</strong> variation<br />
0.11<br />
0.10<br />
The influence <strong>of</strong> this distance is fur<strong>the</strong>r detailed in App<strong>end</strong>ix D. Generally<br />
speaking, if m from Eq. (6.35) is adopted, <strong>the</strong>re is an increase on resistance and<br />
stiffness and decrease on <strong>ductility</strong>.<br />
6.4 CONCLUDING REMARKS<br />
The proposed beam model yields an accurate prediction <strong>of</strong> <strong>the</strong> F-∆ response <strong>of</strong><br />
<strong>bolted</strong> T-stub connections, despite <strong>the</strong> simplifications inherent to a two-
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
dimensional modelling <strong>of</strong> <strong>the</strong> behaviour. These reduced <strong>the</strong> model complexity<br />
to a more reasonable level, when compared to <strong>the</strong> three-dimensional FE modelling.<br />
However, to obtain <strong>the</strong> F-∆ curve, a numerical incremental procedure is<br />
still required and, consequently, <strong>the</strong> model is not suitable for hand calculations.<br />
The dominant effects in both approaches are <strong>the</strong> strain hardening <strong>of</strong> <strong>the</strong> flange<br />
and <strong>the</strong> bolt elongation behaviour, as confirmed by experimental evidence. Ano<strong>the</strong>r<br />
important simplification <strong>of</strong> <strong>the</strong> beam model corresponds to <strong>the</strong> beam section<br />
width, which is kept constant with <strong>the</strong> course <strong>of</strong> loading. As <strong>the</strong> load increases,<br />
<strong>the</strong> flange width tributary to load transmission also increases. The<br />
analysis <strong>of</strong> this variation was not carried out and <strong>the</strong> implementation <strong>of</strong> such<br />
phenomenon is not straightforward nei<strong>the</strong>r. Never<strong>the</strong>less, for those specimens<br />
failing according to a type-11 mode, this issue can be particularly relevant.<br />
The applicability <strong>of</strong> <strong>the</strong> model was well demonstrated within <strong>the</strong> range <strong>of</strong><br />
examples shown above. The behaviour predicted by this model is ra<strong>the</strong>r good<br />
in terms <strong>of</strong> resistance. With respect to <strong>ductility</strong>, it reflects an overestimation <strong>of</strong><br />
test results that is within an acceptable error. Table 6.16 summarizes <strong>the</strong> statistical<br />
parameters (average and coefficient <strong>of</strong> variation) corresponding to <strong>the</strong><br />
sample <strong>of</strong> connections that were analysed above. Two approaches in terms <strong>of</strong><br />
material properties are taken into account: actual properties (Table 6.13) and<br />
nominal properties (Table D.17 in App<strong>end</strong>ix D).<br />
If <strong>the</strong> results are analysed in terms <strong>of</strong> failure types (cf. Table 6.14), <strong>the</strong> predictions<br />
for resistance are accurate for those specimens whose fracture is determined<br />
by <strong>the</strong> bolt. For those specimens failing according to a type-11, <strong>the</strong> results<br />
seem ra<strong>the</strong>r conservative. Concerning <strong>the</strong> predictions for deformation capacity,<br />
<strong>the</strong>se are quite good for failure modes type-11 and -23, even though <strong>the</strong> scatter <strong>of</strong><br />
results for specimens <strong>of</strong> failure type-11 is high (coefficient <strong>of</strong> variation <strong>of</strong> 0.45).<br />
For <strong>the</strong> remaining case (type-23 failure), <strong>the</strong>re is an overestimation <strong>of</strong> results.<br />
The modification for inclusion <strong>of</strong> <strong>the</strong> bolt action provides an enhancement <strong>of</strong><br />
results but introduces an additional complexity. From a design point <strong>of</strong> view, <strong>the</strong><br />
methodology should be fur<strong>the</strong>r simplified so that it can be used in an expedite way,<br />
as Jaspart’s simple proposal. This can be achieved by modelling plasticity phenomena<br />
in <strong>the</strong> flange by means <strong>of</strong> rotational springs at <strong>the</strong> critical sections that capture<br />
<strong>the</strong> overall behaviour.<br />
Table 6.16 Summary <strong>of</strong> <strong>the</strong> proposed beam model from a statistical point <strong>of</strong><br />
view (average ratios and coefficients <strong>of</strong> variation, <strong>the</strong> latter in<br />
italic) for evaluation <strong>of</strong> <strong>the</strong> force-deformation characteristics.<br />
T-stub assembly<br />
Ultimate resistance<br />
Fu.0<br />
Deformation capacity<br />
HR<br />
WP<br />
Actual material properties<br />
1.03 (0.13)<br />
1.05 (0.15)<br />
1.47 (0.38)<br />
1.81 (0.34)<br />
HR Nominal material proper- 0.98 (0.17) 1.73 (0.38)<br />
WP<br />
ties 0.94 (0.23) 1.18 (0.53)<br />
∆u.0<br />
217
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
6.5 REFERENCES<br />
[6.1] European Committee for Standardization (CEN). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[6.2] Jaspart JP. Study <strong>of</strong> <strong>the</strong> semi-rigid behaviour <strong>of</strong> beam-to-column joints<br />
and <strong>of</strong> its influence on <strong>the</strong> stability and strength <strong>of</strong> steel building frames.<br />
PhD <strong>the</strong>sis (in French). University <strong>of</strong> Liège, Liège, Belgium, 1991.<br />
[6.3] Faella C, Piluso V, Rizzano G. Structural semi-rigid connections – <strong>the</strong>ory,<br />
design and s<strong>of</strong>tware. CRC Press, USA, 2000.<br />
[6.4] Piluso V, Faella C, Rizzano G. Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. I:<br />
<strong>the</strong>oretical model. Journal <strong>of</strong> Structural Engineering ASCE; 127(6):686-<br />
693, 2001.<br />
[6.5] Piluso V, Faella C, Rizzano G. Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. II:<br />
model validation. Journal <strong>of</strong> Structural Engineering ASCE; 127(6):694-<br />
704, 2001.<br />
[6.6] Swanson JA. Characterization <strong>of</strong> <strong>the</strong> strength, stiffness and <strong>ductility</strong> behavior<br />
<strong>of</strong> T-stub connections. PhD dissertation, Georgia Institute <strong>of</strong><br />
Technology, Atlanta, USA, 1999.<br />
[6.7] Beg D, Zupančič E, Vayas I. On <strong>the</strong> rotation capacity <strong>of</strong> moment connections.<br />
Journal <strong>of</strong> Constructional Steel Research; 60:601-620, 2004.<br />
[6.8] Jaspart JP. Contributions to recent advances in <strong>the</strong> field <strong>of</strong> steel joints – column<br />
bases and fur<strong>the</strong>r configurations for beam-to-column joints and beam<br />
splices. Aggregation <strong>the</strong>sis. University <strong>of</strong> Liège, Liège, Belgium, 1997.<br />
[6.9] Yee YL, Melchers RE. Moment-rotation curves for <strong>bolted</strong> connections.<br />
Journal <strong>of</strong> Structural Engineering ASCE; 112(3):615-635, 1986.<br />
[6.10] Maquoi R, Jaspart JP. Moment-rotation curves for <strong>bolted</strong> connections:<br />
Discussion <strong>of</strong> <strong>the</strong> paper by Yee YL and Melchers RE. Journal <strong>of</strong> Structural<br />
Engineering ASCE; 113(10):2324-2329, 1986.<br />
[6.11] Gioncu V, Mateescu G, Petcu D, Anastasiadis A. Prediction <strong>of</strong> available<br />
<strong>ductility</strong> by means <strong>of</strong> local plastic mechanism method: DUCTROT<br />
computer program, Chapter 2.1 in Moment resistant connections <strong>of</strong> steel<br />
frames in seismic areas (Ed.: F. Mazzolani). E&FN Spon, London, UK;<br />
95-146, 2000.<br />
[6.12] McGuire W. Steel structures. Prentice-Hall International series in Theoretical<br />
and Applied Mechanics (Eds.: NM Newmark and WJ Hall).<br />
Englewood Cliffs, N.J., USA, 1968.<br />
[6.13] Gioncu V, Mazzolani FM. Ductility <strong>of</strong> seismic resistant steel structures.<br />
Spon Press, London, 2002.<br />
[6.14] Hirt MA, Bez R. Construction métallique – Notions fondamentales et<br />
methods de dimensionnement. Traité de Génie Civil de l’École<br />
polytechnique fédérale de Lausanne, Volume 10. Presses Polytechniques et<br />
Universitaires Romandes, Lausanne, Switzerland, 1994.<br />
[6.15] Lusas 13. Theory manual. Finite element analysis Ltd, Version 13.5.<br />
Surrey, UK, 2003.<br />
218
APPENDIX D: DETAILED RESULTS OBTAINED FROM APPLICATION OF THE<br />
SIMPLIFIED METHODS FOR ASSESSMENT OF THE FORCE-<br />
DEFORMATION RESPONSE OF SINGLE T-STUB CONNECTIONS<br />
D.1 Geometrical and mechanical characteristics <strong>of</strong> <strong>the</strong> specimens<br />
This app<strong>end</strong>ix gives detailed results that were obtained from application <strong>of</strong> <strong>the</strong><br />
simplified methods for assessment <strong>of</strong> <strong>the</strong> F-∆ response <strong>of</strong> single T-stub connections.<br />
For illustration <strong>of</strong> <strong>the</strong> various methodologies presented in Chapter 6,<br />
<strong>the</strong> specimens from <strong>the</strong> database compiled in Chapters 3-5 are employed. The<br />
relevant geometrical and mechanical characteristics <strong>of</strong> those specimens are<br />
summarized in Tables D.1 and D.2, respectively.<br />
D.2 Previous research: exemplification<br />
D.2.1 Evaluation <strong>of</strong> initial stiffness<br />
Table D.3 sets out <strong>the</strong> predictions <strong>of</strong> initial stiffness <strong>of</strong> <strong>the</strong> above specimens by<br />
application <strong>of</strong> <strong>the</strong> procedures proposed by Yee and Melchers [6.9] and subsequently<br />
modified by Jaspart [6.2]. The results show that <strong>the</strong> two approaches<br />
(which differ essentially in <strong>the</strong> location <strong>of</strong> <strong>the</strong> prying forces) yield identical<br />
results, with consistent errors <strong>of</strong> 130%, which seem too high. The table also<br />
shows that <strong>the</strong> scatter <strong>of</strong> results is higher for <strong>the</strong> WP-T-stubs. This result however<br />
may not be significant because in <strong>the</strong>se cases most <strong>of</strong> <strong>the</strong> specimens were<br />
tested experimentally. Thus, <strong>the</strong> determination <strong>of</strong> <strong>the</strong> experimental initial stiffness<br />
is not as straightforward as for <strong>the</strong> specimens tested numerically. It should<br />
be noted that <strong>the</strong> values <strong>of</strong> ke.0 contained in this table were computed using <strong>the</strong><br />
bolt conventional length as defined by Aggerskov – Eq (1.19).<br />
The results after application <strong>of</strong> <strong>the</strong> Swanson’s procedures are summarized<br />
in Table D.4. Again, <strong>the</strong> errors are excessive. Errors well above 100% are not<br />
acceptable. This table also shows that in spite <strong>of</strong> <strong>the</strong> deviations from <strong>the</strong> actual<br />
results, <strong>the</strong> Eurocode 3 prying model yields better results than <strong>the</strong> Kulak<br />
model, also adopted by Swanson. Both models are illustrated in Fig. 6.3. The<br />
two models differ in <strong>the</strong> geometry <strong>of</strong> <strong>the</strong> beam model.<br />
D.2.2 Piecewise multilinear approximation <strong>of</strong> <strong>the</strong> overall response and evaluation<br />
<strong>of</strong> <strong>the</strong> deformation capacity and ultimate resistance<br />
a) Methodology recomm<strong>end</strong>ed by Jaspart<br />
Table D.5 indicates <strong>the</strong> predictions <strong>of</strong> <strong>the</strong> different observed failure modes. The<br />
predicted potential failure type is defined according to <strong>the</strong> expressions pre-<br />
219
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.1 Geometrical characteristics <strong>of</strong> <strong>the</strong> specimens.<br />
Test ID Potential Geometric characteristics (dimensions in [mm])<br />
failure<br />
type<br />
beff tf m m/tf n λ φ<br />
T1 13 40.0 10.7 29.45 2.75 30.00 1.02 12<br />
P1 13 40.0 10.7 34.45 3.22 25.00 0.73 12<br />
P2 13 40.0 10.7 24.45 2.29 30.56 1.25 12<br />
P3 13 35.0 10.7 29.45 2.75 30.00 1.02 12<br />
P4 13 52.5 10.7 29.45 2.75 30.00 1.02 12<br />
P5 23 60.0 10.7 29.45 2.75 30.00 1.02 12<br />
P6 13 35.0 10.7 29.45 2.75 30.00 1.02 12<br />
P7 13 45.0 10.7 29.45 2.75 30.00 1.02 12<br />
P8 13 40.0 11.0 27.10 2.46 33.88 1.25 12<br />
P9 23 40.0 14.0 28.75 2.05 35.94 1.25 12<br />
P10 11 40.0 7.0 30.75 4.39 30.00 0.98 12<br />
P11 23 40.0 14.0 28.75 2.05 30.00 1.04 12<br />
P12 11 40.0 10.7 29.45 2.75 30.00 1.02 16<br />
P13 11 40.0 10.7 29.45 2.75 30.00 1.02 12<br />
P14 11 40.0 10.7 29.45 2.75 30.00 1.02 12<br />
P15 11 40.0 10.7 24.45 2.29 30.56 1.25 16<br />
P16 23 70.0 10.7 29.45 2.75 30.00 1.02 12<br />
P17 13 70.0 10.7 29.45 2.75 30.00 1.02 16<br />
P18 11 70.0 10.7 29.45 2.75 30.00 1.02 20<br />
P19 13 70.0 10.7 29.45 2.75 30.00 1.02 16<br />
P20 23 70.0 14.0 28.75 2.05 30.00 1.04 16<br />
P21 13 92.5 10.7 29.45 2.75 30.00 1.02 20<br />
P22 13 70.0 15.0 32.34 2.16 30.00 0.93 20<br />
P23 23 70.0 18.9 31.59 1.67 30.00 0.95 20<br />
Weld_T1(i) 11 40.0 10.7 37.43 3.50 30.00 0.80 12<br />
Weld_T1(ii) 13 40.0 10.7 33.42 3.12 30.00 0.90 12<br />
Weld_T1(iii) 13 40.0 10.7 30.14 2.82 30.00 1.00 12<br />
WT1 11 45.1 10.3 33.73 3.27 30.00 0.89 12<br />
WT2A 11 45.0 10.3 36.29 3.53 29.90 0.82 12<br />
WT2B 11 45.0 10.3 31.69 3.08 29.90 0.94 12<br />
WT4A 13 74.9 10.4 33.69 3.24 30.00 0.89 12<br />
WT51 23 45.0 10.0 34.39 3.45 30.20 0.88 12<br />
WT53C 13 45.1 10.1 34.34 3.40 30.00 0.87 12<br />
WT53D 11 45.0 10.1 34.24 3.38 30.00 0.88 12<br />
WT53E 11 44.7 10.1 34.26 3.40 30.10 0.88 12<br />
WT7_M12 13 75.6 10.3 33.87 3.28 29.90 0.88 12<br />
WT7_M16 11 74.9 10.3 33.89 3.28 29.80 0.88 16<br />
WT7_M20 11 75.2 10.3 33.81 3.27 29.70 0.88 20<br />
WT57_M12 23 75.0 10.1 34.11 3.38 30.20 0.89 12<br />
WT57_M16 13 75.3 10.2 34.26 3.37 30.10 0.88 16<br />
WT57_M20 11 75.1 10.2 34.27 3.38 30.20 0.88 20<br />
220
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.2 Mechanical characteristics <strong>of</strong> <strong>the</strong> specimens.<br />
Test ID Potential Flange Bolt<br />
failure<br />
type<br />
fy.f<br />
(MPa)<br />
εp.u.f fu.b<br />
(MPa)<br />
δu.b<br />
(mm)<br />
Kb<br />
(N/mm)<br />
T1 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
P1 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
P2 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
P3 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
P4 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
P5 23 431.0 0.284 974.0 0.97 6.92×10 5<br />
P6 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
P7 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
P8 13 431.0 0.284 974.0 0.98 6.78×10 5<br />
P9 23 431.0 0.284 974.0 1.36 5.45×10 5<br />
P10 11 431.0 0.284 974.0 0.75 9.37×10 5<br />
P11 23 431.0 0.284 974.0 1.20 5.57×10 5<br />
P12 11 431.0 0.284 974.0 1.01 1.19×10 6<br />
P13 11 355.0 0.284 974.0 0.97 6.92×10 5<br />
P14 11 275.0 0.284 974.0 0.97 6.92×10 5<br />
P15 11 431.0 0.284 974.0 1.01 1.19×10 6<br />
P16 23 431.0 0.284 974.0 0.97 6.52×10 5<br />
P17 13 431.0 0.284 974.0 1.09 1.11×10 6<br />
P18 11 431.0 0.284 974.0 1.09 1.73×10 6<br />
P19 13 431.0 0.284 974.0 1.09 1.11×10 6<br />
P20 23 431.0 0.284 974.0 1.37 9.54×10 5<br />
P21 13 431.0 0.284 974.0 1.37 1.49×10 6<br />
P22 13 431.0 0.284 974.0 1.52 1.40×10 6<br />
P23 23 431.0 0.284 974.0 1.98 1.14×10 6<br />
Weld_T1(i) 11 431.0 0.284 974.0 0.97 6.92×10 5<br />
Weld_T1(ii) 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
Weld_T1(iii) 13 431.0 0.284 974.0 0.97 6.92×10 5<br />
WT1 11 340.1 0.361 919.9 1.14 1.10×10 6<br />
WT2A 11 340.1 0.361 919.9 1.14 1.10×10 6<br />
WT2B 11 340.1 0.361 919.9 1.14 1.10×10 6<br />
WT4A 13 340.1 0.361 919.9 1.14 1.10×10 6<br />
WT51 23 698.6 0.174 919.9 1.14 1.10×10 6<br />
WT53C 13 698.6 0.174 968.4 4.00 9.14×10 5<br />
WT53D 11 698.6 0.174 1166.0 0.98 1.08×10 6<br />
WT53E 11 698.6 0.174 1196.4 2.80 9.15×10 5<br />
WT7_M12 13 340.1 0.361 919.9 1.14 1.10×10 6<br />
WT7_M16 11 340.1 0.361 919.9 2.60 1.65×10 6<br />
WT7_M20 11 340.1 0.361 919.9 2.60 2.57×10 6<br />
WT57_M12 23 698.6 0.174 919.9 4.00 9.14×10 5<br />
WT57_M16 13 698.6 0.174 919.9 2.60 1.65×10 6<br />
WT57_M20 11 698.6 0.174 919.9 2.60 2.57×10 6<br />
221
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.3 Prediction <strong>of</strong> axial stiffness by application <strong>of</strong> <strong>the</strong> standard Yee and<br />
Melchers procedures and <strong>the</strong> modified proposal <strong>of</strong> Jaspart.<br />
Test ID Num./Exp.<br />
stiffness<br />
222<br />
Yee and Melchers<br />
standard procedure<br />
Modified Yee and<br />
Melchers procedures<br />
ke.0 Ratio ke.0 Ratio<br />
T1 83.54 185.18 2.22 190.80 2.28<br />
P1 63.27 131.03 2.07 133.36 2.11<br />
P2 117.06 280.41 2.40 286.85 2.45<br />
P3 72.62 165.74 2.28 171.77 2.37<br />
P4 97.86 230.23 2.35 234.19 2.39<br />
P5 101.23 255.09 2.52 257.74 2.55<br />
P9 128.47 329.76 2.57 328.23 2.55<br />
P10 43.88 54.04 1.23 58.19 1.33<br />
P12 102.05 201.90 1.98 121.96 1.20<br />
P14 81.97 185.18 2.26 190.80 2.33<br />
P15 128.41 315.17 2.45 330.88 2.58<br />
P16 111.23 287.85 2.59 288.86 2.60<br />
P18 171.57 343.17 2.00 358.98 2.09<br />
P20 181.68 578.78 3.19 565.31 3.11<br />
P23 322.09 870.01 2.70 825.15 2.56<br />
Average 2.32 2.30<br />
Coefficient <strong>of</strong> variation<br />
0.18<br />
0.21<br />
Weld_T1(i) 73.50 104.27 1.42 108.31 1.47<br />
Weld_T1(ii) 88.04 137.43 1.56 142.29 1.62<br />
Weld_T1(iii) 107.29 175.52 1.64 181.03 1.69<br />
WT1g 68.58 138.88 2.03 144.34 2.10<br />
WT1h 73.58 138.88 1.89 144.34 1.96<br />
WT2Aa 64.32 114.67 1.78 119.46 1.86<br />
WT2Ab 61.75 114.67 1.86 119.46 1.93<br />
WT2Ba 63.58 160.23 2.52 166.22 2.61<br />
WT2Bb 79.75 160.23 2.01 166.22 2.08<br />
WT4Aa 75.08 216.22 2.88 219.37 2.92<br />
WT4Ab 86.96 216.22 2.49 219.37 2.52<br />
WT7_M12 91.18 212.53 2.33 215.74 2.37<br />
WT7_M16 116.09 226.41 1.95 234.29 2.02<br />
WT7_M20 137.70 239.59 1.74 251.42 1.83<br />
WT51a 59.62 118.33 1.98 123.71 2.07<br />
WT51b 61.84 118.33 1.91 123.71 2.00<br />
WT53C 64.23 120.20 1.87 124.73 1.94<br />
WT53D 52.90 124.75 2.36 130.04 2.46<br />
WT53E 64.82 119.92 1.85 124.52 1.92<br />
WT57_M12 42.89 186.75 4.35 189.45 4.42<br />
WT57_M16 55.22 206.87 3.75 214.82 3.89<br />
WT57_M20 75.48 214.57 2.84 225.89 2.99<br />
Average 2.23 2.30<br />
Coefficient <strong>of</strong> variation<br />
0.32<br />
0.31
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.4 Prediction <strong>of</strong> axial stiffness by application <strong>of</strong> <strong>the</strong> Swanson procedures.<br />
Test ID Num./Exp.<br />
stiffness<br />
Prediction with b’<br />
from Eq. (6.9)<br />
Prediction with m<br />
from Eq. (6.11)<br />
ke.0 Ratio ke.0 Ratio<br />
T1 83.54 186.06 2.23 164.67 1.97<br />
P1 63.27 136.36 2.16 118.51 1.87<br />
P2 117.06 235.18 2.01 219.72 1.88<br />
P3 72.62 172.30 2.37 152.37 2.10<br />
P4 97.86 214.84 2.20 190.54 1.95<br />
P5 101.23 229.15 2.26 203.48 2.01<br />
P9 128.47 222.73 1.73 213.20 1.66<br />
P10 43.88 83.97 1.91 73.10 1.67<br />
P12 102.05 256.91 2.52 201.46 1.97<br />
P14 81.97 341.76 4.17 280.41 3.42<br />
P15 128.41 237.49 1.85 211.38 1.65<br />
P16 111.23 343.25 3.09 277.12 2.49<br />
P18 171.57 467.64 2.73 331.66 1.93<br />
P20 181.68 412.57 2.27 353.66 1.95<br />
P23 322.09 565.93 1.76 452.25 1.40<br />
Average 2.35 1.99<br />
Coefficient <strong>of</strong> variation<br />
0.26 0.23<br />
Weld_T1(i) 73.50 135.46 1.84 105.75 1.44<br />
Weld_T1(ii) 88.04 149.68 1.70 131.78 1.50<br />
Weld_T1(iii) 107.29 162.48 1.51 158.40 1.48<br />
WT1g 68.58 189.06 2.76 154.29 2.25<br />
WT1h 73.58 189.06 2.57 154.29 2.10<br />
WT2Aa 64.32 172.50 2.68 130.81 2.03<br />
WT2Ab 61.75 172.50 2.79 130.81 2.12<br />
WT2Ba 63.58 197.38 3.10 173.86 2.73<br />
WT2Bb 79.75 197.38 2.47 173.86 2.18<br />
WT4Aa 75.08 266.10 3.54 218.87 2.92<br />
WT4Ab 86.96 266.10 3.06 218.87 2.52<br />
WT7_M12 91.18 263.26 2.89 215.78 2.37<br />
WT7_M16 116.09 345.88 2.98 246.96 2.13<br />
WT7_M20 137.70 452.30 3.28 279.81 2.03<br />
WT51a 59.62 167.26 2.81 136.37 2.29<br />
WT51b 61.84 167.26 2.70 136.37 2.21<br />
WT53C 64.23 163.52 2.55 132.76 2.07<br />
WT53D 52.90 174.20 3.29 141.15 2.67<br />
WT53E 64.82 163.49 2.52 132.71 2.05<br />
WT57_M12 42.89 229.31 5.35 187.65 4.38<br />
WT57_M16 55.22 324.57 5.88 231.40 4.19<br />
WT57_M20 75.48 417.33 5.53 257.03 3.41<br />
Average 3.08 2.41<br />
Coefficient <strong>of</strong> variation<br />
0.37<br />
0.31<br />
223
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.5 Prediction <strong>of</strong> failure modes.<br />
Test Pot. Critical resis- Test ID Pot. Critical resis-<br />
ID failure tance formula<br />
failure tance formula<br />
type Plst. Ultm.<br />
type Plst. Ultm.<br />
T1 13 1 1 or 2 P21 13 1 1<br />
P1 13 1 1 or 2 P22 13 1 or 2 1 or 2<br />
P2 13 1 1 P23 23 2 2<br />
P3 13 1 1 or 2 Weld_T1(i) 11 1 1<br />
P4 13 1 or 2 2 Weld_T1(ii) 13 1 1<br />
P5 23 2 2 Weld_T1(iii) 13 1 1 or 2<br />
P6 13 1 1 WT1 11 1 1<br />
P7 13 1 1 or 2 WT2A 11 1 1<br />
P8 13 1 1 or 2 WT2B 11 1 1<br />
P9 23 2 2 WT4A 13 1 or 2 2<br />
P10 11 1 1 WT7_M12 13 1 or 2 2<br />
P11 23 2 2 WT7_M16 11 1 1<br />
P12 11 1 1 WT7_M20 11 1 1<br />
P13 11 1 1 WT51 23 2 1 or 2<br />
P14 11 1 1 WT53C 13 1 or 2<br />
P15 11 1 1 WT53D 11 1 or 2 1<br />
P16 23 2 2 WT53E 11 1 1<br />
P17 13 1 1 or 2 WT57_M12 23 2 2<br />
P18 11 1 1 WT57_M16 13 1 or 2 1 or 2<br />
P19 13 1 1 or 2 WT57_M20 11 1 1<br />
P20 23 2 2<br />
sented in Chapter 2. For computation <strong>of</strong> Mf.u, Eq. (2.4) recomm<strong>end</strong>ed by<br />
Gioncu et al. is employed [6.11]. This table also indicates <strong>the</strong> critical resistance<br />
formula according to <strong>the</strong> Jaspart methodology (cf. 6.2.1). For application <strong>of</strong> <strong>the</strong><br />
procedures, four different cases are considered regarding <strong>the</strong> resistance formulation<br />
(BF or FBA) and <strong>the</strong> mechanical properties <strong>of</strong> <strong>the</strong> T-stub material. The<br />
complete <strong>characterization</strong> <strong>of</strong> <strong>the</strong> actual material properties <strong>of</strong> <strong>the</strong> various<br />
specimens from <strong>the</strong> database was given in Chapters 3, 4 and 5. The actual<br />
strain hardening modulus, Eh, for <strong>the</strong>se specimens however is always lower<br />
than <strong>the</strong> nominal properties [6.3,6.13]. For steel grade S355, Eh = E/48.2 and<br />
for S275, Eh = E/42.8. No quantitative guidance is given in any <strong>of</strong> <strong>the</strong> references<br />
for steel grade S690. Hence, both actual and nominal values for Eh are<br />
taken into account for those specimens where steel grade S355 and S275 was<br />
employed (S275 was used in specimen P14).<br />
This variation combined with <strong>the</strong> two alternative resistance formulations<br />
yields four different approaches that are summarized in Tables D.6-D.9. The<br />
specimens are divided according to <strong>the</strong> assembly type (HR-T-stubs and WP-T-<br />
stubs).<br />
224
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.6 Prediction <strong>of</strong> ultimate resistance and deformation capacity by using<br />
<strong>the</strong> actual strain hardening modulus <strong>of</strong> <strong>the</strong> flange material and<br />
<strong>the</strong> basic formulation for computation <strong>of</strong> resistance (HR-T-stubs).<br />
Test ID Numerical results Jaspart methodology (Actual Eh; BF)<br />
Ratio<br />
Ratio<br />
Fmax ∆u.0 Fu.0 ∆u.0<br />
(kN) (mm) (kN)<br />
(mm)<br />
T1 103.99 8.70 92.53 0.89 21.54 2.47<br />
P1 91.76 10.77 79.10 0.86 27.33 2.54<br />
P2 116.72 6.18 111.45 0.95 17.00 2.75<br />
P3 95.41 10.17 80.96 0.85 21.02 2.07<br />
P4 115.97 4.68 112.95 0.97 17.17 3.67<br />
P5 130.20 3.63 117.24 0.90 11.24 3.10<br />
P6 95.53 10.06 80.96 0.85 21.02 2.09<br />
P7 111.34 7.56 104.09 0.93 22.06 2.92<br />
P8 112.71 8.08 106.27 0.94 19.18 2.37<br />
P9 131.43 3.31 127.29 0.97 9.31 2.82<br />
P10 76.79 32.75 37.93 0.49 30.29 0.93<br />
P11 121.15 2.94 123.56 1.02 9.42 3.20<br />
P12 154.06 24.22 92.53 0.60 19.88 0.82<br />
P13 93.71 11.38 79.31 0.85 20.30 1.78<br />
P14 86.57 24.15 66.87 0.77 20.21 0.84<br />
P15 171.08 18.02 111.45 0.65 15.00 0.83<br />
P16 125.69 3.06 122.97 0.98 10.93 3.57<br />
P17 192.01 9.29 161.92 0.84 21.77 2.34<br />
P18 266.57 26.07 161.92 0.61 20.39 0.78<br />
P19 186.52 9.29 161.92 0.87 21.77 2.34<br />
P20 225.94 4.06 225.65 1.00 9.84 2.42<br />
P21 281.33 17.67 213.96 0.76 21.37 1.21<br />
P22 305.21 6.40 298.54 0.98 20.76 3.25<br />
P23 346.01 5.22 353.25 1.02 10.43 2.00<br />
Average 0.86 2.21<br />
Coefficient <strong>of</strong> variation<br />
0.17<br />
0.41<br />
For identical assembly types, <strong>the</strong> tables show that <strong>the</strong> model yields identical<br />
results, in terms <strong>of</strong> average ratios. The ultimate resistance predictions are more<br />
accurate if <strong>the</strong> formulation accounting for <strong>the</strong> bolt finite size is employed. Note<br />
that for those specimens whose collapse mode is determined from Eq. (6.6) this<br />
formulation does not apply.<br />
With respect to <strong>the</strong> evaluation <strong>of</strong> deformation capacity, <strong>the</strong> best predictions<br />
are obtained through application <strong>of</strong> <strong>the</strong> basic formulation and assuming <strong>the</strong><br />
nominal strain hardening modulus (Table D.8). The models where <strong>the</strong> actual<br />
strain hardening modulus was used provided an overestimation <strong>of</strong> this property.<br />
On <strong>the</strong> o<strong>the</strong>r hand, if <strong>the</strong> nominal strain hardening modulus is considered<br />
<strong>the</strong> approximation improves significantly.<br />
225
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.6 Prediction <strong>of</strong> ultimate resistance and deformation capacity (WP-Tstubs)<br />
(cont.).<br />
Test ID Num. or Exp. results<br />
Jaspart methodology (Actual Eh; BF)<br />
Fmax ∆u.0 Fu.0 Ratio ∆u.0 Ratio<br />
(kN) (mm) (kN)<br />
(mm)<br />
Weld_T1(i) 92.02 10.85 72.79 0.79 31.33 2.89<br />
Weld_T1(ii) 102.75 8.01 81.54 0.79 26.03 3.25<br />
Weld_T1(iii) 113.10 6.22 81.54 0.72 22.26 3.58<br />
WT1(h) 91.91 14.32 68.28 0.74 18.41 1.29<br />
WT2A(b) 86.82 17.98 62.78 0.72 20.92 1.16<br />
WT2B(b) 97.88 13.09 72.17 0.74 16.63 1.27<br />
WT4A(b) 103.26 4.33 103.55 1.00 12.95 2.99<br />
WT7_M12 100.34 4.60 103.11 1.03 13.30 2.89<br />
WT7_M16 132.34 11.47 113.32 0.86 18.48 1.61<br />
WT7_M20 145.72 9.12 114.05 0.78 17.92 1.96<br />
WT51(b) 97.08 3.96 96.51 0.99 7.87 1.99<br />
WT53C 98.90 4.24 98.89 1.00 6.40 1.51<br />
WT53D 117.36 5.54 100.08 0.85 6.34 1.14<br />
WT53E 115.04 5.26 98.25 0.85 6.36 1.21<br />
WT57_M12 121.87 4.33 120.69 0.99 6.80 1.57<br />
WT57_M16 173.64 5.88 168.21 0.97 6.34 1.08<br />
WT57_M20 241.71 15.98 167.36 0.69 6.17 0.39<br />
Average 0.85 1.87<br />
Coefficient <strong>of</strong> variation<br />
0.14<br />
0.49<br />
b) Methodology recomm<strong>end</strong>ed by Faella, Piluso and Rizzano<br />
To illustrate <strong>the</strong> methodology proposed by Faella et al. [6.3-6.5], six examples<br />
were selected: T1, P16, P18, WT4A, WT7_M20 and WT51. These examples<br />
typify <strong>the</strong> three different failure modes (Type-11, -13 and -23) as well as <strong>the</strong><br />
two assembly types (HR- and WP-T-stubs). In <strong>the</strong> framework <strong>of</strong> this methodology,<br />
however, <strong>the</strong> potential collapse mode is defined differently. As already<br />
mentioned in §6.2.2, Faella et al. assume <strong>the</strong> occurrence <strong>of</strong> three alternative<br />
collapse modes, termed: (i) type-1 if cracking <strong>of</strong> <strong>the</strong> flange material occurs at<br />
<strong>the</strong> two critical sections, at <strong>the</strong> flange-to-web connection, (1) and at <strong>the</strong> bolt<br />
line, (ii) type-2 if cracking <strong>of</strong> <strong>the</strong> flange material occurs at critical section (1)<br />
and, simultaneously, bolt fracture also takes place and (iii) type-3 if a bolt fracture<br />
mechanism arises. The boundaries for <strong>the</strong> occurrence <strong>of</strong> a given mechanism<br />
are indicated in Fig. 6.2. The coefficient βu, which is defined in Eq. (2.1),<br />
with Mf.u given by Eq. (2.5), is compared with βu.lim. Faella et al. suggest two<br />
alternative expressions for βu.lim:<br />
226
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.7 Prediction <strong>of</strong> ultimate resistance and deformation capacity by using<br />
<strong>the</strong> actual strain hardening modulus <strong>of</strong> <strong>the</strong> flange material and<br />
<strong>the</strong> formulation accounting for <strong>the</strong> bolt for computation <strong>of</strong> resistance<br />
(HR-T-stubs).<br />
Test ID Numerical results Jaspart methodology (Actual Eh; FBA)<br />
Fmax ∆u.0 Fu.0 Ratio ∆u.0 Ratio<br />
(kN) (mm) (kN)<br />
(mm)<br />
T1 103.99 8.70 105.79 1.02 22.05 2.53<br />
P1 91.76 10.77 91.97 1.00 30.19 2.80<br />
P2 116.72 6.18 119.60 1.02 12.35 2.00<br />
P3 95.41 10.17 96.36 1.01 25.01 2.46<br />
P4 115.97 4.68 112.95 0.97 11.61 2.48<br />
P5 130.20 3.63 117.24 0.90 11.24 3.10<br />
P6 95.53 10.06 96.36 1.01 25.01 2.49<br />
P7 111.34 7.56 108.65 0.98 14.79 1.96<br />
P8 112.71 8.08 114.85 1.02 15.21 1.88<br />
P9 131.43 3.31 127.29 0.97 9.31 2.82<br />
P10 76.79 32.75 44.90 0.58 35.86 1.10<br />
P11 121.15 2.94 123.56 1.02 9.42 3.20<br />
P12 154.06 24.22 116.02 0.75 24.93 1.03<br />
P13 93.71 11.38 94.40 1.01 24.16 2.12<br />
P14 86.57 24.15 79.59 0.92 24.06 1.00<br />
P15 171.08 18.02 144.50 0.84 19.45 1.08<br />
P16 125.69 3.06 122.97 0.98 10.93 3.57<br />
P17 192.01 9.29 194.44 1.01 23.20 2.50<br />
P18 266.57 26.07 216.96 0.81 27.33 1.05<br />
P19 186.52 9.29 194.44 1.04 23.20 2.50<br />
P20 225.94 4.06 225.65 1.00 9.84 2.42<br />
P21 281.33 17.67 286.69 1.02 28.63 1.62<br />
P22 305.21 6.40 309.55 1.01 11.50 1.80<br />
P23 346.01 5.22 353.25 1.02 10.43 2.00<br />
Average 0.96 2.15<br />
Coefficient <strong>of</strong> variation<br />
0.11<br />
0.34<br />
2λ<br />
βu<br />
.lim =<br />
(D.1)<br />
2λ+ 1<br />
in [6.3] and later [6.4]:<br />
2λ<br />
⎡ dw<br />
⎤<br />
βu.lim = 1− ( 1+<br />
λ)<br />
2λ+ 1<br />
⎢<br />
8n<br />
⎥<br />
⎣ ⎦ (D.2)<br />
which is <strong>the</strong> same as in Eq. (2.2). Table D.10 sets out <strong>the</strong> predictions <strong>of</strong> <strong>the</strong><br />
failure modes by using <strong>the</strong> two above expressions. For fur<strong>the</strong>r comparisons,<br />
reference is made to Eq. (D.2) – last column in Table D.10. Specimen T1 is <strong>the</strong><br />
only case where a change in <strong>the</strong> collapse mode is observed. For compari-<br />
227
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.7 Prediction <strong>of</strong> ultimate resistance and deformation capacity (WP-Tstubs)<br />
(cont.).<br />
Test ID Num. or Exp. results<br />
Jaspart methodology (Actual Eh; FBA)<br />
Fmax ∆u.0 Fu.0 Ratio ∆u.0 Ratio<br />
(kN) (mm) (kN)<br />
(mm)<br />
Weld_T1(i) 92.02 10.85 84.35 0.92 36.31 3.35<br />
Weld_T1(ii) 102.75 8.01 95.61 0.93 30.52 3.81<br />
Weld_T1(iii) 113.10 6.22 104.58 0.92 24.03 3.86<br />
WT1(h) 91.91 14.32 78.79 0.86 21.24 1.48<br />
WT2A(b) 86.82 17.98 71.97 0.83 23.98 1.33<br />
WT2B(b) 97.88 13.09 83.84 0.86 19.32 1.48<br />
WT4A(b) 103.26 4.33 103.55 1.00 9.73 2.25<br />
WT7_M12 100.34 4.60 103.11 1.03 9.88 2.15<br />
WT7_M16 132.34 11.47 135.81 1.03 22.15 1.93<br />
WT7_M20 145.72 9.12 142.39 0.98 22.37 2.45<br />
WT51(b) 97.08 3.96 98.21 1.01 9.54 2.41<br />
WT53C 98.90 4.24 102.51 1.04 9.66 2.28<br />
WT53D 117.36 5.54 115.33 0.98 8.15 1.47<br />
WT53E 115.04 5.26 113.19 0.98 7.33 1.39<br />
WT57_M12 121.87 4.33 120.69 0.99 6.80 1.57<br />
WT57_M16 173.64 5.88 179.87 1.04 9.92 1.69<br />
WT57_M20 241.71 15.98 208.15 0.86 7.67 0.48<br />
Average 0.96 2.08<br />
Coefficient <strong>of</strong> variation<br />
0.07<br />
0.44<br />
son, <strong>the</strong> table also includes <strong>the</strong> critical modes for <strong>the</strong> same examples according<br />
to <strong>the</strong> author (second column) and Jaspart (third and fourth columns).<br />
The application <strong>of</strong> <strong>the</strong> method proposed by <strong>the</strong>se authors requires, in <strong>the</strong><br />
first place, <strong>the</strong> approximation <strong>of</strong> <strong>the</strong> steel flange constitutive law by a quadrilinear<br />
relationship. The actual material properties for <strong>the</strong> example specimens<br />
were defined in terms <strong>of</strong> a piecewise σ-ε law earlier in Chapters 3 and 4. Fig.<br />
D.1 shows those laws in terms <strong>of</strong> natural coordinates and <strong>the</strong> corresponding<br />
quadrilinear approximations.<br />
Next, <strong>the</strong> F-∆ response may be fully characterized. In terms <strong>of</strong> ultimate<br />
conditions, <strong>the</strong> predictions <strong>of</strong> resistance and deformation capacity are summarized<br />
and compared with <strong>the</strong> actual results for <strong>the</strong> various specimens in Table<br />
D.11. These results are obtained by application <strong>of</strong> <strong>the</strong> basic formulation for<br />
evaluation <strong>of</strong> <strong>the</strong> resistance <strong>of</strong> specimens failing according to a type-1 mechanism.<br />
For those specimens whose failure mode is <strong>of</strong> type-2, <strong>the</strong> bolt is also<br />
subjected to plasticity. In this table, <strong>the</strong> bolt plastic deformations, δb.p.u, are<br />
evaluated. It should be noted that, so far, <strong>the</strong> compatibility requirements between<br />
flange and bolt deformations have been disregarded. For specimen T1,<br />
228
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.8 Prediction <strong>of</strong> ultimate resistance and deformation capacity by using<br />
<strong>the</strong> nominal strain hardening modulus <strong>of</strong> <strong>the</strong> flange material<br />
and <strong>the</strong> basic formulation for computation <strong>of</strong> resistance.<br />
Test ID Numerical results Jaspart methodology (Nominal Eh; BF)<br />
Fmax ∆u.0 Fu.0 Ratio ∆u.0 Ratio<br />
(kN) (mm) (kN)<br />
(mm)<br />
T1 103.99 8.70 92.53 0.89 8.98 1.03<br />
P1 91.76 10.77 79.10 0.86 11.39 1.06<br />
P2 116.72 6.18 111.45 0.95 7.09 1.15<br />
P3 95.41 10.17 80.96 0.85 8.76 0.86<br />
P4 115.97 4.68 112.95 0.97 7.23 1.55<br />
P5 130.20 3.63 117.24 0.90 4.84 1.33<br />
P6 95.53 10.06 80.96 0.85 8.76 0.87<br />
P7 111.34 7.56 104.09 0.93 9.20 1.22<br />
P8 112.71 8.08 106.27 0.94 8.00 0.99<br />
P9 131.43 3.31 127.29 0.97 4.01 1.21<br />
P10 76.79 32.75 37.93 0.49 12.63 0.39<br />
P11 121.15 2.94 123.56 1.02 4.05 1.38<br />
P12 154.06 24.22 92.53 0.60 8.29 0.34<br />
P13 93.71 11.38 79.31 0.85 8.43 0.74<br />
P14 86.57 24.15 66.87 0.77 7.44 0.31<br />
P15 171.08 18.02 111.45 0.65 6.25 0.35<br />
P16 125.69 3.06 122.97 0.98 4.70 1.53<br />
P17 192.01 9.29 161.92 0.84 9.08 0.98<br />
P18 266.57 26.07 161.92 0.61 8.50 0.33<br />
P19 186.52 9.29 161.92 0.87 9.08 0.98<br />
P20 225.94 4.06 225.65 1.00 4.23 1.04<br />
P21 281.33 17.67 213.96 0.76 8.91 0.50<br />
P22 305.21 6.40 298.54 0.98 8.66 1.35<br />
P23 346.01 5.22 353.25 1.02 4.48 0.86<br />
Average 0.86 0.93<br />
Coefficient <strong>of</strong> variation 0.17 0.42<br />
Weld_T1(i) 92.02 10.85 72.79 0.79 13.06 1.20<br />
Weld_T1(ii) 102.75 8.01 81.54 0.79 10.85 1.35<br />
Weld_T1(iii) 113.10 6.22 90.42 0.80 9.28 1.49<br />
WT1(h) 91.91 14.32 68.28 0.74 9.29 0.65<br />
WT2A(b) 86.82 17.98 62.78 0.72 10.56 0.59<br />
WT2B(b) 97.88 13.09 72.17 0.74 8.40 0.64<br />
WT4A(b) 103.26 4.33 103.55 1.00 6.62 1.53<br />
WT7_M12 100.34 4.60 103.11 1.03 6.80 1.48<br />
WT7_M16 132.34 11.47 113.32 0.86 9.33 0.81<br />
WT7_M20 145.72 9.12 114.05 0.78 9.05 0.99<br />
Average 0.83 1.07<br />
Coefficient <strong>of</strong> variation<br />
0.13<br />
0.36<br />
229
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.9 Prediction <strong>of</strong> ultimate resistance and deformation capacity by using<br />
<strong>the</strong> nominal strain hardening modulus <strong>of</strong> <strong>the</strong> flange material<br />
and <strong>the</strong> formulation accounting for <strong>the</strong> bolt for computation <strong>of</strong> resistance.<br />
Test ID Numerical results<br />
Fmax ∆u.0<br />
Jaspart methodology (Nominal Eh; FBA)<br />
Fu.0 Ratio ∆u.0 Ratio<br />
(kN) (mm) (kN)<br />
(mm)<br />
T1 103.99 8.70 105.79 1.02 9.24 1.06<br />
P1 91.76 10.77 91.97 1.00 12.62 1.17<br />
P2 116.72 6.18 119.60 1.02 5.25 0.85<br />
P3 95.41 10.17 96.36 1.01 10.43 1.03<br />
P4 115.97 4.68 112.95 0.97 5.01 1.07<br />
P5 130.20 3.63 117.24 0.90 4.84 1.33<br />
P6 95.53 10.06 96.36 1.01 10.43 1.04<br />
P7 111.34 7.56 108.65 0.98 6.31 0.83<br />
P8 112.71 8.08 114.85 1.02 6.44 0.80<br />
P9 131.43 3.31 127.29 0.97 4.01 1.21<br />
P10 76.79 32.75 44.90 0.58 14.95 0.46<br />
P11 121.15 2.94 123.56 1.02 4.05 1.38<br />
P12 154.06 24.22 116.02 0.75 10.39 0.43<br />
P13 93.71 11.38 94.40 1.01 10.03 0.88<br />
P14 86.57 24.15 79.59 0.92 8.86 0.37<br />
P15 171.08 18.02 144.50 0.84 8.11 0.45<br />
P16 125.69 3.06 122.97 0.98 4.70 1.53<br />
P17 192.01 9.29 194.44 1.01 9.72 1.05<br />
P18 266.57 26.07 216.96 0.81 11.39 0.44<br />
P19 186.52 9.29 194.44 1.04 9.72 1.05<br />
P20 225.94 4.06 225.65 1.00 4.23 1.04<br />
P21 281.33 17.67 286.69 1.02 11.94 0.68<br />
P22 305.21 6.40 309.55 1.01 4.97 0.78<br />
P23 346.01 5.22 353.25 1.02 4.48 0.86<br />
Average 0.96 0.91<br />
Coefficient <strong>of</strong> variation 0.11<br />
0.35<br />
Weld_T1(i) 92.02 10.85 84.35 0.92 15.14 1.39<br />
Weld_T1(ii) 102.75 8.01 95.61 0.93 12.73 1.59<br />
Weld_T1(iii) 113.10 6.22 104.58 0.92 10.05 1.61<br />
WT1(h) 91.91 14.32 78.79 0.86 10.73 0.75<br />
WT2A(b) 86.82 17.98 71.97 0.83 12.11 0.67<br />
WT2B(b) 97.88 13.09 83.84 0.86 9.75 0.75<br />
WT4A(b) 103.26 4.33 103.55 1.00 5.05 1.17<br />
WT7_M12 100.34 4.60 103.11 1.03 5.13 1.12<br />
WT7_M16 132.34 11.47 135.81 1.03 11.18 0.98<br />
WT7_M20 145.72 9.12 142.39 0.98 11.29 1.24<br />
Average 0.93 1.13<br />
Coefficient <strong>of</strong> variation<br />
0.08<br />
0.30<br />
230
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Stress (MPa)<br />
900<br />
750<br />
600<br />
450<br />
300<br />
150<br />
0<br />
Actual piecewise true law<br />
Quadrilinear true law<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35<br />
Strain<br />
(a) Flange steel grade S355 (fy.f = 430 MPa) for specimens T1, P16 and P18.<br />
Stress (MPa)<br />
900<br />
750<br />
600<br />
450<br />
300<br />
150<br />
0<br />
Actual piecewise true law<br />
Quadrilinear true law<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35<br />
Strain<br />
(b) Flange steel grade S355 (fy.f = 340 MPa) for specimens WT4A and<br />
WT7_M20.<br />
Stress (MPa)<br />
900<br />
750<br />
600<br />
450<br />
300<br />
150<br />
0<br />
Actual piecewise true law<br />
Quadrilinear true law<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35<br />
Strain<br />
(c) Flange steel grade S690 (fy.f = 698 MPa) for specimen WT51.<br />
Fig. D.1 Quadrilinear approximation <strong>of</strong> <strong>the</strong> actual piecewise flange material<br />
law for application <strong>of</strong> <strong>the</strong> method recomm<strong>end</strong>ed by Faella, Piluso and<br />
Rizzano.<br />
231
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
failing according to a type-2 mechanism, <strong>the</strong> application <strong>of</strong> <strong>the</strong> formulae provided<br />
by Faella et al. yields a negative bolt plastic deformation, which has no<br />
physical meaning.<br />
If now <strong>the</strong> formulation accounting for <strong>the</strong> bolt action for type-1 is considered,<br />
<strong>the</strong> ultimate resistance prediction improves (specimens P18 and<br />
WT7_M20) as well as <strong>the</strong> deformation capacity – Table D.12. In this table, <strong>the</strong><br />
results for specimen T1 are computed by assuming that mode 1 governs collapse,<br />
as ascertained by Eq. (D.1). The results are not so different from those in<br />
Table D.11. With respect to <strong>the</strong> remaining specimens illustrating type-2 collapse<br />
mode, in Table D.12 <strong>the</strong> compatibility requirements between bolt and<br />
flange deformations are accounted for. The actual bolt deformation at fracture<br />
is known. If this value is imposed, a new value for <strong>the</strong> overall T-stub deformation<br />
can be calculated by means <strong>of</strong> linear interpolation. By doing so, <strong>the</strong> predictions<br />
improve.<br />
So far, <strong>the</strong> actual material properties for <strong>the</strong> flange have been employed. As<br />
before, <strong>the</strong> nominal properties for <strong>the</strong> strain hardening range are also taken into<br />
consideration (Table D.13). For this analysis, specimen WT51 that uses S690 is<br />
Table D.10 Prediction <strong>of</strong> failure modes according to Faella and co-authors.<br />
Test ID Potential Crit. resistance formula Pot. failure type ac-<br />
failure type according to Jaspart cording to Faella et al.<br />
Plastic Ultimate Eq. (D.1) Eq. (D.2)<br />
T1 13 1 1 or 2 2 1<br />
P16 23 2 2 2 2<br />
P18 11 1 1 1 1<br />
WT4A 13 1 or 2 2 2 2<br />
WT7_M20 11 1 1 1 1<br />
WT51 23 2 1 or 2 2 2<br />
Table D.11 Prediction <strong>of</strong> ultimate resistance and deformation capacity by using<br />
<strong>the</strong> basic formulation for computation <strong>of</strong> resistance and neglecting<br />
<strong>the</strong> compatibility requirements between flange and bolt<br />
deformations.<br />
Test ID Num. or Exp. results Faella et al. methodology<br />
Fmax ∆u Fu.0 Ratio ∆u.0 Ratio δb.p.u<br />
(kN) (mm) (kN) (mm) (mm)<br />
T1 103.99 8.70 109.17 1.05 9.64 1.11 -12.79<br />
P16 125.69 3.06 128.89 1.03 26.18 8.56 12.82<br />
P18 266.57 26.07 106.19 0.40 9.86 0.38 0.00<br />
WT4A 103.26 4.33 108.07 1.05 25.32 5.85 11.14<br />
WT7_M20 145.72 9.12 130.96 0.90 12.15 1.33 0.00<br />
WT51 97.08 3.96 100.90 1.04 6.46 1.63 1.27<br />
232
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.12 Prediction <strong>of</strong> ultimate resistance and deformation capacity by using<br />
<strong>the</strong> formulation accounting for bolt finite size for computation<br />
<strong>of</strong> resistance <strong>of</strong> specimens failing according to mode 1 and catering<br />
for compatibility requirements between flange and bolt deformations<br />
(specimens from type-2 collapse mode).<br />
Test ID Num. or Exp. results Faella et al. methodology<br />
Fmax ∆u.0 δb.p.fract Fu.0 Ratio ∆u.0 Ratio<br />
(kN) (mm) (mm) (kN)<br />
(mm)<br />
T1 103.99 8.70 0.87 126.39 1.22 10.19 1.17<br />
P16 125.69 3.06 0.87 83.24 0.66 1.96 0.64<br />
P18 266.57 26.07 ⎯ 142.28 0.53 10.24 0.39<br />
WT4A 103.26 4.33 1.08 71.89 0.70 2.95 0.68<br />
WT7_M20 145.72 9.12 ⎯ 163.51 1.12 12.45 1.37<br />
WT51 97.08 3.96 1.08 98.99 1.02 5.71 1.44<br />
Table D.13 Actual properties <strong>of</strong> <strong>the</strong> flange steel grade and nominal properties<br />
according to Faella and co-authors [S355(1) is <strong>the</strong> steel grade from<br />
specimens T1, P16 and P18; S355(2) is <strong>the</strong> steel grade from<br />
specimens WT4A and WT7_M20].<br />
Steel<br />
fy E Eh Eu<br />
(MPa) (GPa) (GPa) (GPa)<br />
εy εh εm εu<br />
S355(1) 430 208 1.74 0.67 0.0021 0.0200 0.1545 0.2872<br />
S355(2) 340 210 2.15 0.48 0.0016 0.0150 0.2240 0.3610<br />
S355 355 210 4.36 0.51 0.0017 0.0166 0.0521 0.8100<br />
Table D.14 Prediction <strong>of</strong> ultimate resistance and deformation capacity with<br />
nominal steel properties by using <strong>the</strong> formulation accounting for<br />
bolt finite size for computation <strong>of</strong> resistance <strong>of</strong> specimens failing<br />
according to mode 1.<br />
Test ID Potential<br />
failure type<br />
Fmax<br />
(kN)<br />
∆u.0<br />
(mm)<br />
δb.p.u<br />
(mm)<br />
T1 2 112.08 28.23 -10.60<br />
P16 2 64.83 1.70 0.87<br />
P18 1 158.05 27.38 ⎯<br />
WT4A 2 60.60 2.53 1.08<br />
WT7_M20 1 224.79 37.21 ⎯<br />
neglected. The results for this new approach are given in Table D.14 (bolt and<br />
flange compatibility is also accounted for). Again, for specimen T1 <strong>the</strong> results<br />
are poor. For <strong>the</strong> remaining cases, <strong>the</strong> results do not improve considerably.<br />
233
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
c) Methodology recomm<strong>end</strong>ed by Beg, Zupančič and Vayas for evaluation <strong>of</strong><br />
<strong>the</strong> deformation capacity<br />
Beg and co-authors proposed simple formulae for <strong>the</strong> assessment <strong>of</strong> <strong>the</strong> deformation<br />
capacity <strong>of</strong> single T-stubs in [6.7]. For <strong>the</strong> examples under analysis,<br />
Table D.15 sets out <strong>the</strong> predictions <strong>of</strong> <strong>the</strong>ir proposals. For <strong>the</strong> specimens failing<br />
according to a type-2 plastic mechanism, a value <strong>of</strong> k = 3.5 is assumed (see Eq.<br />
(6.25)). In general, <strong>the</strong> predictions are not satisfactory. For those specimens<br />
failing according to a type-2 plastic mode, <strong>the</strong> predictions improve, but for <strong>the</strong><br />
remaining cases <strong>the</strong> deviations are not acceptable.<br />
Table D.15 Prediction <strong>of</strong> deformation capacity by means <strong>of</strong> Beg et al. methodology.<br />
Test ID Num. or exp. results<br />
Beg et al. methodology<br />
∆u.0<br />
(mm)<br />
Potential plastic<br />
failure mode<br />
∆u.0<br />
(mm)<br />
Ratio<br />
T1 8.70 1 23.56 2.71<br />
P1 10.77 1 27.56 2.56<br />
P2 6.18 1 19.56 3.17<br />
P3 10.17 1 23.56 2.32<br />
P4 4.68 1 23.56 5.03<br />
P5 3.63 2 4.30 1.18<br />
P6 10.06 1 23.56 2.34<br />
P7 7.56 1 23.56 3.12<br />
P8 8.08 1 21.68 2.68<br />
P9 3.31 2 5.16 1.56<br />
P10 32.75 1 24.60 0.75<br />
P11 2.94 2 5.24 1.78<br />
P12 24.22 1 23.56 0.97<br />
P13 11.38 1 23.56 2.07<br />
P14 24.15 1 23.56 0.98<br />
P15 18.02 1 19.56 1.09<br />
P16 3.06 2 4.32 1.41<br />
P17 9.29 1 23.56 2.54<br />
P18 26.07 1 23.56 0.90<br />
P19 9.29 1 23.56 2.54<br />
P20 4.06 2 5.95 1.46<br />
P21 17.67 1 23.56 1.33<br />
P22 6.40 1 25.87 4.04<br />
P23 5.22 2 9.27 1.78<br />
Average 2.10<br />
Coefficient <strong>of</strong> variation<br />
0.50<br />
234
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.15 Prediction <strong>of</strong> deformation capacity by means <strong>of</strong> Beg et al. methodology<br />
(cont.).<br />
Test ID Num. or exp. results<br />
Beg et al. methodology<br />
∆u.0<br />
(mm)<br />
Potential plastic<br />
failure mode<br />
∆u.0<br />
(mm)<br />
Ratio<br />
Weld_T1(i) 10.85 1 29.95 2.76<br />
Weld_T1(ii) 8.01 1 26.73 3.34<br />
Weld_T1(iii) 6.22 1 24.11 3.87<br />
WT1 14.32 1 26.98 1.88<br />
WT2A 17.98 1 29.03 1.61<br />
WT2B 13.09 1 25.35 1.94<br />
WT4A 4.33 1 26.96 6.23<br />
WT51 3.96 2 5.67 1.43<br />
WT53C 4.24 1 27.47 6.48<br />
WT53D 5.54 1 27.39 4.94<br />
WT53E 5.26 1 27.41 5.21<br />
WT7_M12 4.60 1 27.10 5.89<br />
WT7_M16 11.47 1 27.11 2.36<br />
WT7_M20 9.12 1 27.05 2.97<br />
WT57_M12 4.33 2 12.88 2.97<br />
WT57_M16 5.88 1 27.40 4.66<br />
WT57_M20 15.98 1 27.41 1.72<br />
Average 3.55<br />
Coefficient <strong>of</strong> variation<br />
0.48<br />
D.3 Application <strong>of</strong> <strong>the</strong> proposed model: results for HR-T-stub T1<br />
Specimen T1 was selected for illustration <strong>of</strong> <strong>the</strong> results obtained when <strong>the</strong> proposed<br />
model was applied. The geometrical and mechanical characteristics for<br />
this connection are indicated in Tables D.1-D.2.<br />
First, <strong>the</strong> overall F-∆ response is shown in Fig. D.2. The actual behaviour,<br />
obtained from <strong>the</strong> three-dimensional numerical model, is plotted against <strong>the</strong><br />
simplified response from <strong>the</strong> proposed beam model. Both curves fit well<br />
though <strong>the</strong> simplified curve yields larger <strong>ductility</strong> than <strong>the</strong> real behaviour. This<br />
graph also traces <strong>the</strong> bilinear approximation <strong>of</strong> Jaspart. For this approximation,<br />
<strong>the</strong> FBA was employed for resistance computation (this specimen fails according<br />
to a failure type-13). The actual value for <strong>the</strong> strain hardening modulus was<br />
used. This response deviates significantly from <strong>the</strong> real behaviour. The predictions<br />
from Faella and co-workers are also included (results from Table D.11).<br />
Fig. D.3 shows <strong>the</strong> bolt response. The trilinear model fits <strong>the</strong> numerical<br />
results well. The prying force is plotted against <strong>the</strong> total flange deformation in<br />
Fig. D.4 for both approaches. The two models yield different responses. The<br />
235
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
beam model gives lower results for <strong>the</strong> prying force. The ratios B/F and Q/F<br />
are shown in Fig. D.5. Figs. D.6-D.9 trace <strong>the</strong> beam diagrams <strong>of</strong> b<strong>end</strong>ing moment,<br />
flange deformation, flange rotation and plastic strain, respectively. Four<br />
load levels were chosen: (i) F = 36.0 kN, corresponding to yielding <strong>of</strong> <strong>the</strong><br />
flange at <strong>the</strong> flange-to-web connection; (ii) F = 60.0 kN, corresponding to first<br />
yielding at <strong>the</strong> bolt axis (<strong>the</strong> section at <strong>the</strong> flange-to-web connection is not engaged<br />
in <strong>the</strong> strain hardening domain yet); (iii) F = 88.9 kN, corresponding to<br />
first yielding <strong>of</strong> <strong>the</strong> bolt and (iv) F = 114.5 kN, corresponding to fracture <strong>of</strong> <strong>the</strong><br />
bolt. Fig. D.6 shows that, at ultimate conditions, <strong>the</strong> b<strong>end</strong>ing moments acting at<br />
sections (1) and (2) are similar. Again, <strong>the</strong> plastic deformation <strong>of</strong> <strong>the</strong> flange is<br />
restricted to an area close to <strong>the</strong> critical sections, as shown in Figs. D.8-D.9.<br />
Fig. D.10 traces <strong>the</strong> variation <strong>of</strong> <strong>the</strong> applied load with <strong>the</strong> parameter L/m.<br />
236<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
Actual response<br />
Simplified response (Beam model)<br />
15<br />
0<br />
Bilinear approximation (Jaspart)<br />
Quadrilinear approximation (Faella and co-authors)<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Deformation, ∆ (mm)<br />
Fig. D.2 Specimen T1: force-deformation behaviour as ascertained by <strong>the</strong><br />
different approaches.<br />
Bolt force, B (kN)<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
"Actual" response (3-dim. FE model)<br />
10<br />
0<br />
Simplified response (Beam model)<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Bolt elongation, δ b (mm)<br />
Fig. D.3 Specimen T1: bolt elongation behaviour.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Prying force, Q (kN)<br />
40<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
5<br />
"Actual" response (3-dim. FE model)<br />
Simplified response (Beam model)<br />
0<br />
0 2 4 6 8 10 12 14 16 18<br />
Deformation, ∆ (mm)<br />
Fig. D.4 Specimen T1: prying force behaviour.<br />
Ratio B/F, Q/F<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
B/F (3-dim. FE model) Q/F (3-dim. FE model)<br />
B/F (Beam model) Q/F (Beam model)<br />
0.0<br />
0 15 30 45 60 75 90 105 120<br />
Load, F (kN)<br />
Fig. D.5 Specimen T1: ratio B/F and Q/F.<br />
Mz (Nmm)<br />
1.0E+06<br />
8.0E+05<br />
6.0E+05<br />
4.0E+05<br />
2.0E+05<br />
0.0E+00<br />
-2.0E+05<br />
-4.0E+05<br />
-6.0E+05<br />
-8.0E+05<br />
-1.0E+06<br />
0 5 10 15 20 25 30 35 40 45 50 55 60<br />
Beam length (mm)<br />
Fig. D.6 Specimen T1: flange moment diagram.<br />
F = 36.0 kN F = 60.0 kN<br />
F = 88.9 kN F = 114.5 kN<br />
Mp<br />
Mp<br />
237
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
238<br />
∆/2 (mm)<br />
9<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
F = 36.0 kN<br />
F = 88.9 kN<br />
F = 60.0 kN<br />
F = 114.5 kN<br />
0 5 10 15 20 25 30 35 40 45 50 55 60<br />
Beam length (mm)<br />
Fig. D.7 Specimen T1: flange (half-) deformation diagram.<br />
θz (rad)<br />
0.03<br />
0.00<br />
-0.03<br />
-0.06<br />
-0.09<br />
-0.12<br />
-0.15<br />
-0.18<br />
-0.21<br />
-0.24<br />
-0.27<br />
-0.30<br />
-0.33<br />
0 5 10 15 20 25 30 35 40 45 50 55 60<br />
Beam length (mm)<br />
Fig. D.8 Specimen T1: flange rotation diagram.<br />
Plastic strain<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
-0.1<br />
-0.2<br />
-0.3<br />
-0.4<br />
-0.5<br />
εp.u<br />
F = 36.0 kN<br />
F = 60.0 kN<br />
F = 88.9 kN<br />
F = 114.5 kN<br />
0 5 10 15 20 25 30 35 40 45 50 55 60<br />
Beam length (mm)<br />
Fig. D.9 Specimen T1: flange plastic strain diagram.<br />
F = 36.0 kN<br />
F = 60.0 kN<br />
F = 88.9 kN<br />
F = 114.5 kN<br />
εp.u
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
15<br />
0<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
L/m<br />
Fig. D.10 Specimen T1: length <strong>of</strong> <strong>the</strong> equivalent cantilever.<br />
D.4 Application <strong>of</strong> <strong>the</strong> proposed model: results for WP-T-stub WT1<br />
Specimen WT1 was also selected for fur<strong>the</strong>r detail <strong>of</strong> <strong>the</strong> results extracted from<br />
<strong>the</strong> proposed model. The geometrical and mechanical characteristics for this<br />
connection are indicated in Tables D.1-D.2, as well.<br />
Identical results to <strong>the</strong> above are shown in this section. Fig. D.11 plots <strong>the</strong><br />
alternative predictions for <strong>the</strong> deformation behaviour and compares those with<br />
<strong>the</strong> actual test (experimental) results. The agreement between <strong>the</strong> Faella et al.<br />
quadrilinear approximation and <strong>the</strong> real response is very good. As for <strong>the</strong> beam<br />
model and <strong>the</strong> simple approximation <strong>of</strong> Jaspart, <strong>the</strong> results are good but clearly<br />
underestimate <strong>the</strong> resistance predictions. The <strong>ductility</strong> is also overestimated.<br />
Figs. D.12-D.14 show <strong>the</strong> bolt response and <strong>the</strong> prying behaviour as ascertained<br />
by <strong>the</strong> beam model. No comparisons are established with <strong>the</strong> test results<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30 Actual response (WT1h)<br />
Simplified response (Beam model)<br />
15<br />
0<br />
Bilinear approximation (Jaspart)<br />
Quadrilinear approximation (Faella and co-authors)<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Deformation, ∆ (mm)<br />
Fig. D.11 Specimen WT1: force-deformation behaviour as ascertained by <strong>the</strong><br />
different approaches.<br />
239
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
240<br />
Bolt force, B (kN)<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Bolt elongation, δ b (mm)<br />
Fig. D.12 Specimen WT1: bolt elongation behaviour.<br />
Prying force, Q (kN)<br />
30<br />
25<br />
20<br />
15<br />
10<br />
5<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.13 Specimen WT1: prying force behaviour.<br />
Ratio B/F, Q/F<br />
1.2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
B/F (Beam model) Q/F (Beam model)<br />
0.0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
Load, F (kN)<br />
Fig. D.14 Specimen WT1: ratio B/F and Q/F.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
since <strong>the</strong>se values were not determined experimentally.<br />
Fig. D.15-D.18 give results for <strong>the</strong> flange b<strong>end</strong>ing moment, flange deformation,<br />
flange rotation and plastic strain at four different load levels: (i) F =<br />
26.0 kN, corresponding to yielding <strong>of</strong> <strong>the</strong> flange at <strong>the</strong> flange-to-web connection,<br />
(ii) F = 42.0 kN, corresponding to first yielding at <strong>the</strong> bolt axis, (iii) F =<br />
78.0 kN, corresponding to first yielding <strong>of</strong> <strong>the</strong> bolt and cracking <strong>of</strong> <strong>the</strong> flange<br />
material at section (1) and (iv) F = 85.6 kN, corresponding to cracking <strong>of</strong> <strong>the</strong><br />
flange material at section (1*). Finally, Fig. D.19 shows <strong>the</strong> evolution <strong>of</strong> <strong>the</strong><br />
non-dimensional parameter L/m with increasing loading.<br />
D.5 Prediction <strong>of</strong> <strong>the</strong> nonlinear response <strong>of</strong> <strong>the</strong> above connections using<br />
<strong>the</strong> nominal stress-strain characteristics<br />
If <strong>the</strong> nominal mechanical properties <strong>of</strong> <strong>the</strong> bolt and flange <strong>plate</strong>s are input, <strong>the</strong><br />
Mz (Nmm)<br />
8.0E+05<br />
6.0E+05<br />
4.0E+05<br />
2.0E+05<br />
0.0E+00<br />
-2.0E+05<br />
-4.0E+05<br />
-6.0E+05<br />
F = 26.0 kN F = 42.0 kN<br />
F = 78.0 kN F = 85.6 kN<br />
-8.0E+05<br />
0 5 10 15 20 25 30 35 40 45 50 55 60 65<br />
Beam length (mm)<br />
Fig. D.15 Specimen WT1: flange moment diagram.<br />
∆/2 (mm)<br />
10<br />
9<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
0 5 10 15 20 25 30 35 40 45 50 55 60 65<br />
Beam length (mm)<br />
Fig. D.16 Specimen WT1: flange gap diagram.<br />
F = 26.0 kN F = 42.0 kN<br />
F = 78.0 kN F = 85.6 kN<br />
Mp<br />
Mp<br />
241
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
242<br />
θz (mm)<br />
0.03<br />
0.00<br />
-0.03<br />
-0.06<br />
-0.09<br />
-0.12<br />
-0.15<br />
-0.18<br />
-0.21<br />
-0.24<br />
-0.27<br />
-0.30<br />
-0.33<br />
0 5 10 15 20 25 30 35 40 45 50 55 60 65<br />
Beam length (mm)<br />
Fig. D.17 Specimen WT1: flange rotation diagram.<br />
Plastic strain<br />
0.7<br />
0.5<br />
0.3<br />
0.1<br />
-0.1<br />
-0.3<br />
-0.5<br />
εp.u<br />
F = 26.0 kN<br />
F = 42.0 kN<br />
F = 78.0 kN<br />
F = 85.6 kN<br />
-0.7<br />
0 5 10 15 20 25 30 35 40 45 50 55 60 65<br />
Beam length (mm)<br />
Fig. D.18 Specimen WT1: flange plastic strain diagram.<br />
Load, F (kN)<br />
90<br />
75<br />
60<br />
45<br />
30<br />
15<br />
F = 26.0 kN<br />
F = 42.0 kN<br />
F = 78.0 kN<br />
F = 85.6 kN<br />
εp.u<br />
0<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
L/m<br />
Fig. D.19 Specimen WT1: length <strong>of</strong> <strong>the</strong> equivalent cantilever.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.16 Prediction <strong>of</strong> <strong>the</strong> failure modes.<br />
Test ID Actual determiningfracture<br />
element<br />
Predicted potential<br />
failure<br />
mode.<br />
Mu Eq. Mu Eq.<br />
(2.4)<br />
(2.5)<br />
Determining fracture element<br />
in <strong>the</strong> beam model<br />
(nominal mech. Properties)<br />
T1 Bolt 13 13 Flange, at (1*)<br />
P1 Bolt 13 13 Flange, at (1*)<br />
P2 Bolt 13 13 Bolt<br />
P3 Bolt 13 13 Flange, at (1*)<br />
P4 Bolt 13 13 Bolt<br />
P5 Bolt 23 23 Bolt<br />
P6 Bolt 13 13 Flange, at (1*)<br />
P7 Bolt 13 13 Flange, at (1*)<br />
P8 Bolt 13 13 Flange, at (1*)<br />
P9 Bolt 23 23 Bolt<br />
P10 Flange 11 11 Flange, at (1*)<br />
P11 Bolt 23 23 Bolt<br />
P12 Flange 11 11 Flange, at (1*)<br />
P13 Bolt 13 13 Flange, at (1*)<br />
P14 Flange 11 11 Flange, at (1*)<br />
P15 Flange 11 11 Flange, at (1*)<br />
P16 Bolt 23 23 Bolt<br />
P17 Bolt 13 13 Flange, at (1*)<br />
P18 Flange 11 11 Flange, at (1*)<br />
P19 Bolt 13 13 Flange, at (1*)<br />
P20 Bolt 23 23 Bolt<br />
P21 Bolt 13 13 Flange, at (1*)<br />
P22 Bolt 13 13 Flange, at (1*)<br />
P23 Bolt 23 23 Bolt<br />
Weld_T1(i) Bolt 11 11 Flange, at (1*)<br />
Weld_T1(ii) Bolt 13 13 Flange, at (1*)<br />
Weld_T1(iii) Bolt 13 13 Flange, at (1*)<br />
WT1<br />
Bolt and<br />
flange 11 13<br />
Flange, at (1*)<br />
WT2A<br />
Bolt and<br />
flange 11 13<br />
Flange, at (1*)<br />
WT2B<br />
Bolt and<br />
Flange, at (1*)<br />
flange 13 13<br />
WT4A Bolt 23 23 Bolt<br />
WT7_M12 Bolt 23 23 Bolt<br />
WT7_M16 Flange 11 13 Flange, at (1*)<br />
WT7_M20 Flange 11 11 Flange, at (1*)<br />
243
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
predicted failure modes may slightly change (Tables 6.12 and D.16 – see T-<br />
stubs T1, P1, P6-P8, P22 and Weld_T1(iii)). The responses for this new approach<br />
are traced in Figs. D.20-D.53. These graphs also trace <strong>the</strong> actual response<br />
as well as <strong>the</strong> prediction by using <strong>the</strong> actual material properties. For<br />
most specimens whose failure is determined by <strong>the</strong> bolt (black circle), <strong>the</strong> predictions<br />
<strong>of</strong> deformation capacity improve. If <strong>the</strong> flange governs ultimate collapse<br />
(black square in <strong>the</strong> graphs), <strong>the</strong>n <strong>the</strong> maximum deformation decreases<br />
since <strong>the</strong> nominal ultimate strain is lower than <strong>the</strong> actual value. In terms <strong>of</strong><br />
strength, <strong>the</strong> ultimate resistance is lower now when compared to <strong>the</strong> actual<br />
properties. However, it can be seen from <strong>the</strong> graphs that <strong>the</strong> (nominal) ultimate<br />
resistance is identical to <strong>the</strong> value predicted by <strong>the</strong> actual mechanical properties<br />
if <strong>the</strong> bolt is determinant.<br />
These new predictions at ultimate conditions are summarized in Table D.17<br />
for <strong>the</strong> various specimens. From a design point <strong>of</strong> view, and on average, <strong>the</strong><br />
244<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 13.5 15.0 16.5 18.0<br />
Deformation, ∆ (mm)<br />
Fig. D.20 Specimen T1: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
Actual response<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Deformation, ∆ (mm)<br />
Fig. D.21 Specimen P1: force-deformation behaviour (nominal properties).
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15<br />
Deformation, ∆ (mm)<br />
Fig. D.22 Specimen P2: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
Fig. D.23 Specimen P3: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 13.5 15.0<br />
Deformation, ∆ (mm)<br />
Fig. D.24 Specimen P4: force-deformation behaviour (nominal properties).<br />
245
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
246<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
Deformation, ∆ (mm)<br />
Fig. D.25 Specimen P5: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Deformation, ∆ (mm)<br />
Fig. D.26 Specimen P6: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.27 Specimen P7: force-deformation behaviour (nominal properties).
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.28 Specimen P8: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
Fig. D.29 Specimen P9: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
Actual response<br />
10<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with nominal properties<br />
0 3 6 9 12 15 18 21 24 27 30 33 36<br />
Deformation, ∆ (mm)<br />
Fig. D.30 Specimen P10: force-deformation behaviour (nominal properties).<br />
247
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
248<br />
Load, F (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
Fig. D.31 Specimen P11: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22 24 26<br />
Deformation, ∆ (mm)<br />
Fig. D.32 Specimen P12: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.33 Specimen P13: force-deformation behaviour (nominal properties).
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
Actual response<br />
20<br />
Simplified response (Simple beam model)<br />
10<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22 24 26<br />
Deformation, ∆ (mm)<br />
Fig. D.34 Specimen P14: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
Actual response<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.35 Specimen P15: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
Fig. D.36 Specimen P16: force-deformation behaviour (nominal properties).<br />
249
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
250<br />
Load, F (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Actual response<br />
60<br />
Simplified response (Simple beam model)<br />
30<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
Fig. D.37 Specimen P17: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
280<br />
240<br />
200<br />
160<br />
120<br />
80<br />
Actual response<br />
Simplified response (Simple beam model)<br />
40<br />
0<br />
Simplified response with nominal properties<br />
0 3 6 9 12 15 18 21 24 27 30<br />
Deformation, ∆ (mm)<br />
Fig. D.38 Specimen P18: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Actual response<br />
60<br />
Simplified response (Simple beam model)<br />
30<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
Fig. D.39 Specimen P19: force-deformation behaviour (nominal properties).
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
270<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
Actual response<br />
60<br />
Simplified response (Simple beam model)<br />
30<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
Fig. D.40 Specimen P20: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
320<br />
280<br />
240<br />
200<br />
160<br />
120<br />
Actual response<br />
80<br />
Simplified response (Simple beam model)<br />
40<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
Fig. D.41 Specimen P21: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
Actual response<br />
100<br />
Simplified response (Simple beam model)<br />
50<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14<br />
Deformation, ∆ (mm)<br />
Fig. D.42 Specimen P22: force-deformation behaviour (nominal properties).<br />
251
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
252<br />
Load, F (kN)<br />
450<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
Actual response<br />
100<br />
Simplified response (Simple beam model)<br />
50<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
Deformation, ∆ (mm)<br />
Fig. D.43 Specimen P23: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
Fig. D.44 Specimen Weld_T1(i): force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
Fig. D.45 Specimen Weld_T1(ii): force-deformation behaviour (nominal properties).
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
Fig. D.46 Specimen Weld_T1(iii): force-deformation behaviour (nominal<br />
properties).<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
Actual response<br />
15<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.47 Specimen WT1: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
Actual response<br />
15<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.48 Specimen WT2A: force-deformation behaviour (nominal properties).<br />
253
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
254<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
45<br />
30<br />
Actual response<br />
15<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.49 Specimen WT2B: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
Deformation, ∆ (mm)<br />
Fig. D.50 Specimen WT4A: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with nominal properties<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
Deformation, ∆ (mm)<br />
Fig. D.51 Specimen WT7_M12: force-deformation behaviour (nominal properties).
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.52 Specimen WT7_M16: force-deformation behaviour (nominal properties).<br />
Load, F (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with nominal properties<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
Fig. D.53 Specimen WT7_M20: force-deformation behaviour (nominal properties).<br />
predictions for resistance are very good (ratios <strong>of</strong> 0.96 approximately) but <strong>the</strong>y<br />
overestimate <strong>the</strong> deformation capacity. In addition, <strong>the</strong> predicted failure type<br />
does not always correspond to <strong>the</strong> actual mode. The reason for <strong>the</strong> good resistance<br />
predictions derives from <strong>the</strong> mechanical σ-ε law, which differs significantly<br />
at yielding conditions from <strong>the</strong> nominal law but approximates it at ultimate<br />
conditions.<br />
D.6 Comparative graphs: simple beam model and sophisticated beam<br />
model accounting for <strong>the</strong> bolt action<br />
Figs. D.54-D.75 compare <strong>the</strong> actual response <strong>of</strong> <strong>the</strong> several T-stubs with <strong>the</strong><br />
255
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.17 Prediction <strong>of</strong> deformation capacity and ultimate resistance (nominal<br />
properties <strong>of</strong> steel).<br />
256<br />
Test ID Potential<br />
failure<br />
Actual results Beam model predictions<br />
Fmax ∆u.0 Fu.0 Ratio ∆u.0 Ratio<br />
type (kN) (mm) (kN) (mm)<br />
T1 13 103.99 8.70 104.13 1.00 16.92 1.94<br />
P1 13 91.76 10.77 87.94 0.96 19.30 1.79<br />
P2 13 116.72 6.18 126.50 1.08 14.30 2.31<br />
P3 13 95.41 10.17 91.54 0.96 16.34 1.61<br />
P4 13 115.97 4.68 124.12 1.07 12.42 2.65<br />
P5 23 130.20 3.63 129.04 0.99 9.19 2.53<br />
P6 13 95.53 10.06 91.54 0.96 16.34 1.62<br />
P7 13 111.34 7.56 117.31 1.05 18.03 2.38<br />
P8 13 112.71 8.08 120.92 1.07 16.48 2.04<br />
P9 23 131.43 3.31 144.71 1.10 8.04 2.43<br />
P10 11 76.79 32.75 42.69 0.56 24.44 0.75<br />
P11 23 121.15 2.94 138.24 1.14 6.87 2.34<br />
P12 11 154.06 24.22 104.58 0.68 15.29 0.63<br />
P13 13 93.71 11.38 104.13 1.11 16.92 1.49<br />
P14 11 86.57 24.15 86.88 1.00 17.12 0.71<br />
P15 11 171.08 18.02 130.79 0.76 13.75 0.76<br />
P16 23 125.69 3.06 131.06 1.04 5.99 1.96<br />
P17 13 192.01 9.29 183.01 0.95 17.06 1.84<br />
P18 11 266.57 26.07 184.10 0.69 15.83 0.61<br />
P19 13 186.52 9.29 183.01 0.98 17.06 1.84<br />
P20 23 225.94 4.06 255.70 1.13 8.28 2.04<br />
P21 13 281.33 17.67 243.84 0.87 16.60 0.94<br />
P22 13 305.21 6.40 317.41 1.04 14.35 2.24<br />
P23 23 346.01 5.22 427.41 1.24 10.59 2.03<br />
Average 0.98 1.73<br />
Coefficient <strong>of</strong> variation 0.17 0.38<br />
Weld_T1(i) 11 92.02 10.85 70.00 0.76 9.84 0.91<br />
Weld_T1(ii) 13 102.75 8.01 84.42 0.82 12.41 1.55<br />
Weld_T1(iii) 13 113.10 6.22 101.77 0.90 17.43 2.80<br />
WT1 11 91.91 14.32 84.39 0.92 10.74 0.75<br />
WT2A 11 86.82 17.98 75.00 0.86 9.93 0.55<br />
WT2B 13 97.88 13.09 93.44 0.95 12.63 0.96<br />
WT4A 23 103.26 4.33 112.20 1.09 5.12 1.18<br />
WT7_M12 23 100.34 4.60 111.31 1.11 5.13 1.11<br />
WT7_M16 11 132.34 11.47 140.10 1.06 10.34 0.90<br />
WT7_M20 11 145.72 9.12 140.33 0.96 10.25 1.12<br />
Average 0.94 1.18<br />
Coefficient <strong>of</strong> variation<br />
0.12 0.53
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
120<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.54 Specimen T1.<br />
Load, F (kN)<br />
45<br />
Actual response<br />
30 Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.55 Specimen P1.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
45<br />
30<br />
Actual response<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
120<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.56 Specimen P3.<br />
Deformation, ∆ (mm)<br />
45<br />
Actual response<br />
30 Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)<br />
257
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
258<br />
Load, F (kN)<br />
Fig. D.57 Specimen P5.<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
Actual response<br />
45<br />
30<br />
15<br />
0<br />
Simplified response (Simple<br />
beam model)<br />
Simplified response accounting<br />
for <strong>the</strong> bolt action<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0<br />
80<br />
70<br />
60<br />
50<br />
40<br />
Fig. D.58 Specimen P10.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
30<br />
20<br />
Actual response<br />
Simplified response (Simple beam model)<br />
10<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 3 6 9 12 15 18 21 24 27 30 33 36<br />
160<br />
140<br />
120<br />
100<br />
80<br />
Fig. D.59 Specimen P12.<br />
Deformation, ∆ (mm)<br />
60<br />
Actual response<br />
40 Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22 24 26<br />
Deformation, ∆ (mm)
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
Fig. D.60 Specimen P14.<br />
Load, F (kN)<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30 Actual response<br />
20 Simplified response (Simple beam model)<br />
10<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22 24 26<br />
210<br />
180<br />
150<br />
120<br />
Fig. D.61 Specimen P15.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
90<br />
60<br />
Actual response<br />
Simplified response (Simple beam model)<br />
30<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20<br />
300<br />
250<br />
200<br />
150<br />
Fig. D.62 Specimen P18.<br />
Deformation, ∆ (mm)<br />
100<br />
Actual response<br />
50<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 3 6 9 12 15 18 21 24 27 30<br />
Deformation, ∆ (mm)<br />
259
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
260<br />
Load, F (kN)<br />
Fig. D.63 Specimen P20.<br />
Load, F (kN)<br />
Fig. D.64 Specimen 23.<br />
Load, F (kN)<br />
270<br />
240<br />
210<br />
180<br />
150<br />
120<br />
Actual response<br />
90<br />
60<br />
30<br />
0<br />
Simplified response (Simple<br />
beam model)<br />
Simplified response accounting<br />
for <strong>the</strong> bolt action<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0<br />
Deformation, ∆ (mm)<br />
450<br />
400<br />
350<br />
300<br />
250<br />
200<br />
Actual response<br />
150<br />
Simplified response (Simple<br />
100<br />
50<br />
0<br />
beam model)<br />
Simplified response accounting<br />
for <strong>the</strong> bolt action<br />
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0<br />
120<br />
105<br />
Fig. D.65 Specimen Weld_T1(i).<br />
90<br />
75<br />
60<br />
Deformation, ∆ (mm)<br />
45<br />
Actual response<br />
30 Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
120<br />
105<br />
Fig. D.66 Specimen Weld_T1(ii).<br />
Load, F (kN)<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30 Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
120<br />
105<br />
Fig. D.67 Specimen Weld_T1(iii).<br />
Load, F (kN)<br />
90<br />
75<br />
60<br />
Deformation, ∆ (mm)<br />
45<br />
Actual response<br />
30 Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.68 Specimen WT1.<br />
Deformation, ∆ (mm)<br />
45<br />
30<br />
Actual response<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
261
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
262<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.69 Specimen WT2A.<br />
Load, F (kN)<br />
45<br />
30<br />
Actual response<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.70 Specimen WT2B.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
45<br />
30<br />
Actual response<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20<br />
120<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.71 Specimen WT51.<br />
Deformation, ∆ (mm)<br />
Actual response<br />
45<br />
Simplified response (Simple<br />
30<br />
beam model)<br />
15<br />
0<br />
Simplified response accounting<br />
for <strong>the</strong> bolt action<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
175<br />
150<br />
125<br />
100<br />
Fig. D.72 Specimen WT7_M16.<br />
Load, F (kN)<br />
75<br />
50<br />
Actual response<br />
Simplified response (Simple beam model)<br />
25<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20<br />
200<br />
175<br />
150<br />
125<br />
100<br />
Fig. D.73 Specimen WT7_M20.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
75<br />
Actual response<br />
50 Simplified response (Simple beam model)<br />
25<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20<br />
240<br />
210<br />
180<br />
150<br />
120<br />
Fig. D.74 Specimen WT57_M16.<br />
Deformation, ∆ (mm)<br />
Actual response<br />
90<br />
Simplified response (Simple<br />
60<br />
beam model)<br />
30<br />
0<br />
Simplified response accounting<br />
for <strong>the</strong> bolt action<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
263
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
264<br />
Load, F (kN)<br />
Fig. D.75 Specimen WT57_M20.<br />
270<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90 Actual response<br />
60 Simplified response (Simple beam model)<br />
30<br />
0<br />
Simplified response accounting for <strong>the</strong> bolt action<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Deformation, ∆ (mm)<br />
predictions from <strong>the</strong> simplified approach (bolt modelled as a single extensional<br />
spring) and with a sophistication <strong>of</strong> <strong>the</strong> beam model. This sophistication consists<br />
in assuming that <strong>the</strong> bolt effect can be reproduced with a set <strong>of</strong> extensional<br />
springs along a certain length here taken as <strong>the</strong> bolt diameter. The figures<br />
clearly show that such sophistication improved <strong>the</strong> agreement <strong>of</strong> <strong>the</strong> results<br />
with <strong>the</strong> actual predictions, in most cases. Table D.18 gives <strong>the</strong> actual results<br />
for resistance and deformation capacity and compare <strong>the</strong>m with <strong>the</strong> aforementioned<br />
two-dimensional approaches. Four alternative ultimate conditions are<br />
imposed for <strong>the</strong> sophisticated approach. The determinant conditions however<br />
are ei<strong>the</strong>r fracture <strong>of</strong> <strong>the</strong> bolt at mid-section or cracking <strong>of</strong> <strong>the</strong> material at section<br />
(1*) – values in bold.<br />
D.7 Comparative graphs: influence <strong>of</strong> <strong>the</strong> distance m for <strong>the</strong> WP-T-stubs<br />
The influence <strong>of</strong> <strong>the</strong> geometrical parameter m is evident in <strong>the</strong> graphs from<br />
Figs. D.76-D.89. These graphs compare <strong>the</strong> actual response with <strong>the</strong> beam<br />
model predictions and highlight <strong>the</strong> effect <strong>of</strong> m: increase on resistance and<br />
stiffness (see also Table D.19) and decrease on <strong>ductility</strong>. Tables 6.13 and D.20<br />
set out <strong>the</strong> predictions <strong>of</strong> deformation capacity and show that for <strong>the</strong> original<br />
distance <strong>the</strong>re is an average ratio between experiments and analytical predictions<br />
<strong>of</strong> 1.81 (coefficient <strong>of</strong> variation <strong>of</strong> 0.34). If <strong>the</strong> “new” m is adopted, <strong>the</strong><br />
average ratio drops to 1.02 with a coefficient <strong>of</strong> variation <strong>of</strong> 0.47. In general,<br />
for WP-T-stubs, <strong>the</strong> new value <strong>of</strong> m gives a better agreement with <strong>the</strong> experiments,<br />
particularly for specimens made up <strong>of</strong> S355.
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.18 Comparison <strong>of</strong> <strong>the</strong> predicted values for ultimate resistance and<br />
deformation capacity by applying <strong>the</strong> simple beam model and <strong>the</strong><br />
sophisticated beam model accounting for <strong>the</strong> bolt action.<br />
Test<br />
ID<br />
T1<br />
P1<br />
P3<br />
P5<br />
P10<br />
P12<br />
P14<br />
P15<br />
P18<br />
P<br />
20<br />
P<br />
23<br />
Num. or Exp.<br />
results<br />
Simple beam<br />
model predictions<br />
Beam model accounting<br />
for bolt action<br />
Fmax ∆u.0 Fmax ∆u.0 Fmax<br />
Ratio<br />
∆u.0<br />
(kN) (mm) (kN) (mm) (kN) (mm)<br />
103.99<br />
91.76<br />
95.41<br />
130.20<br />
76.79<br />
154.06<br />
86.57<br />
171.08<br />
266.57<br />
225.94<br />
346.01<br />
8.70<br />
10.77<br />
10.17<br />
3.63<br />
32.75<br />
24.22<br />
24.15<br />
18.02<br />
26.07<br />
4.06<br />
5.22<br />
114.45<br />
103.25<br />
111.96<br />
123.76<br />
50.25<br />
122.43<br />
79.54<br />
153.17<br />
215.75<br />
246.14<br />
408.62<br />
16.76<br />
25.34<br />
24.17<br />
4.63<br />
32.40<br />
19.67<br />
20.49<br />
16.48<br />
20.18<br />
4.03<br />
5.24<br />
Ratio<br />
Ultimate<br />
conditions<br />
98.21 0.94 6.35 0.73 Bolt 1/4<br />
104.04 1.00 8.54 0.98 Flange (1)<br />
113.87 1.09 12.74 1.46 Bolt 1/2<br />
129.79 1.25 20.69 2.38 Flange (1*)<br />
85.78 0.93 8.79 0.82 Bolt 1/4<br />
91.55 1.00 12.10 1.12 Flange (1)<br />
103.27 1.13 19.88 1.85 Bolt 1/2<br />
109.08 1.19 24.23 2.25 Flange (1*)<br />
90.54 0.95 7.42 0.73 Bolt 1/4<br />
94.37 0.99 9.11 0.90 Flange (1)<br />
108.18 1.13 16.06 1.58 Bolt 1/2<br />
116.14 1.22 20.65 2.03 Flange (1*)<br />
117.91 0.91 2.88 0.79 Bolt 1/4<br />
121.71 0.93 3.37 0.93 Bolt 1/2<br />
154.84 1.19 11.82 3.25 Flange (1)<br />
188.42 1.45 23.75 6.54 Flange (1*)<br />
48.36 0.63 12.87 0.39 Flange (1)<br />
52.29 0.68 17.72 0.54 Bolt 1/4<br />
59.17 0.77 27.79 0.85 Flange (1*)<br />
112.97 1.47 134.63 4.11 Bolt 1/2<br />
119.29 0.77 8.02 0.33 Flange (1)<br />
125.30 0.81 10.00 0.41 Bolt 1/4<br />
149.02 0.97 19.22 0.79 Flange (1*)<br />
186.85 1.21 37.05 1.53 Bolt 1/2<br />
72.08 0.83 8.62 0.36 Flange (1)<br />
75.04 0.87 10.31 0.43 Bolt 1/4<br />
88.79 1.03 19.87 0.82 Flange (1*)<br />
95.75 1.11 25.57 1.06 Bolt 1/2<br />
142.15 0.83 5.46 0.30 Flange (1)<br />
146.01 0.85 6.20 0.34 Bolt 1/4<br />
187.94 1.10 16.16 0.90 Flange (1*)<br />
199.91 1.17 19.47 1.08 Bolt 1/2<br />
217.87 0.82 7.42 0.28 Flange (1)<br />
244.55 0.92 11.69 0.45 Bolt 1/4<br />
280.34 1.05 18.69 0.72 Flange (1*)<br />
241.83 1.07 3.03 0.75 Bolt 1/4<br />
247.76 1.10 3.51 0.87 Bolt 1/2<br />
373.18 1.08 3.41 0.65 Bolt 1/4<br />
399.89 1.16 4.00 0.77 Bolt 1/2<br />
265
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
Table D.18 Comparison <strong>of</strong> <strong>the</strong> predicted values for ultimate resistance and<br />
deformation capacity by applying <strong>the</strong> simple beam model and <strong>the</strong><br />
sophisticated beam model accounting for <strong>the</strong> bolt action (cont.).<br />
Test<br />
ID<br />
Weld_<br />
T1(i)<br />
Weld_<br />
T1(ii)<br />
Weld_<br />
T1(iii)<br />
WT1<br />
WT2A<br />
WT2B<br />
WT51<br />
WT7<br />
M16<br />
WT7<br />
M20<br />
WT5<br />
7_M<br />
16<br />
WT5<br />
7_M<br />
20<br />
266<br />
Num. or Exp.<br />
results<br />
Simple beam<br />
model predictions<br />
Beam model accounting<br />
for bolt action<br />
Fmax ∆u.0 Fmax ∆u.0 Fmax<br />
Ratio<br />
∆u.0<br />
(kN) (mm) (kN) (mm) (kN) (mm)<br />
92.02<br />
102.75<br />
113.10<br />
91.91<br />
86.82<br />
97.88<br />
97.08<br />
132.34<br />
145.72<br />
173.64<br />
241.71<br />
10.85<br />
8.01<br />
6.22<br />
14.32<br />
17.98<br />
13.09<br />
3.96<br />
11.47<br />
9.12<br />
5.88<br />
15.98<br />
84.80<br />
101.03<br />
114.21<br />
85.60<br />
75.16<br />
92.51<br />
112.86<br />
140.79<br />
141.17<br />
196.62<br />
196.77<br />
18.27<br />
20.00<br />
18.82<br />
18.51<br />
16.51<br />
19.18<br />
9.43<br />
17.78<br />
17.87<br />
9.13<br />
8.36<br />
Ratio<br />
Ultimate<br />
conditions<br />
80.57 0.88 11.11 1.02 Bolt 1/4<br />
84.55 0.92 14.00 1.29 Flange (1)<br />
88.57 0.96 17.17 1.58 Flange (1*)<br />
96.38 1.05 23.93 2.20 Bolt 1/2<br />
88.34 0.86 8.39 1.05 Bolt 1/4<br />
94.29 0.92 11.55 1.44 Flange (1)<br />
104.18 1.01 17.61 2.20 Bolt 1/2<br />
106.21 1.03 18.97 2.37 Flange (1*)<br />
95.74 0.85 6.44 1.04 Bolt 1/4<br />
103.47 0.91 9.57 1.54 Flange (1)<br />
113.33 1.00 14.16 2.27 Bolt 1/2<br />
125.23 1.11 20.46 3.29 Flange (1*)<br />
82.72 0.90 10.40 0.73 Flange (1)<br />
84.56 0.92 11.53 0.81 Bolt 1/4<br />
90.37 0.98 15.67 1.09 Flange (1*)<br />
111.79 1.22 40.67 2.84 Bolt 1/2<br />
77.64 0.89 12.88 0.72 Flange (1)<br />
79.60 0.92 14.44 0.80 Bolt 1/4<br />
81.50 0.94 16.11 0.90 Flange (1*)<br />
112.53 1.30 69.47 3.86 Bolt 1/2<br />
87.31 0.89 9.42 0.52 Flange (1)<br />
89.21 0.91 10.40 0.58 Bolt 1/4<br />
102.67 1.05 19.83 1.10 Flange (1*)<br />
114.33 1.17 31.74 1.77 Bolt 1/2<br />
104.11 1.07 3.80 0.96 Flange (1)<br />
111.82 1.15 5.07 1.28 Bolt 1/4<br />
119.08 1.23 9.00 2.27 Flange (1*)<br />
124.81 1.29 13.01 3.29 Bolt 1/2<br />
142.96 1.08 10.13 0.88 Flange (1)<br />
158.47 1.20 16.24 1.42 Flange (1*)<br />
176.09 1.33 25.98 2.26 Bolt 1/4<br />
157.51 1.08 8.27 0.91 Flange (1)<br />
177.93 1.22 14.71 1.61 Flange (1*)<br />
183.04 1.05 3.49 0.59 Flange (1)<br />
212.19 1.22 8.35 1.42 Flange (1*)<br />
230.02 1.32 15.13 2.57 Bolt 1/4<br />
210.60 0.87 2.97 0.19 Flange (1)<br />
233.41 0.97 7.56 0.47 Flange (1*)<br />
310.09 1.28 45.68 2.86 Bolt 1/4
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Table D.19 Prediction <strong>of</strong> axial stiffness by using <strong>the</strong> modified proposal for m.<br />
Test ID Num./Exp. Standard m Modified m<br />
stiffness ke.0 Ratio ke.0 Ratio<br />
Weld_T1(i) 73.77 52.36 0.71 59.00 0.80<br />
Weld_T1(ii) 89.12 68.82 0.77 85.26 0.96<br />
Weld_T1(iii) 107.29 87.68 0.82 119.32 1.11<br />
WT1g/h (av.) 71.08 71.61 1.01 84.68 1.19<br />
WT2Aa/b (av.) 61.83 58.97 0.95 66.00 1.07<br />
WT2Ba/b (av.) 79.75 82.62 1.04 103.73 1.30<br />
WT4Aa/b (av.) 86.96 115.86 1.33 136.22 1.57<br />
WT7_M16 60.73 60.73 1.00 71.39 1.18<br />
WT7_M20 64.23 62.12 0.97 72.98 1.14<br />
WT51a/b (av.) 52.90 64.08 1.21 75.47 1.43<br />
WT53C 116.09 117.27 1.01 138.40 1.19<br />
WT53D 137.70 120.60 0.88 142.93 1.04<br />
WT57_M12 85.78 99.95 1.17 116.81 1.36<br />
WT57_M20 150.96 109.09 0.72 128.83 0.85<br />
Average 0.97 1.16<br />
Coefficient <strong>of</strong> variation<br />
0.19<br />
0.18<br />
Table D.20 Prediction <strong>of</strong> ultimate resistance and deformation capacity (WP-Tstubs)<br />
by using <strong>the</strong> modified proposal for m.<br />
Test ID<br />
Num. or Exp. results Modified m<br />
Fmax ∆u.0 Fmax Ratio ∆u.0 Ratio<br />
(kN) (mm) (kN)<br />
(mm)<br />
Weld_T1(i) 92.02 10.85 83.42 0.91 14.57 1.34<br />
Weld_T1(ii) 102.75 8.01 95.86 0.93 10.99 1.37<br />
Weld_T1(iii) 113.10 6.22 108.05 0.96 8.46 1.36<br />
WT1g/h (av.) 91.91 14.32 80.56 0.88 10.27 0.72<br />
WT2Aa/b (av.) 86.82 17.98 74.12 0.85 13.23 0.74<br />
WT2Ba/b (av.) 97.88 13.09 86.75 0.89 8.85 0.68<br />
WT4Aa/b (av.) 103.26 4.33 127.24 1.23 9.36 2.16<br />
WT7_M16 97.08 3.96 107.02 1.10 3.78 0.95<br />
WT7_M20 98.90 4.24 108.12 1.09 3.49 0.82<br />
WT51a/b (av.) 117.36 5.54 111.74 0.95 3.04 0.55<br />
WT53C 132.34 11.47 134.63 1.02 10.74 0.94<br />
WT53D 145.72 9.12 134.73 0.92 10.68 1.17<br />
WT57_M12 121.87 4.33 162.92 1.34 5.38 1.24<br />
WT57_M20 241.71 15.98 185.74 0.77 2.83 0.18<br />
Average 0.99 1.02<br />
Coefficient <strong>of</strong> variation<br />
0.16<br />
0.47<br />
267
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
268<br />
Load, F (kN)<br />
120<br />
105<br />
Fig. D.76 Specimen Weld_T1(i).<br />
Load, F (kN)<br />
90<br />
75<br />
60<br />
45<br />
30<br />
Actual response<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
120<br />
105<br />
Fig. D.77 Specimen Weld_T1(ii).<br />
Load, F (kN)<br />
90<br />
75<br />
60<br />
Deformation, ∆ (mm)<br />
45<br />
30<br />
Actual response<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
120<br />
105<br />
Fig. D.78 Specimen Weld_T1(iii).<br />
90<br />
75<br />
60<br />
Deformation, ∆ (mm)<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20 22<br />
Deformation, ∆ (mm)
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.79 Specimen WT1.<br />
Load, F (kN)<br />
45<br />
30<br />
Actual response<br />
15<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.80 Specimen WT2A.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
45<br />
30<br />
Actual response<br />
15<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20<br />
105<br />
90<br />
75<br />
60<br />
Fig. D.81 Specimen WT2B.<br />
Deformation, ∆ (mm)<br />
45<br />
30<br />
Actual response<br />
15<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)<br />
269
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
270<br />
Load, F (kN)<br />
Fig. D.82 Specimen WT4A.<br />
Load, F (kN)<br />
135<br />
120<br />
105<br />
90<br />
75<br />
60<br />
45<br />
Actual response<br />
30<br />
Simplified response (Simple beam model)<br />
15<br />
0<br />
Simplified response with new m<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
160<br />
140<br />
120<br />
100<br />
Fig. D.83 Specimen WT7_M16.<br />
Load, F (kN)<br />
80<br />
Deformation, ∆ (mm)<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20<br />
160<br />
140<br />
120<br />
100<br />
Fig. D.84 Specimen WT7_M20.<br />
80<br />
Deformation, ∆ (mm)<br />
60<br />
Actual response<br />
40<br />
Simplified response (Simple beam model)<br />
20<br />
0<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20<br />
Deformation, ∆ (mm)
Simplified methodologies: assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> T-stub connections<br />
Load, F (kN)<br />
120<br />
100<br />
80<br />
60<br />
Fig. D.85 Specimen WT51.<br />
Load, F (kN)<br />
40<br />
Actual response<br />
20<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 1 2 3 4 5 6 7 8 9 10<br />
120<br />
100<br />
80<br />
60<br />
Fig. D.86 Specimen WT53C.<br />
Load, F (kN)<br />
Deformation, ∆ (mm)<br />
40<br />
Actual response<br />
20<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 1 2 3 4 5 6 7 8 9 10<br />
120<br />
100<br />
80<br />
60<br />
Fig. D.87 Specimen WT53D.<br />
Deformation, ∆ (mm)<br />
40<br />
Actual response<br />
20<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 1 2 3 4 5 6 7 8 9 10<br />
Deformation, ∆ (mm)<br />
271
Fur<strong>the</strong>r developments on <strong>the</strong> T-stub model<br />
272<br />
Load, F (kN)<br />
210<br />
180<br />
150<br />
120<br />
Fig. D.88 Specimen WT57_M12.<br />
Load, F (kN)<br />
90<br />
60<br />
Actual response<br />
30<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 1 2 3 4 5 6 7 8 9 10<br />
280<br />
240<br />
200<br />
160<br />
Fig. D.89 Specimen WT57_M20.<br />
Deformation, ∆ (mm)<br />
120<br />
80<br />
Actual response<br />
40<br />
0<br />
Simplified response (Simple beam model)<br />
Simplified response with new m<br />
0 2 4 6 8 10 12 14 16 18 20 22 24<br />
Deformation, ∆ (mm)
PART III: MONOTONIC BEHAVIOUR OF BEAM-TO-COLUMN<br />
BOLTED END PLATE CONNECTIONS<br />
273
7 EXPERIMENTAL TESTS ON BOLTED END PLATE CONNEC-<br />
TIONS<br />
7.1 INTRODUCTION<br />
An experimental investigation <strong>of</strong> eight statically loaded ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong><br />
moment connections undertaken at <strong>the</strong> Delft University <strong>of</strong> Technology is described<br />
in this chapter. It provides a better understanding <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong><br />
this joint type up to collapse and complements <strong>the</strong> study on welded T-stubs reported<br />
in Chapter 3.<br />
The specimens were designed to confine failure to <strong>the</strong> <strong>end</strong> <strong>plate</strong> and/or<br />
bolts without development <strong>of</strong> <strong>the</strong> full plastic moment capacity <strong>of</strong> <strong>the</strong> beam<br />
(partial strength joint). The parameters investigated were <strong>the</strong> <strong>end</strong> <strong>plate</strong> thickness<br />
and steel grade. The main objective was <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> ultimate behaviour<br />
<strong>of</strong> <strong>the</strong> components <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing and bolts and eventually <strong>the</strong><br />
proposal <strong>of</strong> sound design rules for this elemental part within <strong>the</strong> framework <strong>of</strong><br />
<strong>the</strong> so-called component method. The description <strong>of</strong> this test programme and<br />
results is given below. Comparisons with <strong>the</strong> code predictions [7.1] are also<br />
drawn.<br />
7.2 DESCRIPTION OF THE TEST PROGRAMME<br />
7.2.1 Test details<br />
The experimental programme essentially comprised four test details (two<br />
specimens for each testing type) on <strong>the</strong> above joint configuration. Two main<br />
parameters were varied in <strong>the</strong> four sets: <strong>the</strong> <strong>end</strong> <strong>plate</strong> thickness, tp and <strong>the</strong> <strong>end</strong><br />
<strong>plate</strong> steel grade. The specimens were fabricated from one column/beam set, as<br />
detailed in Table 7.1. The steel grade specified for <strong>the</strong> beams was S355. Unfortunately,<br />
due to a laboratory misunderstanding, steel grade S235 was ordered<br />
instead. This brought a problem in terms <strong>of</strong> <strong>the</strong> beam resistance that was naturally<br />
lower than expected. Therefore, for <strong>the</strong> critical cases, <strong>the</strong> beam flanges<br />
were stiffened with continuous <strong>plate</strong>s in order to increase <strong>the</strong> beam flange<br />
thickness and minimize <strong>the</strong> chance <strong>of</strong> premature failure. End <strong>plate</strong>s were connected<br />
to <strong>the</strong> beam-<strong>end</strong>s by full strength 45º-continuous fillet welds. The fillet<br />
welds were done in <strong>the</strong> shop in a down-hand position. The procedure involved<br />
manual metal arc welding in which consumable electrodes were used. Basic,<br />
s<strong>of</strong>t, low hydrogen electrodes were used in <strong>the</strong> process. Hand tightened fullthreaded<br />
M20 grade 8.8 bolts in 22 mm diameter drilled holes were employed<br />
in all sets. Two different batches <strong>of</strong> bolts were employed. The first batch <strong>of</strong><br />
275
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
bolts were employed in tests FS1a-b, FS2a-b and FS3a in both tension and<br />
compression zones. The second batch <strong>of</strong> bolts were used to fasten <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
and <strong>the</strong> beam in <strong>the</strong> tension zone in <strong>the</strong> remaining tests.<br />
The geometry <strong>of</strong> <strong>the</strong> specimens is depicted in Figs. 7.1-7.2. The column had<br />
a section pr<strong>of</strong>ile HE340M that was chosen so that it behaves almost as a rigid<br />
element. In addition, for <strong>the</strong> available column, <strong>the</strong> clearance above and below<br />
<strong>the</strong> <strong>end</strong> <strong>plate</strong> was less than 400 mm. However, since this is a rigid column, this<br />
limitation proved not to be severe. Regarding <strong>the</strong> joint geometry, <strong>the</strong> top bolt<br />
Table 7.1 Details <strong>of</strong> <strong>the</strong> test specimens.<br />
Test # Column Beam End Plate<br />
ID Pr<strong>of</strong>ile Steel<br />
grade<br />
Pr<strong>of</strong>ile Steel<br />
grade<br />
tp<br />
(mm)<br />
Steel<br />
grade<br />
FS1 2 HE340M S355 IPE300 S235 10 S355<br />
FS2 2 HE340M S355 IPE300 S235 15 S355<br />
FS3 2 HE340M S355 IPE300 S235 20 S355<br />
FS4 2 HE340M S355 IPE300 S235 10 S690<br />
hp = 400 Hc.up = 400<br />
1200<br />
276<br />
Hc.low = 400<br />
bc = 309<br />
HE340M<br />
hc = 377<br />
tp = 10, 15, 20<br />
5.5 ~ 6<br />
3.5 ~ 4<br />
Lstiffened ~ 500<br />
ts ~ 10<br />
IPE300<br />
ts = 10<br />
Lload = 1000 200<br />
Lbeam = 1200<br />
Fig. 7.1 Geometry <strong>of</strong> <strong>the</strong> specimens (dimensions in [mm]).<br />
hb = 300<br />
bb = 150
LX =<br />
69.65<br />
aw = 5.5 ~ 6<br />
d0 =<br />
22<br />
aw = 3.5 ~ 4<br />
Lcomp =<br />
30.35<br />
bp = 150<br />
e = 30 w = 90 30<br />
1 2<br />
3 4<br />
5 6<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
eX =<br />
30<br />
p = 90<br />
p2-3 = 205<br />
hp = 400<br />
ecomp =<br />
75<br />
aw = 5<br />
aw = 5<br />
150<br />
(a) Detail <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong>. (b) Detail <strong>of</strong> <strong>the</strong> stiffener.<br />
Fig. 7.2 Details <strong>of</strong> <strong>the</strong> specimens (dimensions in [mm]).<br />
row corresponds to specimen WT7_M20 (refer to Chapter 3) from <strong>the</strong> former<br />
test series on isolated T-stubs. All <strong>the</strong> <strong>end</strong> <strong>plate</strong> specimens were designed complying<br />
with <strong>the</strong> Eurocode 3 requirements [7.1] so that <strong>the</strong> components <strong>end</strong> <strong>plate</strong><br />
and bolts in <strong>the</strong> tension zone were <strong>the</strong> determining factor <strong>of</strong> collapse.<br />
7.2.2 Geometrical properties<br />
The actual geometry <strong>of</strong> <strong>the</strong> various connection elements was recorded before<br />
starting <strong>the</strong> test. For <strong>the</strong> various specimens <strong>the</strong> pr<strong>of</strong>iles and <strong>plate</strong>s actual dimensions<br />
and connection geometry are summarized in Table 7.2. These values<br />
are given as an average value <strong>of</strong> <strong>the</strong> several measurements from each series.<br />
Table 7.3 indicates <strong>the</strong> bolts measurements for each test.<br />
7.2.3 Mechanical properties<br />
7.2.3.1 Tension tests on <strong>the</strong> bolts<br />
Two different batches <strong>of</strong> bolts were used in <strong>the</strong> experiments. Having performed<br />
tests from series FS1 and FS2 and test FS3a, it was decided to use a different<br />
batch <strong>of</strong> bolts, from ano<strong>the</strong>r manufacturer as explained later in <strong>the</strong> text. Three<br />
machined bolts from each group were tested in tension in order to determine<br />
<strong>the</strong> mechanical properties <strong>of</strong> <strong>the</strong> bolt material, in accordance with ISO 898-<br />
1:1999(E) [7.2]. The average properties are set out in Table 7.4.<br />
300<br />
277
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 7.2 Actual geometry <strong>of</strong> <strong>the</strong> connection (averaged dimensions, [mm]).<br />
Test ID Column pr<strong>of</strong>ile Stif.<br />
278<br />
hc bc tfc Hc.up Hc.low ts<br />
FS1 175.00 219.00 10.76<br />
FS2<br />
FS3<br />
376.00 307.50 40.21<br />
174.50<br />
177.50<br />
219.50<br />
216.50<br />
10.50<br />
10.46<br />
FS4<br />
174.50 219.50 10.42<br />
Beam pr<strong>of</strong>ile<br />
hb bb tfb twb Lbeam Lload<br />
FS1 300.45 150.50 10.76 7.20 1200.00 1002.50<br />
FS2 301.40 149.60 10.67 7.01 1200.38 1000.25<br />
FS3 301.46 149.75 10.57 7.03 1191.50 992.63<br />
FS4 300.66 149.54 11.86 7.03 1218.75 991.88<br />
End <strong>plate</strong> and connection geometry<br />
hp bp tp e w<br />
FS1 401.04 149.84 10.40 30.01 89.91<br />
FS2 400.84 149.41 15.01 29.76 89.89<br />
FS3 401.40 150.47 20.02 30.27 89.93<br />
FS4 401.69 149.76 10.06 29.94 89.88<br />
eX LX p p2-3 ecomp.<br />
FS1 29.90 69.35 90.03 205.90 76.45<br />
FS2 30.10 69.30 89.98 205.04 75.44<br />
FS3 29.74 68.90 90.14 204.84 76.82<br />
FS4 29.83 69.86 89.95 205.28 76.13<br />
7.2.3.2 Tension tests <strong>of</strong> <strong>the</strong> structural steel<br />
The test programme included two different steel grades for <strong>the</strong> <strong>end</strong> <strong>plate</strong>: S355<br />
and S690. According to <strong>the</strong> European Standards EN 10025 [7.3] and EN 10204<br />
[7.4], <strong>the</strong> steel qualities are S355J0 (ordinary steel) and N-A-XTRA M70<br />
(high-strength steel for <strong>plate</strong>s), respectively. For <strong>the</strong> beam pr<strong>of</strong>ile, steel grade<br />
S235JR was ordered. Table 7.5 summarizes <strong>the</strong> chemical composition <strong>of</strong> <strong>the</strong><br />
different steel grades.<br />
The coupon tension testing <strong>of</strong> <strong>the</strong> structural steel material was performed<br />
according to <strong>the</strong> RILEM procedures [7.5]. The average characteristics are set<br />
out in Table 7.6. In this table <strong>the</strong> values for <strong>the</strong> Young modulus, E, <strong>the</strong> strain<br />
hardening modulus, Est, <strong>the</strong> static yield and tensile stresses, fy and fu, <strong>the</strong> yield<br />
ratio, ρy, <strong>the</strong> strain at <strong>the</strong> strain hardening point, εst, <strong>the</strong> uniform strain, εuni, and<br />
<strong>the</strong> ultimate strain, εu are given. Note that for <strong>the</strong> 10 mm thickness <strong>end</strong> <strong>plate</strong>s,<br />
<strong>the</strong> structural steel is <strong>the</strong> same that had been used in <strong>the</strong> testing <strong>of</strong> <strong>the</strong> isolated<br />
T-stub connections (cf. Chapter 3).
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 7.3 Bolt hole clearance and length (dimensions in [mm]; H-tght:<br />
Hand-tightening; Aft. clps.: after collapse).<br />
Test<br />
ID<br />
#1 #2 #3 #4 #5 #6<br />
d0 21.93 21.98 21.98 21.75 21.98 21.93<br />
FS1a<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
94.00<br />
94.00<br />
94.65<br />
94.00<br />
94.00<br />
94.40<br />
94.10<br />
94.10<br />
94.50<br />
94.25<br />
94.25<br />
94.90<br />
93.00<br />
93.10<br />
93.00<br />
92.90<br />
93.00<br />
93.10<br />
d0 22.05 22.00 22.03 22.05 22.10 21.90<br />
FS1b<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
94.00<br />
94.00<br />
94.95<br />
94.25<br />
94.25<br />
96.00<br />
94.40<br />
94.40<br />
95.40<br />
94.05<br />
94.05<br />
94.85<br />
93.15<br />
93.15<br />
93.00<br />
93.20<br />
93.20<br />
93.15<br />
d0 21.93 22.08 22.00 22.08 22.03 22.03<br />
FS2a<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
94.00<br />
94.02<br />
95.70<br />
93.90<br />
93.94<br />
96.18<br />
94.20<br />
94.20<br />
102.06<br />
93.85<br />
93.85<br />
96.62<br />
92.90<br />
92.94<br />
93.24<br />
92.90<br />
92.96<br />
93.78<br />
d0 22.00 21.93 22.00 21.98 22.00 21.95<br />
FS2b<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
93.90<br />
93.90<br />
95.16<br />
94.30<br />
94.40<br />
97.02<br />
93.90<br />
93.90<br />
101.30<br />
94.12<br />
94.12<br />
96.52<br />
92.86<br />
92.94<br />
93.28<br />
92.78<br />
92.90<br />
93.04<br />
d0 22.95 22.88 22.95 22.98 23.03 22.93<br />
FS3a<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
94.04<br />
94.10<br />
95.56<br />
94.00<br />
94.00<br />
95.10<br />
93.74<br />
93.80<br />
96.04<br />
94.10<br />
94.16<br />
96.12<br />
93.16<br />
93.16<br />
93.48<br />
62.90<br />
92.90<br />
93.44<br />
d0 22.05 21.90 22.00 22.03 21.95 22.03<br />
FS3b<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
92.54<br />
92.54<br />
95.30<br />
92.52<br />
92.52<br />
95.00<br />
92.56<br />
92.56<br />
95.25<br />
92.50<br />
92.50<br />
99.22<br />
92.78<br />
93.00<br />
93.24<br />
93.14<br />
93.14<br />
93.24<br />
d0 22.08 22.00 22.05 21.93 22.00 22.05<br />
FS4a<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
92.46<br />
92.50<br />
94.40<br />
92.45<br />
92.48<br />
93.94<br />
92.54<br />
92.56<br />
99.62<br />
92.52<br />
92.54<br />
102.62<br />
92.70<br />
92.74<br />
93.06<br />
92.68<br />
92.70<br />
93.10<br />
d0 22.03 22.08 21.98 22.00 21.98 22.03<br />
FS4b<br />
Bolt<br />
length<br />
Initial<br />
H-tght<br />
Aft.<br />
clps.<br />
92.40<br />
92.42<br />
94.16<br />
92.38<br />
92.40<br />
94.82<br />
92.32<br />
92.34<br />
100.94<br />
92.38<br />
92.42<br />
100.26<br />
93.04<br />
93.06<br />
93.26<br />
93.06<br />
93.08<br />
93.38<br />
279
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 7.4 Average characteristic values for <strong>the</strong> bolts.<br />
Batch E (MPa) fy (MPa) fu (MPa) εu<br />
1 223166 857.33 913.78 0.184<br />
2 222982 854.31 916.81 0.156<br />
Table 7.5 Chemical composition <strong>of</strong> <strong>the</strong> structural steels according to <strong>the</strong><br />
European standards.<br />
% max. C Mn Si P S N CEV<br />
S235JR 0.17 1.40 ⎯ 0.045 0.045 0.012 0.35<br />
S355J0 0.20 1.60 0.55 0.040 0.045 0.009 0.40<br />
N-A-XTRA M70 0.20 1.60 0.80 0.020 0.010 ⎯ 0.48<br />
Table 7.6 Average characteristic values for <strong>the</strong> structural steels.<br />
280<br />
Specimen Steel<br />
grade<br />
End<br />
<strong>plate</strong><br />
Beam<br />
E<br />
(MPa)<br />
Est<br />
(MPa)<br />
fy<br />
(MPa)<br />
fu<br />
(MPa)<br />
tp = 10 S355 209856 2264 340.12 480.49 0.708<br />
tp = 15 S355 208538 2901 342.82 507.85 0.675<br />
tp = 20 S355 208622 2771 342.62 502.59 0.682<br />
tp = 10 S690 204462 2495 698.55 741.28 0.940<br />
Web S235 208332 1856 299.12 446.25 0.670<br />
Flange S235 209496 1933 316.24 462.28 0.684<br />
Specimen Steel<br />
grade<br />
εst εuni εu<br />
End<br />
<strong>plate</strong><br />
Beam<br />
tp = 10 S355 0.015 0.224 0.361<br />
tp = 15 S355 0.020 0.198 0.475<br />
tp = 20 S355 0.017 0.196 0.457<br />
tp = 10 S690 0.014 0.075 0.174<br />
Web S235 0.016 0.235 0.464<br />
Flange S235 0.016 0.235 0.299<br />
7.2.4 Test arrangement and instrumentation<br />
The main features <strong>of</strong> <strong>the</strong> test apparatus are illustrated in Figs. 7.3a-b. Concerning<br />
<strong>the</strong> T arrangement depicted in Figs. 7.1-7.2, <strong>the</strong> actual connection was rotated<br />
180º for practical reasons. The column was <strong>bolted</strong> to a reaction wall. The<br />
reader should bear in mind that <strong>the</strong> goal <strong>of</strong> <strong>the</strong>se tests was <strong>the</strong> study <strong>of</strong> <strong>the</strong> <strong>end</strong><br />
<strong>plate</strong> in <strong>the</strong> tension zone and <strong>the</strong>refore it had to be ensured that <strong>the</strong> column was<br />
not governing any failure mode.<br />
The load was applied by a 400 kN testing machine (hydraulic jack with<br />
maximum piston stroke <strong>of</strong> ±200 mm), through a purpose-built device (Fig.<br />
ρy
(a) Test apparatus (illustration<br />
with specimen FS2a).<br />
(c) Detail <strong>of</strong> <strong>the</strong> load application device.<br />
Fig. 7.3 Equipment and test specimen.<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
(b) Detail <strong>of</strong> <strong>the</strong> beam and connection zone (illustration<br />
with specimen FS1a).<br />
7.3c) that was clamped to <strong>the</strong> beam at 200 mm from <strong>the</strong> free <strong>end</strong>. A beam<br />
guidance device near <strong>the</strong> loading point was provided to prevent lateral torsional<br />
buckling <strong>of</strong> <strong>the</strong> beam with <strong>the</strong> course <strong>of</strong> loading. For that purpose, a special device<br />
located at 250 mm from <strong>the</strong> load point was attached to <strong>the</strong> specimens<br />
(Figs. 7.3a-b).<br />
The length <strong>of</strong> <strong>the</strong> beam was chosen to ensure a realistic stress pattern developed<br />
at <strong>the</strong> connection, on one hand, and to ensure that fracture <strong>of</strong> <strong>the</strong> several<br />
specimens, i.e. ultimate load, was attained with <strong>the</strong> specific testing machine.<br />
The instrumentation plan is described in Figs. 7.4-7.6 below. The primary<br />
requirements <strong>of</strong> <strong>the</strong> instrumentation were <strong>the</strong> measurement <strong>of</strong> <strong>the</strong> applied load,<br />
<strong>the</strong> relevant displacements <strong>of</strong> <strong>the</strong> connection (e.g. vertical displacement <strong>of</strong> <strong>the</strong><br />
beam, horizontal displacement <strong>of</strong> <strong>the</strong> assembly <strong>end</strong> <strong>plate</strong>-tensile beam flange)<br />
and bolt elongation. The record <strong>of</strong> all measurements was made automatically<br />
with intervals <strong>of</strong> 1 second. The displacements were measured by means <strong>of</strong><br />
281
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
LVDTs located as indicated in Fig. 7.4. These were attached to <strong>the</strong> elements<br />
with special glue. Four LVDTs with an accuracy <strong>of</strong> 0.5% were used to measure<br />
<strong>the</strong> beam vertical displacements (DT1-4). The range <strong>of</strong> <strong>the</strong>se transducers is 480<br />
mm for DT1, 425 mm for DT2 and 200 mm for both DT3 and DT4. The horizontal<br />
displacements <strong>of</strong> <strong>the</strong> assembly <strong>end</strong> <strong>plate</strong>-beam flanges were measured<br />
with 50 mm LVDTs with a precision <strong>of</strong> 0.5% (DT6-7, compression side, DT9-<br />
10, tension side). In order to measure <strong>the</strong> <strong>end</strong> <strong>plate</strong> vertical displacement due to<br />
elongation <strong>of</strong> <strong>the</strong> bolt holes, an additional 35 mm LVDT (DT5 – accuracy <strong>of</strong><br />
0.5%) was attached to <strong>the</strong> lower part <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> (tension side), as illustrated<br />
in Fig. 7.4. To ensure that <strong>the</strong> displacements <strong>of</strong> <strong>the</strong> column could be neglected,<br />
two LVDTs (DT8,11) were attached to <strong>the</strong> back side <strong>of</strong> <strong>the</strong> column.<br />
These transducers could measure up to 1.5 mm displacement with an accuracy<br />
<strong>of</strong> 0.5% as well. This was <strong>the</strong> precision <strong>of</strong> <strong>the</strong> electrical components connected<br />
to <strong>the</strong> data logger.<br />
The bolts deformations were measured with special measuring brackets,<br />
MBs (horseshoe device), as common practice in <strong>the</strong> Stevin Laboratory <strong>of</strong> <strong>the</strong><br />
Delft University <strong>of</strong> Technology. These devices were attached to <strong>the</strong> bolts only<br />
282<br />
Top<br />
DT8<br />
DT11<br />
HE340M<br />
Bottom<br />
Bolts 1/3/5<br />
DT8/11<br />
DT6/7<br />
DT9/10<br />
DT4<br />
DT5<br />
100 200<br />
DT6/9<br />
1000<br />
IPE300<br />
DT3 DT2 DT1<br />
Load<br />
300 300 300<br />
DT5 DT4<br />
DT7/10<br />
DT3 DT2 DT1<br />
Bolts 2/4/6<br />
Fig. 7.4 Location <strong>of</strong> <strong>the</strong> displacement transducers.<br />
Front<br />
Back
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Fig. 7.5 MBs 1,3 and LVDTs 6,9 (illustration with specimen FS4a).<br />
10<br />
19<br />
SG3<br />
SG2<br />
SG1<br />
Back<br />
30 45 45 30<br />
11 13<br />
2 12 1<br />
4 5 6<br />
7 8<br />
4<br />
3<br />
9<br />
14<br />
10<br />
15<br />
Top<br />
3<br />
6 5<br />
30<br />
90<br />
30<br />
250<br />
19<br />
19<br />
Unidirectional strain gauges<br />
xy strain gauges<br />
30<br />
45<br />
13<br />
11 2 12<br />
4 5 6<br />
7 8<br />
14<br />
4<br />
15<br />
9<br />
10<br />
5 19<br />
(a) Sketch <strong>of</strong> <strong>the</strong> location <strong>of</strong> <strong>the</strong> strain gauges on <strong>the</strong> beam and <strong>end</strong> <strong>plate</strong>.<br />
(b) Strain gauges located at <strong>the</strong> beam<br />
and <strong>end</strong> <strong>plate</strong> extension.<br />
Fig. 7.6 Location <strong>of</strong> <strong>the</strong> strain gauges.<br />
30<br />
90<br />
30<br />
(c) Strain gauges located at <strong>the</strong> <strong>end</strong><br />
<strong>plate</strong>.<br />
283
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
on <strong>the</strong> tension side. They could only measure up to 2 mm <strong>of</strong> displacement.<br />
However, <strong>the</strong>y were removed before collapse to prevent damage. Fig. 7.5<br />
shows <strong>the</strong>se devices for bolts 1 and 3.<br />
Finally, strain gauges, SGs, TML (maximum strain 21000 µm/m) were<br />
added to <strong>the</strong> <strong>end</strong> <strong>plate</strong> (back side) in <strong>the</strong> tension zone to provide insight into <strong>the</strong><br />
strain distribution in this zone (Fig. 7.6). In addition, <strong>the</strong> specimens were provided<br />
with strain gauges at <strong>the</strong> top <strong>of</strong> <strong>the</strong> tension beam flange.<br />
For good comparison <strong>of</strong> <strong>the</strong> results, all specimens used <strong>the</strong> same arrangement<br />
for <strong>the</strong> location <strong>of</strong> <strong>the</strong> strain gauges and measuring devices.<br />
7.2.5 Testing procedure<br />
Before installation <strong>of</strong> <strong>the</strong> specimens into <strong>the</strong> testing rig, <strong>the</strong> dimensions <strong>of</strong> <strong>the</strong><br />
<strong>plate</strong>s were recorded and <strong>the</strong> bolts were hand-tightened and measured. The<br />
specimens were <strong>the</strong>n placed into <strong>the</strong> machine and aligned. The bolts were fastened<br />
with an ordinary spanner (45º turn) and measured.<br />
In order to sketch <strong>the</strong> yield line patterns <strong>the</strong> specimens were painted with<br />
chalk. The measurement devices and strain gauges were <strong>the</strong>n connected. Electronic<br />
records started and all <strong>the</strong> equipment was verified.<br />
The specimens were subjected to monotonic tensile force, which was applied<br />
to <strong>the</strong> beam as explained before. The tests were carried out under displacement<br />
control with a constant speed <strong>of</strong> 0.02 mm/s up to collapse <strong>of</strong> <strong>the</strong><br />
specimens. The test itself <strong>the</strong>n started with loading <strong>of</strong> <strong>the</strong> specimen up to<br />
2/3Mj.Rd, which corresponds to <strong>the</strong> <strong>the</strong>oretical elastic limit. Mj.Rd is <strong>the</strong> full plastic<br />
resistance and is determined according to Eurocode 3. Complete unloading<br />
followed on and <strong>the</strong> specimen was <strong>the</strong>n reloaded up to collapse. In this third<br />
phase, <strong>the</strong> test was interrupted at <strong>the</strong> load levels corresponding to 2/3Mj.Rd,<br />
Mj.Rd, at <strong>the</strong> knee-range and after this level each six minutes, equivalent to an<br />
actuator displacement <strong>of</strong> 7.2 mm. The hold on <strong>of</strong> <strong>the</strong> test lasted for three minutes.<br />
The testing procedures adopted for <strong>the</strong> full-scale tests were identical to<br />
those described in Chapter 3 for <strong>the</strong> individual T-stubs.<br />
Four collapse failure modes or a combination <strong>of</strong> those were observed in <strong>the</strong><br />
test: (i) weld cracking, (ii) <strong>plate</strong> cracking, (iii) bolt fracture and (iv) bolt nut<br />
stripping (see Table 7.7). After collapse, <strong>the</strong> bolts were measured again (Table<br />
7.3).<br />
7.3 TEST RESULTS<br />
The results presented in <strong>the</strong> following sections relate to <strong>the</strong> third phase <strong>of</strong> <strong>the</strong><br />
tests, after elimination <strong>of</strong> slippery and after settlement <strong>of</strong> <strong>the</strong> connecting parts.<br />
The plotted graphs refer to <strong>the</strong> applied load, displacement and strain direct<br />
readings and to <strong>the</strong> corresponding b<strong>end</strong>ing moment and deformations.<br />
The b<strong>end</strong>ing moment, M, acting on <strong>the</strong> connection corresponds to <strong>the</strong> ap-<br />
284
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 7.7 Description <strong>of</strong> failure types.<br />
Test ID Mode <strong>of</strong> failure<br />
FS1a Weld failure <strong>of</strong> <strong>the</strong> assembly beam-<strong>end</strong> <strong>plate</strong>, both at <strong>the</strong> flange and<br />
web sides.<br />
FS1b Weld failure <strong>of</strong> <strong>the</strong> assembly beam-<strong>end</strong> <strong>plate</strong>, both at <strong>the</strong> flange and<br />
web sides and <strong>plate</strong> cracking at opposite sides.<br />
FS2a Nut stripping <strong>of</strong> bolt #4 and weld failure along <strong>the</strong> whole <strong>end</strong> <strong>plate</strong><br />
extension width but not at <strong>the</strong> inner part.<br />
FS2b Nut stripping <strong>of</strong> bolts #1 and #4 with no <strong>plate</strong> cracking or weld failure.<br />
FS3a Nut stripping <strong>of</strong> bolts #3 and #4 and some weld failure close to bolt<br />
#3 but without development <strong>of</strong> a crack.<br />
FS3b Nut stripping <strong>of</strong> bolt #3.<br />
FS4a Fracture <strong>of</strong> bolt #4 and some weld failure at <strong>the</strong> <strong>end</strong> <strong>plate</strong> extension<br />
close to bolt #1 but without development <strong>of</strong> a complete crack.<br />
FS4b Fracture <strong>of</strong> bolt #3.<br />
plied load, “Load” multiplied by <strong>the</strong> distance between <strong>the</strong> load application<br />
point and <strong>the</strong> face <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong>, Lload:<br />
M = Load × Lload<br />
(7.1)<br />
The rotational deformation <strong>of</strong> <strong>the</strong> joint, Φ, is <strong>the</strong> sum <strong>of</strong> <strong>the</strong> shear deformation<br />
<strong>of</strong> <strong>the</strong> column web panel zone, γ and <strong>the</strong> connection rotational deformation,<br />
φ, that is defined as <strong>the</strong> change in angle between <strong>the</strong> centrelines <strong>of</strong> beam<br />
and column, θb and θc. In <strong>the</strong>se tests, <strong>the</strong> column hardly deforms as it behaves<br />
as a rigid element. This statement will be validated later in <strong>the</strong> text. Then, both<br />
γ and θc are nought and so:<br />
Φ= φ = θb<br />
(7.2)<br />
The beam rotation is approximately given by (Fig. 7.4):<br />
δDT1 δDT2δDT3<br />
θb = arctan − θb. el = arctan − θb. el = arctan − θb.<br />
el =<br />
900 600 300<br />
(7.3)<br />
δ DT4<br />
= arctan −θbel<br />
.<br />
100<br />
where δDTi are <strong>the</strong> vertical displacements at LVDT DTi and θb.el is <strong>the</strong> beam<br />
elastic rotation. The above expression disregards <strong>the</strong> effect <strong>of</strong> shear deforma-<br />
tion in <strong>the</strong> beam and assumes that <strong>the</strong> vertical displacements <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
are negligible, i.e. δ DT5 ≈ 0 . Some differences in <strong>the</strong> results from DT4 are expected<br />
when compared to <strong>the</strong> remaining LVDTs since it is located closer to <strong>the</strong><br />
<strong>end</strong> <strong>plate</strong>. In this region, <strong>the</strong> beam <strong>the</strong>ory is not valid and <strong>the</strong> stress distribution<br />
is not smooth.<br />
By using <strong>the</strong> above relationships, <strong>the</strong> M-φ curve <strong>of</strong> <strong>the</strong> connection can be<br />
characterized. The main features <strong>of</strong> this curve are: resistance, stiffness and ro-<br />
285
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
tation capacity. In particular, for <strong>the</strong> different tests <strong>the</strong> following characteristics<br />
are assessed [7.6]: <strong>the</strong> knee-range <strong>of</strong> <strong>the</strong> M-φ curve, <strong>the</strong> plastic flexural resistance,<br />
Mj.Rd, <strong>the</strong> maximum b<strong>end</strong>ing moment, Mmax, <strong>the</strong> initial stiffness, Sj.ini, <strong>the</strong><br />
post-limit stiffness, Sj.p-l, <strong>the</strong> rotation corresponding to <strong>the</strong> maximum load level,<br />
φ and <strong>the</strong> rotation capacity, φCd (see Figs. 1.28 and 7.7). The stiffness values<br />
M max<br />
are computed by means <strong>of</strong> linear regression analysis <strong>of</strong> <strong>the</strong> quasi-elastic<br />
branches before and after <strong>the</strong> knee-range.<br />
A brief summary <strong>of</strong> <strong>the</strong> observed collapse failure modes is given in Table<br />
7.7 and some illustrations are given in Fig. 7.8. Failure occurred due to a variety<br />
<strong>of</strong> reasons, but <strong>the</strong> collapse modes always involved <strong>the</strong> components <strong>end</strong><br />
<strong>plate</strong> and bolts in <strong>the</strong> tension zone.<br />
286<br />
B<strong>end</strong>ing moment, M (kNm)<br />
200<br />
160<br />
120<br />
80<br />
Mj.Rd<br />
Sj.p-l<br />
Knee-range<br />
40<br />
Sj.ini<br />
0<br />
Φ MRd Φ Xd<br />
Φ Cd<br />
0 10 20 30 40 50 60 70<br />
Connection rotation φ (mrad)<br />
Fig. 7.7 Moment-rotation characteristics from tests.<br />
(i) General view. (ii) Detail: weld failure,<br />
front side.<br />
Mmax<br />
(iii) Detail: bolts #2-#4<br />
after failure (notice <strong>the</strong><br />
b<strong>end</strong>ing <strong>of</strong> bolt #2).<br />
(a) Specimen FS1b.<br />
Fig. 7.8 Illustration <strong>of</strong> <strong>the</strong> various failure types observed in <strong>the</strong> experiments.
(iv) Detail: <strong>end</strong> <strong>plate</strong> cracking (extension),<br />
back side.<br />
(a) Specimen FS1b (cont.).<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
(v) Elongation <strong>of</strong> <strong>the</strong> bolt holes<br />
in <strong>the</strong> tension zone.<br />
(i) General view <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong>. (ii) Nut stripping <strong>of</strong> bolt #4<br />
(column side).<br />
(iii) Detail <strong>of</strong> <strong>the</strong> weld fracture in <strong>the</strong> tension<br />
zone.<br />
(iv) Detail <strong>of</strong> tension bolts<br />
(bolt #3 nearly fractures).<br />
(b) Specimen FS2a.<br />
Fig. 7.8 Illustration <strong>of</strong> <strong>the</strong> various failure types observed in <strong>the</strong> experiments<br />
(cont.).<br />
287
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
288<br />
(i) Bolt #3.<br />
(ii) Bolt #4.<br />
(c) Specimen FS3a. (d) Specimen FS4b.<br />
Fig. 7.8 Illustration <strong>of</strong> <strong>the</strong> various failure types observed in <strong>the</strong> experiments<br />
(cont.).<br />
7.3.1 Moment-rotation curves<br />
As explained above, <strong>the</strong> M-φ curves for <strong>the</strong> different connections are obtained<br />
from <strong>the</strong> beam vertical displacement readings and <strong>the</strong> applied load. For illustration,<br />
Fig. 7.9 plots <strong>the</strong> load vs. vertical displacement <strong>of</strong> <strong>the</strong> beam for specimen<br />
FS1a. This curve can be converted into a moment-“gross beam rotation” curve<br />
by application <strong>of</strong> Eqs. (7.1) and (7.3) excluding θel, as shown in Fig. 7.10a for<br />
<strong>the</strong> four LVDTs DT1-4. Examination <strong>of</strong> <strong>the</strong>se four curves indicates a good<br />
agreement <strong>of</strong> <strong>the</strong> results obtained for DT1-3 and some deviation for DT4.<br />
These differences have already been explained earlier in <strong>the</strong> text. Therefore,<br />
<strong>the</strong> results from DT1 are kept for fur<strong>the</strong>r analysis. If now <strong>the</strong> beam elastic deformation<br />
is subtracted from <strong>the</strong> “gross rotation” (see Eq. (7.3)), <strong>the</strong> connection<br />
rotation can be completely characterized (Fig. 7.10b). This value is taken as<br />
equal to <strong>the</strong> beam rotation because <strong>the</strong> column rotation, θc, can be disregarded<br />
in comparison with θb (see Fig. 7.11) and also because <strong>the</strong> <strong>end</strong> <strong>plate</strong> vertical<br />
deformation due to <strong>the</strong> bolt hole elongation can be neglected when compared to<br />
<strong>the</strong> δDT1 (see Fig. 7.12). Note that for specimen FS1a <strong>the</strong> slippery at circa 110<br />
kN has to be disregarded.<br />
The M-φ responses for <strong>the</strong> eight connection details are reported in Fig. 7.13.<br />
Almost identical responses are obtained for each set over <strong>the</strong> entire elastoplastic<br />
range. This proves that <strong>the</strong> test procedure and <strong>the</strong> instrumentation setup<br />
adopted for <strong>the</strong> programme were satisfactory. The main features <strong>of</strong> <strong>the</strong> eight<br />
curves are summarized in Table 7.8. All characteristic values are referred to <strong>the</strong><br />
readings from LVDT DT1. In all cases, <strong>the</strong> knee-range domain <strong>of</strong> <strong>the</strong> curves is<br />
alike for <strong>the</strong> same connection detail. The maximum resistance is also similar,
Total applied load (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
DT4 DT3<br />
DT2<br />
DT1<br />
0<br />
0 10 20 30 40 50 60 70<br />
Vertical displacement <strong>of</strong> <strong>the</strong> beam (mm)<br />
Fig. 7.9 Beam vertical displacement readings <strong>of</strong> LVDTs DT1-4 for specimen<br />
FS1a.<br />
B<strong>end</strong>ing moment (kNm)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
DT1<br />
DT4<br />
DT3<br />
DT2<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100<br />
Beam rotation includ. elastic def. (mrad)<br />
(a) Beam rotation computed from <strong>the</strong> displacement readings <strong>of</strong> LVDTs DT1-4<br />
[arctan(δDTi/LDTi)].<br />
B<strong>end</strong>ing moment (kNm)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
Beam rotation including <strong>the</strong> beam elastic<br />
40<br />
deformation<br />
20<br />
0<br />
Connection rotation (equal to <strong>the</strong> beam<br />
rotation)<br />
0 10 20 30 40 50 60 70 80 90 100<br />
Beam rotation θb (mrad)<br />
(b) Beam rotation computed by means <strong>of</strong> Eq. (7.3) from <strong>the</strong> readings <strong>of</strong> DT1.<br />
Fig. 7.10 Beam rotation for specimen FS1a.<br />
289
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
290<br />
Total applied load (kN)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
DT8 (compression<br />
side)<br />
(a) Column horizontal displacements.<br />
B<strong>end</strong>ing moment (kNm)<br />
DT11 (tension<br />
side)<br />
0<br />
-0.20 -0.16 -0.12 -0.08 -0.04 0.00 0.04 0.08 0.12 0.16 0.20<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
Horizontal displac. (column side) (mm)<br />
0<br />
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0<br />
Column rotation θc (mrad)<br />
⎛ ⎛ δDT8 + δ ⎞⎞<br />
DT11<br />
(b) Corresponding column rotations ⎜θc= arctan ⎜ ⎟⎟.<br />
⎜ ⎜ hb − t ⎟⎟<br />
⎝ ⎝ fb ⎠⎠<br />
B<strong>end</strong>ing moment (kNm)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
0.00 0.03 0.06 0.09 0.12 0.15 0.18 0.21 0.24 0.27 0.30<br />
θc/θ b<br />
(c) Ratio between column rotation and beam rotation.<br />
Fig.7.11 Column rotation for specimen FS1a.
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
FS1a<br />
FS4b<br />
0<br />
-7 -6 -5 -4 -3 -2 -1 0 1<br />
Vertical displac. <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> (δ DT5) (mm)<br />
Fig. 7.12 End <strong>plate</strong> vertical displacement for specimens FS1a and FS4b.<br />
(a) Series FS1.<br />
B<strong>end</strong>ing moment (kNm)<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Connection rotation φ (mrad)<br />
FS1a FS1b<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
FS2a FS2b<br />
(b) Series FS2.<br />
Fig. 7.13 Moment-rotation curves for <strong>the</strong> four test series.<br />
291
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
(c) Series FS3.<br />
292<br />
B<strong>end</strong>ing moment (kNm)<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
240<br />
210<br />
180<br />
150<br />
120<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
90<br />
60<br />
30<br />
Connection rotation φ (mrad)<br />
FS3a FS3b<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
FS4a FS4b<br />
(d) Series FS4.<br />
Fig. 7.13 Moment-rotation curves for <strong>the</strong> four test series (cont.).<br />
though in series FS1 and FS3 some differences were observed. In series FS1,<br />
experimental observations show that <strong>the</strong> welding quality in set FS1a is poor,<br />
i.e. <strong>the</strong> welding procedure resulted in a glue weld instead <strong>of</strong> a burnt-in weld.<br />
This induced premature cracking <strong>of</strong> <strong>the</strong> specimen. Regarding series FS3, <strong>the</strong><br />
discrepancies arise because different bolts were employed in <strong>the</strong> two sets and<br />
also because <strong>the</strong>re was a disturbance in test FS3a at a load level <strong>of</strong> 190 kN that<br />
may have had some effect on <strong>the</strong> final results. In terms <strong>of</strong> rotational stiffness,<br />
some differences arise, particularly for Sj.ini in <strong>the</strong> case <strong>of</strong> series FS1 and Sj.p-l<br />
for series FS3. Identical values <strong>of</strong> <strong>the</strong> ratio Sj.ini/Sj.p-l are obtained for <strong>the</strong> four<br />
test types. Exception is made for joint FS3a, which shows some disturbance in<br />
<strong>the</strong> post-limit regime, and <strong>the</strong>refore <strong>the</strong> results are not reliable in this domain.<br />
Now, in terms <strong>of</strong> maximum rotation, <strong>the</strong> values at Mmax are close for all sets<br />
(again, <strong>the</strong> results for FS3a are not reliable in <strong>the</strong> post-limit domain), particularly<br />
for specimen FS2. Higher deviations appear for φCd, especially for series
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 7.8 Main characteristics <strong>of</strong> <strong>the</strong> moment-rotation curves.<br />
Test ID<br />
Resistance [kNm]<br />
Knee-range Mj.Rd Mmax<br />
FS1a 65-112 105.60 ( M Rd<br />
5.81 mrad)<br />
FS1b 68-120 109.30 ( M Rd<br />
6.49 mrad)<br />
FS2a 120-174 165.65 ( M Rd<br />
7.08 mrad)<br />
FS2b 117-181 170.22 ( M Rd<br />
7.74 mrad)<br />
FS3a 112-186 172.27 ( M Rd<br />
7.47 mrad)<br />
FS3b 122-200 192.66 ( M Rd<br />
8.94 mrad)<br />
FS4a 110-170 165.60 ( M Rd<br />
10.24 mrad)<br />
FS4b 110-170 163.52 ( M 9.53 mrad)<br />
φ = 142.76<br />
φ = 161.17<br />
φ = 193.06<br />
φ = 197.31<br />
φ = 202.91<br />
φ = 214.35<br />
φ = 185.32<br />
φ =<br />
Rd<br />
Stiffness [kNm/mrad]<br />
187.67<br />
Ratio<br />
Sj.ini Sj.p-l<br />
FS1a<br />
2 18.19 ( R = 0.9717)<br />
0.84 ( )<br />
FS1b<br />
2 16.84 ( R = 0.9921)<br />
0.74 ( 2<br />
0.9681)<br />
FS2a<br />
2 23.39 ( R = 0.9925)<br />
0.84 ( 2<br />
0.8611)<br />
FS2b<br />
2 22.00 ( R = 0.9968)<br />
0.92 ( 2<br />
0.8405)<br />
FS3a<br />
2 23.23 ( R = 0.9905)<br />
1.81 ( 2<br />
0.8629)<br />
FS3b<br />
2 21.56 ( R = 0.9972)<br />
1.03 ( 2<br />
0.8003)<br />
FS4a<br />
2 16.18 ( R = 0.9936)<br />
0.78 ( 2<br />
0.8004)<br />
FS4b<br />
2 17.15 ( R = 0.9956)<br />
0.74 ( 2<br />
0.8681)<br />
2<br />
R = 0.9384<br />
Sj.ini/Sj.p-l<br />
21.55<br />
R = 22.78<br />
R = 27.93<br />
R = 23.91<br />
R = 12.82<br />
R = 20.96<br />
R = 20.61<br />
R =<br />
Rotation [mrad]<br />
23.29<br />
φXd φCd Mmax<br />
FS1a 18.23 68.91 ( Cd<br />
127.71 kNm)<br />
FS1b 20.00 111.22 ( Cd<br />
70.29 kNm)<br />
FS2a 17.45 82.88 ( Cd<br />
66.00 kNm)<br />
FS2b 19.17 60.89 ( Cd<br />
147.93 kNm)<br />
FS3a 13.75 42.76 ( Cd<br />
108.16 kNm)<br />
FS3b 18.25 48.74 ( Cd<br />
153.10 kNm)<br />
FS4a 19.25 61.69 ( Cd<br />
150.25 kNm)<br />
FS4b 18.33 64.24 ( 158.09 kNm)<br />
φ<br />
M φ = 61.55<br />
M φ = 77.05<br />
M φ = 41.72<br />
M φ = 40.30<br />
M φ = 25.00<br />
M φ = 29.99<br />
M φ = 37.70<br />
M φ = 43.85<br />
Cd<br />
293
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
FS1 and FS2. The differences that are observed in series FS1 have already been<br />
explained above. For series FS2 and FS3, φCd is not well defined since it corresponds<br />
to <strong>the</strong> beginning <strong>of</strong> final unloading <strong>of</strong> <strong>the</strong> test. No actual rupture was<br />
observed in this case. The test was stopped because <strong>the</strong> deformations were already<br />
too high and <strong>the</strong>re was fear <strong>of</strong> damaging <strong>the</strong> equipment if <strong>the</strong> test went<br />
on any fur<strong>the</strong>r.<br />
One connection from each set is now chosen for <strong>the</strong> purpose <strong>of</strong> a comparative<br />
study. In all <strong>the</strong> cases, <strong>the</strong> assembly <strong>end</strong> <strong>plate</strong>-bolts is <strong>the</strong> main source <strong>of</strong><br />
connection deformability. Fig. 7.14 compares <strong>the</strong> rotational behaviour <strong>of</strong> <strong>the</strong><br />
four test types and shows an increase in resistance and rotational stiffness and a<br />
loss <strong>of</strong> rotation capacity with <strong>the</strong> <strong>end</strong> <strong>plate</strong> thickness (FS1, FS2 and FS3). The<br />
effect <strong>of</strong> <strong>the</strong> steel grade is identical (FS1 and FS4).<br />
294<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
FS3b<br />
FS4b<br />
FS2a<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
FS1b<br />
Fig. 7.14 Comparison <strong>of</strong> <strong>the</strong> moment-rotation curves for <strong>the</strong> four test series.<br />
7.3.2 Behaviour <strong>of</strong> <strong>the</strong> tension zone<br />
7.3.2.1 End <strong>plate</strong> deformation behaviour<br />
The most significant characteristic describing <strong>the</strong> overall <strong>end</strong> <strong>plate</strong> deformation<br />
behaviour in <strong>the</strong> tension zone is <strong>the</strong> F-∆ response. The test setup does not allow<br />
a direct measurement <strong>of</strong> <strong>the</strong> force at <strong>the</strong> component level but <strong>the</strong> information<br />
ga<strong>the</strong>red from LVDTs DT9 and DT10 permits a full <strong>characterization</strong> <strong>of</strong> <strong>the</strong> <strong>end</strong><br />
<strong>plate</strong> deformation behaviour. These transducers are attached to <strong>the</strong> beam flange<br />
and <strong>the</strong>y measure <strong>the</strong> gap between <strong>the</strong> <strong>end</strong> <strong>plate</strong> and <strong>the</strong> column flange (see<br />
Fig. 7.4). As an example, Fig. 7.15a traces <strong>the</strong> moment-gap response obtained<br />
for DT9 and DT10 for specimens FS1b and FS4a and indicates a good agreement<br />
over <strong>the</strong> whole loading history. For comparison, Fig. 7.15b shows that<br />
<strong>the</strong>se measurements are also identical for <strong>the</strong> two sets from one test type.
B<strong>end</strong>ing moment (kNm)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
FS1b: DT9 FS1b: DT10<br />
FS4a: DT9 FS4a: DT10<br />
0<br />
0 3 6 9 12 15 18 21 24 27 30 33<br />
End <strong>plate</strong> (horiz.) deformation (δ DT9-10) (mm)<br />
(a) Comparison <strong>of</strong> <strong>the</strong> responses for <strong>the</strong> two devices (DT9, DT10) for tests<br />
FS1b and FS4a.<br />
B<strong>end</strong>ing moment (kNm)<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
FS1a FS1b<br />
0 3 6 9 12 15 18 21 24 27 30 33<br />
Horizontal displac. (tension side) (mm)<br />
(b) Comparison <strong>of</strong> <strong>the</strong> responses for <strong>the</strong> two tests from series FS1 (deformations<br />
from DT9).<br />
Fig. 7.15 End <strong>plate</strong> deformation in <strong>the</strong> tension zone for several specimens.<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
FS3b<br />
FS4b<br />
FS2a<br />
FS1b<br />
0<br />
0 3 6 9 12 15 18 21 24 27 30 33<br />
End <strong>plate</strong> (horiz.) deformation (δ DT9) (mm)<br />
Fig. 7.16 Comparison <strong>of</strong> <strong>the</strong> moment-<strong>end</strong> <strong>plate</strong> deformation curves for <strong>the</strong><br />
four test series.<br />
295
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Fig. 7.16 compares <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformation behaviour for <strong>the</strong> four connection<br />
details. The deformability <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> increases for smaller values<br />
<strong>of</strong> tp and lower steel grades. This behaviour is identical to <strong>the</strong> connection rotation,<br />
as expected, since <strong>the</strong> components <strong>end</strong> <strong>plate</strong> and bolts are <strong>the</strong> main<br />
sources <strong>of</strong> connection deformability. Fig. 7.17 illustrates <strong>the</strong> evolution <strong>of</strong> <strong>the</strong><br />
<strong>end</strong> <strong>plate</strong> deformation response with <strong>the</strong> applied load for <strong>the</strong> specific case <strong>of</strong><br />
FS4b and Figs. 7.17d and 7.18 compare <strong>the</strong> collapse conditions for <strong>the</strong> four test<br />
types.<br />
A comparative analysis <strong>of</strong> <strong>the</strong> influence <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformability over<br />
<strong>the</strong> connection rotational behaviour is plotted in <strong>the</strong> graph <strong>of</strong> Fig. 7.19. For series<br />
FS2, FS3 and FS4 where <strong>the</strong> bolts mainly determine failure, ei<strong>the</strong>r by fracture<br />
or by stripping, <strong>the</strong> shape <strong>of</strong> <strong>the</strong> curves is identical. In series FS1 where<br />
<strong>end</strong> <strong>plate</strong> cracking and weld fracture are engaged in <strong>the</strong> collapse mode, <strong>the</strong><br />
shape <strong>of</strong> <strong>the</strong> curve is slightly different. Even so, <strong>the</strong>se curves clearly demonstrate<br />
that <strong>the</strong> ratio between <strong>end</strong> <strong>plate</strong> deformation behaviour is higher for<br />
lower <strong>end</strong> <strong>plate</strong> thickness values and lower steel grades.<br />
(i) General view. (ii) Tension zone.<br />
(a) Load = 80 kN (<strong>the</strong>oretical elastic limit; elastic branch <strong>of</strong> <strong>the</strong> M-φ curve).<br />
(i) General view. (ii) Tension zone.<br />
(b) Load = 120 kN (<strong>the</strong>oretical plastic resistance; knee-range branch <strong>of</strong> <strong>the</strong> M-φ<br />
curve).<br />
Fig. 7.17 Evolution <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformations until failure conditions for<br />
test series FS4b.<br />
296
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
(i) General view. (ii) Tension zone.<br />
(c) Load = 162 kN (post-limit branch <strong>of</strong> <strong>the</strong> M-φ curve).<br />
(i) General view. (ii) Tension zone.<br />
(d) Load = 188 kN (maximum load attained during <strong>the</strong> test).<br />
(i) General view. (ii) Tension zone.<br />
(e) Collapse conditions.<br />
Fig. 7.17 Evolution <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformations until failure conditions for<br />
test series FS4b (cont.).<br />
Finally, Fig. 7.20 shows an alternative procedure for computation <strong>of</strong> <strong>the</strong><br />
connection deformation from <strong>the</strong> readings <strong>of</strong> <strong>the</strong> horizontal LVDTs, in <strong>the</strong><br />
compression and tension zone <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> (e.g. specimen FS1a). As expected,<br />
<strong>the</strong> agreement between both procedures is excellent.<br />
297
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
(a) Specimen FS1a. (b) Specimen FS2a. (c) Specimen FS3b.<br />
Fig. 7.18 Comparison <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformations at failure conditions for<br />
test series FS1-3.<br />
298<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
FS3b<br />
FS2a<br />
FS4b<br />
FS1b<br />
0<br />
0.00 0.03 0.06 0.09 0.12 0.15 0.18 0.21 0.24 0.27 0.30<br />
Ratio δ DT9/φ (mm/mrad)<br />
Fig. 7.19 Comparison <strong>of</strong> <strong>the</strong> ratio <strong>end</strong> <strong>plate</strong> deformation vs. connection rotation<br />
for <strong>the</strong> four test series.<br />
B<strong>end</strong>ing moment (kNm)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
Connection rotation as defined above<br />
Connection rotation computed from DT9<br />
20<br />
0<br />
Connection rotation computed from DT10<br />
0 10 20 30 40 50 60 70 80 90 100<br />
Connection rotation (mrad)<br />
Fig. 7.20 Comparison <strong>of</strong> <strong>the</strong> moment-rotation curve for test FS1a by using alternative<br />
definitions <strong>of</strong> connection rotation.
7.3.2.2 Yield line patterns<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Figs. 7.21 and 7.22 depict <strong>the</strong> yield line patterns <strong>of</strong> <strong>the</strong> inner tension bolt #3 for<br />
specimens FS1b and FS2b at collapse conditions. These patterns could be<br />
sketched because <strong>the</strong> specimens were painted with chalk. Clearly, for series<br />
FS1 <strong>the</strong> yielding <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> in this area spreads to <strong>the</strong> compression bolt,<br />
whilst for FS2, with a thicker <strong>plate</strong>, <strong>the</strong>re is a small amount <strong>of</strong> plasticity in <strong>the</strong><br />
<strong>end</strong> <strong>plate</strong>.<br />
(a) Load = 93 kN. (b) Load = 151 kN. (c) Collapse<br />
conditions.<br />
Fig. 7.21 Yield line patterns around <strong>the</strong> inner tension bolt for different load<br />
levels (e.g. specimen FS1b).<br />
(a) Load = 130 kN. (b) Load = 188 kN. (c) Near collapse conditions.<br />
Fig. 7.22 Yield line patterns around <strong>the</strong> inner tension bolt for different load<br />
levels (e.g. specimen FS2b).<br />
7.3.2.3 Bolt elongation behaviour<br />
The experimental results demonstrate that <strong>the</strong> two rows <strong>of</strong> tension bolts carry<br />
unequal forces (Fig. 7.23): <strong>the</strong> inner tension bolts carry a larger proportion <strong>of</strong><br />
<strong>the</strong> load than <strong>the</strong> outer bolts. This conclusion is also supported by <strong>the</strong> graphs<br />
299
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
shown in Fig. 7.24 that compare <strong>the</strong> ratio between <strong>the</strong> bolt elongation and <strong>the</strong><br />
gap <strong>end</strong> <strong>plate</strong>-column flange. This ratio increases for <strong>the</strong> inner tension bolts.<br />
The graphs also highlight <strong>the</strong> influence <strong>of</strong> <strong>the</strong> bolt tension deformation on <strong>the</strong><br />
overall behaviour with <strong>the</strong> increase <strong>of</strong> tp and steel grade. This conclusion is in<br />
line with <strong>the</strong> above observations.<br />
300<br />
Total applied load (kN)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
MB1 MB2<br />
30<br />
0<br />
MB3 MB4<br />
0.00 0.15 0.30 0.45 0.60 0.75 0.90 1.05 1.20 1.35 1.50<br />
(Tension) Bolt elongation (mm)<br />
Fig. 7.23 Bolt elongation behaviour (e.g. specimen FS4b).<br />
End pl. (hor.) def. (δ DT9) (mm)<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
FS1b<br />
FS4b<br />
FS2b<br />
FS3a<br />
0<br />
0.00 0.03 0.06 0.09 0.12 0.15 0.18 0.21 0.24 0.27 0.30<br />
δ b/δ DT9 (mm/mm)<br />
(a) Bolt #1.<br />
Fig. 7.24 Comparison <strong>of</strong> <strong>the</strong> “nondimensional” bolt elongation behaviour for<br />
<strong>the</strong> four specimen types.<br />
7.3.2.4 Strain behaviour<br />
This section illustrates some <strong>of</strong> <strong>the</strong> experimental strain results. Unfortunately,<br />
<strong>the</strong> travel range <strong>of</strong> <strong>the</strong> gauges used for recording <strong>the</strong> strains was <strong>of</strong>ten exceeded<br />
before <strong>the</strong> connection failure occurred and in many specimens, <strong>the</strong> gauges were<br />
damaged in early stages <strong>of</strong> loading. In some cases, <strong>the</strong> strain gauges were not
(b) Bolt #2.<br />
(c) Bolt #3.<br />
End pl. (hor.) def. (δ DT9) (mm)<br />
End pl. (hor.) def. (δ DT9) (mm)<br />
End pl. (hor.) def. (δ DT9) (mm)<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
FS1b<br />
FS4b<br />
FS2b<br />
FS3a<br />
0<br />
0.00 0.03 0.06 0.09 0.12 0.15 0.18 0.21 0.24 0.27 0.30<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
FS1b<br />
δ b/δ DT9 (mm/mm)<br />
FS4b<br />
FS2b<br />
0<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50<br />
14<br />
12<br />
10<br />
8<br />
6<br />
4<br />
2<br />
FS1b<br />
δ b/δ DT9 (mm/mm)<br />
FS4b<br />
FS2b<br />
FS3a<br />
0<br />
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50<br />
δ b/δ DT9 (mm/mm)<br />
(d) Bolt #4.<br />
Fig. 7.24 Comparison <strong>of</strong> <strong>the</strong> “nondimensional” bolt elongation behaviour for<br />
<strong>the</strong> four specimen types (cont.).<br />
FS3a<br />
301
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
attached correctly and consequently <strong>the</strong>ir results are not trustworthy.<br />
Anyhow, some results can be retained for future comparisons. Fig. 7.25<br />
shows <strong>the</strong> results obtained in different gauges (for <strong>the</strong>ir location, please refer to<br />
Fig. 7.6). These results also allow an assessment <strong>of</strong> <strong>the</strong> yield line patterns<br />
(hogging, SG11 and sagging yield lines, SG5, 7 and 9).<br />
302<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
FS2a<br />
FS3b<br />
FS4b<br />
FS1b<br />
0<br />
0 3000 6000 9000 12000 15000 18000 21000 24000<br />
Strain (SG5) (µm/m)<br />
(a) Strains at SG5, located at <strong>the</strong> <strong>end</strong> <strong>plate</strong> extension.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
0<br />
FS3b<br />
FS1b<br />
FS4b<br />
FS2a<br />
Limit <strong>of</strong> <strong>the</strong> strain gauges<br />
Limit <strong>of</strong> <strong>the</strong> strain gauges<br />
0 3000 6000 9000 12000 15000 18000 21000 24000<br />
Strain (SG7) (µm/m)<br />
(b) Strains at SG7, located at <strong>the</strong> inner <strong>end</strong> <strong>plate</strong> side, near <strong>the</strong> beam tension<br />
flange.<br />
Fig. 7.25 Comparison <strong>of</strong> some strain results obtained for <strong>the</strong> four specimen<br />
types at different strain gauges.<br />
7.4 DISCUSSION OF TEST RESULTS<br />
Eurocode 3 gives quantitative rules for <strong>the</strong> prediction <strong>of</strong> <strong>the</strong> joint flexural plastic<br />
resistance and initial rotational stiffness. These structural properties are<br />
evaluated below by using <strong>the</strong> actual geometrical characteristics from Table 7.2
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
FS3b<br />
FS4b<br />
FS2a<br />
0<br />
0 3000 6000 9000 12000 15000 18000 21000 24000<br />
Strain (SG9) (µm/m)<br />
FS1b<br />
(c) Strains at SG9, located at <strong>the</strong> inner <strong>end</strong> <strong>plate</strong> side, near <strong>the</strong> beam web.<br />
Total applied load (kN)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
30<br />
Limit <strong>of</strong> <strong>the</strong> strain gauges<br />
Limit <strong>of</strong> <strong>the</strong> strain gauges<br />
0<br />
-24000 -21000 -18000 -15000 -12000 -9000 -6000 -3000 0<br />
Strain (SG11) (µm/m)<br />
(d) Strains at SG11, located at <strong>the</strong> bolt axis (<strong>end</strong> <strong>plate</strong> extension).<br />
Fig. 7.25 Comparison <strong>of</strong> some strain results obtained for <strong>the</strong> four specimen<br />
types at different strain gauges (cont.).<br />
and <strong>the</strong> mechanical properties from Tables 7.4 and 7.6. The recomm<strong>end</strong>ations<br />
on rotation capacity are also verified to investigate if <strong>the</strong>re is enough rotation<br />
capacity according to <strong>the</strong> code. The provisions are compared with <strong>the</strong> test results.<br />
7.4.1 Plastic flexural resistance<br />
According to Eurocode 3, <strong>the</strong> joint plastic flexural resistance is evaluated by<br />
means <strong>of</strong> Eq. (1.60). As <strong>the</strong> overall connection behaviour was dominated by<br />
<strong>the</strong> <strong>end</strong> <strong>plate</strong> and bolts, <strong>the</strong> computation <strong>of</strong> Fti.Rd relies on <strong>the</strong> T-stub idealization<br />
<strong>of</strong> <strong>the</strong> tension zone that can fail according to <strong>the</strong> three possible plastic collapse<br />
mechanisms.<br />
FS4b<br />
FS1b<br />
FS3b<br />
FS2a<br />
303
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 7.9 Evaluation <strong>of</strong> <strong>the</strong> resistance <strong>of</strong> <strong>the</strong> test specimens (<strong>the</strong> experimental<br />
values correspond to <strong>the</strong> average <strong>of</strong> <strong>the</strong> two tests per connection<br />
detail; Ratio = [Theory/Experiments]).<br />
Test<br />
ID<br />
304<br />
Row 1 Row 2<br />
h1 Ft1.Rd Plastic h2 Ft2.Rd Plastic<br />
Mj.Rd<br />
(mm) (kN) mode (mm) (kN) mode (kNm)<br />
Ratio<br />
FS1 334.52 83.86 Type-1 244.49 202.34 Type-1 77.52 0.72<br />
FS2 335.26 176.07 Type-1 245.28 297.87 Type-2 132.09 0.79<br />
FS3 335.34 274.06 Type-2 245.20 389.01 Type-2 187.29 1.03<br />
FS4 334.76 161.16 Type-1 244.81 287.94 Type-2 124.44 0.76<br />
Application <strong>of</strong> <strong>the</strong> procedure detailed in §1.6.1.2 provides <strong>the</strong> results presented<br />
in Table 7.9. It is worth mentioning that <strong>the</strong> predicted yield line patterns<br />
(double curvature for <strong>the</strong> bolt row located at <strong>the</strong> <strong>end</strong> <strong>plate</strong> extension and side<br />
yielding near <strong>the</strong> beam flange) are in line with <strong>the</strong> experimental observations<br />
(cf. Figs. 7.22-7.23 for <strong>the</strong> inner bolt row, for instance). By comparing <strong>the</strong> code<br />
predictions with <strong>the</strong> experiments, <strong>the</strong>y are within <strong>the</strong> knee-range bounds but<br />
below <strong>the</strong> experimental values <strong>of</strong> flexural resistance.<br />
7.4.2 Initial rotational stiffness<br />
The initial rotational stiffness was evaluated according to <strong>the</strong> Eurocode 3 procedure,<br />
as explained in §1.6.1.1. For simplicity, z was taken as equal to <strong>the</strong> distance<br />
from <strong>the</strong> centre <strong>of</strong> compression to a mid point between <strong>the</strong> two bolt rows<br />
in tension [7.1].<br />
Table 7.10 sets out <strong>the</strong> predicted values for <strong>the</strong> initial stiffness and compares<br />
<strong>the</strong>m with <strong>the</strong> experiments. The ratio between <strong>the</strong> predicted values and<br />
<strong>the</strong> experiments shows that Eurocode overestimates this property. The differences<br />
may derive from <strong>the</strong> fact that <strong>the</strong> expression as presented in <strong>the</strong> code was<br />
calibrated for a certain range <strong>of</strong> joints. The particular joints that were tested<br />
were not ‘balanced’, i.e. <strong>the</strong>re was a much weaker component than <strong>the</strong> remaining<br />
ones. This situation is unlikely to occur in common joints for which <strong>the</strong> expression<br />
was calibrated.<br />
7.4.3 Rotation capacity<br />
The experimental values <strong>of</strong> rotation capacity and corresponding moment values<br />
for <strong>the</strong> various tests are set out in Table 7.8. It can be easily seen that test FS1,<br />
which employs a thinner <strong>end</strong> <strong>plate</strong> and steel grade S355, presents higher <strong>ductility</strong><br />
than <strong>the</strong> remaining tests.<br />
Application <strong>of</strong> <strong>the</strong> Eurocode 3 guidelines to <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> rota-
Experimental tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 7.10 Evaluation <strong>of</strong> <strong>the</strong> initial rotational stiffness <strong>of</strong> <strong>the</strong> test specimens<br />
(<strong>the</strong> experimental values correspond to <strong>the</strong> average <strong>of</strong> <strong>the</strong> two tests<br />
per connection detail; Ratio = [Theory/Experiments]).<br />
Test ID keff.1 keff.2 keq<br />
(kN/mm)<br />
kcws kcwc<br />
FS1 225.023 333.57 541.95 2718.64 4867.62<br />
FS2 375.03 453.37 816.22 2717.60 4983.72<br />
FS3 2453.12 496.04 942.50 2711.50 5052.06<br />
FS4 202.18 315.65 500.25 2710.23 4857.34<br />
Test ID z<br />
(mm)<br />
Sj.ini<br />
(kNm/mrad)<br />
Ratio<br />
FS1 289.51 34.66 1.98<br />
FS2 289.62 46.76 2.06<br />
FS3 290.27 51.76 2.31<br />
FS4 290.41 32.77 1.97<br />
Table 7.11 Verification <strong>of</strong> <strong>the</strong> recomm<strong>end</strong>ations for rotation capacity.<br />
Test ID tp<br />
(mm)<br />
Maximum tp<br />
(mm)<br />
FS1 10.40 11.80 Yes.<br />
FS2 15.01 11.75 No.<br />
FS3 20.02 11.76 No.<br />
FS4 10.06 8.25 No.<br />
tion capacity [7.1] – cf. §1.6.2 – shows that <strong>the</strong> first condition is guaranteed for<br />
all specimens (<strong>the</strong> joint moment resistance is governed by <strong>the</strong> resistance <strong>of</strong> <strong>the</strong><br />
<strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing), whilst <strong>the</strong> second condition (Eq. (1.68)) is only fulfilled<br />
for specimens FS1 (Table 7.11). Though <strong>the</strong>se recomm<strong>end</strong>ations are only valid<br />
for steel grades up to S460, <strong>the</strong>y were also applied to series FS4 that includes<br />
<strong>end</strong> <strong>plate</strong>s from grade S690.<br />
7.5 CONCLUDING REMARKS<br />
Tests on eight ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> moment connections were conducted under<br />
static loading. All specimens were designed to trigger failure in <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
ra<strong>the</strong>r than in <strong>the</strong> beam or <strong>the</strong> column. The following conclusions can be drawn<br />
from <strong>the</strong> test programme:<br />
1. The joint moment resistance increases with <strong>the</strong> increase <strong>of</strong> <strong>end</strong> <strong>plate</strong> thickness<br />
and with <strong>the</strong> yield stress <strong>of</strong> <strong>the</strong> <strong>plate</strong>;<br />
305
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
2. The joint initial rotational stiffness also increases with <strong>the</strong> <strong>end</strong> <strong>plate</strong> thickness,<br />
but <strong>the</strong> sensitivity to <strong>the</strong> thickness variation is not as noticeable as for resistance.<br />
The steel grade has little influence if any on this property;<br />
3. The joint post-limit rotational stiffness is identical for all specimens, i.e. <strong>the</strong><br />
variation with <strong>end</strong> <strong>plate</strong> thickness or steel grade is not significant;<br />
4. The Eurocode 3 proposals give safe approaches for <strong>the</strong> prediction <strong>of</strong> <strong>the</strong> joint<br />
resistance but overestimate <strong>the</strong> joint initial stiffness in this particular case;<br />
5. The available rotation capacity and hence <strong>the</strong> joint <strong>ductility</strong> decreases with<br />
<strong>the</strong> <strong>plate</strong> thickness (series FS1, FS2 and FS3) and with <strong>the</strong> <strong>plate</strong> steel grade<br />
(FS1 and FS4);<br />
6. In terms <strong>of</strong> <strong>the</strong> verification <strong>of</strong> sufficient rotation capacity, Eurocode 3 gives<br />
safe criteria but perhaps too conservative [7.7]. For instance, in terms <strong>of</strong> overall<br />
rotation capacity, specimens from series FS2 and FS4 exhibit rotation values <strong>of</strong><br />
40 mrad.<br />
7.6 REFERENCES<br />
[7.1] European Committee for Standardization (CEN). PrEN 1993-1-8:2003,<br />
Eurocode 3: Design <strong>of</strong> steel structures, Part 1.8: Design <strong>of</strong> joints, Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[7.2] International Standard ISO 898-1:1999(E). Mechanical properties <strong>of</strong><br />
fasteners made <strong>of</strong> carbon steel and alloy steel – Part 1: Bolts, screws and<br />
studs, August 1999, Switzerland, 1999.<br />
[7.3] European Committee for Standardization (CEN). PrEN 10025:2000E:<br />
Hot rolled products <strong>of</strong> structural steels, September 2000, Brussels, 2000.<br />
[7.4] European Committee for Standardization (CEN). EN 10204:1995E: Metallic<br />
products, October 1995, Brussels, 1995.<br />
[7.5] RILEM draft recomm<strong>end</strong>ation. Tension testing <strong>of</strong> metallic structural<br />
materials for determining stress-strain relations under monotonic and<br />
uniaxial tensile loading. Materials and Structures; 23:35-46, 1990.<br />
[7.6] Weynand K. Sicherheits-und Wirtsschaftlichkeitsuntersuchungen zur<br />
anw<strong>end</strong>ung nachgiebiger anschlüsse im stahlbau. PhD <strong>the</strong>sis (in German).<br />
University <strong>of</strong> Aachen, Aachen, Germany, 1996.<br />
[7.7] Girão Coelho AM, Bijlaard F, Simões da Silva L. Experimental assessment<br />
<strong>of</strong> <strong>the</strong> <strong>ductility</strong> <strong>of</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections. Engineering<br />
Structures (in print), 2004.<br />
306
8 DUCTILITY OF BOLTED END PLATE CONNECTIONS<br />
8.1 INTRODUCTION<br />
The methodology developed in this chapter provides a <strong>characterization</strong> <strong>of</strong> <strong>the</strong><br />
full nonlinear M-Φ behaviour <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections. The assessment<br />
<strong>of</strong> <strong>the</strong> available joint rotation is addressed in particular. Kuhlmann and Kühnemund<br />
[8.1] assume that <strong>the</strong> available joint rotation should be taken as <strong>the</strong> total<br />
joint rotation or rotation capacity, ΦCd. In <strong>the</strong> context <strong>of</strong> <strong>the</strong> component<br />
method, for a direct computation <strong>of</strong> <strong>the</strong> joint rotation capacity <strong>the</strong> following<br />
steps have to be fulfilled [8.2-8.3]: (i) <strong>the</strong> F-∆ curve <strong>of</strong> each joint component<br />
up to failure is modelled, (ii) <strong>the</strong> weakest joint component, i.e. <strong>the</strong> component<br />
with lower resistance, is identified, (iii) <strong>the</strong> plastic engagement <strong>of</strong> <strong>the</strong> remaining<br />
components is determined, (iv) <strong>the</strong> global displacements <strong>of</strong> <strong>the</strong> individual<br />
components at <strong>the</strong> level <strong>of</strong> maximum resistance are evaluated to finally (v) determine<br />
ΦCd.<br />
Literature suggests that most <strong>of</strong> <strong>the</strong> joint rotation in thin <strong>end</strong> <strong>plate</strong>s comes<br />
from <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformation [8.4-8.5]. The tests described in Chapter 7 confirm<br />
this statement. For really thin <strong>end</strong> <strong>plate</strong>s, <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformation would<br />
be sufficient to characterize <strong>the</strong> M-Φ curve since it becomes <strong>the</strong> weakest joint<br />
component. In general, ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections are characterized by <strong>the</strong><br />
participation <strong>of</strong> two or more components to <strong>the</strong> joint plastic deformation, as<br />
highlighted by Faella et al. [8.2]. In <strong>the</strong> framework <strong>of</strong> <strong>the</strong> component method,<br />
in this joint configuration <strong>the</strong> following sources <strong>of</strong> deformability for <strong>characterization</strong><br />
<strong>of</strong> <strong>the</strong> rotation capacity are identified (Fig. 8.1): column web in shear<br />
(cws), column web in compression (cwc), column web in tension (cwt), column<br />
flange in b<strong>end</strong>ing (cfb), <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing (epb) and bolts in tension<br />
(bt). Components beam web and flange in compression (bfc) and beam web in<br />
tension (bwt) are not taken into account in <strong>the</strong> model since <strong>the</strong>y basically provide<br />
a resistance limitation [8.6]. Components column flange in b<strong>end</strong>ing, <strong>end</strong><br />
<strong>plate</strong> in b<strong>end</strong>ing and bolts in tension are modelled as equivalent T-stubs, as already<br />
explained. The full M-Φ response is characterized from <strong>the</strong> F-∆ curve <strong>of</strong><br />
<strong>the</strong> joint components, which are assembled into an appropriate mechanical<br />
model. Chapter 1 (§1.6) discusses alternative component models that are illustrated<br />
in Fig. 8.2. If most <strong>of</strong> <strong>the</strong> joint rotation comes in fact from <strong>the</strong> subassembly<br />
<strong>end</strong> <strong>plate</strong>-bolts, <strong>the</strong> several models yield identical solutions since <strong>the</strong><br />
only deformable components are <strong>the</strong> <strong>end</strong> <strong>plate</strong> in b<strong>end</strong>ing and <strong>the</strong> bolts in tension<br />
(Fig. 8.3). Consequently, <strong>the</strong> component models illustrated above are<br />
equivalent, as shown in Fig. 8.3. This is <strong>the</strong> case <strong>of</strong> <strong>the</strong> tested joints that were<br />
reported in Chapter 7.<br />
307
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
308<br />
cws<br />
cwt<br />
cwc<br />
T-stub idealization<br />
Fig. 8.1 Basic joint components <strong>of</strong> an ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connection with<br />
two bolt rows in tension.<br />
z<br />
(cws)<br />
(cwc)<br />
(cwt.1)<br />
(cwt.2)<br />
T-stubs (row 1)<br />
(cfb.1)<br />
(cfb.2)<br />
(epb.1)<br />
(epb.2)<br />
T-stubs<br />
(row 2)<br />
(bt.1)<br />
(bt.2)<br />
Φ<br />
Φ<br />
M<br />
bwt<br />
bfc<br />
M<br />
(cwt.1)<br />
(cwt.2)<br />
z<br />
T-stubs (row 1)<br />
(cfb.1)<br />
(cfb.2)<br />
(epb.1)<br />
(epb.2)<br />
T-stubs<br />
(row 2)<br />
(cws) (cwc)<br />
(a) Mechanical model adopted in Eurocode 3. (b) UC component model.<br />
(cwt.1)<br />
(cwt.2)<br />
(cws) (cwc)<br />
T-stubs (row 1)<br />
(cfb.1)<br />
(cfb.2)<br />
(epb.1)<br />
(epb.2)<br />
T-stubs<br />
(row 2)<br />
(c) Innsbruck mechanical model.<br />
Fig. 8.2 Alternative component models for analysis <strong>of</strong> <strong>the</strong> rotational behaviour<br />
<strong>of</strong> an ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connection (two bolt rows in tension).<br />
(bt.1)<br />
(bt.2)<br />
Φ<br />
Φ<br />
M<br />
(bt.1)<br />
(bt.2)<br />
Φ<br />
Φ<br />
M
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
The following sections address <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> rotational behaviour<br />
<strong>of</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections up to failure. The full M-Φ curve is derived<br />
by using a computational tool, NASCon [8.7]. This s<strong>of</strong>tware allows for a<br />
multilinear definition <strong>of</strong> <strong>the</strong> deformation behaviour <strong>of</strong> components and uses <strong>the</strong><br />
spring model illustrated in Fig. 8.2b. Since <strong>ductility</strong> is such an important property<br />
in a partial strength scenario, particular attention is given to this issue.<br />
The available experimental tests (Chapter 7) were basically aimed at <strong>the</strong> investigation<br />
<strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> behaviour. Therefore, <strong>the</strong> proposed methodology is<br />
illustrated and validated only for this connection type. However, from a <strong>the</strong>oretical<br />
point <strong>of</strong> view, <strong>the</strong> procedure can be applied to any beam-to-column joint<br />
configuration, as long as <strong>the</strong> F-∆ response <strong>of</strong> each component can be predicted<br />
with sufficient accuracy. This research work focuses on those components<br />
(cws)<br />
(cwt.1)<br />
(cwt.2)<br />
(cfb.1)<br />
(cfb.2)<br />
(cwc)<br />
(epb.1)<br />
(epb.2)<br />
(cws) (cwc)<br />
(cwt.1)<br />
(cwt.2)<br />
(cws) (cwc)<br />
(bt.1)<br />
(bt.2)<br />
(cwt.1)<br />
(cwt.2)<br />
Φ<br />
Φ<br />
(cfb.1)<br />
(cfb.2)<br />
M<br />
(epb.1)<br />
(epb.2)<br />
(cfb.1)<br />
(cfb.2)<br />
(bt.1)<br />
(bt.2)<br />
(epb.1)<br />
(epb.2)<br />
(bt.1)<br />
(bt.2)<br />
Φ<br />
Φ<br />
M<br />
Φ<br />
Φ<br />
M<br />
T-stub (row 1)<br />
(epb.1) (bt.1)<br />
(epb.2) (bt.2)<br />
T-stub (row 2)<br />
Fig. 8.3 Equivalence <strong>of</strong> component models for analysis <strong>of</strong> <strong>the</strong> rotational behaviour<br />
<strong>of</strong> tested joints.<br />
Φ<br />
Φ<br />
M<br />
309
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
modelled by equivalent T-stubs. Chapter 3 describes some experiments performed<br />
on WP-T-stubs. A three-dimensional FE model has been proposed in<br />
Chapter 4 for <strong>the</strong> assessment <strong>of</strong> <strong>the</strong> F-∆ behaviour <strong>of</strong> this component. Chapter<br />
6 proposes a simplified beam model for <strong>the</strong> assessment <strong>of</strong> <strong>the</strong> overall deformation<br />
behaviour <strong>of</strong> individual T-stubs and also describes alternative simplified<br />
methods recomm<strong>end</strong>ed by o<strong>the</strong>r authors. These methodologies are used below<br />
for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> behaviour.<br />
8.2 MODELLING OF BOLT ROW BEHAVIOUR THROUGH EQUIVALENT T-<br />
STUBS<br />
The T-stub idealization <strong>of</strong> <strong>the</strong> tension zone <strong>of</strong> a connection consists is substituting<br />
this zone for T-stub sections <strong>of</strong> appropriate effective length (Fig. 8.4).<br />
These T-stub sections are connected by <strong>the</strong>ir flange to a rigid foundation (halfmodel)<br />
and subjected to a uniformly distributed force acting in <strong>the</strong> web <strong>plate</strong><br />
[8.6]. The extension <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> and <strong>the</strong> portion between <strong>the</strong> beam flanges<br />
are modelled as two separate equivalent T-stubs (Fig. 8.4). On <strong>the</strong> column side,<br />
two situations have to be analysed: (i) <strong>the</strong> bolt rows act individually or (ii) <strong>the</strong><br />
bolt rows act in combination (Fig. 8.4).<br />
To define <strong>the</strong> effective length, <strong>the</strong> complex pattern <strong>of</strong> yield lines that occurs<br />
around <strong>the</strong> bolts is converted into a simple equivalent T-stub. This effective<br />
length does not represent any actual length <strong>of</strong> <strong>the</strong> connection. The typical observed<br />
yield-line pattern in thin <strong>end</strong> <strong>plate</strong>s is shown in Figs. 8.5 and 8.6, for<br />
two different cases: (i) <strong>end</strong> <strong>plate</strong> with one bolt row below <strong>the</strong> tension beam<br />
flange and (ii) <strong>end</strong> <strong>plate</strong> with two bolt rows below <strong>the</strong> flush line, respectively.<br />
For thicker <strong>end</strong> <strong>plate</strong>s, <strong>the</strong> patterns may not develop fully as <strong>the</strong> bolt elongation<br />
behaviour may govern <strong>the</strong> overall behaviour. For <strong>end</strong> <strong>plate</strong>s with more than<br />
one bolt row below <strong>the</strong> flush line, <strong>the</strong> cases <strong>of</strong> individual and combined bolt<br />
row behaviour have to be taken into consideration, as illustrated in Fig. 8.6.<br />
8.3 APPLICATION TO BOLTED EXTENDED END PLATE CONNECTIONS<br />
The above procedures are applied to <strong>the</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections from<br />
Chapter 7 that were tested monotonically up to failure. In <strong>the</strong>se examples, <strong>the</strong><br />
tension zone on <strong>the</strong> <strong>end</strong> <strong>plate</strong> side that is idealized as a T-stub was always critical.<br />
The remaining joint components behaved elastically until collapse.<br />
8.3.1 Component <strong>characterization</strong><br />
The four joint configurations FS1-FS4 comprise two bolt rows in tension. For<br />
each test detail, on <strong>the</strong> <strong>end</strong> <strong>plate</strong> side, two equivalent T-stubs are identified<br />
(Figs. 8.4-8.5). For fur<strong>the</strong>r reference, <strong>the</strong>se two T-stubs are designated by “T-<br />
310
eff.ep.r1<br />
Bolt<br />
row 1<br />
End <strong>plate</strong> side<br />
beff.cf.r(1+2)<br />
Bolt row 2<br />
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Bolt rows 1<br />
and 2<br />
beff.ep.r2<br />
Bolt row<br />
1, individually<br />
Bolt row<br />
2, individually<br />
Column side<br />
Fig. 8.4 T-stub idealization <strong>of</strong> an ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> <strong>bolted</strong> connection with<br />
two bolt rows in tension.<br />
stub top” and “T-stub bottom”, for bolt rows 1 and 2, respectively (cf. Figs.<br />
8.4-8.5). The <strong>characterization</strong> <strong>of</strong> <strong>the</strong>se components in terms <strong>of</strong> F-∆ behaviour is<br />
performed by means <strong>of</strong> four alternative procedures (Table 8.1): (i) experimentally,<br />
(ii) numerically (three-dimensional FE model), (iii) analytically (simple<br />
beam model) and (iv) simplified bilinear approximation proposed by Jaspart<br />
[8.8]. The experimental results are not available for all equivalent T-stubs and<br />
<strong>the</strong> numerical model is not implemented for all T-stubs, as shown in Table 8.1.<br />
For <strong>the</strong> equivalent T-stubs top from joints FS1 and FS4, <strong>the</strong> tests on WT7_M20<br />
and WT57_M20, respectively, provide an experimental F-∆ curve that can be<br />
beff.fc.r1<br />
beff.fc.r2<br />
311
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
312<br />
Bolt row 1 beff .1 r = bep<br />
2<br />
Bolt row 2 b . 2=<br />
αm<br />
eff r ep<br />
Hogging yield-line<br />
Sagging yield-line<br />
(a) Plot. (b) Illustration: spec. FS1b.<br />
Fig. 8.5 Typical yield-line pattern in thin ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong>s with two bolt<br />
rows in tension.<br />
p2-3<br />
Bolt row 1<br />
Bolt row 2<br />
Hogging yield-line<br />
Sagging yield-line<br />
p2-3<br />
beff.r1<br />
acting in<br />
combination<br />
Influence <strong>of</strong><br />
bolt row 2<br />
(0.5beff.r2 + 0.5p)<br />
beff.r(2+3)<br />
Influence <strong>of</strong><br />
bolt row 3<br />
(0.5beff.r3 + 0.5p)<br />
Fig. 8.6 T-stub idealization <strong>of</strong> an ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> with three bolt rows in tension<br />
with <strong>the</strong> bolts below <strong>the</strong> tension beam flange acting in combination.
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 8.1 Alternative procedures for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> T-stub response.<br />
Test ID Equivalent<br />
Characterization procedures<br />
T-stub Experimental Numerical Beam Jaspart<br />
model approximation<br />
Top �<br />
� � �<br />
FS1<br />
(WT7_M20)<br />
Bot. ⎯ ⎯ � �<br />
FS2<br />
Top<br />
Bot.<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
�<br />
�<br />
�<br />
�<br />
FS3<br />
Top<br />
Bot.<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
�<br />
�<br />
�<br />
�<br />
Top �<br />
⎯ � �<br />
FS4<br />
(WT57_M20)<br />
Bot. ⎯ ⎯ � �<br />
used for component <strong>characterization</strong>. The individual T-stub specimens do not<br />
correspond exactly to <strong>the</strong> equivalent T-stubs top as <strong>the</strong> bolt properties are different.<br />
However, no major differences are expected. For application <strong>of</strong> <strong>the</strong> recomm<strong>end</strong>ations<br />
<strong>of</strong> Jaspart [8.8], <strong>the</strong> bolt deformability is associated to that <strong>of</strong><br />
<strong>the</strong> <strong>end</strong> <strong>plate</strong>.<br />
The effective length <strong>of</strong> <strong>the</strong> different components is defined according to<br />
Eurocode 3 [8.6] and is summarized in Table 8.2. The actual geometric properties<br />
<strong>of</strong> <strong>the</strong> joints are used (Table 7.2). Figs. 8.7-8.10 illustrate <strong>the</strong> T-stub responses<br />
for <strong>the</strong> various configurations and with <strong>the</strong> alternative methodologies.<br />
Table 8.3 sets out <strong>the</strong> predictions <strong>of</strong> ultimate resistance and deformation capacity<br />
<strong>of</strong> <strong>the</strong> above equivalent T-stubs, as ascertained by <strong>the</strong> different procedures.<br />
The experimental results correspond in fact to experimental failure (see Figs.<br />
8.7a and 8.10a). Concerning <strong>the</strong> numerical predictions for <strong>the</strong> equivalent Tstub<br />
top for joint FS1, <strong>the</strong> values that are indicated in <strong>the</strong> table do not account<br />
for any reduction <strong>of</strong> <strong>the</strong> failure <strong>ductility</strong> <strong>of</strong> <strong>the</strong> HAZ (see also §5.4).<br />
The graphs in Figs. 8.7-8.10 also plot <strong>the</strong> experimental <strong>end</strong> <strong>plate</strong> deformation<br />
behaviour, which is obtained directly from <strong>the</strong> measurement <strong>of</strong> <strong>the</strong> displacement<br />
<strong>of</strong> <strong>the</strong> tension beam flange with <strong>the</strong> course <strong>of</strong> loading. The corresponding<br />
force level is evaluated indirectly, Ft = M/z, whereby z is <strong>the</strong> lever<br />
arm determined from Eq. (1.59). In <strong>the</strong>se graphs, this force Ft acting at <strong>the</strong><br />
level <strong>of</strong> <strong>the</strong> tension beam flange was divided equally by <strong>the</strong> two bolt rows. This<br />
procedure gives a reasonable agreement with <strong>the</strong> predictions for <strong>the</strong> T-stub top<br />
but deviates from <strong>the</strong> predicted behaviour for <strong>the</strong> T-stub bottom in <strong>the</strong> same<br />
case (e.g. specimen FS2, Fig. 8.8). In fact, <strong>the</strong> division <strong>of</strong> <strong>the</strong> tensile force by<br />
<strong>the</strong> two bolt rows modelled as two equivalent T-stubs seems more appropriate<br />
for <strong>the</strong> top T-stub, ra<strong>the</strong>r than <strong>the</strong> bottom T-stub. The equivalent T-stub top<br />
shares <strong>the</strong> tensile beam flange whereas <strong>the</strong> T-stub bottom shares <strong>the</strong> beam web.<br />
313
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Therefore, <strong>the</strong> force acting at <strong>the</strong> web <strong>of</strong> <strong>the</strong> T-stub top is directly related to <strong>the</strong><br />
tensile force Ft. This is not true for <strong>the</strong> bottom T-stub. Fur<strong>the</strong>rmore, <strong>the</strong> assumption<br />
<strong>of</strong> an equal division <strong>of</strong> Ft by <strong>the</strong> two bolt rows is questionable. Consequently,<br />
<strong>the</strong> graphs that were traced are merely illustrative and should be regarded<br />
as such.<br />
The experimental deformation <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> at <strong>the</strong> tensile beam flange<br />
level is obtained from <strong>the</strong> readings <strong>of</strong> <strong>the</strong> LVDTs. Table 8.3 indicates <strong>the</strong>se<br />
values at failure (in bold). They are directly related to <strong>the</strong> equivalent T-stub top<br />
Table 8.2 Effective length <strong>of</strong> <strong>the</strong> equivalent T-stubs.<br />
beff<br />
Test ID<br />
(mm) FS1 FS2 FS3 FS4<br />
T-stub top 74.92 74.71 75.24 74.88<br />
T-stub bot. 205.77 202.67 202.73 206.42<br />
(a) T-stub top.<br />
314<br />
Fep.r1 (kN)<br />
Fep.r2 (kN)<br />
270<br />
225<br />
180<br />
135<br />
90<br />
End <strong>plate</strong> deformation (exp.)<br />
Exp. results WT7_M20<br />
45<br />
0<br />
Numerical FE results<br />
Beam model<br />
Jaspart approximation<br />
0 2 4 6 8 10 12 14 16 18 20<br />
400<br />
350<br />
300<br />
250<br />
200<br />
∆ep.r1 (mm)<br />
150<br />
100<br />
.<br />
End <strong>plate</strong> deformation (exp.)<br />
Beam model<br />
50<br />
0<br />
Jaspart approximation<br />
0 2 4 6 8 10 12 14 16 18 20<br />
(b) T-stub bottom.<br />
Fig. 8.7 Equivalent T-stubs for joint FS1.<br />
∆ep.r2 (mm)
(a) T-stub top.<br />
Fep.r1 (kN)<br />
Fep.r2 (kN)<br />
350<br />
300<br />
250<br />
200<br />
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
150<br />
100<br />
End <strong>plate</strong> deformation (exp.)<br />
Beam model<br />
50<br />
0<br />
Jaspart approximation<br />
0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 13.5 15.0<br />
700<br />
600<br />
500<br />
400<br />
∆ep.r1 (mm)<br />
300<br />
200<br />
.<br />
End <strong>plate</strong> deformation (exp.)<br />
100<br />
0<br />
Beam model<br />
Jaspart approximation<br />
0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 13.5 15.0<br />
(b) T-stub bottom.<br />
Fig. 8.8 Equivalent T-stubs for joint FS2.<br />
Fep.r1 (kN)<br />
600<br />
500<br />
400<br />
300<br />
∆ep.r2 (mm)<br />
200<br />
End <strong>plate</strong> deformation (exp.)<br />
100<br />
0<br />
Beam model<br />
Jaspart approximation<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
(a) T-stub top.<br />
Fig. 8.9 Equivalent T-stubs for joint FS3.<br />
∆ep.r1 (mm)<br />
315
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
316<br />
Fep.r2 (kN)<br />
900<br />
800<br />
700<br />
600<br />
500<br />
400<br />
300<br />
.<br />
200<br />
End <strong>plate</strong> deformation (exp.)<br />
100<br />
0<br />
Beam model<br />
Jaspart approximation<br />
0 1 2 3 4 5 6 7 8 9 10 11 12<br />
∆ep.r2 (mm)<br />
(b) T-stub bottom.<br />
Fig. 8.9 Equivalent T-stubs for joint FS3 (cont.).<br />
(a) T-stub top.<br />
Fep.r1 (kN)<br />
Fep.r2 (kN)<br />
350<br />
300<br />
250<br />
200<br />
150<br />
End <strong>plate</strong> deformation (exp.)<br />
100<br />
Exp. results WT57_M20<br />
Beam model<br />
50<br />
0<br />
Jaspart approximation<br />
0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 13.5 15.0 16.5<br />
480<br />
400<br />
320<br />
240<br />
∆ep.r1 (mm)<br />
160<br />
.<br />
End <strong>plate</strong> deformation (exp.)<br />
80<br />
0<br />
Beam model<br />
Jaspart approximation<br />
0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 13.5 15.0 16.5<br />
∆ep.r2 (mm)<br />
(b) T-stub bottom.<br />
Fig. 8.10 Equivalent T-stubs for joint FS4.
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 8.3 Assessment <strong>of</strong> <strong>the</strong> ultimate conditions <strong>of</strong> <strong>the</strong> equivalent T-stubs by<br />
means <strong>of</strong> <strong>the</strong> proposed alternative <strong>characterization</strong> procedures.<br />
Test ID Equiva-<br />
Characterization procedures<br />
lent Experimental Numeri- Beam Jaspart<br />
T-stub<br />
cal model approx.<br />
Evaluation <strong>of</strong> Fep.ri.u (kN)<br />
FS1<br />
Top<br />
Bot.<br />
105.29<br />
⎯<br />
177.53<br />
⎯<br />
137.80<br />
360.18<br />
137.68<br />
275.70<br />
FS2<br />
Top<br />
Bot.<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
316.00<br />
375.79<br />
273.52<br />
366.47<br />
FS3<br />
Top<br />
Bot.<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
526.93<br />
865.40<br />
324.28<br />
448.49<br />
FS4<br />
Top<br />
Bot.<br />
207.97<br />
⎯<br />
⎯<br />
⎯<br />
182.99<br />
439.82<br />
195.17<br />
310.90<br />
Evaluation <strong>of</strong> ∆ep.ri.u (kN)<br />
FS1<br />
Top<br />
Bot.<br />
27.42<br />
⎯<br />
9.35<br />
⎯<br />
12.68<br />
⎯<br />
9.35<br />
10.20<br />
7.04<br />
4.62<br />
FS2<br />
Top<br />
Bot.<br />
14.55<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
8.87<br />
10.83<br />
5.87<br />
4.83<br />
FS3<br />
Top<br />
Bot.<br />
11.79<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
⎯<br />
11.21<br />
10.69<br />
3.77<br />
3.77<br />
FS4<br />
Top<br />
Bot.<br />
16.03<br />
⎯<br />
11.76<br />
⎯<br />
⎯<br />
⎯<br />
4.02<br />
7.53<br />
4.39<br />
5.34<br />
deformation capacity. The predictions do not compare well to <strong>the</strong> experiments<br />
as <strong>the</strong>y clearly underestimate <strong>the</strong> deformation capacity, particularly for <strong>the</strong><br />
thinner <strong>end</strong> <strong>plate</strong>s (ratios between <strong>the</strong> beam model predictions and <strong>the</strong> actual<br />
results from <strong>the</strong> LVDTs range from 3 to 4 for FS1 and FS4). For <strong>the</strong> thicker<br />
<strong>end</strong> <strong>plate</strong>s <strong>the</strong> agreement improves considerably. In fact, for specimen FS3 <strong>the</strong><br />
beam model predictions are quite accurate.<br />
The nondimensional analysis <strong>of</strong> <strong>the</strong>se equivalent T-stubs, at <strong>the</strong> top bolt<br />
row, at failure, i.e. in terms <strong>of</strong> <strong>the</strong> component <strong>ductility</strong> index, ϕep.r1, is summarized<br />
in Table 8.4 (BM: beam model; JBA: Jaspart approximation; Num: Numerical<br />
results for T-stub top and beam model for T-stub bottom; Exp: Experimental<br />
results for T-stub top and beam model for T-stub bottom). These<br />
indexes are evaluated from Eq. (1.39). The values in italic correspond to <strong>the</strong> ratios<br />
to <strong>the</strong> experimental results for <strong>the</strong> <strong>end</strong> <strong>plate</strong> deformation, at <strong>the</strong> beam<br />
flange level. Generally speaking, <strong>the</strong> predictions given by <strong>the</strong> beam model are<br />
good, showing a pronounced underestimation for FS4, which uses S690, and a<br />
clear overestimation for FS3. The average error is 14% but <strong>the</strong> coefficient <strong>of</strong><br />
variation is significant (0.75). Jaspart [8.8], on <strong>the</strong> o<strong>the</strong>r hand, gives estimations<br />
with an average error <strong>of</strong> 25% but <strong>the</strong> scatter <strong>of</strong> results is lower, with a co-<br />
317
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
efficient <strong>of</strong> variation <strong>of</strong> 0.26. The experimental results for <strong>the</strong> single T-stub are<br />
available for specimens FS1 (WT7_M20) and FS4 (WT57_M20). The corresponding<br />
indexes show some deviations from <strong>the</strong> actual joint results.<br />
The results just described are again analysed in <strong>the</strong> following sections in<br />
order to establish some criteria regarding <strong>the</strong> <strong>ductility</strong> requirements for <strong>the</strong><br />
overall joint behaviour.<br />
Table 8.4 Evaluation <strong>of</strong> <strong>the</strong> equivalent “T-stub top” component <strong>ductility</strong> index<br />
(<strong>characterization</strong> <strong>of</strong> <strong>the</strong> T-stubs for evaluation <strong>of</strong> <strong>the</strong> analytical<br />
response: BM – beam model, JBA – Jaspart bilinear approximation,<br />
Num – numerical FE model, Exp – T-stubs top characterized<br />
experimentally).<br />
ϕep.r1<br />
Test ID<br />
Average Coeff.<br />
FS1 FS2 FS3 FS4<br />
var.<br />
Experimental 27.98 15.99 16.15 11.06 ⎯ ⎯<br />
BM<br />
18.33<br />
(0.66)<br />
13.86<br />
(0.87)<br />
25.48<br />
(1.58)<br />
3.72<br />
(0.34) 0.86 0.75<br />
JBA<br />
20.71<br />
(0.74)<br />
16.31<br />
(1.02)<br />
11.42<br />
(0.71)<br />
6.01<br />
(0.54) 0.75 0.26<br />
Num<br />
23.92<br />
(0.86)<br />
⎯ ⎯ ⎯<br />
⎯ ⎯<br />
Exp<br />
18.33<br />
(0.66)<br />
⎯ ⎯ 15.08<br />
(1.36) 1.01 0.50<br />
Analytical<br />
8.3.2 Evaluation <strong>of</strong> <strong>the</strong> nonlinear moment-rotation response<br />
The full M-Φ joint response is evaluated using <strong>the</strong> s<strong>of</strong>tware NASCon [8.7].<br />
This s<strong>of</strong>tware is a computational implementation <strong>of</strong> <strong>the</strong> component method.<br />
The model file is written by means <strong>of</strong> <strong>the</strong> user-fri<strong>end</strong>ly “Connection Assistant”<br />
tool. All <strong>the</strong> details <strong>of</strong> <strong>the</strong> joint and joint components are specified in this file<br />
(see Fig. 8.11 for illustration). The multilinear component behaviour is input in<br />
this file. The model is <strong>the</strong>n imported by NASCon to generate <strong>the</strong> M-Φ curve<br />
(Fig. 8.12). A displacement control-based strategy was selected (Fig. 8.13). Finally,<br />
<strong>the</strong> overall M-Φ curve can be visualized (Fig. 8.14).<br />
The various curves are shown in <strong>the</strong> graphs from Figs. 8.15-8.18 and are<br />
compared with <strong>the</strong> experiments. The graphs trace <strong>the</strong> responses obtained in<br />
NASCon for <strong>the</strong> different <strong>characterization</strong> processes described in <strong>the</strong> previous<br />
section. The critical component is also indicated in <strong>the</strong> graphs as well as <strong>the</strong><br />
governing part (flange or bolt). Whenever <strong>the</strong> components are characterized<br />
with <strong>the</strong> bilinear approximation proposed by Jaspart [8.8], <strong>the</strong> critical failure<br />
mode at ultimate conditions (1, 2 or 3) is indicated. Note that for different<br />
<strong>characterization</strong> processes, <strong>the</strong> determinant T-stub for rotation capacity can<br />
change (e.g. joint FS1 and <strong>the</strong> beam model or <strong>the</strong> bilinear approximation pro-<br />
318
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Fig. 8.11 Modelling <strong>of</strong> <strong>the</strong> connection and component behaviour (e.g. FS1).<br />
Fig. 8.12 Model loading (e.g. FS1).<br />
319
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Fig. 8.13 NASCon strategy selection and prescribed loading (e.g. FS1).<br />
Table 8.5 Evaluation <strong>of</strong> initial stiffness (experimental results correspond to<br />
<strong>the</strong> average results <strong>of</strong> <strong>the</strong> two tests).<br />
Sj.ini<br />
Test ID<br />
Average Coeff.<br />
(kNm/mrad) FS1 FS2 FS3 FS4<br />
var.<br />
Experimental 17.52 22.69 22.39 16.67 ⎯ ⎯<br />
Eurocode 3<br />
34.66<br />
(1.98)<br />
46.76<br />
(2.06)<br />
51.76<br />
(2.31)<br />
32.77<br />
(1.97) 2.08 0.08<br />
BM<br />
30.78<br />
(1.76)<br />
45.26<br />
(1.99)<br />
54.78<br />
(2.45) 1.96 1.96 0.18<br />
JBA<br />
34.45<br />
(1.97)<br />
47.65<br />
(2.10)<br />
53.23<br />
(2.38) 2.10 2.10 0.10<br />
Num<br />
30.29<br />
(1.73)<br />
⎯ ⎯ ⎯<br />
⎯ ⎯<br />
Exp<br />
29.51<br />
(1.68)<br />
⎯ ⎯ 42.47<br />
(2.55) 2.12 0.29<br />
Analytical<br />
posed by Jaspart for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> T-stubs – Figs. 8.15a-b).<br />
Table 8.5 summarizes <strong>the</strong> characteristics <strong>of</strong> <strong>the</strong> curves in terms <strong>of</strong> initial<br />
stiffness. Again, <strong>the</strong> values in italic correspond to <strong>the</strong> ratio to <strong>the</strong> experiments.<br />
320
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
(a) Behaviour <strong>of</strong> component T-stub top (which determines ultimate conditions).<br />
(b) Behaviour <strong>of</strong> component T-stub bottom.<br />
Fig. 8.14 Moment-rotation curve (e.g. FS1).<br />
321
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
322<br />
B<strong>end</strong>ing moment (kNm)<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
FS1a<br />
60<br />
FS1b<br />
40<br />
20<br />
0<br />
NASCon prediction (T-stub<br />
top critical - flange)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(a) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> beam model.<br />
B<strong>end</strong>ing moment (kNm)<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
FS1a<br />
60<br />
FS1b<br />
40<br />
20<br />
0<br />
NASCon prediction (T-stub<br />
bottom critical - mode 2U)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(b) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> Jaspart bilinear model.<br />
B<strong>end</strong>ing moment (kNm)<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
FS1a<br />
FS1b<br />
40<br />
20<br />
0<br />
NASCon prediction (T-stub<br />
top critical - flange)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(c) Equivalent T-stub top characterized numerically (three-dimensional model).<br />
Fig. 8.15 Moment-rotation curve for joint FS1.
B<strong>end</strong>ing moment (kNm)<br />
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
180<br />
160<br />
140<br />
120<br />
100<br />
80<br />
FS1a<br />
60<br />
FS1b<br />
40<br />
20<br />
0<br />
NASCon prediction (T-stub<br />
top critical - flange)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(d) Equivalent T-stub top characterized experimentally.<br />
Fig. 8.15 Moment-rotation curve for joint FS1 (cont.).<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
FS2a<br />
90<br />
60<br />
FS2b<br />
30<br />
0<br />
NASCon prediction (T-stub<br />
top critical - flange)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(a) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> beam model.<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
FS2a<br />
90<br />
60<br />
FS2b<br />
30<br />
0<br />
NASCon prediction (T-stub<br />
top critical - mode 2U)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(b) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> Jaspart bilinear model.<br />
Fig. 8.16 Moment-rotation curve for joint FS2.<br />
323
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
324<br />
B<strong>end</strong>ing moment (kNm)<br />
400<br />
350<br />
300<br />
250<br />
200<br />
FS3a<br />
150<br />
100<br />
FS3b<br />
50<br />
0<br />
NASCon prediction (T-stub<br />
bottom critical - bolt)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(a) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> beam model.<br />
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
150<br />
120<br />
FS3a<br />
90<br />
60<br />
FS3b<br />
30<br />
0<br />
NASCon prediction (T-stub<br />
top critical - mode 2U)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(b) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> Jaspart bilinear model.<br />
Fig. 8.17 Moment-rotation curve for joint FS3.<br />
B<strong>end</strong>ing moment (kNm)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
FS4a<br />
FS4b<br />
30<br />
0<br />
NASCon prediction (T-stub<br />
top critical - flange)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(a) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> beam model.<br />
Fig. 8.18 Moment-rotation curve for joint FS4.
B<strong>end</strong>ing moment (kNm)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
FS4a<br />
FS4b<br />
30<br />
0<br />
NASCon prediction (T-stub<br />
top critical - mode 1U)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(b) Equivalent T-stubs characterized by means <strong>of</strong> <strong>the</strong> Jaspart bilinear model.<br />
B<strong>end</strong>ing moment (kNm)<br />
210<br />
180<br />
150<br />
120<br />
90<br />
60<br />
FS4a<br />
FS4b<br />
30<br />
0<br />
NASCon prediction (T-stub<br />
top critical - flange and bolt)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
(c) Equivalent T-stub top characterized experimentally and T-stub bottom<br />
characterized by means <strong>of</strong> <strong>the</strong> beam model.<br />
Fig. 8.18 Moment-rotation curve for joint FS4 (cont.).<br />
In general, <strong>the</strong> analytical predictions overestimate <strong>the</strong> initial stiffness in comparison<br />
with <strong>the</strong> experiments, Sj.ini, (e.g. specimen FS1 – Sj.ini.Exp = 17.52<br />
kNm/mrad, Sj.ini.BM = 30.78 kNm/mrad = 1.76 Sj.ini.Exp). This is quite straightforward<br />
from <strong>the</strong> statistical analysis <strong>of</strong> <strong>the</strong> ratios to <strong>the</strong> experiments also presented<br />
in Table 8.5 in italic.<br />
The examination <strong>of</strong> <strong>the</strong> curves also shows that <strong>the</strong> analytical predictions for<br />
resistance can also be slightly overestimated for some specimens, particularly<br />
in <strong>the</strong> plastic domain (e.g. FS3, Fig. 8.17) though for thinner <strong>end</strong> <strong>plate</strong>s <strong>the</strong> predictions<br />
are good (e.g. FS1, FS4, Figs. 8.15 and 8.18).<br />
The rotation capacity is clearly underestimated by <strong>the</strong> analytical methods,<br />
even for <strong>the</strong> cases <strong>of</strong> FS1 and FS4 with <strong>the</strong> experimental component <strong>characterization</strong>.<br />
Table 8.6 sets out <strong>the</strong> rotation predictions (experimental and analytical;<br />
values in italic represent <strong>the</strong> ratio to <strong>the</strong> experimental values). Experimen-<br />
325
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
tally, two rotation values were evaluated: <strong>the</strong> rotation corresponding to maximum<br />
load level, Φ M , and <strong>the</strong> rotation capacity, ΦCd (see also Table 7.8).<br />
max<br />
Analytically, <strong>the</strong> rotation capacity is attained when <strong>the</strong> first component reaches<br />
failure. The experimental values in Table 8.6 are <strong>the</strong> averaged values between<br />
<strong>the</strong> tests for each configuration, except for FS1 and FS3 for which <strong>the</strong> value <strong>of</strong><br />
tests “b” are adopted. This table also indicates <strong>the</strong> critical component for each<br />
methodology (EPX: cracking at <strong>the</strong> extension <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong>; BNSo+i: bolt<br />
nut stripping <strong>of</strong> <strong>the</strong> outer and inner bolt; BNSi: bolt nut stripping <strong>of</strong> <strong>the</strong> inner<br />
bolt; BTi: inner bolt in tension; Tt-fl: T-stub top, flange; Tb-b: T-stub bottom,<br />
bolt; Tt-fl+b: T-stub top, flange and bolt; 1U: mode 1 critical at ultimate conditions;<br />
2U: mode 2 critical at ultimate conditions).<br />
The statistical investigation <strong>of</strong> <strong>the</strong> results shows that <strong>the</strong> application <strong>of</strong> <strong>the</strong><br />
beam model for <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> individual T-stubs provides an average<br />
ratio to <strong>the</strong> experiments <strong>of</strong> 0.40 with a coefficient <strong>of</strong> variation <strong>of</strong> 0.58. The<br />
predictions obtained from application <strong>of</strong> Jaspart’s approximation [8.8] yield a<br />
lower value for <strong>the</strong> average ratio but also a lower coefficient <strong>of</strong> variation. Never<strong>the</strong>less,<br />
when both approaches are compared in terms <strong>of</strong> failure predictions, <strong>the</strong><br />
beam model gives a better agreement with <strong>the</strong> experimental observations.<br />
The joint <strong>ductility</strong> properties are fur<strong>the</strong>r analysed in <strong>the</strong> following sections.<br />
Table 8.6 Comparison <strong>of</strong> <strong>the</strong> predictions <strong>of</strong> rotation capacity <strong>of</strong> <strong>the</strong> various<br />
joints and failure modes.<br />
φCd<br />
Test ID<br />
Average Coeff.<br />
(mrad) FS1 FS2 FS3 FS4<br />
var.<br />
Experimental<br />
111.22<br />
EPX<br />
71.89<br />
BNSo+i<br />
48.74<br />
BNSi<br />
62.97<br />
BTi<br />
⎯ ⎯<br />
29.20 28.80 35.00 13.50<br />
BM (0.26) (0.40) (0.72) (0.21) 0.40 0.58<br />
Tt-fl Tt-fl Tb-b Tt-fl<br />
20.96 19.44 13.52 14.56<br />
JBA (0.19) (0.27) (0.28) (0.23) 0.24 0.17<br />
2U 2U 2U 1U<br />
39.84 ⎯ ⎯ ⎯ ⎯ ⎯<br />
Num (0.36)<br />
Tt-fl<br />
29.52 ⎯ ⎯ 36.80<br />
Exp (0.27)<br />
(0.58) 0.43 0.52<br />
Tt-fl Tt-fl+b<br />
Analytical<br />
8.3.3 Evaluation <strong>of</strong> <strong>the</strong> rotation capacity according to o<strong>the</strong>r authors<br />
Having discussed <strong>the</strong> results obtained from <strong>the</strong> author’s methodology in terms<br />
326
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
<strong>of</strong> predictions <strong>of</strong> rotation capacity, <strong>the</strong> proposals from o<strong>the</strong>r researchers are<br />
now analysed. The verifications on <strong>ductility</strong> requirements for <strong>the</strong>se specimens<br />
according to Eurocode 3 have already been carried out in Chapter 7. The main<br />
conclusions are summarized in Table 8.7.<br />
Three alternative procedures for evaluation <strong>of</strong> <strong>the</strong> rotation capacity are illustrated.<br />
These procedures were proposed by Adegoke and Kemp [8.5], for<br />
thin <strong>end</strong> <strong>plate</strong>s, Beg et al. [8.3] and Zoetemeijer [8.9]. This latter method is restricted<br />
to those cases where type-2 plastic failure mode is critical and consequently<br />
it is only validated by specimen FS3, for which <strong>the</strong> plastic failure mode<br />
<strong>of</strong> both equivalent T-stubs is <strong>of</strong> type-2. The three methodologies have been described<br />
in Chapter 1, §1.6.2. Table 8.8 sets out <strong>the</strong> main results for <strong>the</strong> above<br />
procedures. In general, <strong>the</strong> rotation capacity is underestimated.<br />
The application <strong>of</strong> <strong>the</strong> methodology proposed by Adegoke and Kemp [8.5]<br />
requires <strong>the</strong> definition <strong>of</strong> <strong>the</strong> location <strong>of</strong> <strong>the</strong> neutral axis <strong>of</strong> <strong>the</strong> connection at<br />
plastic and ultimate conditions (cf. §1.6.2 and references [8.5,8.10]). This location<br />
was defined from <strong>the</strong> results obtained through application <strong>of</strong> <strong>the</strong> UC mechanical<br />
model. This method reflects <strong>the</strong> t<strong>end</strong>ency observed in <strong>the</strong> experiments:<br />
<strong>the</strong> rotation capacity decreases with <strong>the</strong> <strong>plate</strong> thickness, for identical<br />
<strong>plate</strong> steel grades. For <strong>the</strong> specimen with steel S690, <strong>the</strong> rotation capacity is<br />
overestimated. However, <strong>the</strong> scope <strong>of</strong> <strong>the</strong> method is restricted to current steel<br />
grades and consequently <strong>the</strong> latter results are just illustrative. The method proposed<br />
by <strong>the</strong>se authors yields an average ratio to <strong>the</strong> experiments <strong>of</strong> 0.53 with a<br />
Table 8.7 Verification <strong>of</strong> <strong>the</strong> recomm<strong>end</strong>ations for rotation capacity according<br />
to Eurocode 3 (values in [mm]).<br />
Test<br />
Maximum Critical component governing Verification?<br />
ID<br />
tp<br />
tp<br />
<strong>the</strong> joint resistance<br />
FS1 10.40 11.80 End <strong>plate</strong> in b<strong>end</strong>ing Yes.<br />
FS2 15.01 11.75 End <strong>plate</strong> in b<strong>end</strong>ing No.<br />
FS3 20.02 11.76 End <strong>plate</strong> in b<strong>end</strong>ing No.<br />
FS4 10.06 8.25 End <strong>plate</strong> in b<strong>end</strong>ing No.<br />
Table 8.8 Analytical evaluation <strong>of</strong> <strong>the</strong> rotation capacity according to o<strong>the</strong>r<br />
authors.<br />
φCd<br />
Test ID<br />
Average Coeff.<br />
(mrad) FS1 FS2 FS3 FS4<br />
var.<br />
Experimental 111.22 71.89 48.74 62.97 ⎯ ⎯<br />
Adegoke and 31.66 22.71 17.47 72.88<br />
Kemp (0.28) (0.32) (0.36) (1.16) 0.53 0.79<br />
Beg et al. 48.40 47.88 105.40 49.34<br />
(0.44) (0.67) (2.16) (0.78) 1.01 0.77<br />
Zoetemeijer ⎯ ⎯ 17.53<br />
(0.36)<br />
⎯ ⎯ ⎯<br />
327
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
(ra<strong>the</strong>r high) coefficient <strong>of</strong> variation <strong>of</strong> 0.79.<br />
Beg and co-authors’ proposals [8.3] do not reproduce well <strong>the</strong> actual behaviour.<br />
In fact, for specimen FS3 that employs a thicker <strong>end</strong> <strong>plate</strong>, <strong>the</strong> predictions<br />
are <strong>the</strong> highest. Though <strong>the</strong> averaged ratio in this case is unitary, <strong>the</strong> high coefficient<br />
<strong>of</strong> variation indicates that <strong>the</strong> methodology is not sufficiently accurate.<br />
Finally, <strong>the</strong> predictions for FS3 by applying <strong>the</strong> Zoetemeijer’s proposals<br />
[8.9] underestimate <strong>the</strong> experimental results.<br />
8.3.4 Characterization <strong>of</strong> <strong>the</strong> joint <strong>ductility</strong><br />
A joint <strong>ductility</strong> index has been proposed in Chapter 1 and it has been defined<br />
as follows:<br />
ΦCd<br />
ϑ j = (8.1)<br />
ΦM<br />
Rd<br />
Essentially, it relates <strong>the</strong> rotation value at ultimate conditions with a rotation<br />
value attained in a plastic situation. In Eq. (8.1), <strong>the</strong> values <strong>of</strong> ΦCd and Φ M Rd<br />
were adopted. However, in this work o<strong>the</strong>r distinct values <strong>of</strong> rotation have been<br />
defined: ΦXd, corresponding to <strong>the</strong> rotation at which <strong>the</strong> moment first reaches<br />
Mj.Rd, and φ M , <strong>the</strong> rotation value at maximum load (see Figs. 1.28 and 7.7).<br />
max<br />
Tables 8.9 and 8.10 evaluate <strong>the</strong> experimental joint <strong>ductility</strong> indexes and explore<br />
<strong>the</strong> above differences. In <strong>the</strong>se examples, <strong>the</strong> joint rotation and <strong>the</strong> connection<br />
rotation are equal and so <strong>the</strong> latter values are indicated. As expected, if<br />
<strong>the</strong> index is related to φ (ϑj.Rd), usually lower than φXd, its value is greater<br />
328<br />
M Rd<br />
than if related to φXd (ϑj.Xd). The differences between <strong>the</strong> two indexes vary be-<br />
tween 47% in FS4a and 68% in FS1b. Two possibilities are considered in terms<br />
φ (at Mmax). The<br />
<strong>of</strong> <strong>the</strong> rotation at ultimate conditions: φCd (failure) and M max<br />
<strong>ductility</strong> indexes are naturally bigger in <strong>the</strong> first case, with deviations that vary<br />
between 31% in FS1b and 50% in FS2a. Again, in <strong>the</strong>se comparisons, <strong>the</strong> test<br />
results corresponding to specimens FS1a and FS3a should be excluded.<br />
Ano<strong>the</strong>r relevant aspect that warrants some attention relates to <strong>the</strong> indexes<br />
values within <strong>the</strong> same test series. This aspect is restricted to series FS2 and<br />
FS4. The differences between <strong>the</strong> two tests can diverge 8% for ϑj.Xd (at failure)<br />
and 33% for <strong>the</strong> same index, for specimens FS4 and FS2, respectively. These<br />
differences, however, are not consistent for <strong>the</strong> alternative definitions and<br />
within <strong>the</strong> same test series.<br />
Analytically, <strong>the</strong> joint <strong>ductility</strong> indexes are also evaluated. Table 8.6 sets<br />
out <strong>the</strong> analytical predictions for rotation capacity. From <strong>the</strong> analytical point <strong>of</strong><br />
view, <strong>the</strong>se are also <strong>the</strong> values at Mmax except when <strong>the</strong> experimental charac-<br />
terization <strong>of</strong> <strong>the</strong> T-stub top is input (e.g. Fig. 8.15 for specimens FS1). These<br />
values are again repeated in Table 8.11 along with <strong>the</strong> rotation values for φ .<br />
M Rd
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Table 8.9 Experimental evaluation <strong>of</strong> <strong>the</strong> joint <strong>ductility</strong> indexes at φ M .<br />
Rd<br />
Test ID Rotation values at <strong>the</strong> KR [mrad] Ductility indexes at Mmax<br />
φKR.inf<br />
φ φKR.sup<br />
M Rd<br />
φ ϑj.inf ϑj.Rd ϑj.sup<br />
M max<br />
FS1a 3.0 5.8 17.5 61.6 20.53 10.62 3.52<br />
FS1b 4.2 6.5 25.0 77.1 18.36 11.86 3.08<br />
FS2a 6.5 7.1 20.0 41.7 6.42 5.87 2.09<br />
FS2b 6.3 7.7 20.5 40.3 6.40 5.23 1.97<br />
FS3a 5.5 7.5 15.0 25.0 4.55 3.33 1.67<br />
FS3b 5.5 8.9 18.0 30.0 5.45 3.37 1.67<br />
FS4a 6.9 10.2 21.0 37.7 5.46 3.70 1.80<br />
FS4b 6.9 9.5 21.6 43.8 6.35 4.61 2.03<br />
Test ID Rotation values at <strong>the</strong> KR [mrad] Ductility indexes at failure<br />
φKR.inf<br />
φ φKR.sup φCd ϑj.inf ϑj.Rd ϑj.sup<br />
M Rd<br />
FS1a 3.0 5.8 17.5 68.9 22.97 11.88 3.94<br />
FS1b 4.2 6.5 25.0 111.2 26.48 17.11 4.45<br />
FS2a 6.5 7.1 20.0 82.9 12.75 11.68 4.15<br />
FS2b 6.3 7.7 20.5 60.9 9.67 7.91 2.97<br />
FS3a 5.5 7.5 15.0 42.8 7.78 5.71 2.85<br />
FS3b 5.5 8.9 18.0 48.7 8.85 5.47 2.71<br />
FS4a 6.9 10.2 21.0 61.7 8.94 6.05 2.94<br />
FS4b 6.9 9.5 21.6 64.2 9.30 6.76 2.97<br />
Table 8.10 Experimental evaluation <strong>of</strong> <strong>the</strong> joint <strong>ductility</strong> indexes at φXd.<br />
Test ID Rotation values at <strong>the</strong> KR [mrad] Ductility indexes at Mmax<br />
φKR.inf φXd φKR.sup<br />
φ ϑj.inf ϑj.Xd ϑj.sup<br />
M max<br />
FS1a 3.0 18.2 17.5 61.6 20.53 3.38 3.52<br />
FS1b 4.2 20.0 25.0 77.1 18.36 3.86 3.08<br />
FS2a 6.5 17.5 20.0 41.7 6.42 2.38 2.09<br />
FS2b 6.3 19.2 20.5 40.3 6.40 2.10 1.97<br />
FS3a 5.5 13.8 15.0 25.0 4.55 1.81 1.67<br />
FS3b 5.5 18.2 18.0 30.0 5.45 1.65 1.67<br />
FS4a 6.9 19.2 21.0 37.7 5.46 1.96 1.80<br />
FS4b 6.9 18.3 21.6 43.8 6.35 2.39 2.03<br />
Test ID Rotation values at <strong>the</strong> KR [mrad] Ductility indexes at failure<br />
φKR.inf φXd φKR.sup φCd ϑj.inf ϑj.Xd ϑj.sup<br />
FS1a 3.0 18.2 17.5 68.9 22.97 3.79 3.94<br />
FS1b 4.2 20.0 25.0 111.2 26.48 5.56 4.45<br />
FS2a 6.5 17.5 20.0 82.9 12.75 4.74 4.15<br />
FS2b 6.3 19.2 20.5 60.9 9.67 3.17 2.97<br />
FS3a 5.5 13.8 15.0 42.8 7.78 3.10 2.85<br />
FS3b 5.5 18.2 18.0 48.7 8.85 2.68 2.71<br />
FS4a 6.9 19.2 21.0 61.7 8.94 3.21 2.94<br />
FS4b 6.9 18.3 21.6 64.2 9.30 3.51 2.97<br />
329
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
For fur<strong>the</strong>r comparisons, this is <strong>the</strong> relevant rotation at plastic conditions. Table<br />
8.12 evaluates <strong>the</strong> joint <strong>ductility</strong> index for <strong>the</strong> various joints (cf. rotation values<br />
in Table 8.11). For <strong>the</strong> analytical procedures, <strong>the</strong> <strong>ductility</strong> index was evaluated<br />
at <strong>the</strong> analytical value for rotation capacity but with respect to <strong>the</strong> analytical<br />
and experimental values <strong>of</strong><br />
330<br />
φ (ϑj.Rd.anl and ϑj.Rd.exp, respectively), as shown<br />
M Rd<br />
in Table 8.12 (cf. rotation values in Tables 8.9 and 8.11). For <strong>the</strong> various processes<br />
<strong>the</strong> index ϑj.Rd.anl is bigger than ϑj.Rd.exp.<br />
The analytical predictions <strong>of</strong> <strong>the</strong> <strong>ductility</strong> index are quite severe, particularly<br />
for <strong>the</strong> thinner <strong>end</strong> <strong>plate</strong> specimens, FS1 and FS4 (2 nd -4 th columns, Table<br />
8.12 and 7 th column on Table 8.9). This situation also results from <strong>the</strong> underestimation<br />
<strong>of</strong> <strong>the</strong> T-stub component <strong>ductility</strong> itself (e.g. FS4, Fig. 8.10 and Table<br />
8.3). The analytical predictions for deformation capacity <strong>of</strong> <strong>the</strong> individual Tstubs<br />
are ra<strong>the</strong>r conservative, as seen above. Consequently, <strong>the</strong> rotation capacity<br />
<strong>of</strong> <strong>the</strong> overall joint, which is calculated from <strong>the</strong> individual components<br />
contribution, is also underestimated. On <strong>the</strong> contrary, for specimen FS3 that<br />
uses a 20 mm thick <strong>end</strong> <strong>plate</strong>, <strong>the</strong> <strong>ductility</strong> index is overestimated (2 nd on Table<br />
8.12 and 7 th column on Table 8.9).<br />
Table 8.11 Analytical predictions <strong>of</strong> rotation <strong>of</strong> <strong>the</strong> various joints (in [mrad]).<br />
Test<br />
ID<br />
Analytical predictions<br />
BM JA Num Exp<br />
φ M φCd φ Rd<br />
M φCd φ Rd<br />
M φCd φ Rd<br />
M φCd<br />
Rd<br />
FS1 3.0 29.2 2.4 21.0 4.6 39.8 3.3 29.5<br />
FS2 3.4 28.8 3.0 19.4 ⎯ ⎯ ⎯ ⎯<br />
FS3 4.4 35.0 3.4 13.5 ⎯ ⎯ ⎯ ⎯<br />
FS4 5.0 13.5 4.0 14.6 ⎯ ⎯ 4.4 36.8<br />
Table 8.12 Analytical evaluation <strong>of</strong> <strong>the</strong> joint <strong>ductility</strong> indexes.<br />
Test<br />
Analytical predictions<br />
ID BM JA Num Exp<br />
ϑj.Rd.anl ϑj.Rd.exp ϑj.Rd.anl ϑj.Rd.exp ϑj.Rd.anl ϑj.Rd.exp ϑj.Rd.anl ϑj.Rd.exp FS1 9.73 4.49 8.75 3.23 8.65 6.12 8.94 4.54<br />
FS2 8.47 3.89 6.47 2.62 ⎯ ⎯ ⎯ ⎯<br />
FS3 7.95 3.93 3.97 1.52 ⎯ ⎯ ⎯ ⎯<br />
FS4 2.70 1.36 3.65 1.47 ⎯ ⎯ 8.36 3.72<br />
8.4 DISCUSSION<br />
The rotational behaviour <strong>of</strong> <strong>bolted</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> beam-to-column connections<br />
was evaluated in <strong>the</strong> context <strong>of</strong> <strong>the</strong> component method. The methodology
Ductility <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
was restricted to joints whose behaviour was governed by <strong>the</strong> <strong>end</strong> <strong>plate</strong> modelled<br />
as equivalent T-stubs in tension. It has been shown that <strong>the</strong> overall M-Φ<br />
response can be modelled fairly accurately provided that <strong>the</strong> T-stub component<br />
F-∆ behaviour is well characterized. Because <strong>ductility</strong> is such an important<br />
characteristic in connection performance, <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> joint rotation<br />
capacity, i.e. <strong>the</strong> available joint rotation, was addressed with greater depth. In<br />
order to meet <strong>the</strong> <strong>ductility</strong> requirements, <strong>the</strong> required joint rotation, Φj.req must<br />
be less or equal to <strong>the</strong> available joint rotation, Φj.avail:<br />
Φ ≤Φ (8.2)<br />
jreq . javail .<br />
It is generally accepted that a minimum <strong>of</strong> 40-50 mrad ensures “sufficient rotation<br />
capacity” <strong>of</strong> a <strong>bolted</strong> joint in a partial strength scenario [8.11]. Tables 8.9<br />
and 8.10 show that joints FS2 and FS4 also guarantee this condition at maximum<br />
load. Therefore, <strong>the</strong> Eurocode 3 current provisions seem too conservative<br />
as far as rotational capacity is concerned (Table 8.7). This study affords some<br />
basis for <strong>the</strong> proposal <strong>of</strong> some additional criteria on this topic.<br />
From <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> <strong>ductility</strong> indexes in Table 8.9 (top half <strong>of</strong> <strong>the</strong> table),<br />
computed at maximum load, a minimum joint <strong>ductility</strong> index <strong>of</strong> 4.0 seems<br />
appropriate in order to ensure “sufficient rotation capacity”. This limitation<br />
should be set in conjunction with an absolute minimum value <strong>of</strong> 40 mrad and is<br />
valid for steel grade S355. For steel grade S690 similar criteria might be established.<br />
However, <strong>the</strong> T-stub component in isolation has to be fur<strong>the</strong>r explored<br />
for higher steel grades because <strong>of</strong> <strong>the</strong> inherent specificities. In addition, <strong>the</strong><br />
analytical procedure has to be calibrated with o<strong>the</strong>r joint specimens since <strong>the</strong><br />
joint <strong>ductility</strong> indexes are not yet accurate enough (cf. Tables 8.9 and 8.12).<br />
Naturally, as <strong>the</strong> above mentioned joints were designed to confine failure to<br />
<strong>the</strong> <strong>end</strong> <strong>plate</strong> and bolts, <strong>the</strong> deformation behaviour is exclusively dep<strong>end</strong>ent on<br />
<strong>the</strong>se two components that form an equivalent T-stub. Therefore, <strong>the</strong> conclusions<br />
are only valid if <strong>the</strong> T-stub determines collapse. In this case it would be<br />
preferable to set a criterion in terms <strong>of</strong> <strong>the</strong> component <strong>ductility</strong> index, ra<strong>the</strong>r<br />
than <strong>the</strong> joint <strong>ductility</strong> index. However, it is found out that <strong>the</strong> information contained<br />
in Table 8.4 is not sufficient and can even be contradictory. For instance,<br />
take specimen FS3 as an example. In terms <strong>of</strong> joint <strong>ductility</strong> index (Tables 8.9<br />
and 8.12), it is quite lower than <strong>the</strong> remaining cases. As for <strong>the</strong> single T-stub,<br />
<strong>the</strong> <strong>ductility</strong> index is higher than for specimen FS2 or FS4. This situation may<br />
arise in <strong>the</strong> definition <strong>of</strong> <strong>the</strong> equivalent T-stub itself. For specimens FS1 or<br />
FS4, corresponding to thin <strong>end</strong> <strong>plate</strong>s, <strong>the</strong> predictions for rotation capacity are<br />
underestimating but <strong>the</strong> ratio to <strong>the</strong> experiments is consistent (see Table 8.6<br />
and <strong>the</strong> BM <strong>characterization</strong>). In both specimens, where <strong>the</strong> equivalent T-stubs<br />
are governed by a type-1 plastic mode, <strong>the</strong> whole yield line pattern is likely to<br />
develop. For <strong>the</strong> o<strong>the</strong>r two cases, type-2 “plastic” failure mode is also present<br />
and <strong>the</strong>refore <strong>the</strong> complete pattern may not develop fully. This means that <strong>the</strong><br />
actual T-stub effective width may be different from beff in Table 8.2. Consequently,<br />
<strong>the</strong> T-stub response for assessment <strong>of</strong> <strong>the</strong> joint rotational behaviour<br />
would also be different.<br />
331
Monotonic behaviour <strong>of</strong> beam-to-column <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections<br />
Finally, although it has been shown that deemed-to-satisfy criteria for sufficient<br />
rotation capacity stated in Eurocode 3 are overconservative, <strong>the</strong> establishment<br />
<strong>of</strong> more accurate criteria still requires fur<strong>the</strong>r work.<br />
8.6 REFERENCES<br />
[8.1] Kuhlmann U, Kühnemund F. Ductility <strong>of</strong> semi-rigid steel joints. In: Proceedings<br />
<strong>of</strong> <strong>the</strong> International Colloquium on Stability and Ductility <strong>of</strong> Steel<br />
Structures (SDSS 2002) (Ed.: M. Ivanyi), Budapest, Hungary; 363-370,<br />
2002.<br />
[8.2] Faella C, Piluso V, Rizzano G. Structural Semi-Rigid Connections –<br />
Theory, Design and S<strong>of</strong>tware. CRC Press, USA, 2000.<br />
[8.3] Beg D, Zupančič E, Vayas I. On <strong>the</strong> rotation capacity <strong>of</strong> moment connections.<br />
Journal <strong>of</strong> Constructional Steel Research; 60:601-620, 2004.<br />
[8.4] Zandonini R, Zanon P. Experimental analysis <strong>of</strong> <strong>end</strong> <strong>plate</strong> connections.<br />
In: Proceedings <strong>of</strong> <strong>the</strong> First International Workshop on Connections in Steel<br />
Structures, Behaviour, Strength and Design (Eds.: R. Bjorhovde, J.<br />
Brozzetti and A. Colson), Cachan, France; 40-51, 1988.<br />
[8.5] Adegoke IO, Kemp AR. Moment-rotation relationships <strong>of</strong> thin <strong>end</strong> <strong>plate</strong><br />
connections in steel beams. In: Proceedings <strong>of</strong> <strong>the</strong> International Conference<br />
on Advances in Structures, ASSCCA’03 (Eds.: G.J. Hancock, M.A.<br />
Bradford, T.J. Wilkinson, B. Uy and K.J.R. Rasmussen), Sydney, Australia;<br />
119-124, 2003.<br />
[8.6] European Committee for Standardization (CEN). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[8.7] Borges LAC. Probabilistic evaluation <strong>of</strong> <strong>the</strong> rotation capacity <strong>of</strong> steel<br />
joints. MsC <strong>the</strong>sis. University <strong>of</strong> Coimbra, Coimbra, Portugal, 2003.<br />
[8.8] Jaspart JP. Study <strong>of</strong> <strong>the</strong> semi-rigid behaviour <strong>of</strong> beam-to-column joints<br />
and <strong>of</strong> its influence on <strong>the</strong> stability and strength <strong>of</strong> steel building frames.<br />
PhD <strong>the</strong>sis (in French). University <strong>of</strong> Liège, Liège, Belgium, 1991.<br />
[8.9] Zoetemeijer P. Summary <strong>of</strong> <strong>the</strong> research on <strong>bolted</strong> beam-to-column<br />
connections. Report 25-6-90-2. Faculty <strong>of</strong> Civil Engineering, Stevin<br />
Laboratory – Steel Structures, Delft University <strong>of</strong> Technology, 1990.<br />
[8.10] Kemp AR, Ne<strong>the</strong>rcot DA. Required and available rotations in continuous<br />
composite beams with semi-rigid connections. Journal <strong>of</strong> Constructional<br />
Steel Research; 57:375-400, 2001.<br />
[8.11] Grecea D, Statan A, Ciutina A, Dubina D. Rotation capacity <strong>of</strong> MR<br />
beam-to-column joints under cyclic loading. In: Proceedings <strong>of</strong> <strong>the</strong> Fifth<br />
International ECCS/AISC Workshop on Connections in Steel Structures,<br />
Innovative steel connections, Amsterdam, The Ne<strong>the</strong>rlands; 2004 (to be<br />
published).<br />
332
9 CONCLUSIONS AND RECOMMENDATIONS<br />
9.1 CONCLUSIONS<br />
The primary goal <strong>of</strong> this dissertation was to develop a methodology for <strong>the</strong><br />
<strong>characterization</strong> <strong>of</strong> <strong>the</strong> full nonlinear rotational behaviour <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong><br />
beam-to-column steel connections based on <strong>the</strong> component method. Because <strong>of</strong><br />
<strong>the</strong> emphasis recently placed on <strong>the</strong> design <strong>of</strong> joints within <strong>the</strong> partial<br />
strength/semi-rigid approach, special attention was addressed to <strong>the</strong> <strong>characterization</strong><br />
<strong>of</strong> <strong>the</strong> <strong>ductility</strong> <strong>of</strong> this joint type. The scope <strong>of</strong> <strong>the</strong> research was restricted<br />
to <strong>end</strong> <strong>plate</strong> connections for which <strong>the</strong> collapse was governed by <strong>the</strong><br />
tension zone idealized by means <strong>of</strong> T-stubs.<br />
This goal was achieved by firstly conducting a comprehensive experimental<br />
test programme <strong>of</strong> thirty-two individual T-stubs that were supplemented by robust<br />
FE analyses. The research on T-stubs constitutes a reliable database for<br />
validation <strong>of</strong> a simplified analytical (beam) model for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> F-<br />
∆ behaviour <strong>of</strong> T-stub connections. This investigation drew particular attention<br />
to <strong>the</strong> assemblies made up <strong>of</strong> welded <strong>plate</strong>s that model <strong>the</strong> <strong>end</strong> <strong>plate</strong> behaviour<br />
in <strong>the</strong> context <strong>of</strong> <strong>the</strong> T-stub idealization. Additionally, eight monotonic fullscale<br />
tests on <strong>end</strong> <strong>plate</strong> connections were conducted to analyse <strong>the</strong> ultimate response<br />
<strong>of</strong> this joint type and assess <strong>the</strong>ir behaviour from a <strong>ductility</strong> point <strong>of</strong><br />
view. The tests showed that <strong>end</strong> <strong>plate</strong> connections can achieve rotation capacity<br />
provided that <strong>the</strong> <strong>end</strong> <strong>plate</strong> is a “weak link” relative to <strong>the</strong> bolts.<br />
There are some original contributions in this research work:<br />
1. A detailed review on <strong>the</strong> state-<strong>of</strong>-art <strong>of</strong> <strong>the</strong> <strong>characterization</strong> <strong>of</strong> <strong>the</strong> M-Φ behaviour<br />
<strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong> beam-to-column steel connections which highlighted<br />
<strong>the</strong> current methodologies and Eurocode 3 provisions [9.1];<br />
2. A comprehensive test programme on WP-T-stub connections that constitutes<br />
a database <strong>of</strong> experimental results on this simple connection. Previous research<br />
work on this assembly type is not documented in technical literature. Piluso et<br />
al. [9.2] refer a single test on a WP-T-stub to validate an analytical methodology.<br />
This test programme provided insight into <strong>the</strong> actual behaviour <strong>of</strong> this<br />
simple connection, failure modes and deformation capacity. The main parameters<br />
affecting <strong>the</strong> deformation response <strong>of</strong> WP-T-stubs were identified and <strong>the</strong>ir<br />
influence on <strong>the</strong> overall behaviour <strong>of</strong> <strong>the</strong> connection was qualitatively and<br />
quantitatively assessed. The role <strong>of</strong> <strong>the</strong> welding and <strong>the</strong> presence <strong>of</strong> transverse<br />
stiffeners were also tackled. Additionally, <strong>the</strong> behaviour <strong>of</strong> WP-T-stubs and<br />
HR-T-stubs was confronted in order to clarify <strong>the</strong> main differences between<br />
both assembly-types;<br />
3. Documentation <strong>of</strong> <strong>the</strong> problems with <strong>the</strong> welding consumables and <strong>the</strong> pro-<br />
333
cedures is made. During <strong>the</strong> experiments on WP-T-stubs, some <strong>of</strong> <strong>the</strong> specimens<br />
showed early damage <strong>of</strong> <strong>the</strong> <strong>plate</strong> material near <strong>the</strong> weld toe due to <strong>the</strong><br />
effect <strong>of</strong> <strong>the</strong> welding consumable that induced premature cracking and reduced<br />
<strong>the</strong> overall deformation capacity. A solution to this problem was given by setting<br />
requirements to <strong>the</strong> weld metal to be used;<br />
4. Advanced FE modelling was conducted on HR-T-stubs and WP-T-stubs. A<br />
robust three-dimensional model that encompasses material and geometrical<br />
nonlinearities and contact friction phenomenon was developed. The model<br />
provided qualitative and quantitative understanding about <strong>the</strong> T-stub behaviour.<br />
It may also be used as a benchmark for FE modelling <strong>of</strong> <strong>bolted</strong> <strong>end</strong> <strong>plate</strong><br />
steel connections;<br />
5. Although no new models were developed in this work, some problems with<br />
existing models were identified and some modifications were tested (e.g. <strong>the</strong><br />
modification on <strong>the</strong> definition <strong>of</strong> <strong>the</strong> distance m for WP-T-stubs on Chapter 6);<br />
6. The completion and documentation <strong>of</strong> monotonic tests on <strong>bolted</strong> <strong>end</strong> <strong>plate</strong><br />
connections in b<strong>end</strong>ing, up to failure;<br />
7. Tests on <strong>bolted</strong> connection employing high-strength steel grade S690 were<br />
carried out. There is growing demand for high-strength steels in construction<br />
and insufficient knowledge on <strong>the</strong>se steel grades. In this work some results on<br />
<strong>bolted</strong> connections that use S690 are given to provide additional information<br />
on this subject;<br />
8. A methodology for <strong>characterization</strong> <strong>of</strong> <strong>the</strong> rotational response <strong>of</strong> a joint<br />
based on <strong>the</strong> component method was implemented and calibrated against experimental<br />
results. The methodology was restricted to joints whose behaviour<br />
was governed by <strong>the</strong> <strong>end</strong> <strong>plate</strong> modelled as equivalent T-stubs in tension. The<br />
results <strong>of</strong> this particular study along with <strong>the</strong> conclusions drawn from <strong>the</strong><br />
analysis <strong>of</strong> individual T-stubs afforded some basis for <strong>the</strong> proposal <strong>of</strong> some criteria<br />
for <strong>the</strong> verification <strong>of</strong> sufficient rotation capacity. The proposal was made<br />
in terms <strong>of</strong> a non-dimensional parameter, <strong>the</strong> joint <strong>ductility</strong> index. Naturally,<br />
this limitation was set in conjunction with an absolute minimum value <strong>of</strong> 40<br />
mrad. This proposal was restricted to S355 as it was recognized <strong>the</strong> data were<br />
insufficient for higher steel grades.<br />
Several conclusions are drawn from this research work:<br />
1. The prediction <strong>of</strong> failure should be based upon a deformation-based criterion<br />
ra<strong>the</strong>r than a resistance-based parameter. However, for consistency with Eurocode<br />
3 that uses <strong>the</strong> β-ratio at design conditions to predict <strong>the</strong> critical “plastic”<br />
collapse mode, in this work a similar ratio βu (at ultimate conditions) was<br />
brought in, to identify <strong>the</strong> potential fracture mode. Naturally, this brings some<br />
inconsistencies with <strong>the</strong> observed and <strong>the</strong> predicted failure type;<br />
2. The experimental-numerical work on <strong>the</strong> T-stub behaviour (both assembly<br />
types) identifies <strong>the</strong> major contributions <strong>of</strong> <strong>the</strong> overall T-stub deformation: <strong>the</strong><br />
flange flexural deformation and <strong>the</strong> tension bolt elongation. Usually, a higher<br />
deformation capacity <strong>of</strong> <strong>the</strong> T-stub is expected if <strong>the</strong> flange cracking governs<br />
<strong>the</strong> collapse instead <strong>of</strong> bolt fracture. The cracking associated to <strong>the</strong> flange<br />
mechanism, in <strong>the</strong> case <strong>of</strong> <strong>the</strong> welded <strong>plate</strong>s assembly, also dep<strong>end</strong>s on struc-<br />
334
Conclusions and recomm<strong>end</strong>ations<br />
tural constraint conditions and modifications in <strong>the</strong> mechanical properties in<br />
<strong>the</strong> HAZ, particularly those linked to <strong>the</strong> presence <strong>of</strong> residual stresses;<br />
3. During <strong>the</strong> experiments, <strong>the</strong> importance <strong>of</strong> <strong>the</strong> correct selection <strong>of</strong> electrodes<br />
and welding procedures in <strong>the</strong> case <strong>of</strong> <strong>the</strong> testing <strong>of</strong> WP-T-stubs was highlighted.<br />
It has been shown that <strong>the</strong> use <strong>of</strong> evenmatch s<strong>of</strong>t low hydrogen electrodes<br />
ensures a ductile behaviour;<br />
4. In general, bolts fail in tension before stripping. The stripping <strong>of</strong> <strong>the</strong> bolt<br />
threads and/or nut is not likely to occur in most cases. In <strong>the</strong> full-scale tests, <strong>the</strong><br />
nut stripping phenomenon occurred in four tests. In <strong>the</strong> experimental investigation<br />
on individual T-stubs <strong>the</strong> same problem was observed in one test. In fact,<br />
this phenomenon is ra<strong>the</strong>r frequent in practice. Research indicates that when<br />
<strong>the</strong> nut hardness is below a certain level (89 Rockwell B or 180 Brinell) <strong>the</strong>re<br />
is a risk <strong>of</strong> stripping. This phenomenon limits <strong>the</strong> <strong>ductility</strong> performance <strong>of</strong> <strong>the</strong><br />
whole joint and <strong>the</strong>refore it should be avoided. A solution to this problem can<br />
be given by setting requirements to <strong>the</strong> hardness and strength properties <strong>of</strong> <strong>the</strong><br />
nut;<br />
5. A two-dimensional beam model for assessment <strong>of</strong> <strong>the</strong> T-stub behaviour was<br />
developed. It retains all <strong>the</strong> relevant behavioural characteristics. To obtain <strong>the</strong><br />
F-∆ curve, a numerical incremental procedure is required and, consequently, <strong>the</strong><br />
model is not suitable for hand calculations. However, it clearly simplifies <strong>the</strong> process<br />
<strong>of</strong> behaviour <strong>characterization</strong> when compared to <strong>the</strong> three-dimensional FE approach<br />
or <strong>the</strong> experimental technique. The applicability <strong>of</strong> <strong>the</strong> model was well<br />
demonstrated within <strong>the</strong> range <strong>of</strong> examples shown in <strong>the</strong> text. The behaviour<br />
predicted by this model is ra<strong>the</strong>r good in terms <strong>of</strong> resistance. With respect to<br />
<strong>ductility</strong>, it reflects an overestimation <strong>of</strong> test results that is within an acceptable<br />
error. These differences may derive from a great sensitivity <strong>of</strong> <strong>the</strong> model to<br />
strain hardening parameters and bolt <strong>ductility</strong>. Additionally, <strong>the</strong> model encompasses<br />
a major simplification regarding <strong>the</strong> T-stub width, which is kept constant<br />
with <strong>the</strong> course <strong>of</strong> loading. It is well known that as <strong>the</strong> load increases, <strong>the</strong> flange<br />
width tributary to load transmission also increases. The implementation <strong>of</strong> such a<br />
variation is not straightforward. Ideally, <strong>the</strong> T-stub breadth should vary with <strong>the</strong><br />
loading and this variation should be dep<strong>end</strong>ent on <strong>the</strong> failure mode as well;<br />
6. Concerning <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> F-∆ response <strong>of</strong> T-stubs by means <strong>of</strong> o<strong>the</strong>r<br />
simplified methodologies, <strong>the</strong> bilinear approximation proposed by Jaspart [9.3] is<br />
accurate in terms <strong>of</strong> curve mimicry. However, <strong>the</strong> prediction <strong>of</strong> <strong>the</strong> potential failure<br />
mode is sometimes incorrect;<br />
7. The T-stub idealization <strong>of</strong> <strong>end</strong> <strong>plate</strong> behaviour is reliable in <strong>the</strong> elasticyielding<br />
domain. When strain hardening is present such idealization should be<br />
re-evaluated, especially in terms <strong>of</strong> effective width that is clearly different from<br />
<strong>the</strong> initial elastic behaviour;<br />
8. The methodology developed for <strong>the</strong> evaluation <strong>of</strong> <strong>the</strong> M-Φ curve <strong>of</strong> joints<br />
and its <strong>ductility</strong>, in particular, mainly dep<strong>end</strong>s on <strong>the</strong> T-stub idealization <strong>of</strong> <strong>end</strong><br />
<strong>plate</strong> behaviour as <strong>the</strong> joints were designed to confine failure to <strong>the</strong> <strong>end</strong> <strong>plate</strong><br />
and bolts. As a result, <strong>the</strong> conclusions drawn in Chapter 8 are only valid if <strong>the</strong><br />
335
collapse is determined by <strong>the</strong> T-stub component. The <strong>characterization</strong> <strong>of</strong> <strong>the</strong> Tstub<br />
behaviour and failure modes is <strong>the</strong>refore crucial. Two simplified methodologies<br />
were implemented for that purpose: (i) <strong>the</strong> proposed beam model and<br />
(ii) <strong>the</strong> bilinear approximation proposed by Jaspart. The outcomes were quite<br />
good, in general. However, <strong>the</strong> prediction <strong>of</strong> <strong>the</strong> failure modes was more accurate<br />
in <strong>the</strong> first case, as already explained;<br />
9. This methodology provided satisfactory results in terms <strong>of</strong> joint <strong>ductility</strong>,<br />
perhaps too conservative. However, a correcting factor can be defined to improve<br />
<strong>the</strong> results. This work does not permit <strong>the</strong> establishment <strong>of</strong> such a correction<br />
due to lack <strong>of</strong> data. Fur<strong>the</strong>rmore, <strong>the</strong> conclusions for series FS2 and FS3<br />
are quite limiting as <strong>the</strong> governing failure mode was <strong>the</strong> nut stripping <strong>of</strong> <strong>the</strong> inner<br />
bolts. This phenomenon should be avoided as explained and thus fur<strong>the</strong>r<br />
investigation on <strong>the</strong> behaviour <strong>of</strong> <strong>the</strong>se specimens is required;<br />
10. As already mentioned above, a minimum joint <strong>ductility</strong> index <strong>of</strong> 4.0 was<br />
proposed in order to ensure “sufficient rotation capacity”. Additionally, an absolute<br />
minimum rotation value <strong>of</strong> 40 mrad should also be guaranteed. It would<br />
have been preferable to set a criterion in terms <strong>of</strong> <strong>the</strong> T-stub component <strong>ductility</strong><br />
index, ra<strong>the</strong>r than <strong>the</strong> joint <strong>ductility</strong> index. However, <strong>the</strong>re was not enough<br />
data to make such a proposal;<br />
11. For steel grade S690, similar criteria for rotation capacity should be established.<br />
However, <strong>the</strong> T-stub component in isolation has to be fur<strong>the</strong>r explored<br />
for higher steel grades because <strong>of</strong> <strong>the</strong> inherent specificities;<br />
12. With reference to <strong>the</strong> <strong>end</strong> <strong>plate</strong> behaviour modelled as equivalent T-stubs<br />
(Chapter 8), <strong>the</strong> results for specimens FS2 and FS3 could be fur<strong>the</strong>r improved<br />
if <strong>the</strong> effective width <strong>of</strong> <strong>the</strong> T-stub bottom was reduced. The suggestion for this<br />
reduction is based on experimental observations <strong>of</strong> <strong>the</strong> yielded portions <strong>of</strong> <strong>the</strong><br />
<strong>end</strong> <strong>plate</strong> below <strong>the</strong> tension beam flange. If <strong>the</strong> following T-stub breadth:<br />
beff. red. bot = mep + eep + 0.5dh+<br />
m2<br />
(9.1)<br />
( FS 2)<br />
is implemented, <strong>the</strong>n for <strong>the</strong> above specimens, b = mm and<br />
336<br />
eff . red. bot 122.37<br />
( FS 3)<br />
b eff . red. bot = 124.07 mm (0.60 and 0.61 times <strong>the</strong> original value obtained from<br />
Eurocode 3, respectively – cf. Table 8.2). If <strong>the</strong> equivalent T-stub response is<br />
re-evaluated with <strong>the</strong>se changes (beam model <strong>characterization</strong>), <strong>the</strong> corresponding<br />
joints M-Φ curves will fit <strong>the</strong> experiments better, as shown in Figs.<br />
9.1 and 9.2. From a resistance point <strong>of</strong> view, <strong>the</strong> results are clearly improved.<br />
Also, <strong>the</strong> failure mode is compliant with experimental evidence. In terms <strong>of</strong><br />
<strong>ductility</strong>, <strong>the</strong> results do not vary significantly, though. This problem is probably<br />
linked to <strong>the</strong> T-stub idealization itself and so additional research should be carried<br />
out.<br />
9.2 FUTURE RESEARCH<br />
Some relevant issues were exposed during this investigation that warrant fur-
B<strong>end</strong>ing moment (kNm)<br />
240<br />
210<br />
180<br />
Conclusions and recomm<strong>end</strong>ations<br />
150<br />
FS2a<br />
120<br />
90<br />
FS2b<br />
60<br />
30<br />
0<br />
NASCon original prediction (Tstub<br />
top critical - flange)<br />
NASCon prediction (T-stub<br />
bottom critical - bolt)<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
Fig. 9.1 Moment-rotation curve for joint FS2 (T-stub <strong>characterization</strong> by means<br />
<strong>of</strong> <strong>the</strong> beam model and reduced effective length <strong>of</strong> <strong>the</strong> T-stub bottom).<br />
B<strong>end</strong>ing moment (kNm)<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
FS3a<br />
FS3b<br />
0<br />
0 10 20 30 40 50 60 70 80 90 100 110 120<br />
Connection rotation φ (mrad)<br />
NASCon original prediction<br />
(T-stub bottom critical - bolt)<br />
NASCon prediction (T-stubs<br />
top - flange & bottom critical<br />
- bolt)<br />
Fig. 9.2 Moment-rotation curve for joint FS3 (T-stub <strong>characterization</strong> by means<br />
<strong>of</strong> <strong>the</strong> beam model and reduced effective length <strong>of</strong> <strong>the</strong> T-stub bottom).<br />
<strong>the</strong>r consideration. They are listed below and are proposed as future research:<br />
1. The bolt force-elongation curve that was proposed in Chapter 6 for <strong>the</strong> bolt<br />
response simplified modelling requires fur<strong>the</strong>r investigation as far as fullthreaded<br />
bolts are concerned. This curve was derived for short-threaded bolts.<br />
This work clearly shows that for full-threaded bolts <strong>the</strong> predictions <strong>of</strong> bolt fracture<br />
overestimate <strong>the</strong> overall results. The formula for evaluation <strong>of</strong> <strong>the</strong> bolt<br />
fracture should include explicitly <strong>the</strong> ratio between <strong>the</strong> bolt shank and threaded<br />
lengths. Additionally, <strong>the</strong>re should be a resistance limitation as it was observed<br />
that <strong>the</strong> bolt force at fracture could be as high as 1.30Bu, whereby Bu is <strong>the</strong> bolt<br />
tensile strength, evaluated in engineering stresses;<br />
2. A clarification <strong>of</strong> <strong>the</strong> definition <strong>of</strong> <strong>the</strong> distance m is needed. Chapters 3-5<br />
337
gave experimental and numerical results for <strong>the</strong> stress and strain results on WP-<br />
T-stubs and showed that <strong>the</strong> yield lines near <strong>the</strong> flange-to-web connection<br />
would potentially develop at <strong>the</strong> <strong>end</strong> <strong>of</strong> <strong>the</strong> fillet weld. This would change <strong>the</strong><br />
expression for computation <strong>of</strong> <strong>the</strong> distance m. According to Eurocode 3, m in<br />
<strong>the</strong>se cases is defined as follows:<br />
m = d − 0.8 2aw (9.2)<br />
Chapter 6 compared <strong>the</strong> beam model results obtained when this distance was<br />
employed with those obtained from:<br />
m = d − 2aw (9.3)<br />
which are fur<strong>the</strong>r improved. The latter definition is more compliant with <strong>the</strong><br />
observations and should be regarded as a possible modification. Additional<br />
work on this subject is essential;<br />
3. Fur<strong>the</strong>r research on <strong>the</strong> T-stub idealization <strong>of</strong> <strong>the</strong> <strong>end</strong> <strong>plate</strong> behaviour is required.<br />
Three-dimensional FE analysis may be helpful for investigating this<br />
specific topic. The numerical results presented in this research work can be<br />
used as benchmarks for validation <strong>of</strong> <strong>the</strong> global joint model. Naturally, <strong>the</strong> experimental<br />
results are also essential for <strong>the</strong> calibration procedures. The establishment<br />
<strong>of</strong> more appropriate rules for <strong>the</strong> definition <strong>of</strong> <strong>the</strong> effective equivalent<br />
T-stub width, particularly in <strong>the</strong> post-limit regime, are fundamental. The experiments<br />
can not provide enough results for this analysis. Advanced FE modelling<br />
provides all <strong>the</strong> necessary data and will be carried out by <strong>the</strong> author as a<br />
follow up study to this investigation.<br />
9.3 REFERENCES<br />
[9.1] European Committee for Standardization (CEN). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage<br />
49 draft, May 2003, Brussels, 2003.<br />
[9.2] Piluso V, Faella C, Rizzano G. Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. II:<br />
Model validation. Journal <strong>of</strong> Structural Engineering ASCE; 127(6):694-<br />
704, 2001.<br />
[9.3] Jaspart JP. Study <strong>of</strong> <strong>the</strong> semi-rigid behaviour <strong>of</strong> beam-to-column joints<br />
and <strong>of</strong> its influence on <strong>the</strong> stability and strength <strong>of</strong> steel building frames.<br />
PhD <strong>the</strong>sis (in French). University <strong>of</strong> Liège, Liège, Belgium, 1991.<br />
338
LIST OF REFERENCES<br />
Adegoke IO, Kemp AR. (2003). Moment-rotation relationships <strong>of</strong> thin <strong>end</strong> <strong>plate</strong><br />
connections in steel beams. In: Proceedings <strong>of</strong> <strong>the</strong> International Conference on<br />
Advances in Structures (ASSCCA’03) (Eds.: G.J. Hancock, M.A. Bradford,<br />
T.J. Wilkinson, B. Uy and K.J.R. Rasmussen), Sydney, Australia; 119-124.<br />
(In: Chapters 1, 8)<br />
Aggarwal AK. (1994). Comparative tests on <strong>end</strong><strong>plate</strong> beam-to-column connections.<br />
Journal <strong>of</strong> Constructional Steel Research; 30:151-175. (In: Chapter 1)<br />
Aggerskov H. (1976). High-strength <strong>bolted</strong> connections subjected to prying. Journal<br />
<strong>of</strong> <strong>the</strong> Structural Division ASCE; 102(ST1):161-175. (In: Chapters 1, 4)<br />
Aggerskov H. (1977). Analysis <strong>of</strong> <strong>bolted</strong> connections subjected to prying. Journal<br />
<strong>of</strong> <strong>the</strong> Structural Division ASCE; 103(ST11):2145-2163. (In: Chapter 1)<br />
Aribert JM, Lachal A. (1977). Étude élasto-plastique par analyse des contraintes<br />
de la compression locale sur l’âme d’un pr<strong>of</strong>ilé. Construction Métallique; 4:51-<br />
66. (In: Chapter 1)<br />
Aribert JM, Lachal A, Dinga ON. (1999). Modélisation du comportement<br />
d’assemblages métalliques semi-rigides de type pouter-pouteau boulonnés par<br />
platine d’extremité. Construction Métallique; 1:25-46. (In: Chapter 1)<br />
Aribert JM, Lachal A, El Nawawy O. (1981). Modélisation élasto-plastique de la<br />
résistance d’un pr<strong>of</strong>ilé en compression locale. Construction Métallique; 2:3-26.<br />
(In: Chapter 1)<br />
Aribert JM, Lachal A, Moheissen M. (1990). Interaction du voilement et de la<br />
résistance plastique de l’âme d’un pr<strong>of</strong>ilé laminé soumis à une double<br />
compression locale (nuance dácier allant jusqu’à FeE460. Construction<br />
Métallique; 2:3-23. (In: Chapter 1)<br />
Astaneh A. (1985). Procedure for design and analysis <strong>of</strong> hanger-type connections.<br />
Engineering Journal AISC; 22(2):63-66. (In: Chapter 1)<br />
Bahaari MR, Sherbourne AN. (1994). Computer modelling <strong>of</strong> an ext<strong>end</strong>ed <strong>end</strong><strong>plate</strong><br />
<strong>bolted</strong> connection. Computers and Structures; 52(5):879-893. (In: Chapter<br />
1)<br />
Bahaari MR, Sherbourne AN. (1996). 3D simulation <strong>of</strong> <strong>bolted</strong> connections to unstiffened<br />
columns-II: Ext<strong>end</strong>ed <strong>end</strong><strong>plate</strong> connections. Journal <strong>of</strong> Constructional<br />
Steel Research; 40(3):189-223. (In: Chapter 1)<br />
Bahaari MR, Sherbourne AN. (1996). Structural behavior <strong>of</strong> <strong>end</strong>-<strong>plate</strong> <strong>bolted</strong><br />
connections to stiffened columns. Journal <strong>of</strong> Structural Engineering ASCE;<br />
122(8):926-935. (In: Chapter 1)<br />
Bahaari MR, Sherbourne AN. (1997). Finite element prediction <strong>of</strong> <strong>end</strong> <strong>plate</strong> <strong>bolted</strong><br />
339
connection behaviour II: Analytic formulation. Journal <strong>of</strong> Structural Engineering<br />
ASCE; 123(2):165-175. (In: Chapter 1)<br />
Ballio G, Mazzolani FM. (1983). Theory and design <strong>of</strong> steel structures. Chapman<br />
and Hall, London, UK. (In: Chapter 1)<br />
Bang KS, Kim WY. (2002). Estimation and prediction <strong>of</strong> <strong>the</strong> HAZ s<strong>of</strong>tening in<br />
<strong>the</strong>rmomechanically controlled-rolled and accelerated-cooled steel. Welding<br />
Journal; 81(8):174S-179S. (In: Chapter 4)<br />
Ba<strong>the</strong> KJ. (1982). Finite element procedures in engineering analysis. Prentice-<br />
Hall, Englewood Cliffs, New Jersey, USA. (In: Chapter 4)<br />
Ba<strong>the</strong> KJ, Wilson EL. (1976). Numerical methods in finite element analysis. Prentice-Hall,<br />
Englewood Cliffs, New Jersey, USA. (In: Chapter 4)<br />
Beg D, Zupančič E, Vayas I. (2004). On <strong>the</strong> rotation capacity <strong>of</strong> moment connections.<br />
Journal <strong>of</strong> Constructional Steel Research; 60:601-620. (In: Chapters 1,<br />
6, 8)<br />
Borges LAC. (2003). Probabilistic evaluation <strong>of</strong> <strong>the</strong> rotation capacity <strong>of</strong> steel<br />
joints. MSc <strong>the</strong>sis. University <strong>of</strong> Coimbra, Coimbra, Portugal. (In: Chapters 1,<br />
8)<br />
Bose B, Sarkar S, Bahrami M. (1996). Ext<strong>end</strong>ed <strong>end</strong><strong>plate</strong> connections: comparison<br />
between three-dimensional nonlinear finite-element analysis and full-scale<br />
destructive tests. Structural Engineering Review; 8(4):315-328. (In: Chapter 1)<br />
Bursi OS. (1991). An experimental-numerical method for <strong>the</strong> modelling <strong>of</strong> plastic<br />
failure mechanisms <strong>of</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> steel connections. Structural Engineering<br />
Review; 3:111-119. (In: Chapter 1)<br />
Bursi OS. (1995). A refined finite element model for T-stub steel connections.<br />
Cost C1, Numerical simulation group, Doc. C1WD6/95-07. (In: Chapter 4)<br />
Bursi OS, Jaspart JP. (1997). Benchmarks for finite element modelling <strong>of</strong> <strong>bolted</strong><br />
steel connections. Journal <strong>of</strong> Constructional Steel Research; 43(1):17-42. (In:<br />
Chapters 2, 4)<br />
Bursi OS, Jaspart JP. (1997). Calibration <strong>of</strong> a finite element model for <strong>bolted</strong> <strong>end</strong><br />
<strong>plate</strong> steel connections. Journal <strong>of</strong> Constructional Steel Research; 44(3):225-<br />
262. (In: Chapters 1-2)<br />
Bursi OS, Jaspart JP. (1998). Basic issues in <strong>the</strong> finite element simulation <strong>of</strong> ext<strong>end</strong>ed<br />
<strong>end</strong>-<strong>plate</strong> connections. Computers and Structures; 69:361-382. (In:<br />
Chapter 1)<br />
Choi CK, Chung GT. (1996). Refined three-dimensional finite element model for<br />
<strong>end</strong> <strong>plate</strong> connection. Journal <strong>of</strong> Structural Engineering ASCE; 122(11):1307-<br />
1316. (In: Chapter 1)<br />
Crisfield M. (1973). Large deflection elasto-plastic buckling analysis <strong>of</strong> <strong>plate</strong>s<br />
using finite elements. TRRL Report LR 593, Transport and Road Research<br />
Laboratory, Department <strong>of</strong> <strong>the</strong> Environment. Crowthorne, UK. (In: Chapter 4)<br />
Crisfield M. (1997). Non-linear finite element analysis <strong>of</strong> solids and structures,<br />
Vol. 1 – Essentials. John Wiley & Sons Ltd. Chichester, UK. (In: Chapter 4)<br />
Crisfield M. (1997). Non-linear finite element analysis <strong>of</strong> solids and structures,<br />
Vol. 2 – Advanced topics. John Wiley & Sons Ltd. Chichester, UK. (In: Chapter<br />
4)<br />
340
Davies AC. (1992). The science and practice <strong>of</strong> welding – welding science and<br />
technology – Vol. I. Cambridge University Press, Cambridge, UK. (In: Chapter<br />
3)<br />
Davison JB, Kirby PA, Ne<strong>the</strong>rcot DA. (1987). Rotational stiffness characteristics<br />
<strong>of</strong> steel beam-to-column connections. Journal <strong>of</strong> Constructional Steel Research;<br />
8:17-54. (In: Chapter 1)<br />
Davison JB, Kirby PA, Ne<strong>the</strong>rcot DA. (1987). Effect <strong>of</strong> lack <strong>of</strong> fit on connection<br />
restraint. Journal <strong>of</strong> Constructional Steel Research; 8:55-69. (In: Chapter 1)<br />
Douty RT, McGuire W. (1965). High strength moment connections. Journal <strong>of</strong><br />
Structural Division ASCE; 91(ST2):101-128. (In: Chapter 1)<br />
European Committee for Standardization (CEN). (1994). EN 499:1994E: Welding<br />
consumables – Covered electrodes for manual metal arc welding <strong>of</strong> non alloy<br />
and fine grain steels - Classification, December 1994, Brussels. (In: Chapter 3)<br />
European Committee for Standardization (CEN). (1995). EN 10204:1995E: Metallic<br />
products, October 1995, Brussels. (In: Chapters 3, 7)<br />
European Committee for Standardization (CEN). (2000). prEN 10025:2000E: Hot<br />
rolled products <strong>of</strong> structural steels, September 2000, Brussels. (In: Chapters 3,<br />
7)<br />
European Committee for Standardization (CEN). (2003). prEN 1993-1-8:2003,<br />
Part 1.8: Design <strong>of</strong> joints, Eurocode 3: Design <strong>of</strong> steel structures. Stage 49<br />
draft, May 2003, Brussels. (In: Chapters 1-9)<br />
Faella C, Piluso V, Rizzano G. (2000). Structural semi-rigid connections – <strong>the</strong>ory,<br />
design and s<strong>of</strong>tware. CRC Press, USA. (In: Chapters 1-2, 4, 6, 8)<br />
Gebbeken N, Wanzek T, Petersen, C. (1997). Semi-rigid connections, T-stub<br />
model – Report on experimental investigations. Report 97/2. Institut für<br />
Mechanik und Static, Universität des Bundeswehr München, Munich, Germany.<br />
(In: Chapter 4)<br />
Gioncu V, Mateescu G, Petcu D, Anastasiadis A. (2000). Prediction <strong>of</strong> available<br />
<strong>ductility</strong> by means <strong>of</strong> local plastic mechanism method: DUCTROT computer<br />
program, Chapter 2.1 in Moment resistant connections <strong>of</strong> steel frames in seismic<br />
areas (Ed.: F. Mazzolani). E&FN Spon, London, UK; 95-146. (In: Chapters<br />
2, 4, 6)<br />
Gioncu V, Mazzolani FM. (2002). Ductility <strong>of</strong> seismic resistant steel structures.<br />
Spon Press, London, UK. (In: Chapters 3, 6)<br />
Girão Coelho AM. (1999). Equivalent elastic models for <strong>the</strong> analysis <strong>of</strong> steel<br />
joints. MSc <strong>the</strong>sis (in Portuguese). University <strong>of</strong> Coimbra, Coimbra, Portugal.<br />
(In: Chapter 1)<br />
Girão Coelho AM. (2002). Material data <strong>of</strong> <strong>the</strong> <strong>plate</strong> sections <strong>of</strong> <strong>the</strong> welded T-stub<br />
specimens. Internal report, Steel and Timber Section, Faculty <strong>of</strong> Civil Engineering,<br />
Delft University <strong>of</strong> Technology. (In: Chapter 4)<br />
Girão Coelho AM, Bijlaard F, Simões da Silva L. (2002). On <strong>the</strong> behaviour <strong>of</strong><br />
<strong>bolted</strong> <strong>end</strong> <strong>plate</strong> connections modelled by welded T-stubs. In: Proceedings <strong>of</strong><br />
<strong>the</strong> Third European Conference on Steel Structures (Eurosteel) (Eds.: A. Lamas<br />
and L. Simões da Silva), Coimbra, Portugal, 907-918. (In: Chapter 2)<br />
341
Girão Coelho AM, Bijlaard F, Simões da Silva L. (2002). On <strong>the</strong> deformation<br />
capacity <strong>of</strong> beam-to-column <strong>bolted</strong> connections. Document ECCS-TWG 10.2-<br />
02-003, European Convention for Constructional Steelwork – Technical<br />
Commit-tee 10, Structural connections (ECCS-TC10). (In: Chapters 2, 4)<br />
Girão Coelho AM, Bijlaard F, Simões da Silva L. (2002). Experimental research<br />
work on T-stub connections made up <strong>of</strong> welded <strong>plate</strong>s. Document ECCS-<br />
TWG 10.2-217, European Convention for Constructional Steelwork – Technical<br />
Committee 10, Structural Connections (ECCS-TC10). (In: Chapter 3)<br />
Girão Coelho AM, Simões da Silva L. (2002). Numerical evaluation <strong>of</strong> <strong>the</strong> <strong>ductility</strong><br />
<strong>of</strong> a <strong>bolted</strong> T-stub connection. In: Proceedings <strong>of</strong> <strong>the</strong> Third International<br />
Conference on Advances in Steel Structures (ICASS’02) (Eds.: S.L. Chan,<br />
F.G. Teng and K.F.Chung), Hong Kong, China, 277-284. (In: Chapter 3)<br />
Girão Coelho AM, Bijlaard F, Gresnigt N, Simões da Silva L. (2004). Experimental<br />
assessment <strong>of</strong> <strong>the</strong> behaviour <strong>of</strong> <strong>bolted</strong> T-stub connections made up <strong>of</strong><br />
welded <strong>plate</strong>s. Journal <strong>of</strong> Constructional Steel Research; 60:269-311. (In:<br />
Chapter 3)<br />
Girão Coelho AM, Bijlaard F, Simões da Silva L. (2004). Experimental assessment<br />
<strong>of</strong> <strong>the</strong> <strong>ductility</strong> <strong>of</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections. Engineering Structures<br />
(in print). (In: Chapter 7)<br />
Gomes FCT, Neves LFC, Silva LAPS, Simoes RAD. (1995). Numerical simulation<br />
<strong>of</strong> a T-stub. Cost C1, Numerical simulation group, Doc. C1WG6/95. (In:<br />
Chapter 4)<br />
Grecea D, Statan A, Ciutina A, Dubina D. (2004). Rotation capacity <strong>of</strong> MR<br />
beam-to-column joints under cyclic loading. In: Proceedings <strong>of</strong> <strong>the</strong> Fifth International<br />
ECCS/AISC Workshop on Connections in Steel Structures, Innovative<br />
steel connections, Amsterdam, The Ne<strong>the</strong>rlands; (to be published). (In:<br />
Chapter 8)<br />
Hinton E, Owen DR. (1979). An introduction to finite element computations. Pineridge<br />
Press Limited, Swansea, UK. (In: Chapter 4)<br />
Hirt MA, Bez R. (1994). Construction métallique: notions fondamentales et methods<br />
de dimensionnement. Traité de Génie Civil de l’École polytechnique<br />
fédérale de Lausanne, Volume 10. Presses Polytechniques et Universitaires<br />
Romandes, Lausanne, Switzerland. (In: Chapters 4, 6)<br />
Holmes M, Martin LH. (1983). Analysis and design <strong>of</strong> structural connections:<br />
reinforced concrete and Steel. Ellis Horwood Limited, Chichester, UK. (In:<br />
Chapter 1)<br />
Huber G, Tschemmernegg F. (1996). Component characteristics, Chapter 4 in<br />
Composite Steel-Concrete Joints in Braced Frames for Buildings (Ed.: D.<br />
Anderson), COST C1, Brussels, Luxembourg; 4.1-4.49. (In: Chapter 1)<br />
Huber G, Tschemmernegg F. (1998). Modelling <strong>of</strong> beam-to-column joints. Journal<br />
<strong>of</strong> Constructional Steel Research; 45:199-216. (In: Chapter 1)<br />
Huber G. (1999). Nicht-lineare berechnungen von verbundquerschnitten und<br />
biegeweichen knoten. PhD Thesis (in English), University <strong>of</strong> Innsbruck, Innsbruck,<br />
Austria. (In: Chapter 1)<br />
342
International Standard ISO 898-1:1999(E). (1999). Mechanical properties <strong>of</strong> fasteners<br />
made <strong>of</strong> carbon steel and alloy steel – Part 1: Bolts, screws and studs,<br />
August 1999, Switzerland. (In: Chapter 7)<br />
Janss J, Jaspart JP, Maquoi R. (1988). Experimental study <strong>of</strong> <strong>the</strong> non-linear behaviour<br />
<strong>of</strong> beam-to-column <strong>bolted</strong> joints. In: Proceedings <strong>of</strong> <strong>the</strong> First International<br />
Workshop on Connections in Steel Structures, Behaviour, Strength and Design<br />
(Eds.: R. Bjorhovde, J. Brozzetti and A. Colson), Cachan, France; 26-32. (In:<br />
Chapter 1)<br />
Jaspart JP. (1991). Study <strong>of</strong> <strong>the</strong> semi-rigid behaviour <strong>of</strong> beam-to-column joints and<br />
<strong>of</strong> its influence on <strong>the</strong> stability and strength <strong>of</strong> steel building frames. PhD <strong>the</strong>sis<br />
(in French). University <strong>of</strong> Liège, Liège, Belgium. (In: Chapters 1, 6, 8-9)<br />
Jaspart JP. (1994). Numerical simulation <strong>of</strong> a T-stub – experimental data. Cost C1,<br />
Numerical simulation group, Doc. C1WD6/94-09. (In: Chapter 4)<br />
Jaspart JP. (1997). Contributions to recent advances in <strong>the</strong> field <strong>of</strong> steel joints –<br />
column bases and fur<strong>the</strong>r configurations for beam-to-column joints and beam<br />
splices. Aggregation <strong>the</strong>sis. University <strong>of</strong> Liège, Liège, Belgium. (In: Chapters<br />
1, 6)<br />
Jaspart JP, Maquoi R. (1994). Prediction <strong>of</strong> <strong>the</strong> semi-rigid and partial strength<br />
properties <strong>of</strong> structural joints. Proceedings <strong>of</strong> <strong>the</strong> Annual Technical Meeting on<br />
Structural Stability Research, Lehigh, USA; 177-191. (In: Chapter 1)<br />
Jenkins WM, Tong CS, Prescott AT. (1986). Moment-transmitting <strong>end</strong><strong>plate</strong> connections<br />
in steel construction, and a proposal basis for flush <strong>end</strong><strong>plate</strong> design.<br />
The Structural Engineer; 64A(5):121-132. (In: Chapter 1)<br />
Kato B, McGuire W. (1973). Analysis <strong>of</strong> T-stub flange-to-column connections.<br />
Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE; 99(ST5):865-888. (In: Chapter 1)<br />
Kemp AR, Ne<strong>the</strong>rcot DA. (2001). Required and available rotations in continuous<br />
composite beams with semi-rigid connections. Journal <strong>of</strong> Constructional<br />
Steel Research; 57:375-400. (In: Chapter 8)<br />
Krishnamurthy N. (1980). Modelling and prediction <strong>of</strong> steel <strong>bolted</strong> connection<br />
behaviour. Computers and Structures; 11:75-82. (In: Chapter 1)<br />
Krishnamurthy N, Graddy DE. (1976). Correlation between 2- and 3-dimensional<br />
finite element analysis <strong>of</strong> steel <strong>bolted</strong> <strong>end</strong>-<strong>plate</strong> connections. Computers and<br />
Structures; 6:381-389. (In: Chapter 1)<br />
Krishnamurthy N, Huang HT, Jeffrey PK, Avery LK. (1979). Analytical M-θ<br />
curves for <strong>end</strong>-<strong>plate</strong> connections. Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE;<br />
105(ST1):133-145. (In: Chapter 1)<br />
Kuhlmann U, Davison JB, Kattner M. (1998). Structural systems and rotation capacity.<br />
In: Proceedings <strong>of</strong> <strong>the</strong> International Conference on Control <strong>of</strong> <strong>the</strong> Semi-<br />
Rigid Behaviour <strong>of</strong> Civil Engineering Structural Connections (Ed.: R. Maquoi),<br />
Liège, Belgium; 167-176. (In: Chapter 1)<br />
Kuhlmann U, Kühnemund F. (2000). Rotation capacity <strong>of</strong> steel joints: verification<br />
procedure and component tests. In: Proceedings <strong>of</strong> <strong>the</strong> NATO Advanced Research<br />
Workshop: The paramount role <strong>of</strong> joints into <strong>the</strong> reliable response <strong>of</strong><br />
structures (Eds.: C.C. Baniotopoulos and F. Wald), Nato Science series, Klu-<br />
343
wer Academic Publishers, Dordrecht, The Ne<strong>the</strong>rlands; 363-372. (In: Chapter<br />
1)<br />
Kuhlmann U, Kühnemund F. (2002). Ductility <strong>of</strong> semi-rigid steel joints. In: Proceedings<br />
<strong>of</strong> <strong>the</strong> International Colloquium on Stability and Ductility <strong>of</strong> Steel<br />
Structures (SDSS 2002) (Ed.: M. Ivanyi), Budapest, Hungary; 363-370. (In:<br />
Chapters 1, 8)<br />
Kühnemund F. (2003). On <strong>the</strong> verification <strong>of</strong> <strong>the</strong> rotation capacity <strong>of</strong> semi-rigid<br />
joints in steel structures. PhD Thesis (in German), University <strong>of</strong> Stuttgart,<br />
Stuttgart, Germany. (In: Chapter 1)<br />
Kukreti AR, Ghasseimieh M, Murray JM. (1990). Behaviour and design <strong>of</strong> large<br />
capacity moment <strong>end</strong> <strong>plate</strong>s. Journal <strong>of</strong> Structural Engineering ASCE; 116(3):<br />
809-828. (In: Chapter 1)<br />
Kukreti AR, Murray JM, Abolmaali A. (1987). End <strong>plate</strong> connection momentrotation<br />
relationship. Journal <strong>of</strong> Constructional Steel Research; 8:137-157. (In:<br />
Chapter 1)<br />
Kukreti AR, Murray JM, Ghasseimieh M. (1989). Finite element modelling <strong>of</strong><br />
large capacity stiffened steel tee-hanger connections. Computers and Structures;<br />
32(2):409-422. (In: Chapter 1)<br />
Loureiro AJR. (2002). Effect <strong>of</strong> heat input on plastic deformation <strong>of</strong> underma<strong>the</strong>d<br />
welds. Journal <strong>of</strong> Materials Processing Technology; 128:240-249. (In: Chapter<br />
4)<br />
Lusas 13. (2001). Element reference manual. Finite element analysis Ltd, Version<br />
13.2. Surrey, UK. (In: Chapter 4)<br />
Lusas 13. (2001). Modeller reference manual. Finite element analysis Ltd, Version<br />
13.2. Surrey, UK. (In: Chapter 4)<br />
Lusas 13. (2001). Solver reference manual. Finite element analysis Ltd, Version<br />
13.2. Surrey, UK. (In: Chapter 4)<br />
Lusas 13. (2001). Theory manual. Finite element analysis Ltd, Version 13.2. Surrey,<br />
UK, 2001. (In: Chapter 4)<br />
Lusas 13. (2003). Theory manual. Finite element analysis Ltd, Version 13.5. Surrey,<br />
UK. (In: Chapter 6)<br />
Mann AP, Morris LJ. (1979). Limit design <strong>of</strong> ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong> connections.<br />
Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE; 105(ST3):511-526. (In: Chapter 1)<br />
Maquoi R, Jaspart JP. (1986) Moment-rotation curves for <strong>bolted</strong> connections: Discussion<br />
<strong>of</strong> <strong>the</strong> paper by Yee YL and Melchers RE. Journal <strong>of</strong> Structural Engineering<br />
ASCE; 113(10):2324-2329. (In: Chapter 6)<br />
McGuire W. (1968). Steel structures. Prentice-Hall International series in Theoretical<br />
and Applied Mechanics (Eds.: NM Newmark and WJ Hall). Englewood<br />
Cliffs, N.J., USA. (In: Chapter 6)<br />
Mistakidis ES, Baniotopoulos CC, Bisbos CD, Panagiotopoulos PD. (1996). A 2-<br />
D numerical model for <strong>the</strong> analysis <strong>of</strong> steel T-stub connections. Cost C1, Numerical<br />
simulation group, Doc. C1WD6/96-09. (In: Chapter 4)<br />
Mistakidis ES, Baniotopoulos CC, Bisbos CD, Panagiotopoulos PD. (1997). Steel<br />
T-stub connections under static loading: an effective 2-D numerical model.<br />
344
Journal <strong>of</strong> Constructional Steel Research; 44(1-2):51-67. (In: Chapter 4)<br />
Nair RS, Birkemoe PC, Munse WH. (1974). High strength bolts subject to tension<br />
and prying. Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE; 100(ST2):351-372. (In:<br />
Chapter 1)<br />
Ne<strong>the</strong>rcot DA, Zandonini R. (1989). Methods <strong>of</strong> prediction <strong>of</strong> joint behaviour:<br />
beam-to-column connections, Chapter 2 in Structural connections, stability and<br />
strength (Ed.: R. Narayanan). Elsevier Applied Science, London; 23-62. (In:<br />
Chapter 1)<br />
Owen DRJ, Hinton E. (1980). Finite elements in plasticity, <strong>the</strong>ory and practice.<br />
Pineridge Press Limited, Swansea, UK. (In: Chapter 4)<br />
Packer JA, Morris LJ. (1977). A limit state design method for <strong>the</strong> tension region <strong>of</strong><br />
<strong>bolted</strong> beam-to-column connections. The Structural Engineer; 55(10):446-458.<br />
(In: Chapter 1)<br />
Piluso V, Faella C, Rizzano G. (2001). Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. I:<br />
Theoretical model. Journal <strong>of</strong> Structural Engineering ASCE; 127(6):686-693.<br />
(In: Chapters 1-2, 6)<br />
Piluso V, Faella C, Rizzano G. (2001). Ultimate behavior <strong>of</strong> <strong>bolted</strong> T-stubs. II:<br />
Model validation. Journal <strong>of</strong> Structural Engineering ASCE; 127(6):694-704,.<br />
(In: Chapters 1, 4, 6, 9)<br />
RILEM draft recomm<strong>end</strong>ation. (1990). Tension testing <strong>of</strong> metallic structural materials<br />
for determining stress-strain relations under monotonic and uniaxial tensile<br />
loading. Materials and Structures; 23:35-46. (In: Chapters 3, 7)<br />
Rodrigues DM, Menezes LF, Loureiro A, Fernandes JV. (2004). Numerical study<br />
<strong>of</strong> <strong>the</strong> plastic behaviour in tension <strong>of</strong> welds in high strength steels. International<br />
Journal <strong>of</strong> Plasticity; 20:1-18. (In: Chapters 3-4)<br />
Sherbourne AN , Bahaari MR. (1994). 3D simulation <strong>of</strong> <strong>end</strong>-<strong>plate</strong> <strong>bolted</strong> connections.<br />
Journal <strong>of</strong> Structural Engineering ASCE; 120(11):3122-3136. (In: Chapter<br />
1)<br />
Simo JC, Rifai MS. (1990). A class <strong>of</strong> mixed assumed strain methods and <strong>the</strong><br />
method <strong>of</strong> incompatible modes. International Journal for Numerical Methods<br />
in Engineering; 29:1595-1638. (In: Chapter 4)<br />
Simões da Silva L, Calado L, Simões R, Girão Coelho A. (2000). Evaluation <strong>of</strong><br />
<strong>ductility</strong> in steel and composite beam-to-column joints: analytical evaluation.<br />
In: Connections in Steel Structures IV: Steel Connections in <strong>the</strong> New Millennium<br />
(Ed.: R. Leon), Roanoke, USA (available on CD). (In: Chapter 1)<br />
Simões da Silva LAP, Girão Coelho AM. (2000). Mode interaction in non-linear<br />
models for steel and steel-concrete composite structural connections. In: Proceedings<br />
<strong>of</strong> <strong>the</strong> Third International Conference on Coupled Instabilities in<br />
Metal Structures (CIMS’2000) (Eds.: D. Camotim, D. Dubina and J. Rondal),<br />
Lisbon, Portugal; 605-614. (In: Chapter 1)<br />
Simões da Silva LAP, Girão Coelho AM. (2001). A <strong>ductility</strong> model for steel connections.<br />
Journal <strong>of</strong> Constructional Steel Research; 57:45-70. (In: Chapter 1)<br />
Simões da Silva LAP, Girão Coelho AM, Neto EL. (2000). Equivalent postbuckling<br />
models for <strong>the</strong> flexural behaviour <strong>of</strong> steel connections. Computers<br />
and<br />
345
Structures; 77:615-624. (In: Chapter 1)<br />
Simões da Silva LAP, Girão Coelho AM, Simões RAD. (2001). Analytical<br />
evaluation <strong>of</strong> <strong>the</strong> moment-rotation response <strong>of</strong> beam-to-column composite<br />
joints under static loading. Steel and Composite Structures; 1(2):245-268,<br />
2001. (In: Chapter 1)<br />
Simões da Silva L, Santiago A, Vila Real P. (2002). Post-limit stiffness and <strong>ductility</strong><br />
<strong>of</strong> <strong>end</strong> <strong>plate</strong> beam-to-column steel joints. Computers and Structures;<br />
80:515-531. (In: Chapter 1)<br />
Swanson JA. (1999). Characterization <strong>of</strong> <strong>the</strong> strength, stiffness and <strong>ductility</strong> behavior<br />
<strong>of</strong> T-stub connections. PhD dissertation, Georgia Institute <strong>of</strong> Technology,<br />
Atlanta, USA. (In: Chapters 1, 4-6)<br />
Swanson JA, Kokan DS, Leon RT. (2002). Advanced finite element modelling <strong>of</strong><br />
<strong>bolted</strong> T-stub connection components. Journal <strong>of</strong> Constructional Steel Research;<br />
58:1015-1031. (In: Chapter 4)<br />
Thornton WA. (1985). Prying action – a general treatment. Engineering Journal<br />
AISC; 22(2):67-75. (In: Chapter 1)<br />
van der Vegte GJ, Makino Y, Sakimoto T. (2002). Numerical research on single<strong>bolted</strong><br />
connections using implicit and explicit solution techniques. Memoirs <strong>of</strong><br />
<strong>the</strong> Faculty <strong>of</strong> Engineering Kumamoto University, XXXXVII(1):19-44. (In:<br />
Chapter 4)<br />
Vasarhelyi DD, Chiang KC. (1967). Coefficient <strong>of</strong> friction in joints <strong>of</strong> various<br />
steel. Journal <strong>of</strong> Structural Division ASCE; 93(ST4):227-243. (In: Chapter 4)<br />
Virdi KS. (1999). Guidance on good practice in simulation <strong>of</strong> semi-rigid connections<br />
by <strong>the</strong> finite element method. In: Numerical simulation <strong>of</strong> semi-rigid connections<br />
by <strong>the</strong> finite element method (Ed.: K.S. Virdi). COST C1, Report <strong>of</strong><br />
working group 6 – Numerical simulation, Brussels; 1-12. (In: Chapter 4)<br />
Wanzek T, Gebbeken N. (1999). Numerical aspects for <strong>the</strong> simulation <strong>of</strong> <strong>end</strong> <strong>plate</strong><br />
connections. In: Numerical simulation <strong>of</strong> semi-rigid connections by <strong>the</strong> finite<br />
element method (Ed.: K.S. Virdi). COST C1, Report <strong>of</strong> working group 6 –<br />
Numerical simulation, Brussels; 13-31. (In: Chapter 4)<br />
Weynand K, Jaspart JP, Steenhuis M. (1995). The stiffness model <strong>of</strong> revised Annex<br />
J <strong>of</strong> Eurocode 3. In: Proceedings <strong>of</strong> <strong>the</strong> Third International Workshop on<br />
Connections in Steel Structures III (Eds.: R. Bjorhovde, A. Colson and R. Zandonini),<br />
Trento, Italy; 441-452. (In: Chapter 1)<br />
Weynand K. (1996). Sicherheits-und Wirtsschaftlichkeitsuntersuchungen zur anw<strong>end</strong>ung<br />
nachgiebiger anschlüsse im stahlbau. PhD <strong>the</strong>sis (in German). University<br />
<strong>of</strong> Aachen, Aachen, Germany. (In: Chapters 1, 7)<br />
Witteveen J, Stark JWB, Bijlaard FSK, Zoetemeijer P. (1982). Welded and <strong>bolted</strong><br />
beam-to-column connections. Journal <strong>of</strong> <strong>the</strong> Structural Division ASCE;<br />
108(ST2):433-455. (In: Chapter 1)<br />
Yee YL, Melchers RE. (1986). Moment-rotation curves for <strong>bolted</strong> connections.<br />
Journal <strong>of</strong> Structural Engineering ASCE; 112(3):615-635. (In: Chapters 1, 6)<br />
Zajdel M. (1997). Numerical analysis <strong>of</strong> <strong>bolted</strong> tee-stub connections. TNO-Report<br />
97-CON-R-1123. (In: Chapter 4)<br />
346
Zandonini R, Zanon P. (1988). Experimental analysis <strong>of</strong> <strong>end</strong> <strong>plate</strong> connections. In:<br />
Proceedings <strong>of</strong> <strong>the</strong> First International Workshop on Connections in Steel<br />
Structures, Behaviour, Strength and Design (Eds.: R. Bjorhovde, J. Brozzetti<br />
and A. Colson), Cachan, france; 40-51. (In: Chapters 1, 8)<br />
Zoetemeijer P. (1974). A design method for <strong>the</strong> tension side <strong>of</strong> statically loaded,<br />
<strong>bolted</strong> beam-to-column connections. Heron; 20(1):1-59. (In: Chapter 1)<br />
Zoetemeijer P. (1990). Summary <strong>of</strong> <strong>the</strong> research on <strong>bolted</strong> beam-to-column connections.<br />
Report 25-6-90-2. Faculty <strong>of</strong> Civil Engineering, Stevin Laboratory –<br />
Steel Structures, Delft University <strong>of</strong> Technology. (In: Chapters 1-2, 8)<br />
Zoetemeijer P, Munter H. (1983). Ext<strong>end</strong>ed <strong>end</strong> <strong>plate</strong>s with disappointing rotation<br />
capacity – Test results and analysis. Stevin Laboratory Report 6-83-13. Faculty<br />
<strong>of</strong> Civil Engineering, Delft University <strong>of</strong> Technology. (In: Chapter 1)<br />
Zoetemeijer P, Munter H. (1983). Proposal for <strong>the</strong> standardization <strong>of</strong> ext<strong>end</strong>ed <strong>end</strong><br />
<strong>plate</strong> connections based on test results – Test and analysis. Stevin Laboratory<br />
Report 6-83-23. Faculty <strong>of</strong> Civil Engineering, Delft University <strong>of</strong> Technology.<br />
(In: Chapter 1)<br />
Zupančič E, Beg D, Vayas I. (2002). Deformation capacity <strong>of</strong> components <strong>of</strong> moment<br />
resistant connections. European Convention for Constructional Steelwork<br />
– Technical Committee 10, Structural Connections (ECCS-TC10), Document<br />
ECCS-TWG 10.2-02-005. (In: Chapter 1)<br />
347