SEI-november-2010-journal

SEI Volume 20 | Number 4 | November 2010 

STRUCTURAL ENGINEERING 

INTERNATIONAL 

International Association for Bridge and Structural Engineering (IABSE) 

Fibre Reinforced 

Polymer Composites

Bridge Information Modeling – 

It’s everything to do with the bridge. 

Bridge Information Modeling is a compelling new methodology for project delivery that dramatically improves bridge quality and 

reduces risk. Using the Bentley bridge solution for BrIM, engineers develop and employ an unprecedented depth of information 

about the bridge as they streamline the entire bridge development process, design through construction engineering. 

Organizations achieve the highest level of real-time team collaboration using the Bentley solution. They get the right information 

in the right hands and ultimately empower project stakeholders to collectively deliver the best-quality, most constructible bridges 

possible. They also deliver an invaluable bridge information asset to the owner for lifetime operation of the bridge. 

With our leading bridge design and engineering systems, RM Bridge and LEAP Bridge, projects evolve in a synthesis of modeling, 

analysis, and design. Engineers perform an array of intensive analyses of any condition, from seismic impact to wind dynamics and 

even rolling stock analysis – for every stage of construction. 

Join the world’s top engineering consultancies. Choose the Bentley bridge solution for BrIM – and the satisfaction of delivering safe, 

sustainable bridges on time and on budget. 

Only Bentley can take you there. www.bentley.com/SEI 

© 2010 Bentley Systems, Incorporated. Bentley, the “B” Bentley logo, and LEAP are either registered or unregistered trademarks or service marks of Bentley Systems, Incorporated or one of its direct or indirect whollyowned 

subsidiaries. Other brands and product names are trademarks of their respective owners.

idges - linking people ® 

Bandra Worli Sea Link, India 

Equipped with 254 DISKTRON disc bearings (Vmax = 

14,000 kN) and 26 mageba TENSA ® MODULAR expansion 

joints of types LR2 to LR6 (max. movement: 480mm). 

Canada 

China 

Croatia 

Denmark 

Finland 

France 

Germany 

Hong Kong 

Poland 

Portugal 

Slovenia 

South Korea 

Switzerland 

USA 

structural bearings seismic devices 

expansion joints monitoring & services 

Switzerland www.mageba.ch

Sika’s Advanced Technologies 

for Bridge Engineering 

Sika ® ViscoCrete ® 

High performance concretes using 

Sika ® ViscoCrete ® Admixture 

Technology 

• Self compacting 

• Fast /Controlled setting 

• Watertight and corrosion resistant 

SikaGrout ® 

Long lasting cable ducts using 

SikaGrout ® PT-Technology 

• Factory quality controlled 

• Waterproof 

• Corrosion resistant 

Sika ® CarboDur ® 

Structural strengthening using 

Sika ® CarboDur ® Technology 

• CFRP plate systems 

• Fabric systems 

• Patented prestressing systems 

• CarboShear ® system 

Sika Services AG 

Business Unit Constractors 

CH-8330 Pfäffikon 

Switzerland 

Phone +41 58 436 23 79 

Fax: +41 58 436 23 77 

www.sika.com 

Innovation & 

Consistency 

since 

1910

Contents 4/2010 

Information on SEI 

www.iabse.org/sei 

SEI Advisory Board 

J.-E. Breen, W.F. Chen, Y. Fujino, 

N.J. Gimsing, P.R. Head, M. A. Hirt, 

D.A. Nethercot, M.J. Priestley, 

J. Schlaich, P. Taylor, M. Virlogeux, 

J.C. Walraven, H.F. Xiang. 

SEI Editorial Board 

H.H. Snijder, Chair; A. Schumacher, 

Vice Chair; M.G. Bruschi, A. Frangi, 

N.P. Hoej, K. Sugiura, D.Xu 

Correspondents 

China: D. Xu. Denmark: M. Braestrup. 

Egypt: F. Saad. Finland: 

M.-K. Söderqvist. France: B. Godart. 

Germany: U. Kuhlmann. India: V. Kumar. 

Italy: G. Bignotti, L. Ceriolo. 

Japan: K. Fujita. Korea: H.K. Kim. 

Norway: L. Toverud. Poland: W. Radomski. 

Russia: S.V. Mozalev. Sweden: 

H. Sundquist. Thailand: E. Limsuwan. 

UK: D.K. Doran. USA: J. Burns, 

D. Frangopol. 

Publisher 

IABSE 

ETH Zurich 

8093 Zurich, Switzerland 

Tel: 41-44-633 2647 

Fax: 41-44-633 1241 

secretariat@iabse.org 

www.iabse.org 

Publications Manager 

Brindarica Bose, IABSE 

Advertising Inquiries 

Sissel Niggeler, IABSE 

Published 

Quarterly: 1 Feb., 1 May, 1 Aug., 1 Nov. 

Subscription 2011 

Included in IABSE Membership. 

220 CHF: Individual Subscription 

630 CHF: Institutional Subscription 

Available through subscription agencies. 

ISSN 1016-8664, E-ISSN 1683-0350 

Copyright © IABSE. All rights reserved. 

Opinions and positions expressed in signed 

articles are those of the authors and are not 

necessarily those of Structural Engineering 

International or IABSE. 

Front cover: 

Wolchul Mountain Bridge, Korea 

See article on page 405 

Structural Engineering International 

International Association for Bridge and Structural Engineering 

Editorial 

Strengthening IABSE; P. L. Popovic; USA 361 

Special Feature: Fibre Reinforced Polymer Composites 

Scientific Papers* 

Introduction: Fiber Reinforced Polymer (FRP) Composites; A. Schumacher; Switzerland, 

M. D. G. Pulido; Spain 362 

Glass Fibre Reinforced Polymer Pultruded Flexural Members: Assessment of Existing 

Design Methods; J. R. Correia, F. Branco, J. Gonilha, N. Silva, D. Camotim; Portugal 362 

Effects of Hygrothermal Ageing on the Mechanical Properties of Glass-Fibre-Reinforced 

Polymer Pultruded Profiles; J. R. Correia, S. Cabral-Fonseca, A. Carreiro, R. Costa, 

M. P. Rodrigues, I. Eusébio, F. Branco; Portugal 370 

Evaluation of a Life Prediction Model and Environmental Effects of Fatigue for Glass Fiber 

Composite Materials; D. B. Dittenber, G. V. S. Hota; USA 379 

A Composite Bridge is Favoured by Quantifying Ecological Impact; R. A. Daniel; The Netherlands 385 

Experimental Assessment of Bond Behaviour of Fibre-Reinforced Polymers on Brick Masonry; 

E. Garbin, M. Panizza, M. Valluzzi; Italy 392 

Bridges with Glass Fibre–Reinforced Polymer Decks: The Road Bridge in Friedberg, Germany; 

J. Knippers, E. Pelke, M. Gabler, D. Berger; Germany 400 

Technical Reports 

Current and Future Applications of Glass-Fibre-Reinforced Polymer Decks in Korea; S. W. Lee, 

K. J. Hong, S. Park; Korea 405 

Field Issues Associated with the Use of Fiber-Reinforced Polymer Composite Bridge Decks 

and Superstructures in Harsh Environments; L. N. Triandafilou, J. S. O’Connor; USA 409 

Examples of Applications of Fibre Reinforced Plastic Materials in Infrastructure in Spain; 

A. Bansal, J. F. M. Cano, B. O. O. Muñoz, C. Paulotto; Spain 414 

Fiber-Reinforced Polymer Decks for Movable Bridges; R. D. Bottenberg; USA 418 

Glass Fiber Reinforced Polymer Strengthening and Evaluation of Railroad Bridge Members; 

G. V. S. Hota, P. V. Vijay, R. S. Abhari; USA 423 

Design of the St Austell Fibre-Reinforced Polymer Footbridge, UK; J. Shave, S. Denton, 

I. Frostick; UK 427 

General 

Scientific Papers* 

Aluminium Structures in Building and Civil Engineering Applications; F. Soetens; The Netherlands 430 

Glass Tensegrity Trusses; M. Froli, L. Lani; Italy 436 

A Simplified Serviceability Assessment of Footbridge Dynamic Behaviour Under Lateral 

Crowd Loading; L. Bruno, F. Venuti; Italy 442 

Technical Reports 

The Construction of the Main Bridge of the Yichang Yangtze River Railway Bridge in China; 

Y. Zhou, L. Zhang; China 447 

Static and Dynamic Analysis of the “Piedra Movediza” Replica Rock, Argentina; M. I. Montanaro, 

M. H. Peralta, N. Ercoli, M. L. Godoy, I. Rivas; Argentina 451 

Footbridge Studenci over the Drava River in Maribor, Slovenia; V. Markelj; Slovenia 454 

Sanchaji Bridge: Three-Span Self-Anchored Suspension Bridge, China; G. Dai, X. Song, N. Hu; China 458 

The First Extradosed Bridge in Slovenia; V. Markelj; Slovenia 462 

Recent PhD Abstracts 468 

Eminent Structural Engineer 

Christian Menn—Bridge Designer and Builder; E. Brühwiler; Switzerland 470 

Panorama 

IABSE Annual Meetings 473 

Predrag (Pete) Popovic, USA, New President of IABSE 473 

IABSE Awards 2010 474 

The IABSE Foundation Anton Tedesko Medal 479 

IABSE Symposium Venice, September 22–24, 2010 479 

Dhaka Conference ‘Advances in Bridge Engineering-II’ 482 

Calendar of Events and IABSE Members’ Business Cards 483 

IABSE Membership Application Form 484 

*Peer-reviewed papers 

Abstracting and Indexing: This publication is abstracted in Cambridge Scientific Abstracts under CSA Civil 

Engineering Abstracts; Emerald Abstracts; Construction and Building Abstracts (CBA); CAB Abstracts; INSPEC; 

and is included in EBSCOhost and SwetsWise Online Content. For SEI content Photocopying, Electronic usage, in 

the USA: Contact Copyrights Clearance Centre (CCC) at www.copyrights.com 

In rest of the world: Contact IABSE, at secretariat@iabse.org 

Structural Engineering International 4/2010 359

Building Asia together 

Holcim is building the very foundations of modern life and as a leading supplier of building 

materials in Asia we are strongly committed to the region. 

Global expertise and know-how, local market excellence and can-do attitude provide the 

strongest foundation for future growth. As with the Phu My Bridge, opened in September 

2009, that brought long awaited relief to Ho Chi Minh City, Vietnam. 

That’s what it takes to build with confidence in the most dynamic region in the world. We do 

this with respect for both the environment and the local communities where we operate. 

www.holcim.com 

Strength. Performance. Passion.

Strengthening IABSE 

This issue of Structural Engineering International contains a series of articles 

on strengthening structures with fiber-reinforced polymers (FRP). It seems 

appropriate that this Editorial should also be about strengthening—in particular 

strengthening IABSE. 

The Executive Committee under the leadership of Jacques Combault has been 

working to update IABSE’s long-range plan and is expecting to complete its work 

by the end of 2010. The plan will outline strategies to strengthen IABSE. 

IABSE offers a well-established network for the comprehensive dissemination of 

knowledge to structural engineers worldwide through excellent conferences and 

publications such as Structural Engineering International (SEI) and Structural 

Engineering Documents (SED). Technical Commissions, Working Groups, and 

E-learning present additional platforms for exchange of technical information 

and ideas. However, despite these strengths, IABSE membership has not grown. 

Currently, IABSE has about 3600 members in 100 countries. 

There are several obvious reasons why our membership level has plateaued, 

including: competition from national technical societies and international societies 

focused on particular structure types or materials; lack of awareness among potential 

members of IABSE and the benefits it offers; high costs to attend conferences 

and pay membership fees for many members from developing countries; limited 

number of young engineers and students joining IABSE; and of course, the current 

world economic crisis causing a decrease in membership renewals. This latter trend 

could also have an impact over time on IABSE finances. 

To spur membership growth and in effect “strengthen” IABSE, we need to increase 

the visibility and relevance of IABSE to structural engineers from around the world. 

Stronger IABSE finances will follow increased membership. Our goal should be to 

attract 1000 new IABSE members over the next three years or to increase the net 

membership by at least 500. This will require all of our involvement and special 

efforts by National Groups. Specific strategies to address these challenges will be 

communicated as our long term plan is finalized. 

I am optimistic that together we will, despite the current economic environment, be 

able to strengthen and improve our organization over the next several years and 

that we will enjoy this journey together. 

Predrag L. Popovic 

President, IABSE 

Structural Engineering International 4/2010 Editorial 361

Introduction: Fiber Reinforced Polymer (FRP) Composites 

Fiber reinforced polymer (FRP) composites can be considered 

a new class of construction material when compared 

with classical materials such as steel, concrete, timber and 

masonry. The relatively recent and growing interest in FRP 

in the domain of structural engineering can be traced to its 

advantageous properties ranging from a very high strengthto-weight 

ratio, electromagnetic neutrality, excellent fatigue 

behaviour, to superior durability including corrosion resistance. 

These properties have, in turn, lead to a broad spectrum 

of application that can be divided into two general 

categories: all-FRP members or structures in new construction 

or in the replacement of existing structural elements, 

and FRP components in the repair and rehabilitation of 

damaged or deteriorating structures. 

Structural Engineering International received an overwhelming 

response from around the world to its call for 

papers on the topic of FRP structures and strengthening of 

structures using FRP. The number of abstracts submitted, 

and subsequent high-quality papers received, has prompted 

the extension of this Special Edition over two issues—the 

present issue, as well as the coming May 2011 issue. In this 

first issue, six Scientific Papers on topics including existing 

design method assessments for FRP members, durability, 

environmental and fatigue issues for glass fiber reinforced 

polymer composites (GFRP), ecological advantages of FRP 

as compared with other materials, bond issues related to 

the use of FRP in the strengthening of masonry structures, 

and GFRP decks for bridges are presented. The Scientific 

Papers are complemented by six Technical Reports ranging 

from descriptions on the innovative use of FRPs in bridge 

decks to the application of GFRP in the strengthening of 

rail road bridges. 

Dr. Ann Schumacher, Vice-Chair SEI Editorial Board, 

Swiss Institute for Steel Construction, Switzerland 

Prof. M. Dolores G. Pulido, Chair WG 2 - Fiber 

Reinforced Polymer (FRP) Structures, 

Spanish National Research Council – Instituto 

CC Eduardo Torroja, Spain 

Glass Fibre Reinforced Polymer Pultruded Flexural 

Members: Assessment of Existing Design Methods 

João R. Correia, Prof. Dr, Technical Univ. of Lisbon, Instituto Superior Técnico/ICIST, Civil Eng. and Architecture, 

Lisbon, Portugal; Fernando Branco, Prof. Dr, Technical Univ. of Lisbon, Instituto Superior Técnico/ICIST, Civil Eng. and 

Architecture, Lisbon, Portugal; José Gonilha, Civil Eng., Technical Univ. of Lisbon, Instituto Superior Técnico/ICIST, 

Civil Eng. and Architecture, Lisbon, Portugal; Nuno Silva, Civil Eng., Technical Univ. of Lisbon, Instituto Superior 

Técnico/ICIST, Civil Eng. and Architecture, Lisbon, Portugal; Dinar Camotim, Prof. Dr, Technical Univ. of Lisbon, 

Instituto Superior Técnico/ICIST, Civil Eng. and Architecture, Lisbon, Portugal. Contact: jcorreia@civil.ist.utl.pt 

Abstract 

Glass fibre reinforced polymer (GFRP) 

pultruded profiles are being increasingly 

used in bridge and building construction 

as an alternative to traditional 

materials because of their several 

favourable properties that include high 

strength, low self-weight, short installation 

times, low maintenance requirements 

and improved durability. In spite 

of these advantageous characteristics, 

there are some factors delaying the 

widespread use of GFRP pultruded 

profiles in civil infrastructure, one of 

which is the lack of widely accepted 

design codes. This paper presents the 

Peer-reviewed by international experts 

and accepted for publication 

by SEI Editorial Board 

Paper received: February 19, 2010 

Paper accepted: July 28, 2010 

results of analytical, experimental and 

numerical investigations on the structural 

behaviour of GFRP pultruded 

profiles, the objective of which was to 

evaluate the relative accuracy of existing 

design methods. A survey of analytical 

formulae available for the design 

of GFRP pultruded flexural members 

at both service and ultimate limit states 

is first presented. Subsequently, results 

of a test programme carried out at 

Instituto Superior Técnico (IST) are 

briefly discussed—the experiments 

included material characterization tests 

and full-scale flexural tests on I-section 

simply supported beams and cantilevers. 

These tests allowed for the evaluation 

of the service behaviour of GFRP 

flexural members and some of their 

most relevant failure mechanisms and 

respective ultimate loads. Results from 

experimental tests are compared with 

those obtained from analytical formulae 

and numerical models in order to 

evaluate the relative accuracy of existing 

design methods. 

Keywords: GFRP pultruded profiles; 

service behaviour; local buckling; 

global buckling; design methods; analytical 

formulae; numerical models. 

Introduction 

The limited durability of structures 

made with traditional materials and 

their consequent rehabilitation costs, 

which have substantially increased in 

the past few years, have been promoting 

the development of new structural 

materials that are less prone to corrosion, 

lighter and easier to erect. In this 

context, in the last two decades, fibre 

reinforced polymer (FRP) materials 

in general, and glass fibre reinforced 

polymer (GFRP) pultruded profiles 

in particular, have found a growing 

number of applications in buildings 

362 Scientific Paper Structural Engineering International 4/2010

and bridges, in both new constructions 

and rehabilitation of degraded 

infrastructures. 1–10 

GFRP pultruded profiles have great 

potential as structural materials, presenting 

several advantages over traditional 

materials because of their 

high strength-to-weight ratio, low 

self-weight, electromagnetic transparency, 

possibility of being produced 

with any cross section, ease of installation, 

low maintenance requirements 

and improved durability under aggressive 

environments. 11 The drawback, 

in addition to the initial costs, lack of 

competitiveness for mainstream applications 

and the concerns regarding 

their behaviour under fire, 12,13 is that 

there are still no generally accepted 

design codes or guidelines available 

for civil engineering practitioners. As 

a consequence, at present, most structural 

designs are based on manufacturers’ 

design guides, often presented in 

a tabular format, which are sometimes 

incomplete and over-conservative. 

The Eurocomp Design Code and 

Handbook, 14 published in 1996, provides 

design recommendations for 

polymer composites in general, but 

this non-normative document does not 

specifically address pultruded members. 

In 2002, the European Committee 

for Standardization (CEN) released 

the EN 13706 standard, 15 a normative 

document that merely defines two 

classes of pultruded profiles (associated 

with minimum values of material 

properties), not providing any 

design guidance. In 2007, the Italian 

National Research Council published 

the first national design guidelines 

for structures made of pultruded profiles; 

16 however, these specifications 

are mandatory only in Italy. It is also 

worth mentioning that most textbooks 

on the mechanics of composite materials 

and composite structures refer to 

aerospace and mechanical engineering 

applications—with the exception 

of a recent publication by Bank, 17 

which provides a comprehensive set of 

design rules for FRP structures, written 

in a civil engineering format. 

Before a comprehensive and widely 

accepted set of design rules and recommendations 

can be established 

for the use of GFRP pultruded profiles, 

further research work is needed 

to obtain in-depth understanding 

of their structural behaviour and 

to provide additional validation for 

the design methods that have been 

proposed. 

This paper presents the results of analytical, 

experimental and numerical 

investigations on the structural behaviour 

of GFRP pultruded profiles, the 

objective of which was to evaluate the 

relative accuracy of existing design 

methods. A survey of analytical formulae 

that have been suggested for 

the design of GFRP pultruded flexural 

members, for both service and 

ultimate limit states, is first presented. 

Subsequently, results of a test programme 

carried out at IST are briefly 

discussed—the experiments included 

material characterization tests on 

small-scale coupons and full-scale flexural 

tests on I-sections of simply supported 

GFRP beams and cantilevers. 

These tests, which are described in 

detail in Refs. [18, 19], allowed evaluation 

of the service behaviour of GFRP 

flexural members and some of their 

most relevant failure mechanisms and 

respective ultimate loads. The results 

from these experimental tests are then 

compared with predictions obtained 

from both analytical formulae and 

numerical models, in order to evaluate 

the relative accuracy of existing design 

methods. 

Design Methods for GFRP 

Flexural Members 

The design of structures made of 

GFRP pultruded profiles can be performed 

in much the same way as that 

of steel structures, provided that some 

necessary adaptations are taken into 

account, the most important of which 

are the orthotropic nature and linear 

elastic behaviour of the GFRP 

material. 

Thereafter, the structural design of 

standard GFRP profiles can be performed 

on the basis of either analytical 

beam models or shell and/or solid 

finite element (FE) models. For the 

former approach, which is most currently 

used in the design of GFRP 

frames and trusses, a simplified equivalent 

isotropic behaviour is assumed. 

For the latter approach, the orthotropic 

nature of the GFRP material is 

explicitly taken into account. 

Serviceability Limit States 

For service limit states design, the 

bending deflections of pultruded 

flexural members can be determined 

with a reasonable accuracy using 

analytical beam models based on the 

Timoshenko beam theory, that is, 

taking into account the shear contribution 

to overall deformation. In fact, 

shear deformations can be relatively 

important owing to the high elasticto-shear 

moduli ratio. For example, 

the elastic short-term deflection of a 

simply supported beam with a point 

load at midspan (similar to the beams 

whose experiments are reported later) 

can be calculated using Eq. (1), 

3 

P⋅L 

P⋅L 

δ = + 

48⋅E 

⋅I 

4⋅G 

⋅A 

full 

x 

full 

w 

(1) 

where d is the midspan deflection, P 

is the applied load, L is the span, I x is 

the second moment of area about the 

strong axis x, A w is the web(s) cross section 

and E full and G full are the full-scale 

longitudinal elastic and shear moduli 

for an equivalent isotropic behaviour, 

which can be determined on the basis 

of experiments (see next section). 

In order to evaluate long-term deflections 

in pultruded beams, it is necessary 

to address the viscoelastic response 

associated with the polymeric nature 

of the matrix properly. Therefore, 

time-dependent deformations due to 

sustained loads must be calculated taking 

into account the viscoelastic values 

of the full-scale moduli in Eq. (1). In 

Ref. [17], Bank presents a set of creep 

moduli and creep rate exponents recommended 

for design, which were 

obtained from long-term creep tests 

using the linearized version of Findley’s 

creep theory. 

Ultimate Limit States 

For ultimate limit states design, the fact 

that GFRP flexural members can theoretically 

collapse due to several failure 

modes must be taken into account. For 

the most commonly produced geometries 

(thin-walled open sections), 

the following failure mechanisms can 

occur: (a) flexural (tensile or compressive) 

failure; (b) web shear failure; (c) 

web transverse crushing; (d) local buckling; 

and (e) lateral-torsional buckling. 

Flexural Failure 

The bending moment associated with 

flexural failure of a pultruded member 

(M u ) can be calculated using Eq. (2), 

Mu = σ 

x , u⋅W 

(2) 

x 

where s x,u is the longitudinal failure 

stress (either compressive or tensile) 

of the GFRP material and W x is the 

cross-section elastic modulus about the 

strong axis. It is worth mentioning that 

flexural failure, due to compressive 

Structural Engineering International 4/2010 Scientific Paper 363

crushing or tensile rupture, is not 

likely to occur for most common pultruded 

shapes, unless local buckling is 

prevented by an adequate stiffening 

system. 17 

Shear Failure 

The critical shear force (V u ) of a pultruded 

flexural member can be calculated 

using Eq. (3): 

V 

u 

τ 

u⋅I 

= 

S 

x 

x 

⋅t 

≈τ ⋅A 

(3) 

u 

where t u is the in-plane shear strength 

of the pultruded material, S x is the first 

moment of area about the strong axis, 

t is the laminate (web/flanges) thickness 

and A v is the shear area which, 

for most common profiles, corresponds 

to the web(s) of the profile. It should 

be noted that, similar to flexural failure, 

shear failure of the web material 

due to in-plane shear stresses seldom 

occurs, 17 as the strength of current 

cross sections and spans is dominated 

by buckling phenomena. 

Web Transverse Crushing 

The web(s) of GFRP pultruded 

beams can fail because of transverse 

crushing basically at two locations: 

(a) in the supports and (b) under concentrated 

loads. The critical crushing 

force (F crush u ) can be determined using 

Eq. (4): 

F 

crush 

u 

v 

c 

≈σ , 

⋅A 

(4) 

y u 

eff 

where s c y,u can be taken as the transverse 

compressive strength and A eff is 

the effective cross section of the web 

subjected to the concentrated load, 

that is, the area of the web directly 

subjected to the support reaction or 

concentrated load. In order to avoid 

this failure mechanism, the lengths of 

the supports or loading patches can be 

increased and, in addition, web stiffeners 

can be used. 

Local Buckling due to In-Plane 

Compression 

Local buckling is an instability phenomenon 

characterized by transverse 

(flexural) bending of the member 

walls while the axis remains basically 

undeformed. Besides the high widthto-thickness 

ratios typically exhibited 

by thin-walled members made of any 

material (e.g. steel), GFRP pultruded 

profiles exhibit an added susceptibility 

to local buckling because of their 

reduced in-plane moduli. For the most 

common doubly symmetric profiles, 

the critical local buckling stress of 

flanges under compression can be 

determined using two alternative 

design formulae, derived by Kollár 20 

and by Mottram 21 —Eqs. (5) and (6), 

respectively, 

(Kollár 20 ) 

local 

σ cr 

= 

( ) 

2 

b / 2 ⋅ 

⎛ 

⎜ 

⎝ 

f 

1 

t 

L 

7 1 412 

f 

+ , ⋅ 

× 

D ⋅ D 

T 

ξ I-flange 

⎞ 

+ 12 ⋅D S⎟ 

(5) 

⎠ 

local 

where σ cr 

is the critical local buckling 

stress, b f is the flange width, t f is 

the flange thickness, D L , D T and D S 

are the longitudinal, transverse and 

shear flexural rigidities of the flange 

plate and x I-flange is the coefficient of 

edge restraint (assuming the flange is 

the critical wall); (Mottram 21 ) 

σ 

local 

cr 

2 2 

π ⋅tf 

= × 

2 

( b / 2 

f ) 

⎡ 

2 

⎛ b ⎞ E ⋅ E ⎤ 

f 

L T 

⎢⎜ 

045 , + 

2 ⎟ 

⎥ 

⎢⎝ 

4a 

⎠ 12( 1−ν ⋅ν 

⎣ 

L T) 

⎥ 

⎦ 

(6) 

where E L and E T are the in-plane 

longitudinal and transverse moduli, 

n L and n T are the major and minor 

Poisson’s ratios and a is the length of 

the buckle half-wavelength which, for 

I-section profiles, is suggested 21 to be 

taken as 3b f . 

With regard to the above-mentioned 

formulae, it should be mentioned 

that the use of Kollar’s design equations 

involves knowing all the in-plane 

properties (including the in-plane 

shear modulus) of both the web(s) 

and the flanges. Mottram’s alternative 

simplified procedure makes use of the 

flange’s properties only. 

Lateral-Torsional Buckling 

The critical lateral-torsional buckling 

stress for homogeneous doubly symmetric 

open profiles can be determined 

using the well-known Eurocode 

3 equation, adapted to the GFRP 

material orthotropy—Eq. (7): 

global 

Cb 

σ = × 

cr 

Sx 

2 

4 2 

π ⋅E ⋅I ⋅G ⋅J 

π ⋅ E ⋅I 

⋅C 

L y LT 

L y w 

+ 

2 

2 2 

( k ⋅ L 

f b) 

( k ⋅ L ) ( k ⋅ L 

f b w b) 

(7) 

where C b is a coefficient accounting 

for moment variation along the 

beam length, S x is the section modulus 

about the strong axis, I y is the second 

moment of area about the weak axis, 

J is the torsional constant, C w is the 

warping constant, k f is the effective 

length coefficient for flexural buckling 

about the weak axis (k f = 1,0 for simply 

supported beams, such as those tested 

in the experiments reported herein), 

k w is the effective length coefficient 

for torsional buckling of the section (in 

general, for simply supported beams, 

k w can be taken as 1,0) and L b is the 

unbraced length of the beam. 

For the particular case of cantilevers 

loaded at the shear centre of 

their extremity section, the critical 

lateral-torsional buckling load can be 

predicted using design formulae proposed 

by Timoshenko and Gere, 22 

assuming no warping at the fixed end 

and adapted to the GFRP material 

orthotropy—Eq. (8): 

EIG J 

global L y LT 

Pcr 

= γ 2 

(8) 

2 

L 

where P cr is the critical lateral-torsional 

buckling load, and g 2 is a dimensionless 

factor depending on the torsional 

and warping rigidities. 

Experimental Assessment of 

the Design Methods 

As already mentioned, the establishment 

of consensual design approaches 

is dependent on further validation of 

the existing design methods and, most 

likely, on the development of new 

methodologies. In order to contribute 

to achieving this goal, a research effort 

was conducted at IST, which consisted 

of a fairly extensive experimental 

investigation, described in detail 

in Refs. [18, 19], whose results were 

then compared with several different 

types of numerical simulations. 23 

In this study, the experimental results 

obtained in the above investigation 

are used to assess the accuracy of the 

design methods described earlier. 

The experimental investigation 

involved pultruded GFRP I-beams (a) 

made of an isophthalic polyester matrix 

reinforced with E-glass fibre rovings 

and mats (inorganic content of 62%, 

by weight) and (b) exhibiting the following 

nominal dimensions: web height 

of 200 mm, flange width of 100 mm 

and thickness of 10 mm. The experimental 

study comprised (a) material 

characterization tests, to evaluate the 

mechanical properties and response of 

the GFRP material; (b) flexural tests 

on simply supported beams, aimed at 


evaluating their behaviour under service 

and failure conditions (including 

web transverse crushing and local and 

lateral-torsional buckling); and (c) flexural 

tests on cantilevers to investigate 

their lateral-torsional buckling behaviour 

and failure under tip-point loads 

applied at different locations. Each test 

type is addressed individually in the 

following sections. After providing the 

experimental set-up and procedure, the 

relevant results are outlined and used to 

assess the quality (accuracy and safety) 

of the corresponding design method. 

Material Characterization Tests 

Interlaminar shear tests (ASTM 

D2344) were first conducted on 

specimens with nominal dimensions 

of 9,8 × 20,0 × 60,0 mm 3 , applying a 

concentrated load at the centre of a 

45,0 mm span, in order to determine 

the interlaminar shear strength (F sbs ). 

Three-point bending tests (ISO 14125) 

were then performed on specimens 

with nominal dimensions of 9,8 × 15,0 

× 300 mm 3 , tested in the longitudinal 

direction (L), in order to determine 

the flexural strength (s fu,L ), the elastic 

modulus in bending (E f,L ) and the 

strain at failure (e fu,L ). Tensile tests 

(ISO 527-1,4) were also performed, 

using specimens with nominal dimensions 

of 9,8 × 15,0 × 350 mm 3 , loaded 

in their longitudinal direction, allowing 

measurement of the tensile strength 

(s tu,L ), the strain at failure (e tu,L ), the 

elastic modulus in tension (E t,L ) and 

the Poisson’s ratio (n LT ). Finally, compressive 

tests (ASTM D695) were carried 

out on specimens with nominal 

dimensions of 9,8 × 12,7 × 39,0 mm 3 , 

in order to determine, for both longitudinal 

(L) and transverse (T) directions, 

the compressive strength (s cu,L and 

s cu,T ), the strain at failure (e cu,L and 

e cu,T ) and the elastic modulus in compression 

(E c,L and E c,T ). 

In all mechanical tests, the material 

generally exhibited linear-elastic 

behaviour until failure, a typical feature 

of the GFRP material. 18,19 The 

failure modes observed in the different 

mechanical tests are illustrated in 

Fig. 1. Table 1 presents a summary of 

the mechanical properties obtained in 

these tests (which will be later used 

as input data in the analytical and 

numerical design methods), namely, 

the ultimate stress (s u ), the elastic 

modulus (E), the strain at failure (e u ), 

the Poisson’s ratio (n LT ), the interlaminar 

shear strength (F sbs ) and the 

in-plane shear strength (τ u ), the latter 

obtained from tensile tests on double 

lap bolted joints. 18 It is worth mentioning 

that the behaviour exhibited by 

coupons extracted from both the web 

and the flanges was similar. The different 

mechanical properties in tension, 

flexure and compression, together 

with the material orthotropy, are also 

outlined. 

Property/test 

and direction 

Longitudinal 

flexure 

Flexural Behaviour of Simply 

Supported Beams 

Test Set-up and Results 

This experimental series consisted of 

four beams with different spans and 

lateral bracing systems, all subjected to 

a point load at midspan. Beams V1 and 

V2 were both tested in a 4,00 m span, 

while beams V3 and V4 were tested in 

spans of 1,44 and 1,00 m, respectively. 

In beam V1, in order to prevent lateraltorsional 

instability, a lateral bracing 

system was used along the beam span 

(Fig. 2). All the other beams were laterally 

unrestrained. Load was applied 

using a hydraulic jack that transmitted 

the load to the top flange of the GFRP 

profile through square steel spreading 

plates of side 0,08 m; a metallic sphere 

was placed between the two spreading 

plates, in order to avoid any transverse 

loading. In beam V2, in order to 

investigate the influence of the loading 

system in restraining the beam at midspan, 

successive changes were introduced 

in the loading system. 18,19 The 

supports of all the beams were made 

of 0,05 m diameter steel rollers with 

L ongitudinal 

tension 

Longitudinal 

compression 

Transverse 

compression 

s u (MPa) 624,6 ± 26,9 475,5 ± 25,5 375,8 ± 67,9 122,0 ± 15,4 

E (GPa) 26,9 ± 1,3 32,8 ± 0,9 26,4 ± 1,9 7,4 ± 0,4 

e u (10 −3 ) 24,9 ± 1,3 15,4 ± 1,5 17,0 ± 2,5 21,5 ± 1,7 

n xy (–) — 0,28 — — 

Interlaminar shear strength, F sbs = 35,0 ± 3,9 MPa. 

In-plane shear strength, t u = 38,7 ± 5,6 MPa. 

Full-scale properties (equivalent isotropic behaviour): E full = 38,3 GPa; G full = 3,58 GPa. 

Table 1: Mechanical properties of the GFRP profile from coupon (average and standard 

deviation values) and full-scale testing 

(a) 

(b) (c) (d) 

Fig. 1: Failure modes: (a) interlaminar shear; (b) flexure; (c) tension; and (d) compression 


Fig. 2: Beam V1—deformation on the brink 

of collapse 

0,08 m long top steel plates; both end 

supports allowed for free rotation and 

one of them also allowed for longitudinal 

sliding. 

All beams presented linear-elastic 

behaviour up to failure. 18,19 The fullscale 

elastic constants of the GFRP 

profile were estimated on the basis of 

the method proposed by Bank, 24 which 

involves performing a linear regression 

analysis of the slope of the load-deflection 

curves for varying spans. This 

analysis provided a longitudinal elastic 

modulus (E full ) of 38,3 GPa and a shear 

modulus (G full ) of 3,58 GPa. One can 

readily note that the elastic constants 

provided by mechanical tests on smallscale 

specimens (E t,x = 32,8 GPa; E f,x = 

26,9 GPa, cf. Table 1) may differ considerably 

from those obtained in fullscale 

tests. Such variation is mainly due 

to the inhomogeneous constitution 

of both the GFRP laminates and the 

overall cross section and also to differences 

in the experimental set-up. 

Failure of beam V1 occurred due to 

local buckling of the top flange, for 

a midspan deflection of 107,3 mm 

(about 1/37 of the span, Fig. 2) and a 

load of 60,2 kN, which corresponded to 

a longitudinal maximum stress of 268,2 

MPa. Failure occurred with delamination 

of the top flange and web-top 

flange separation in the vicinity of 

midspan (Fig. 3), followed by web 

transverse bending. This test showed 

the importance of the local buckling 

phenomenon in members under compression, 

such as the flanges of bended 

beams. In fact, at failure, the maximum 

longitudinal stress was about 56 and 

71% of the tensile and compressive 

material strengths, respectively. 

In several iterations of the flexural 

test of beam V2, failure was always 

triggered by lateral-torsional buckling 

(Fig. 4). The different test set-ups 

proved to have a significant influence 

on the buckling load, which varied 

from 13,0 to 20,7 kN. The lowest buckling 

load (minimum restriction introduced 

by the load application system at 

midspan) corresponded to a maximum 

longitudinal stress of 58,0 MPa, showing 

the importance of global instability 

in slender unrestrained beams. 

Failure of beams V3 and V4 was due 

to crushing of the web at midspan 

(Fig. 5) under the applied load, 

and occurred for loads of 88,2 and 

107,5 kN, respectively. Crushing failure 

of the web was followed by the 

development of longitudinal cracks 

in the web–top flange junction. The 

above-mentioned failure loads correspond 

to maximum transverse compressive 

stresses in the web (under the 

applied load) of 112,6 and 137,1 MPa, 

calculated using Eq. (4). 

Assessment of Design Methods 

In addition to the analytical formulae 

presented earlier, FE models of all 

Fig. 4: Beam V2—lateral-torsional buckling 

tested GFRP beams were developed 

using the FE program SAP2000. 25 The 

web and flanges of the GFRP profiles 

were modelled using adapted DKQ 

(discrete Kirchhoff quadrilateral) 

shell elements, consisting of four-node 

rectangular/triangular elements with 

bilinear interpolation functions. 25,26 

Linear-elastic orthotropic material 

behaviour was assumed on the basis 

of the results of experiments (cf. 

Table 1). The actual supporting conditions 

were simulated with node constraints. 

Both linear static and linear 

buckling analyses were carried out— 

Fig. 6 shows the buckled configuration 

of beam V1. 

For the serviceability behaviour, the 

deflections of all tested beams could 

be back-calculated with a very high 

accuracy (maximum and average relative 

errors among the tested beams of 

3 and 1%, respectively) on the basis of 

Timoshenko beam theory and using 

the calculated full-scale elastic constants 

in Eq. (1). It is also worth mentioning 

that in all tested beams, the 

shear contribution to deformation was 

significant: 12,6% for the relatively 

slender beams V1 and V2; 41,1 and 

59,1% for the less slender beams V3 

and V4, respectively. 

With regard to the failure behaviour, 

the experimental critical load of beam 

V1 (60,2 kN) compared reasonably 

well with numerical (53,9 kN, relative 

difference of −10,5%) and analytical 

predictions, the latter obtained using 

the two alternative design formulae 

presented by Kollár 20 (58,0 kN, −3,7%) 

and Mottram 21 (56,8 kN, −5,6%). 

Analytical predictions were computed 

on the basis of Eqs. (5) and (6), respectively, 

using the coupon material properties 

(cf. Table 1); in these calculations, 

a standard value of ν T = 0,10 was considered 

and it was assumed that the 

in-plane shear modulus, G LT , is equal 

to the full-scale shear modulus, G full ). 

The differences between experimental 

and predicted critical loads are very 

Fig. 3: Beam V1—local buckling failure 

Fig. 5: Beam V3—web crushing under applied load 


(Fig. 8)—firstly, in the curves corresponding 

to the horizontal deflection of 

both flanges (d 2 and d 3 ) and secondly, 

in the curve describing the vertical 

deflection of the shear centre (d 1 ). 

As expected, the critical bucking load 

decreased with increasing span and, for 

each span the highest critical load was 

obtained when the load was applied 

at the centre of the bottom flange 

(BF), while the lowest critical load was 

obtained when loading at the centre 

of the top flange (TF), as illustrated in 

Fig. 9. For the shortest span of 2,0 m, 

maximum longitudinal stresses varied 

between 53,3 and 125,0 MPa, while for 

the longest span of 4,0 m those stresses 

varied between 24,7 and 44,7 MPa. 

Fig. 6: FE model of beam V1—local buckling configuration 

small, particularly if the relatively high 

coefficients of variation exhibited by 

GFRP material properties are taken 

into account. In principle, because of 

the geometric imperfections of the 

material, one would expect the experimental 

results to be below analytical/ 

numerical predictions; the fact that 

predictions are slightly lower than 

the experimental critical load, has to 

be attributed to the above-mentioned 

material inhomogeneity and, eventually, 

to some slight restriction introduced 

by the loading system. 

For beam V2, the minimum experimental 

critical load (13,0 kN) differed 

quite significantly from numerical 

(5,0 kN) and analytical (4,7 kN) predictions, 

obtained respectively with the 

above-mentioned FE model and using 

Eq. (7), again computed using coupon 

material properties (cf. Table 1). For 

this beam, the restriction (friction) 

introduced by the loading system had 

a very significant effect in preventing 

the triggering of lateral-torsional buckling. 

Therefore, in order to understand 

this instability mechanism better and 

to assess the accuracy of analytical and 

numerical design tools, it was decided 

to perform tests on cantilevers, for 

which it is easier to prevent the loading 

system from restraining deformations 

(see next section). 

For beams V3 and V4, comparison of 

the maximum transverse compressive 

stresses in the web under the applied 

load (112,6 and 137,1 MPa, respectively, 

obtained by dividing the applied 

load by the area of the web directly 

under the metal plates positioned 

below the hydraulic jacks) with the 

compressive strength in the transverse 

direction, obtained from the material 

characterization tests (122,0 MPa, cf. 

Table 1), allows justifying the observed 

failure mode. 

Flexural Behaviour of Cantilevers 

Test Set-up and Results 

In this experimental series, three different 

spans of 2,0, 3,0 and 4,0 m were 

tested and, for each span, the load was 

applied in three alternative positions 

of the free end cross section: at the 

centre of the top flange (TF), at the 

centroid or shear centre (SC) and at 

the centre of the bottom flange (BF). 

Load was applied using a dead-load 

system, consisting of a metal bucket 

filled with metal plates and water, 

which was suspended from the free end 

cross section of the GFRP cantilevers, 

at the three predefined positions. The 

vertical support of the cantilevers was 

made of a thick steel plate connected 

to a transverse metal beam, which was 

placed over the top flange of the profile 

using four Dywidag bars. Horizontal 

deflections in the support section were 

restrained by means of metallic plates 

and sets of metallic bolts placed on 

both sides of the web. 

All cantilevers tested presented lateraltorsional 

buckling—Fig. 7 illustrates the 

buckled configuration of a 4,0 m span 

cantilever loaded at the SC. This global 

instability could easily be distinguished 

in the load-deflection behaviour 

Fig. 7: Lateral-torsional buckling of a 4,0 m 

span cantilever (load at SC) 

Load (kN) 

2,25 

2,00 

1,75 

1,50 

1,25 

1,00 

d 3 

0,75 

0,50 

d 1 

0,25 

d 2 

0,00 

0 20 40 60 80 

Deflection (mm) 

Fig. 8: Load-deflection curves for a 4,0 

span cantilever (load at SC) 

Critical load (kN) 

Fig. 9: Critical load as a function of the 

span, for different load positions 

100 

18 

16 

BF — experiment 

SC — experiment 

14 

TF — experiment 

12 

10 

8 

SC — analytical 

BF — numerical 

SC — numerical 

TF — numerical 

6 

4 

2 

0 

1,5 2,0 2,5 3,0 3,5 4,0 4,5 

Cantilever span (m) 


supplying the GFRP profiles used in the 

experimental investigations. 

References 

Fig. 10: FE model of a 4,0 m span cantilever (load at TF) —global buckling configuration 

Assessment of Design Methods 

Shell FE models of all tested cantilevers, 

similar to those described for the simply 

supported beams, were developed 

with the program SAP2000. Figure 10 

illustrates the buckling mode for a 

4,0 m span cantilever loaded at the TF. 

Figure 9 shows the comparison 

between experimental critical loads 

and numerical predictions using the 

aforementioned FE models (for all 

three load positions) and analytical 

predictions (computed only for shear 

centre loading) using Eq. (8). 

Average relative errors for analytical 

and numerical predictions are 4,7 

and 12,9%, respectively. Therefore, 

it seems fair to claim a reasonable 

agreement between experimental critical 

loads and both design methods— 

predictions unarguably exhibit a similar 

pattern of variation with both the 

span and the load position. Figure 9 

also shows that, in general, experimental 

critical loads are lower than predictions, 

which may be attributed to the 

effect of geometrical imperfections in 

the profile and, in addition, to the not 

completely fixed restraint condition at 

the support. 

Conclusion 

The flexural behaviour of GFRP 

pultruded profiles presents several differences 

when compared to traditional 

materials, at both material and structural 

levels. On one hand, contrary to 

steel that yields and concrete that cracks, 

in general, GFRP profiles present 

linear-elastic behaviour until failure, 

which usually occurs with large deformations. 

On the other hand, the design 

of GFRP members is often governed 

by deformability restrictions, due to the 

low elastic modulus in the longitudinal 

direction and also due to the contribution 

of shear to the global deformation. 

In general, design at ultimate limit states 

is not governed by material strength, as 

failure is usually due to local or global 

buckling phenomena, with stress levels 

in service that are relatively low, in spite 

of the high strength exhibited by the 

GFRP material. 

The comparison between experimental 

results and predicted behaviour with the 

FE models and analytical formulae presented 

in this paper for a limited range 

of failure scenarios shows that numerical 

and analytical tools available for 

the design of GFRP pultruded flexural 

members are reasonably accurate. In 

particular, for service design, the bending 

deformations of GFRP beams can 

be readily calculated with Timoshenko’s 

beam theory, using full-scale elastic 

constants and assuming an equivalent 

isotropic behaviour. At ultimate limit 

states, failure mechanisms associated 

with material crushing and local or 

global buckling can be easily computed, 

using laminate material properties, 

obtained through coupon testing. 

Acknowledgements 

The authors wish to acknowledge the support 

of FCT, ICIST and Agência da Inovação 

(Grant No. 2009/003456) for funding the 

research and also STEP and ALTO for 

[1] Keller T. Recent all-composite and hybrid 

fibre-reinforced polymer bridges and buildings. 

Prog. Struct. Eng. Mater. 2001; 3(2): 132–140. 

[2] Keller T. Use of Fibre Reinforced Polymers 

in Bridge Construction. Structural Engineering 

Documents, vol. 7. IABSE: Zurich, 2003. 

[3] Burgoyne C. Advanced composites in civil 

engineering in Europe, Struct. Eng. Int. 1999; 

9(4): 267–273. 

[4] Braestrup M. Footbridge constructed from 

glass-fiber reinforced profiles, Denmark. Struct. 

Eng. Int. 1999; 9(4): 256–258. 

[5] GangaRao H, Craigo II CA. Fiber reinforced 

composite bridge decks in the USA. Struct. Eng. 

Int. 1999; 9(4): 286–288. 

[6] Keller T. Towards structural forms for composite 

fibre materials. Struct. Eng. Int. 1999; 9(4): 

297–300. 

[7] Sobrino JA, Pulido MDG. Towards advanced 

composite material footbridges. Struct. Eng. Int. 

2002; 12(2): 84–86. 

[8] Cheng L, Karbhari VM. New bridge systems 

using FRP composites and concrete: a state-ofthe-art 

review. Prog. Struct. Eng. Mater. 2006; 

8(4): 143–154. 

[9] Neto ABS, La Rovere HL. Composite concrete/GFRP 

slabs for footbridge deck systems. 

Compos. Struct. 2010; 92(10): 2554–2564. 

[10] Hollaway LC. A review of the present and 

future utilisation of FRP composites in the civil 

infrastructure with reference to their important 

in-service properties. Const. Build. Mater., in 

press, doi:10.1016/j.conbuildmat.2010.04.062. 

[11] Correia JR, Cabral-Fonseca S, Branco 

FA, Ferreira J, Eusébio MI, Rodrigues MP. 

Durability of glass fibre reinforced polyester 

(GFRP) pultruded profiles for construction 

applications. Mech. Compos. Mater. 2006; 42(4): 

325–338. 

[12] Correia JR, Branco FA, Ferreira J, Cabral- 

Fonseca S, Rodrigues JPC. Lifetime performance 

of GFRP pultruded profiles for structural 

applications. IABSE Symposium Improving 

Infrastructure—Bringing People Closer Worldwide, 

Weimar, 2007. 

[13] Correia JR. GFRP Pultruded Profiles in 

Civil Engineering: Hybrid Solutions, Bonded 

Connections and Fire Behaviour, PhD Thesis, 

IST, Technical University of Lisbon, 2008. 

[14] Clarke JL (ed.). Structural Design of 

Polymer Composites—EuroComp Design Code 

and Handbook. E & FN Spon: London, 1996. 

[15] CEN. EN 13706: Reinforced Plastics 

Composites—Specifications for Pultruded 

Profiles. Part 1: Designation; Part 2: Methods of 

Test and General Requirements; Part 3: Specific 

Requirements. European Committee for 

Standardisation: Brussels, 2002. 

[16] National Research Council of Italy. Guide 

for the Design and Construction of Structures 

made of FRP Pultruded Elements, Advisory 

Committee on Technical Recommendations for 

Construction: Roma, 2008. 


[17] Bank LC. Composites for Construction: 

Structural Design with FRP Materials. Wiley: 

Hoboken, NJ, 2006. 

[18] Correia JR. GFRP Pultruded Profiles: 

The Use of GFRP-concrete Hybrid Beams 

in Construction, MSc Thesis, IST, Technical 

University of Lisbon, 2004 (in Portuguese). 

[19] Correia JR, Branco FA, Silva NMF, 

Camotim D, Silvestre N. First-order, buckling 

and post-buckling behaviour of GFRP pultruded 

beams—part 1: experimental study, submitted for 

publication. Contact author for details: jcorreia@ 

civil.ist.utl.pt 

[20] Kollár LP. Local buckling of fiber reinforced 

plastic composite structural members with open 

and closed cross sections. J. Struct. Eng. 2003; 

129(11): 1503–1513. 

[21] Mottram JT. Determination of critical load 

for flange buckling in concentrically loaded pultruded 

columns. Compos. B: Eng. 2004; 35(1): 

35–47. 

[22] Timoshenko SP, Gere JM. Theory of Elastic 

Stability, McGraw-Hill: New York, 1963. 

[23] Silva NMF, Camotim D, Silvestre N, Correia 

JR, Branco FA. First-order, buckling and 

post-buckling behaviour of GFRP pultruded 

beams—part 2: numerical simulation, submitted 

for publication. Contact author for details: 

jcorreia@civil.ist.utl.pt 

[24] Bank LC. Flexural and shear moduli of fullsection 

fiber reinforced plastic (FRP) pultruded 

beams. J. Testing Eval. 1989; 17(1): 40–45. 

[25] Computers and Structures Inc. SAP 2000 

User’s Manual (version 11.0), 2007. 

[26] Zienkiewicz OC, Taylor RL. The finite 

element method—Volume 2. Solid and fluid 

mechanics. Dynamics and Non-Linearity. 

McGraw-Hill: London, 1991. 


Effects of Hygrothermal Ageing on the Mechanical 

Properties of Glass-Fibre-Reinforced Polymer Pultruded 

Profiles 

João R. Correia, Prof., Dr, Technical Univ. of Lisbon, Instituto Superior Técnico/ICIST, Civil Eng. and Architecture, Lisbon, Portugal; 

Susana Cabral-Fonseca, Dr, Eng., Laboratório Nacional de Engenharia Civil, Lisbon, Portugal; Ana Carreiro, Civil Eng., Technical 

Univ. of Lisbon, Instituto Superior Técnico, Civil Eng. and Architecture, Lisbon, Portugal; Ricardo Costa, Civil Eng., Technical Univ. 

of Lisbon, Instituto Superior Técnico, Civil Eng. and Architecture, Lisbon, Portugal; Maria Paula Rodrigues, Dr, Eng., Laboratório 

Nacional de Engenharia Civil, Lisbon, Portugal; Isabel Eusébio, Dr, Eng., Laboratório Nacional de Engenharia Civil, Lisbon, Portugal; 

Fernando Branco, Prof., Dr, Technical Univ. of Lisbon, Instituto Superior Técnico/ICIST, Civil Eng. and Architecture, Lisbon, 

Portugal. Contact: jcorreia@civil.ist.utl.pt 

Abstract 

This paper presents the results of an experimental study on the physical and 

mechanical changes suffered by glass-fibre-reinforced polymer (GFRP) pultruded 

profiles, made of either unsaturated polyester or vinylester resins, after 

accelerated hygrothermal ageing. Specimens from both types of profiles, comprising 

identical fibre contents and architectures, were subjected to: (a) immersion 

in demineralized water; (b) immersion in saltwater at temperatures of 20, 40 

and 60°C for 12 months and; (c) continuous condensation at 40°C for 9 months. 

Batches of test specimens from both profiles, conditioned in those accelerated 

exposure environments, were periodically monitored with respect to: (a) mass 

changes; (b) variation in glass transition temperature evaluated through dynamic 

mechanical analysis (DMA) and; (c) degradation of mechanical properties, 

assessed by means of tensile, flexural and interlaminar shear tests. 

Keywords: GFRP; unsaturated polyester matrix; vinylester matrix; pultruded 

profiles; hygrothermal ageing; mechanical properties. 

Introduction 

Fibre-reinforced polymer (FRP) materials 

in general, and glass-fibre-reinforced 

polymer (GFRP) pultruded 

profiles in particular, are being used 

increasingly in civil engineering applications 

as an alternative to traditional 

materials, such as steel, reinforced concrete 

and timber. This growing acceptance 

of FRP structures, particularly in 

corrosive applications, can be attributed 

to their improved durability and low 

maintenance requirements, in addition 

to other intrinsic advantageous properties 

of advanced composite materials 

that include high strength, lightness 

and low thermal conductivity. 1,2 

In regard to durability, the long-term 

use of FRP materials in vessels, pipelines, 

storage tanks and chemicalresistant 

equipment of the oil industry 

provides evidence of their improved 

performance in relatively harsh and 






corrosive environments, when compared 

to traditional materials. However, 

for civil engineering applications; owners, 

designers and contractors request 

comprehensive and validated data on 

durability, since the service life of mainstream 

structures is generally expected 

to exceed 50 years. As most FRP civil 

engineering structures are quite recent 3 

and research already carried out on this 

topic is still limited, such information 

correlating the effects of environmental 

degradation on the physical, chemical 

and mechanical properties of FRPs 

is currently not available. The development 

of reliable degradation models, 

similar to those already available for 

traditional materials, involves gathering 

such comprehensive data on durability. 

It is also worth mentioning that comparative 

studies on the performance of 

alternative matrix formulations used in 

FRP materials are also scarce. In this 

context, paradoxically, the widespread 

acceptance of FRP materials is delayed 

because of concerns about durability. 

In this regard, several authors have 

recently identified durability as one 

of the most critical gap between perceived 

need for information and available 

information, and as a crucial area 

of focus for future research on FRP 

materials. 4,5 

This paper presents results of an 

experimental study on the physical and 

mechanical changes suffered by GFRP 

pultruded profiles, made of either 

unsaturated polyester or vinylester 

resins with identical fibre contents and 

architectures, following accelerated 

hygrothermal ageing. Specimens from 

both types of profiles were subjected 

to immersion in demineralized water 

and saltwater for temperatures of 20, 

40 and 60°C and, in addition, to continuous 

condensation at 40°C, simulating 

the ageing conditions in wet 

environments (e.g. placed under water 

or subjected to high levels of moisture), 

in coastal areas or where the 

use of de-icing salts is common. Other 

environmental degradation agents, 

not investigated in the present study, 

include acid or alkaline fluids, thermal 

cycles, freeze–thaw cycles, ultraviolet 

radiation and elevated temperature. 4 

Batches of test specimens from both 

types of profiles, placed in the abovementioned 

degradation environments, 

were periodically removed and monitored 

regarding: (a) mass changes; 

(b) variation in glass transition temperature 

evaluated through dynamic 

mechanical analysis (DMA) and; (c) 

degradation of mechanical properties 

in tension, flexure and shear. 

This paper provides extensive data 

on the effects of hygrothermal ageing 

on the performance of GFRP pultruded 

profiles, thereby contributing 

to shortening of the above-mentioned 

gap between perceived need for information 

and available information on 

durability. In addition, as similar fibre 

contents/architectures were used in 

this study, results obtained allow for a 

direct comparison between the performances 

of unsaturated polyester and 

vinylester resins. 


Experimental Programme 

Materials 

The material studied was obtained 

from two commercial GFRP pultruded 

tubular profiles (50 × 50 mm, 

thickness 5 mm). This material consists 

of alternating layers of unidirectional 

E-glass fibre rovings and strand mats 

embedded in either unsaturated polyester 

resin (profile “UP”) or vinylester 

resin (profile “VE”). The former resin 

is used in most structural applications 

when there are no particular requirements 

in terms of environmental 

aggressiveness, whereas the latter is 

often selected for applications in relatively 

harsh or corrosive environments. 

The two profiles were produced with 

the same glass fibre content and architecture 

(Figs. 1 and 2), thus allowing 

comparison of the durability performance 

of the polyester and vinylester 

resins used in these off-the-shelf standard 

profiles. 

Initial Characterization 

The chemical, physical and mechanical 

characterization of both types of materials 

was carried out using the following 

techniques: 

1. Chemical composition: Infrared 

spectra of the materials were studied 

in the 450 to 4000 cm −1 region, 

according to ASTM E 1252 standard. 

6 For these measurements, 

powder samples scraped from the 

surfaces of test specimens were 

mixed with dry spectroscopic-grade 

potassium bromide and pressed into 

pellets. 32 scans were collected and 

averaged at a spectral resolution of 

4 cm −1 , in a Thermo Scientific Nicolet 

spectroscope. The glass fibre content 

was determined by the calcination 

method described in ASTM D 3171 

standard. 7 

2. Physical properties: Density was 

measured according to ISO 1183 

standard 8 (immersion method). 

Glass transition temperature (T g ) 

was determined by DMA, in accordance 

with ISO 6721 standard. 9 

Three-point bending type clamped 

specimens of 5 × 15 × 60 mm were 

tested at constant frequency of 1 Hz 

and strain amplitude of 15 μm, using 

a DMA analyser. The analysis was 

carried out from room temperature 

up to 200°C, at a rate of 2°C/min. 

Three replicates were tested for 

each type of material. 

3. Mechanical properties: Tensile tests 

were conducted according to ISO 

527—parts 1 and 5 standard 10 in 

rectangular test specimens (5 × 25 

× 300 mm), without end tabs, using 

an universal testing machine with 

a load capacity of 100 kN. Threepoint 

bending flexural tests were 

performed according to ISO 14125 

standard, 11 in rectangular test specimens 

(5 × 15 × 150 mm) with a span 

of 100 mm using a system constituted 

by a hydraulic press with a 10 

kN load capacity. Interlaminar shear 

tests were carried out in accordance 

with ASTM D 2344 standard 12 in 

rectangular test specimens (5 × 10 × 

30 mm), loaded in a 20 mm span with 

the same system used in the bending 

tests. Compressive properties were 

determined according to ASTM D 

695 standard 13 in rectangular specimens 

(5 × 10 × 30 mm). 

Exposure Environments 

In order to study the potential degradation 

of the two types of profiles in 

typical environments of civil engineering 

applications, test specimens were 

subjected to the exposure conditions 

described in Table 1. 

The experimental procedures used 

in the immersion ageing conditions, 

both in water and in saltwater, were 

based on ISO 175 standard, 14 with the 

concentration of salt in the saltwater 

medium being in agreement with 

ASTM D 1141 standard 15 —for both 

media, the cut edges of the test specimens 

were completely immersed. The 

ageing performed in the continuous 

condensation chamber was carried out 

according to the procedures described 

in ISO 6270 standard. 16 

1 mm 

Mats Rovings Mats Rovings 

VE profile 

UP profile 

Fig. 1: Cross section and fibres architecture of UP and VE profiles (UP = Unsaturated 

polyester resin; VE = Vinylester resin) 

Fig. 2: Outer mats and inner rovings of a burnt laminate (VE profile) 

Experimental Characterization after 

Hygrothermal Ageing 

After exposure to the different ageing 

conditions described in Table 1, batches 

of aged test specimens obtained from 

each type of profile were subjected 

to the following characterization 

techniques: 

1. Mass changes: Control specimens 

with geometry similar to that of specimens 

used in DMA were removed 

periodically from the different 

exposure environments in order to 

evaluate their mass changes. After 

removal from the exposure environments, 

the surface of the specimens 

was dried with a cloth in order to 

remove any residual free moisture. 

Specimens were then immediately 

weighed using a 0,0001 g precision 

scale. 

2. Dynamic mechanical analysis: The 

T g was measured according to the 

same procedure used in the initial 

characterization tests. Three replicates 

were tested for each type of 


material and ageing condition (duration 

and exposure environment). 

3. Mechanical behaviour: Tensile, flexural 

and interlaminar shear tests 

were performed according to the 

above-mentioned standards. At least 

five replicates were tested in the longitudinal 

direction for each material 

and ageing condition. 

Excluding the study of the mass 

changes, after being removed from 

the different exposure environments 

and prior to further testing, specimens 

were placed inside polyethylene bags. 

These were hermetically closed, in 

order to maintain the moisture content 

of the material, and then placed inside 

a room with temperature controlled at 

20 (±2)°C. Prior to testing, specimens 

were removed from the polyethylene 

bags and immediately tested without 

any further conditioning. 

Results and Discussion 

Initial Characterization 

The results of initial chemical, physical 

and mechanical characterization of 

both profiles are listed in Table 2. 

Chemical composition determined by 

Fourier transform infrared spectroscopy 

(FTIR) (Fig. 3) showed little differences 

between the two materials 

analysed. In fact, the spectra of both 

profiles were quite similar. The localizations 

and intensity of the peaks confirmed 

the presence of ester, as well 

as aromatic and aliphatic structures, 

which are common in the molecular 

structure of both unsaturated polyester 

and vinylester resins. FTIR spectra 

also showed the existence of calcium 

carbonate (as filler) and silica (from 

the glass fibres). 

The glass fibre content and density 

of the VE profile were slightly higher 

than those of the UP profile. On the 

other hand, the T g (obtained from both 

the storage modulus, E′, and the loss 

factor, tand) of the VE profile was 

lower than that of the UP profile. From 

the mechanical aspect point of view, 

the UP and VE profiles started to lose 

their initial performance at temperatures 

above 108 and 99°C, respectively 

(Fig. 4). 

With regard to the mechanical behaviour, 

in all characterization tests (tension, 

flexure, interlaminar shear and 

compression), both types of profiles 

exhibited a well defined and typical 

linear elastic behaviour up to failure. A 

comparative analysis of the mechanical 

behaviour exhibited by both profiles 

shows that they are quite similar in 

their tensile properties (both strength, 

s tu,x , and modulus, E t,x ) and interlaminar 

shear strength (s u,sbs ). However, 

in flexure, the VE profile showed 

superior initial performance, for both 

strength (s fu,x ,) and stiffness (E f,x ). 

Owing to the relatively high scatter in 

the results obtained for the compressive 

strength (s cu,x ), it was decided to 

Type of exposure Duration Conditions 

Immersion in water 

(W-20), (W-40), (W-60) 

Immersion in saltwater 

(S-20), (S-40), (S-60) 

Continuous condensation 

(CC-40) 

Table 1: Exposure ageing conditions 

Absorbance Absorbance 

1,0 

0,8 

0,6 

0,4 

0,2 

0,0 

1,0 

0,8 

0,6 

0,4 

0,2 

0,0 

UP profile 

VE profile 

3, 6, 9 and 12 months 

3, 6 and 9 months 

Fig. 3: FTIR spectra of UP and VE profiles 

3000 2000 1000 

Wavenumbers (cm –1 ) 

Composition: demineralized water 

Temperatures: 20 (±2)°C, 

40 (±1)°C and 60 (±1)°C 

Composition: 35 g/L NaCl 

Temperatures: 20 (±2)°C, 

40 (±1)°C and 60 (±1)°C 

Temperature: 40 (±2)°C 

Relative humidity: 100% 

Property Test method Profile UP Profile VE 

Chemical 

composition 

FTIR 

FTIR spectra consistent with unsaturated 

polyester or vinylester, with presence 

of calcium carbonate and silica 

Glass fibre 

content (%) 

Calcination 68,4 ± 1,8 68,7 ± 0,4 

Density (g/cm 3 ) Immersion 1,869 ± 0,113 2,028 ± 0,052 

T g (°C) DMA E′ initial 107,9 ± 10,8 98,6 ± 7 

tand 146 ± 2,3 126,9 ± 2,3 

Mechanical 

properties 

Tension s tu,x (MPa) 406 ± 31 393 ± 51 

E t,x (GPa) 37,6 ± 2,6 38,9 ± 4,1 

Flexure s fu,x (MPa) 417 ± 65 537 ± 73 

E f,x (GPa) 20,0 ± 6,9 28,4 ± 3,4 

Interlaminar 

shear 

s u,sbs (MPa) 38,5 ± 2,7 39,2 ± 4,2 

Compression s cu,x (MPa) 280 ± 123 360 ± 131 

Table 2: Initial physical, chemical and mechanical properties 

exclude the compressive properties 

from this durability study. 

Moisture Uptake with Hygrothermal 

Ageing 

Figure 5 illustrates the mass variation 

exhibited by both profiles for 

the different immersion media (solutions 

of demineralized water, W, 

and saltwater, S) and temperatures 


E′(GPa) 

(a) 

E′(GPa) 

(b) 

Mass variation (%) 

1,50 

1,25 

1,00 

0,75 

0,50 

0,25 

0,00 

0 

(a) 

Mass variation (%) 

(b) 

25 

20 

15 

10 

5 

0 

25 

Un-aged 

25 

20 

15 

10 

5 

0 

1,50 

1,25 

1,00 

0,75 

0,50 

0,25 

0,00 

UP profile 

VE profile 

0 

50 75 

UP profile 

1000 2000 3000 4000 5000 6000 7000 8000 9000 

VE profile 

0 

100 125 150 175 200 

Temperature (°C) 


Exposure period (h) 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

1000 2000 3000 4000 5000 6000 7000 8000 9000 

Exposure period (h) 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

Fig. 5: Mass variation of UP (a) and VE (b) profiles for different hygrothermal ageing 

conditions 

0,2 

0,16 

0,12 

0,08 

0,04 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

T g,onset 

0 

25 50 75 100 125 150 175 200 

Un-aged W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

Fig. 4: DMA 3-point bending curves of UP (a) and VE (b) profiles before and after 

hygrothermal ageing 

0,2 

0,16 

0,12 

0,08 

0,04 

Tan d (–) 

Tan d (–) 

(20, 40 and 60°C), as well as in continuous 

condensation (CC) at 40°C. 

Figure 5 shows that the evolution of all 

mass variation curves follow roughly 

a Fickian response (i.e. a fast initial 

mass gain that slows as saturation 

approaches), with rates of mass uptake 

increasing with temperature, particularly 

in the beginning of the exposure. It 

can also be seen that, for similar ageing 

conditions (immersion media and temperatures), 

the comparison of the mass 

variation exhibited by both profiles 

depicts significant differences, with the 

mass uptake for the VE profile being 

considerably lower than that exhibited 

by the UP profile for all hygrothermal 

ageing conditions; these differences, 

already reported by Chin et al. 17 , stem 

mainly from the distinct water absorption 

capacities of both resin systems, in 

particular, the higher hydrolytic stability 

of VE resins. Figure 5 also shows 

that, for similar temperatures, mass 

uptake in saltwater was always lower 

than that in demineralized water, for 

both UP and VE profiles. Finally, one 

can readily observe that the increasing 

immersion temperature does not 

have a direct correlation with the 

increased level of mass uptake and 

this result should be attributed to the 

potential mass loss by extraction of 

low molecular components, an effect 

that is expected to increase with the 

immersion temperature. In fact, weight 

changes in these ageing processes usually 

result from a balance between the 

water uptake due to moisture ingress 

and the loss of material. When compared 

with the immersion in demineralized 

water at 40°C, under continuous 

condensation at 40°C both materials 

exhibited a higher initial weight gain, 

although for longer periods, weight 

gains for those environments became 

quite similar. 

DMA after Hygrothermal Ageing 

Figure 4 shows the results of DMA 

after 12 months of immersion and 9 

months of continuous condensation— 

for each condition, only one curve 

corresponding to a representative 

specimen is plotted. The left axis represents 

the variation of E′ curves with 

temperature, which exhibit a characteristic 

“step” in the glass transition 

region; the right axis shows the corresponding 

tand curves, which present a 

typical peak in that region. 

The variation in the behaviour exhibited 

by the E′ curves at the transition 

region reflects mainly the changes 


in the polymer matrix performance, 

which progresses from a glassy state 

to an elastomeric state, characteristic 

of its viscoelastic nature. In fact, the 

reinforced material (in this case, the 

glass fibres) does not suffer a stiffness 

reduction in this temperature range. 

Therefore, the DMA technique indicates 

the contribution of the viscoelastic 

nature of the matrix for the overall 

behaviour of the composite, and in the 

present study, this information is useful 

to help understand the actual influence 

of the matrix nature on the durability 

of the composite. In addition, the quality 

of the fibre–matrix interface can 

also influence DMA results. 

Figure 6 plots, in summary, the variation 

of the T g of both profiles (mean 

value ± standard deviation, determined 

based on the onset of E′), as a function 

of the type and duration of exposure. 

For the UP profile, immersion in 

both demineralized water and saltwater 

caused, in general, a decrease 

in the value of T g . It is seen that the 

maximum reduction in T g occurred for 

specimens immersed in demineralized 

water, whereas the minimum reduction 

corresponds to immersion in saltwater. 

After 12 months of exposure, 

T g seems to increase with the immersion 

temperature for both media; for 

saltwater immersion at 60°C, the T g 

becomes even higher than that of the 

un-aged material, most likely due to 

a post-curing phenomenon, induced 

Tg (E') [°C] 

(a) 

Tg (E') [°C] 

(b) 

140 

120 

100 

80 

60 

40 

20 

0 

140 

120 

100 

80 

60 

40 

20 

0 

UP profile 

VE profile 

by the increased temperature. For the 

UP profile, the tand curves for continuous 

condensation, and immersion 

in demineralized water at 60°C show 

the appearance of a second “peak” at 

higher temperatures (Fig. 4). The occurrence 

of this second peak, associated to 

the widening of its base, suggests that 

the ageing of the material involves a 

plasticization mechanism. The occurrence 

of two “peaks” in the tanδ curve 

may be attributed to the different 

mobility of two kinds of segments in 

the polymeric matrix, caused by their 

different extents of plasticization. 

For the VE profile, the variation of T g 

was less dependent on the immersion 

temperature and, in general, its variation 

was less significant than that verified 

in the UP profile. The tand curves 

for the VE profile did not show any 

widening, suggesting that the molecular 

structure did not suffer significant 

changes. The only exceptions were the 

immersions at 60°C in both media, in 

which an asymmetry could be observed 

in the configuration of the tand curves, 

near their maximum value. This result is 

consistent with the lower water uptake 

ability exhibited by this material, when 

compared with the UP profile. 

Water uptake by unsaturated polyester 

and vinylester composites is known to 

cause plasticization in the short term 

and hydrolysis over the long term 

through attack of the ester linkages. 18 

As the ester group is located in the 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

Initial 3 months 6 months 9 months 12 months 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 


Fig. 6: Glass transition temperature of UP (a, c) and VE (b, d) profiles after 

hygrothermal ageing 

middle of the molecular structure of 

polyester, and in the ends of the molecular 

structure of vinylester, in principle, 

the later resin is more resistant to the 

above-mentioned plasticization mechanisms. 

Both these phenomena induce 

higher levels of molecular mobility, 

resulting in a consequent decrease in 

the T g , although such decrease can 

often be offset through residual curing 

of the resins in aqueous media. These 

competing phenomena result in fluctuations 

in the T g as a function of the 

exposure period; in the experiments 

reported herein, such behaviour was 

shown, in particular, by the UP profile. 

Mechanical Performance after 

Hygrothermal Ageing 

Tensile Properties 

The results obtained from tensile tests 

on both materials, namely, the tensile 

strength and the tensile modulus as 

a function of time and hygrothermal 

conditions, are presented in Fig. 7. 

Figure 7 shows that for all ageing conditions 

and for both materials there 

was an overall decrease in the average 

tensile strength with the duration of 

exposure (the only exception was the 

VE profile after 12 months of exposure 

in demineralized water at 20°C). 

As expected, the level of degradation 

of the tensile strength increased consistently 

with the temperature of the 

immersion medium, with maximum 

reductions occurring at 60°C—after 12 

months of exposure, the lowest levels 

of retention were 64 and 75% for the 

UP and VE profiles, both immersed 

in demineralized water. The general 

higher aggressiveness of demineralized 

water compared to saltwater was 

consistent with previous investigations 

(e.g. those reported by Van de Velde 

and Kiekens 19 ) and could be attributed 

to osmotic effects. For all hygrothermal 

ageing conditions and periods of exposure, 

the tensile strength retention of 

the VE profile was consistently higher 

than that of the UP profile. Chu et al. 20 

reported tensile strength reductions 

in pultruded E-glass vinylester laminates 

that follow the overall trend of 

the present tests—the strength retention 

presented by those authors was 

considerably smaller, but so was the 

thickness of the tested material and, 

consequently, the maximum moisture 

uptake. 

The variation exhibited by the average 

tensile modulus with the exposure 

period (Fig. 7), which was much 

more irregular and associated with 


Tensile strength (MPa) 

(a) 

Tensile modulus (GPa) 

(c) 

500 

UP profile 

450 

400 

350 

300 

250 

200 

150 

100 

50 

0 

W-20 W-40 W-60 S-20 

55 

50 

45 

40 

35 

30 

25 

20 

15 

10 

5 

0 

S-40 S-60 CC-40 


Tensile modulus (GPa) Tensile strength (MPa) 

(b) 

500 

450 

400 

350 

300 

250 

200 

150 

100 

50 

0 

VE profile 

W-20 W-40 W-60 S-20 

S-40 S-60 CC-40 


UP profile 

55 

50 

VE profile 

45 

40 

35 

30 

25 

20 

15 

10 

5 

0 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

Initial 3 months 6 months 9 months 12 months (d) Initial 3 months 6 months 9 months 12 months 

Fig. 7- Tensile strength and modulus of UP (a, c) and VE (b, d) profiles after hygrothermal ageing 

higher coefficients of variation (when 

compared to the tensile strength), 

makes it more difficult to establish 

systematic analyses and comparisons 

between the two materials. However, 

for all exposure conditions and durations, 

one can conclude that the stiffness 

retention was considerably higher 

than the corresponding strength retention—the 

minimum levels of retention 

after 12 months were 76% for the UP 

profile (saltwater at 20°C) and 85% 

for the VE profile (saltwater at 60°C). 

In addition, and similar to the tensile 

strength, the stiffness retention of the 

VE profile was always higher than that 

exhibited by the UP profile. Finally, 

when compared to strength, stiffness 

retention appeared to be much more 

insensitive to the temperature and 

composition of the immersion media. 

For both profiles the strength and stiffness 

retention of specimens immersed 

in demineralized water at 40°C was 

comparable to that of specimens under 

continuous condensation (particularly 

for the VE profile) and this result 

agrees well with the water uptake 

measurements. 

Flexural Properties 

The results obtained from flexural 

tests, namely the flexural strength and 

the flexural modulus, on both materials 

are presented as a function of time 

and hygrothermal conditions in Fig. 8. 

In general, the level of degradation of 

the flexural strength of both profiles 

increased with the temperature of the 

immersion medium, being maximum 

at 60°C—this result was in agreement 

with the behaviour already reported 

for the tensile strength and with results 

obtained by other authors with GFRP 

pultruded profiles. 19,21 As for the tensile 

strength, for similar temperatures, 

strength reduction in demineralized 

water was usually lower than that in 

saltwater (it should be mentioned that 

this latter result differs from those 

reported by Liao et al. 22 ) Unlike the 

behaviour exhibited in the tensile tests 

(essentially dominated by the fibres), 

in flexure (also significantly influenced 

by the matrix and the fibre–matrix 

interface) the VE profile did not present 

a better mechanical performance 

than the UP profile for all ageing 

conditions; in fact, for most exposure 

conditions and durations, the strength 

retention of the VE profile was considerably 

smaller than that of the UP profile—after 

12 months of exposure, the 

lowest levels of retention were 66% 

for the UP profile and 58% for the 

VE profile, both immersed in demineralized 

water. This result may be due, 

at least to some extent, to some postcuring 

effect on the UP profile, which 

was identified in the DMA tests, and, in 

addition, to the considerable scatter of 

the results obtained in the initial characterization 

flexural tests. In this regard, 

it is worth mentioning that Kootsookos 

and Mouritz 23 also reported higher 

flexural strength retention in moulded 

glass–polyester laminates, compared 

with glass–vinylester laminates having 

similar fibre architectures. 

Flexural modulus after 12 months of 

exposure decreased in all ageing conditions, 

for both profiles (Fig. 8). Similar 

to strength, in general, stiffness retention 

in demineralized water was lower 

than that in saltwater and, in addition, 

the UP profile presented better 

performance than the VE profile—the 

lowest levels of retention were 69% 

for the UP profile and 58% for the VE 

profile, both immersed in demineralized 

water. Similar to the tensile modulus, 

the flexural stiffness retention 

appeared to be more insensitive to the 

temperature of the immersion media, 

when compared to strength. 

As for tensile performance, the variation 

of flexural properties in specimens 

subjected to continuous condensation 

was roughly similar to that of 

specimens immersed in demineralized 

water at 40°C. 

Interlaminar Shear Strength 

Figure 9 illustrates the variation of the 

interlaminar shear strength (a matrix 

dominated property) of both profiles 

as a function of the hygrothermal conditions 

and the period of exposure. 

Figure 9 shows that very significant 

reductions occurred in the interlaminar 


600 

500 

400 

300 

200 

100 

0 

(a) 

35 

30 

25 

20 

15 

10 

5 

0 

Flexural strength (MPa) 

Flexural modulus (GPa) 

UP profile 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

700 

VE profile 

600 

500 

400 

300 

200 

100 

0 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

Initial 3 months 6 months 9 months 12 months (b) Initial 3 months 6 months 9 months 12 months 

UP profile 

0 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

W-20 W-40 W-60 S-20 S-40 S-60 CC-40 

(c) 

Initial 3 months 6 months 9 months 12 months (d) Initial 3 months 6 months 9 months 12 months 

Fig. 8– Flexural strength and modulus of UP (a, c) and VE (b, d) profiles after hygrothermal ageing 

Flexural modulus (GPa) Flexural strength (MPa) 

35 

VE profile 

30 

25 

20 

15 

10 

5 

shear strength of both profiles as a 

function of time. As for the other two 

mechanical tests, and similar to results 

reported earlier by Kharbari, 24 the 

interlaminar shear strength retention 

decreased consistently with the immersion 

temperature. After 12 months 

of exposure at 60°C, the strength 

retention in demineralized water was 

approximately 55% for the UP profile 

and 61% for the VE profile. For 

saltwater immersion, the corresponding 

values were slightly higher, with 

strength retentions of 60 and 62% for 

the UP and VE profiles, respectively. 

As for the two other mechanical tests, 

the variation of the interlaminar shear 

strength for immersion in demineralized 

water at 40°C was roughly 

analogous to that under continuous 

condensation. Strength reductions 

exhibited by the VE profile immersed 

in demineralized water at the three 

different temperatures followed a 

trend similar to those reported by Chu 

et al. 20 —these authors obtained higher 

strength retentions but, as already discussed, 

the results of the two studies 

are not directly comparable. Finally, 

the better performance exhibited by 

the VE profile for all hygrothermal 

ageing conditions and periods of exposure 

is outlined. 

Conclusion 

This paper presented results of an 

ongoing research project on the 

environmental degradation suffered 

by GFRP pultruded profiles made of 

either UP or VE resins, with similar 

fibre contents and architectures. On 

the basis of results obtained for an 

exposure of 12 months in demineralized 

water and saltwater at 20, 40 and 

60°C, as well as 9 months in continuous 

condensation at 40°C, the following 

conclusions can be arrived at: 

Interl. shear strength (MPa) Interl. shear strength (MPa) 

(a) 

50 

UP profile 

45 

40 

35 

30 

25 

20 

15 

10 

5 

0 

W-20 W-40 

(b) 

50 

45 

40 

35 

30 

25 

20 

15 

10 

5 

0 


VE profile 

W-20 

W-40 

W-60 

W-60 

1. The water uptake capacity of GFRP 

profiles and their temperature 

dependency are strongly dependent 

on the nature of the polymeric 

matrix—for similar ageing conditions, 

the VE profile exhibited considerably 

lower mass uptake than 

the UP profile. 

2.The UP profile was the one that 

presented signs of plasticization in 

S-20 

S-20 

S-40 

S-40 

S-60 

S-60 

CC-40 

CC-40 


Fig. 9: Interlaminar shear strength of UP (a) and VE (b) profiles after hygrothermal 

ageing 


DMA, and as a consequence of such 

plasticization mechanism the values 

of T g in aged specimens suffered a 

general reduction. In spite of this, 

for some ageing conditions, namely 

for higher temperatures, such reduction 

was offset because of the occurrence 

of resin post-curing. These 

competing phenomena resulted in 

fluctuations in the T g as a function 

of the period of exposure. For the 

VE profile, variations of T g were 

much lower, with DMA not suggesting 

any appreciable changes in the 

molecular structure. 

3. The mechanical properties of GFRP 

profiles constituted by both types 

of resins were noticeably affected, 

even for immersion at 20°C (Fig. 10). 

For most of the hygrothermal ageing 

conditions and periods of exposure, 

the retention of both tensile strength 

(a fibre-dominated mechanical 

property) and interlaminar shear 

strength (matrix dominated) of the 

VE profile was considerably higher 

than that of the UP profile—degradation 

increased with the temperature 

of the immersion medium, with 

demineralized water being generally 

more aggressive than saltwater. In 

flexure, the tendency of the results 

was not so clear and, to some extent 

was contradictory to results of the 

other mechanical tests, as the VE 

profile showed a generally worse 

performance than the UP profile. It 

is believed that flexural results may 

have been influenced by the postcuring 

phenomen on observed in the 

UP profile. 

4. The above-mentioned degradation 

was mainly due to physical phenomena, 

such as plasticization of the 

polymeric matrix, since no appreciable 

chemical degradation was 

detected through FTIR analyses. 25,26 

Nevertheless, this degradation may 

influence the use of GFRP profiles 

in wet environments (structures 

placed underwater or subjected to 

Property retention (%) 

120 

100 

80 

60 

40 

20 

0 

Tensile 

strength 

Tensile 

modulus 

high levels of moisture) and especially 

in tropical zones (where, in 

addition, temperatures are high), 

where the use of different resin 

systems and/or superficial protections 

(such as paintings or gel coats) 

shall be considered for improved 

performance. 

The tendencies stated in the aforementioned 

conclusions will be assessed 

and eventually confirmed in the forthcoming 

experiments to be carried out 

within this research project. Additional 

experiments are also being carried out 

in order to evaluate the reversibility of 

the degradation suffered by both profiles—in 

these new experiments, specimens 

will be submitted to mechanical 

tests after being dried, as against the 

present experiments, in which they 

were tested in a saturated state. The 

next steps will also include the development 

of analytical models in order 

to simulate the observed degradation 

suffered by both types of profiles. 


The authors wish to acknowledge the support 

of FCT, ICIST and Agência da Inovação 

(Grant No. 2009/003456) for funding the 

research and also STEP and ALTO for supplying 

the GFRP profiles used in the experimental 

investigations. 

References 

[1] Correia JR. GFRP pultruded profiles in 

civil engineering: hybrid solutions, bonded connections 

and fire behaviour, PhD Thesis, IST, 

Technical University of Lisbon, 2008. 

[2] Correia JR, Cabral-Fonseca S, Branco FA, 

Ferreira J, Eusébio MI, Rodrigues MP. Durability 

of glass fibre reinforced polyester (GFRP) pultruded 

profiles for construction applications. 

Mech. Compos. Mater. 2006; 42(4): 325–338. 

[3] Keller T, Bai Y, Vallée T. Long-term performance 

of a glass-fiber-reinforced polymer truss 

bridge. J. Compos. Const. 2007; 11(1): 99–108. 

[4] Kharbari VM, Chin JW, Hunston D, 

Benmokrane B, Juska T, Morgan R, Lesko JJ, 

Flexural 

strength 

Flexural 

modulus 

UP-W-20 VE-W20 UP-S-20 VE-S-20 

Int. shear 

strength 

Fig. 10: Retention of mechanical properties after 12 months of immersion at 20°C 

Sorathia U, Reynaud D. Durability gap analysis 

for fiber-reinforced polymer composites in 

civil infrastructure. J. Compos. Const. 2003; 7: 

238–247. 

[5] Harries KA, Porter MA, Busel JP. FRP materials 

and concrete research needs. Concr. Int., 

ACI 2003; October: 69–74. 

[6] ASTM~E~1252. Standard Practice for General 

Techniques for Obtaining Spectra Qualitative 

Analysis. American Society for Testing and 

Materials, West Conshohocken, PA, 1998. 

[7] ASTM~D~3171. Standard Test Methods for 

Constituent Content of Composite Materials. 

American Society for Testing and Materials, 

West Conshohocken, PA, 2009. 

[8] ISO~1183—Plastics. Methods for 

Determining the Density of Non-Cellular Plastics. 

Part 1: Immersion Method, Liquid Pyknometer 

Method And Titration Method. International 

Organization for Standardization, Brussels, 2004. 

[9] ISO~6721—Plastics. Determination of Dynamic 

Mechanical Properties. Part 1: General Principles; 

Part 5: Flexural Vibration- Non-Resonance 

Method. International Organization for 

Standardization, Brussels, 2001. 

[10] ISO~527—Plastics. Determination of Tensile 

Properties. Part 1: General Principles; Part~5: Test 

Conditions for Unidirectional Fiber-Reinforced 

Plastic Composites, International Standards 

Organization, Brussels, 1997. 

[11] ISO~14125. Fiber-Reinforced Plastic Composites—Determination 

of Flexural Properties, 

International Standards Organization, 

Brussels, 1998. 

[12] ASTM~D~2344. Standard Test for Short- 

Beam Strength of Polymer Matrix Composite 

Materials and Their Laminates, American Society 

for Testing and Materials, West Conshohocken, 

PA, 2006. 

[13] ASTM~D~695. Standard Test Method for 

Compressive Properties of Rigid Plastics. 



[14] ISO~175—Plastics. Methods of Test for the 

Determination of the Effects of Immersion in 

Liquid Chemicals. International Organisation 

for Standardization, Brussels, 1999. 

[15] ASTM~D~1141. Standard Practice for 

the Preparation of Substitute Ocean Water. 



[16] ISO~6270—Paints and varnishes. Determination 

of Resistance to Humidity. Part 1: 

Continuous Condensation, International 

Standards Organization, Brussels, 1998. 

[17] Chin JW, Nguyen T, Aouadi K. Effects of 

environmental exposure on fiber- reinforced 

plastic (FRP) materials used in construction. 

J. Compos. Technol. Res. 1997; 19(4): 

205–213. 

[18] Chu W, Karbhari VM. Effect of water 

sorption on performance of pultruded E-glass/ 

vinylester composites. J. Mater. Civil Eng. 2005; 

17(1): 63–71. 

[19] Van de Velde K, Kiekens P. Effects of chemical 

environments on pultruded E-glass reinforced 

polyesters. J. Compos. Technol. Res. 2001; 

23(2): 92–101. 


[20] Chu W, Wu L, Kharbari VM. Durability 

evaluation of moderate temperature cured 

E-glass/vinlyester systems. Compos. Struct. 2004; 

66(1–4): 367–376. 

[21] Nishizaki I, Meiarashi S. Long-term deterioration 

of GFRP in water and moist environment. 

J. Compos. Const. 2002; 6(1): 21–27. 

[22] Liao K, Schultheisz CR, Hunston DL. 

Effects of environmental aging on the properties 

of pultruded GFRP. Compos. B: Eng. 1999; 30(5): 

485–493. 

[23] Kootsookos A, Mouritz AP. Seawater durability 

of glass- and carbon-polymer composites. 

Compos. Sci. Technol. 2004; 64(10–11): 1503–1511. 

[24] Kharbari VM. E-Glass/vinylester composites 

in aqueous environments: effects on shortbeam 

shear strength. J. Compos. Const. 2004; 

8(2): 148–156. 

[25] Costa R. Durability of Glass Fibre 

Reinforced Polyester Pultruded Profiles. MSc 

Thesis, IST, Technical University of Lisbon, 2009 

(in Portuguese). 

[26] Carreiro A. Durability of Glass Fibre 

Reinforced Vinylester Pultruded Profiles. MSc 

Thesis, IST, Technical University of Lisbon, 2010 

(in Portuguese). 

International Association for Bridge 

and Structural Engineering 

IABSE Conference Cairo Egypt 

May 7-9, 2012 

Global Thinking in Structural Engineering: 

Recent Achievements 

Organised by 

The Egyptian Group of IABSE 

The Egyptian Society of Engineers 

More information: secretariat@iabse.org 


Evaluation of a Life Prediction Model and Environmental 

Effects of Fatigue for Glass Fiber Composite Materials 

David Brian Dittenber, Ph.D. Student; GangaRao V.S. Hota, Prof., Director of Constructed Facilities Center; Civil Eng. Dept., 

West Virginia University, USA. Contact: ddittenb@mix.wvu.edu 

Abstract 

Fiber-reinforced polymer (FRP) composites are being utilized in an increasing 

number of applications in structural industries. Establishing values for longterm 

fatigue response is essential for widespread use of FRP composites. The 

variety of available fibers, matrix materials, and manufacturing processes makes 

the fatigue response difficult to predict without extensive empirical testing. 

A proposed fatigue life prediction model uses the internal strain energy release 

rate as the metric for predicting fatigue life from a minimum of data points. The 

objective of this research was to apply the above model to fatigue data for various 

composite coupons and components in order to evaluate its applicability in 

predicting fatigue life. The model was found to be able to regularly fit and predict 

fatigue data within 5% log error at both coupon and component levels. The 

effects of environmental conditions, including 12 MPa pressurized absorption 

and fatiguing in salt water and elevated temperatures, were also explored for a 

glass/vinyl ester FRP. The results of this research can be used to aid in the design 

of numerous structural FRP applications, such as windmill blades, bridge decks, 

or deep sea piping. 

Keywords: FRP composites; fatigue life prediction; temperature effects; saltwater 

environment; component fatigue; strain energy. 

Introduction 

Fiber-reinforced polymer (FRP) 

composites offer an opportunity to 

revolutionize and rehabilitate infrastructure 

because of their advantages 

over traditional structural materials: 

higher strength, higher fatigue and 

impact resistance, higher corrosive and 

environmental resistance, longer service 

life, lower installation and maintenance 

costs, and more consistent 

performance. 1 

FRPs are composites that have reinforcement 

in the form of fibers and 

matrix in the form of a polymeric 

material, that is, epoxy, vinyl ester, or 

polyester. The most common fiber 

reinforcements are glass and carbon 

fibers, but the higher cost of carbon 

generally prohibits their use for infrastructure. 

Glass fiber-reinforced polymers 

(GFRPs) are lightweight and 

have a good combination of mechanical 

performance per unit cost. 

Performing extensive experimental 

work is one way to fully characterize 




Paper received: March 14, 2010 

Paper accepted: May 21, 2010 

the performance of a material, but 

fatigue testing is time consuming and 

costly. Therefore, observed behavioral 

trends need to be combined with 

established physical relationships in 

order to produce mathematical models 

to predict fatigue life with reasonable 

accuracy. Many researchers have developed 

fatigue life prediction models for 

composites in the past few decades, 

but none have been widely adapted in 

industry. 

Fatigue in Composites 

When the weakest location in a composite 

fails, the surrounding fiber/ 

matrix interface experiences increased 

local stresses, which can lead to fatigue 

damage. The weakest location is typically 

caused by material defects, such 

as misaligned fibers, voids, or resin-rich 

regions. As fatigue damage progresses, 

cracks in either the resin or the resin/ 

matrix interface develop between the 

fibers, causing many to become overloaded 

and fail. In order to predict 

the fatigue behavior of a material, the 

remaining strength or the remaining 

number of cycles, the material properties 

and the environmental/loading 

conditions must be related to the various 

damage modes that the material is 

likely to experience. 

When developing a fatigue life prediction 

model, it is important to evaluate 

how the model is able to handle a variety 

of materials and test conditions. 

Several researchers have compiled a 

large composite material fatigue database 

consisting of data from fatigue 

testing on wind turbine blade materials. 

2,3 Over 190 polymeric composite 

materials with over 12 000 individual 

coupon tests were conducted. Their 

research was focused on materials 

with lay-up combinations of 0°, ±45°, 

and 0°/±45° fabrics, manufactured by 

either hand lay-up or resin transfer 

molding (RTM). 

Fatigue Model Development 

The proposed GFRP fatigue life 

model 4 uses the internal strain energy 

of the material as the damage metric; 

this energy is expended due to damage 

in the forms of matrix cracking, 

fiber/matrix interface failure, delaminations, 

or fiber breakage before rupture. 

Strain energy was chosen as the 

damage metric because of its ability 

to represent multiple damage modes 

through one measurement and its high 

sensitivity to damage accumulation 

due to the squaring of the stress/strain 

term. While other researchers 5 have 

previously used the strain energy to 

predict fatigue life, this more recent 

model differs in the following ways: 

1. It is laminate-derived instead of 

lamina-derived, meaning no ply 

mechanics calculations need to be 

made. 

2. It focuses on utilizing a minimum 

number of experimental tests to 

determine material coefficients. 

3. It relates the strain energy back to 

the stress versus number of cycles 

curve in order to provide physical 

meaning. 

4. It is intended for use in industry, 

therefore attempts to simplify the 

process while still providing a reasonable 

fatigue life prediction. 

The expenditure of strain energy 

occurs in three stages (Fig. 1): Stage I 

is a steep curve of energy loss as the 


Expended energy (U) 

Failure 

Stage I Stage II (Linear) Stage III 

N 

f 

Uo 

Uo 

2 dU 

dN 

2 a σ σ 

= ( ) 

= 

Model Evaluation 

b 

( ( max ult ) ) 

(10) 

Fatigue test results were selected from 

the composite material fatigue database 

2 using the following controlling 

criteria: 

Cycles to failure (N) 

Fig. 1: Three stages of energy expenditure in fatigue 

material is initially loaded, and generally 

lasts for about 15% of the fatigue 

life; Stage II is nearly linear, with the 

slope a characteristic of the material 

and testing conditions, and generally 

lasts for about 75% of the fatigue life; 

Stage III is again a steep curve leading 

to material rupture over the last 10% 

of the fatigue life. 4,6 

Equation (1) describes the rate of 

the release of strain energy, U j , with 

respect to the number of cycles, N j , 

as a function of the mean strain, e m , 

the strain range, e r , the amount of 

expended strain energy corresponding 

to the current cycle count, and the particular 

composite material type, C t . 4 If 

the material type is kept constant, then 

this relationship supposes that the rate 

of release of strain energy with respect 

to the number of cycles is a function of 

the loading conditions: 

dU 

j 

= 

m r t 

dN f ( , , U C j, (1) 

) 

j 

The strain energy at any cycle of a 

fatigue test can be determined from 

the deflection (or strain) and loading 

data; for a tension–tension fatigue test, 

this is shown in Eq. (2), where P j is 

the load and d j is the deflection at any 

cycle j: 

U 

P j 

j 

j 

= 

(2) 

2 

To account for more complex material 

properties, it would be acceptable to use 

different versions of the strain energy 

equation. The strain energy density, W, 

could be calculated as shown in Eq. (3) 

for elastic plane stress problems. 5 

( ) 

W = 

x x 

+ 

y y 

+ 

xy xy 

/2 (3) 

The strain energy model calculates 

the initial strain energy U 0 using the 

mean loading stress (s mean ) just prior 

to initiating fatigue, as shown in Eq. 

(4), where A is the cross sectional area 

of the sample, L is the gage length, 

and E static is the static modulus of 

elasticity. 4 

2 

AL 

U0 

= mean 

(4) 

2Estatic 

After analyzing experimental strain 

energy release results for several GFRP 

fatigue tests, it was observed that the 

strain energy expended at ~90% of the 

fatigue life (the end of Stage II) was 

consistently close to 1,5 times the initial 

strain energy, U 0 . The data point at 

the end of Stage II can therefore be 

defined as (N f , U f ), as shown in Eqs. 

(5) and (6). 4 

U 15 f 

, U 

(5) 

0 

Nf 

09 , N 

(6) 

ult 

The energy release rate per cycle is 

relatively constant within Stage II and 

is dependent on the material type and 

loading conditions. Plotting the normalized 

strain against the energy release 

rate, the data can be curve-fitted with 

a power law as shown below, where a 

and b are the fatigue coefficients. 

b 

dU 

dN = a ⎛ ε ⎞ 

⎝ ⎜ ε ⎠ 

⎟ 

max (7) 

ult 

Because of the linearity of the Stage II 

energy release, dU/dN can be approximated 

as ΔU/ΔN [Eq. (8)]. If we assume 

that the strain ratio can be approximated 

as equivalent to the stress ratio 

due to a linear relationship between 

the stress and strain of the material 

[Eq. (9)], N f can be calculated as shown 

in Eq. (10). 

dU 

U 

15 , U U 

= = 

N N 

N 

U 

0 0 0 

d 

f 

2Nf 

 

 

max 

ult 

= 

(8) 

 

max 

(9) 

 

ult 

1. Tension–tension testing 

2. Stress ratio (s min /s max ), R = 0,1 

3. Minimum three load levels (between 

25 and 75% of max stress) 

4. Minimum five tests run to failure 

with at least 100 cycles each 

5. E-glass fibers 

6. Fatigue rate ≤ 20 Hz 

7. Generally ordered results (if s 1 > s 2 , 

then N 1 < N 2 ). 

After applying these criteria, test 

results for 109 different composites 

(1254 individual tests) were analyzed. 

Each material listing in the database 

included its static modulus of elasticity, 

static ultimate strength, and 

coupon dimensions, with stress load 

levels and number of cycles to failure 

for each coupon. Polyester was 

the most common matrix material, 

but results for vinyl ester, epoxy, and 

thermoplastic matrices were included. 

Most of the laminates selected were 

manufactured by RTM or lay-up techniques 

but included a variety of fiber 

architectures. 

For each composite, the fatigue coefficients 

(a and b) were obtained through 

a power regression on the criteriaselected 

data after plotting the results 

of Eqs. (8) and (9) (Fig. 2). Using the 

fatigue coefficients, the curve generated 

by Eq. (10) was plotted against 

the data on a log scale (see Fig. 3). 

In order to assess the accuracy of the 

model in fitting data, curves of ±5% 

log(N f ) and ±10% log(N f ) were generated, 

with any data lying within these 

envelopes considered to be reasonably 

well-modeled (Fig. 3). The number of 

data points within each error envelope 

was then tallied for each sample. 

Results 

Composite Material Fatigue 

Database 

The results of the fatigue database 2 

analysis of the model are shown in 

Table 1. Since data scatter is as likely 

as a poor-fitting model to result in data 


Energy release rate 

ΔU/ΔN in N-mm/cycle 

Normalized max stress (s max /s ult ) 

10 

9 

8 

ΔU/ΔN 

7 

Power (ΔU/ΔN) 

6 

5 

y = 2610,44 × 17,67 

4 

R 2 = 0,989 

3 

2 

1 

0 

0 0,2 0,4 0,6 0,8 

Fig. 2: Typical mate rial energy release rate 

0,85 

0,75 

0,65 

0,55 

0,45 

0,35 

Normalized applied stress (s max /s ult ) 

0,25 

0 2 4 6 8 10 

Cycles to failure (Log N) 

Fig. 3: Typical log error analysis (on glass/vinyl ester composite) 

Actual 

Model fit 

Upper limit (+10%) 

Lower limit (−10%) 



Total Within 5% log error Within 10% log error 

Number Number Percentage (%) Number Percentage (%) 

Full data set 1260 1034 82,1 1222 97 

R 2 > 0,9 1201 1003 83,5 1171 97,5 

R 2 > 0,95 1099 935 85,1 1078 98,1 

R 2 > 0,98 623 565 90,7 621 99,7 

Table 1: Log error anal ysis results 

points located outside of these error 

envelopes, results were also considered 

for different correlation coefficient 

(R 2 ) values obtained from the initial 

power regression. A higher R 2 value 

would indicate less scatter; using R 2 > 

0,98 (Table 1) the model is able to fit 

over 90% of the fatigue data to within 

the ±5% error envelope. 

Once it had been shown that the 

model provided good accuracy at fitting 

the data, another analysis was run 

to assess its ability to predict coupon 

fatigue life. Only composites with polyester 

matrices and a total of nine data 

points were considered, resulting in 14 

laminates and a total of 126 fatigue 

tests. The same curve equation and 

error envelopes were generated for 

each composite using several different 

fatigue coefficients. The power regression 

was performed using anywhere 

from two to nine logically selected 

data points and the number of data 

points within the error envelopes was 

tallied. A curve was also generated 

for each composite using only a single 

data point (~50% ultimate strength) 

and the loose correlation of Eq. (11). 

( 

) 

b = 20 1 fiber volume fraction (11) 

Grouping and analyzing the data by one 

or more of the following characteristics 

at a time produced no noticeable 

trends in the values of either the a 

or b coefficients: matrix type, common 

fiber architecture, manufacturing 

process, material thickness, material 

cross-sectional area, modulus of elasticity, 

testing rate, and ultimate tensile 

strength. The differences in matrices 

may have had a slight impact on the 

b-coefficient (with epoxies averaging 

11,8, polyesters 11,9, and vinyl esters 

12,1), but was not large enough to be 

significant and did not account for the 

high degree of variation within each of 

those groups. 

The linear approximation of Eq. (11) 

was obtained by running regressions 

on plots of the fiber volume fraction 

versus the b-coefficient for several 

smaller groupings of RTM polyester-matrix 

data (e.g. unidirectional 

samples, 0°/45° samples, etc.). Plotting 

the b-coefficient against the fiber volume 

fraction still does not account for 

the high amount of scatter (Fig. 4), 

but does allow for a reasonable average 

to be obtained. An attempt was 

made to normalize the fiber content 

with respect to the 0° direction before 

plotting against the b-coefficient, but it 

did not provide any better an approximation. 

Equation (11) provides an 

approximation of the b-coefficient for 

a majority of samples only if they have 

polyester matrices and were manufactured 

by RTM; the error can be much 

higher for those samples with other 

matrices or manufacturing methods. 

The percentages of the 126 data points 

that fell within the ±5 and ±10% error 

envelopes using between one and nine 

data points to obtain the fatigue coefficients 

are shown in Fig. 5. 

Glass/Vinyl Ester Composite 

The same logarithmic error envelope 

analysis was carried out using fatigue 

test results from a glass/vinyl ester composite 

material. Six tension– tension 

fatigue tests were run at six different 

stress levels between 35 and 70% of 

the ultimate static strength. 

After performing a similar curve fit 

analysis as was conducted on the database 

materials, it was determined that 

the material had an R 2 value of 0,995 

for the power regression and that all 

of the six data points lay within the 

±5% error envelope (Fig. 3). After 

assessing the curve prediction analysis, 

it was found that at least five of 

the data points fell within the ±5% 

error envelope using between two 

and six samples to obtain the fatigue 


-Coefficient 

20 

18 

16 

14 

12 

10 

8 

6 

4 

2 

y = −20,428x + 19,535 

R 2 = 0,2612 

0 

0,20 0,25 0,30 0,35 0,40 0,45 0,50 0,55 

Fig. 4: L inear regression on b-coefficient versus fiber volume fraction (0/±45 fibers) 

Average % of samples within error 

100% 

95% 

90% 

85% 

80% 

75% 

70% 

65% 

60% 

55% 

50% 

Number of samples a-Coefficient b-Coefficient R 2 value Within 5% Within 10% 

used 

2 5,969 14,245 1 6 6 

3 5,998 14,246 1 6 6 

4 4,990 14,122 0,998 5 6 

5 5,006 14,085 0,997 5 6 

6 6,616 14,39 0,995 6 6 

Table 2: Model coefficients and prediction analysis for glass/vinyl ester composite 

coefficients (Table 2). An appropriate 

b- approximation has not yet been 

determined for this resin type, so a 

single-sample fatigue life prediction 

could not be obtained. 

Component Data 

Generally, few fatigue tests are run 

to failure on composite components 

owing to the difficulty and expense 

of full-scale testing. One researcher 7 

ran six bending fatigue tests each on 

two glass/vinyl ester composite beams 

(one an I-beam, the other a box 

beam). Using his data and an adaptation 

of U 0 for strain and bending [see 

Eq. (12), where I is the moment of 

inertia of the sample, L is again the 

gage length, and c is the distance from 

the neutral axis], a fit and prediction 

Fiber volume fraction 

Fig. 5: Erro r from varying number of samples used to plot the prediction curve 

0,60 

5% log error 

10% log error 

1 2 3 4 5 6 7 8 9 

Number of samples used (out of 9) 

analysis could be run for full-scale 

components. Segments of the beams, 

each of dimensions 120 mm × 120 mm 

× 6,5 mm were fatigue tested over a 

span of 1,83 m. 

mean 

ILEstatic 

U0 

= 2 

2 

(12) 

6c 

The R 2 values on the power regression 

for the box and I-beams were 

0,967 and 0,985, respectively. All 

12 of the data points lay within the 

±5% error envelopes; in fact, they 

were also all within curves of ±2,5% 

log(N f ) (Fig. 6). The prediction analysis 

revealed that using anywhere 

between three and six data points 

to obtain the fatigue coefficients 

resulted in all six data points being 

within the ±2,5% error envelope for 

both beams. If only two data points 

were used to obtain the fatigue coefficients, 

five of the six data points 

were within the ±2,5% error envelope 

and all six were within the 

±5% error envelope. An appropriate 

b- approximation has not yet been 

determined for component fatigue, 

so a single-sample fatigue life prediction 

could not be obtained. 

Environmental Effects on 

Fatigue in Composites 

Both increased temperature and the 

presence of moisture have similar 

effects on polymeric composites—both 

induce stress by swelling and both relax 

stress by softening. 8 Several researchers 

9 have shown that the fatigue damage 

accumulation sequence was essentially 

the same at each temperature level 

they tested (ranging between −20 and 

150°C) but the rate of accumulation 

increased with temperature. 

Fatigue tests on the tested glass/vinyl 

ester composite coupons were conducted 

under varying environmental 

conditions in both bending and tension–tension 

fatigue with an R-ratio 

(s min /s max ) of 0,2. The environmental 

conditions included the absence or 

presence of salt water during testing, 

elevated temperatures, and accelerated 

immersion preconditioning (at 

12 MPa), and testing was performed 

at several stress levels (from 35 to 

70% of ultimate stress). While it has 

been shown 10 that the absence or 

presence of salt often has little effect 

on the performance of composites, 

salt water was used for the testing in 

order to most closely simulate ocean 

conditions. 

In bending fatigue, the most common 

failure mode was delamination of the 

outer layers on the tension side of the 

sample, believed to be initiated by 

microcracking at the midpoint. The 

presence of salt water reduced the 

fatigue life of the material to 84% of 

the value of a dry test at room temperature 

and a stress level of 63% of 

ultimate load, and to 54% of the value 

of a dry test at room temperature and 

a stress level of 50%. The effect of 

temperature on the life of the material 

was found to be approximately linear 

(with a higher temperature leading to 

a reduction in fatigue life), at least on 

those samples that were also tested in 

a salt water environment (see Fig. 7). 

The tests confirmed that the fatigue 

life of the material is decreased by the 


Normalized max strain (e max /e ult ) 

Fig. 6: I-beam fa tigue model fit 


0,85 

0,80 

0,75 

0,70 

0,65 

0,60 

0,55 

Actual 

Prediction 



0,50 

Upper limit (+2,5%) 

Lower limit (−2,5%) 

0,45 

4 4,5 5 5,5 6 

70 

60 

50 

40 

30 

20 

10 

Cycles to failure 

Linear (cycles to failure) 

Fig. 7: Temperature ef fect on bending fatigue 

presence of salt water, elevated temperature, 

or increased loading level. 

The failure mode for all of the tension–tension 

fatigue tests seemed to 

be a combination of lamina separation 

and fiber breaking. The presence of salt 

water in the 50% stress level sample 

reduced the fatigue life of the material 

to 43% of the value of a dry test, 

while at 63% no salt water effect was 

observed—this makes it clear that salt 

water has a larger effect over time, as 

the 63% stress level tension test only 

lasted for a few hours. As compared to 

bending fatigue data under the same 

stress levels, the tension fatigue tests 

exhibited much fewer cycles to failure 

and were more susceptible to environmental 

effects. This increased susceptibility 

in tension fatigue is likely due 

to the higher overall stress on fiber/ 

resin interfaces and the increased tensioned-surface 

area, allowing greater 

permeability. 

Through the absorption tests, the salt 

water saturation point of the material 

was determined to be between 0,10 

and 0,17% of the weight. The slight 

increase in immersion temperature 

Cycles to failure (Log N) 

0 

0 5000 10 000 15 000 20 000 25 000 30 000 35 000 

Number of cycles to failure 

(16°C) resulted in a decrease in the 

number of cycles to failure in all 

three tests by 67%. It appears that 

room temperature immersion conditioning 

could reduce the fatigue 

life of the material to 50 to 65% of 

the fatigue life of the pre-immersion 

material, while 38°C immersion conditioning 

could reduce the fatigue 

life of the material to 15 to 25%. 

More tests need to be run at different 

stress levels before the strain 

energy model can be applied to the 

tests run under different environmental 

conditions. 

Conclusion 

y = −0,0019x + 74,587 

R 2 = 0,985 

The strain energy fatigue model 

appears to provide both a good fit and 

a good prediction for the fatigue life 

of GFRP composite materials. The 

large amount of data analyzed from 

the composite material fatigue database 

indicates the ability of the model 

to fit a variety of GFRP materials with 

80 to 90% of the tests falling within 

±5% of the log number of cycles to 

failure. The model was also shown to 

be able to predict the fatigue life of 

polyester GFRPs to within ±5% of 

the log number of cycles to failure 

using only two experimental values 

with a success rate of over 75%; using 

three increased the success rate to 

over 82%. It appears that the model 

can predict values nearly as accurately 

with nine samples as it can with 

only two or three samples, illustrating 

how it only requires a minimum 

number of experimental data points. 

The results of the analysis of the component 

fatigue tests suggest that the 

model is able to fit and predict the 

fatigue life for a full-scale composite 

component as accurately as for a coupon 

while maintaining the same level 

of simplicity. The model was able to 

fit and predict the fatigue life of both 

of the components tested to within 

±2,5% of the log number of cycles to 

failure. 

The introduction of moisture or elevated 

temperatures has been shown 

to significantly decrease the fatigue 

life of the glass/vinyl ester composite 

material. This effect seems to be particularly 

linear for temperature changes 

and is relative to the testing length for 

the presence of salt water. Pre-immersion 

appears to produce a consistent 

reduction in fatigue life, depending on 

the immersion temperature. 

The strain energy fatigue model provides 

a simple method to predict 

fatigue life to within acceptable levels 

of accuracy in many structures, such as 

windmill blades, bridge decks, or deepsea 

pipes. Additional work should 

focus on developing material and testing 

condition-based approximations 

for b-coefficients, as well as continuing 

to test the model against componentscale 

and environmental fatigue test 

results. 

References 

[1] Liang R, GangaRao HV. Applications of 

fiber reinforced polymer composites. In Polymer 

Composites III, 2004 Creese R, GangaRao 

HV (eds). DEStech: Lancaster (PA), 2004; 

173–187. 

[2] Samborsky DD, Mandell JF. DOE/MSU 

Composite Material Fatigue Database, Version 

18.1, 2005. 

[3] Mandell JF, Samborsky DD. DOE/MSU 

Composite Material Fatigue Database: Test 

Methods, Materials, and Analysis. Contractor 

Report SAND97-3002, Albuquerque, NM: 

Sandia National Laboratories, 1997. 

[4] Natarajan V, GangaRao HV, Shekar V. 

Fatigue response of fabric-reinforced polymeric 

composites. J. Compos. Mater. 2005; 39(17): 

1541–1559. 


[5] Ellyin F, El-Kadi H. A fatigue failure criterion 

for fiber reinforced composite laminae. 

Composite Structure 1990; 15: 61–74. 

[6] Dittenber DB, GangaRao HV. Evaluation of 

fatigue life prediction model for GFRP composite 

materials. Proceedings of SPE-ANTEC 2010; 

2010 May 16–20; Orlando (FL), United States; 

2010. 

[7] Nagaraj V. Static and Fatigue Response of 

Pultruded FRP Beams without and with Splice 

Connections. College of Engineering and 

Mineral Resources [thesis]. Morgantown (WV), 

West Virginia University, 1994. 

[8] Weitsman Y. Moisture in composites: sorption 

and damage. In Fatigue of Composite Materials. 

Reifsnider, KL (ed.). Elsevier: Amsterdam, 1990; 

385–429. 

[9] Khan R, Khan Z, Al-Sulaiman F, Merah 

N. Fatigue life estimates in woven carbon 

fabric/epoxy composites at non-ambient temperatures. 

J. Compos. Mater. 2002; 36(22): 

2517–2535. 

[10] McBagonluri F, Garcia K, Hayes M, 

Verghese KNE, Lesko JJ. Characterization of 

fatigue and combined environment on durability 

performance of glass/vinyl ester composite 

for infrastructure applications. Int. J. Fatigue 

1999; 22: 53–64. 

Further Information 

1. Constructed Facilities Center at 

West Virginia University. http://www. 

cemr.wvu.edu/cfc/. 

2. Department of Energy/Montana 

State University Composite Material 

Fatigue Database. http://www.coe. 

montana.edu/composites/. 


A Composite Bridge is Favoured by Quantifying 

Ecological Impact 

Ryszard A. Daniel, PhD, Eng., Ministry of Transport, Public Works and Water Management, Division of Infrastructure, Utrecht, 

The Netherlands. Contact: richard.daniel@rws.nl 

Abstract 

Carrying traffic loads is not the only objective of bridge designers nowadays. 

Other demands include constructing a bridge in a sustainable way, which reduces 

pollution and other harm to the environment. In The Netherlands, the government 

responds to such demands by promoting technologies and materials that 

decrease the environmental impact of construction projects. 

An assessment of that impact is, however, quite complex for bridge projects. 

The existing analytical methods, such as life-cycle analysis (LCA), require an 

extensive data input. Moreover, their results are more reliable for relatively 

simple products of short life cycles, for example, door or window frames, than 

for complex construction projects. In construction projects, the life cycles cannot 

be determined with the same precision and the materials are usually chosen 

in the very early stage of design. As a result, the data required by the LCA are 

often incomplete or even disputable. Therefore, there is a demand for ecological 

analysis methods that enable quick scanning of several material options, require 

less-extensive data input and are hardly, or not, vulnerable to arbitrariness. 

Keywords: FRP structures; eco-analysis, material choice; sustainable material; 

sustainable bridge; energy input; exergy; emissions; pollution data. 

This paper answers the above-mentioned 

demand by presenting a method 

for ecological material selection for 

a bridge. It shows a way to quantify 

the environmental impacts of possible 

material choices in comparable 

terms and to assess those choices with 

respect to their impact. The method 

was first developed and applied for 

the quay footbridges in the Noordland 

inner harbour, province of Zeeland, 

The Netherlands. Five material options 

were considered: structural steel, stainless 

steel, composites (fibre-reinforced 

polymers, FRPs), aluminium and reinforced 

concrete. The analysis allowed 

evaluating these options in terms of 

three crucial ecological indicators: 

energy consumption, pollution to air 

and pollution to water. 

The ecological analysis was performed 

along with the costs and 

service-life assessment. The comperforming 

remarkably well since 

then, validating the computed ecological 

and other indicators. Its good 

performance suggests the possible 

construction of more similar footbridges 

in that area in future. This 

paper presents a comparison of those 

indicators for the material options 

considered, and discusses these and 

some selected problems of the ecological 

analyses. 

The applied ecological analysis has 

been presented on various occasions 

since the bridge construction. 1–3 Yet, 

it still evokes much interest because 

of the importance of environmental 

engineering in relation to, for example, 

climatic processes. This paper aims to 

respond to that interest, giving more 

details of both the applied ecological 

analysis and the constructed FRP 

bridge. 

Introduction 

puted performances of all the material 

options considered resulted in 

an advice to construct a bridge of 

pultruded FRP profiles (Fig. 1). The 

customer followed that advice. It was 

the first bridge constructed using this 

technology in The Netherlands. The 

bridge was assembled and brought 

into service in 2001. It has been 

Project Objectives and Scope 

The Dutch province of Zeeland is 

a costal area in the south-western 

delta of the rivers Rhine, Meuse and 

Scheldt. High exposure to sea water, 

wind loads and chloride corrosion 

form part of the usual design specifications. 

At the end of 1999, the Regional 




Paper received: January 14, 2010 


Fig. 1: Installation of the Noordland inner harbour footbridge 


North 

Sea 

B 

NL 

Authority for Public Works and Water 

Management ordered an investigation 

on construction materials for a footbridge 

in the Noordland inner harbour 

that forms part of the Eastern Scheldt 

Storm Surge Barrier complex. The 

bridge provides a double span access 

to a mooring pontoon (Fig. 2). The 

new bridge was to replace the old steel 

bridge that was largely deteriorated by 

corrosion after only 35 years of service. 

This was not surprising, considering the 

extreme conditions at that location. 

The service load of the bridge is 400 

kN/m 2 . Other loads are wind, snow, 

glitter ice, and so on. There is no navigation 

under the bridge. The support 

level to pontoon varies because of the 

tides. The allowable span deflection 

is limited to 1/250. The customer was 

interested in comparing the performances 

of the first four bridge materials 

from the following list: 

– structural steel (with coating); 

– stainless steel; 

– synthetic material (composite); 

– aluminium; 

– concrete. 

The fifth material was investigated 

later for the sake of completeness. The 

weight of a concrete bridge made it 

unfit for a pontoon support. Timber 

was also not an interesting option 

because of its maintenance requirements, 

combustibility and short service 

life at this particular location. 

Nonetheless, it certainly can be considered—also 

with respect to the environment—in 

other bridge projects. In this 

paper, timber is not included, because 

the considerations that determine its 

environmental performances are of a 

different nature. An important criterion 

is, for example, sustainable forest 

management. 4 It is difficult to quantify 

such criteria in a manner that allows 

for a comparison with other materials. 

The performances of each option had to 

be quantified in terms of the following 

D 

A 

A 

13,5 

Fig. 2: Bridge location and dimensions (Units: m) 

13,5 

Pontoon 

1,60 

A - A 

1,10 

four criteria: construction costs, maintenance 

costs, service life and environmental 

impact. Aesthetics was not a 

prior concern at this desolate location. 

Maintenance and service life appeared 

to show a strong correlation. It was, 

therefore, agreed to impose a uniform 

service life of 50 years on all material 

options. This period reflects the current 

design views in The Netherlands. 

In this way, the number of assessment 

criteria was reduced to three, which 

simplified the analysis. 

Construction and maintenance costs 

are quite common criteria in engineering; 

therefore, only the final results are 

presented. To quantify the environmental 

impact, however, an investigation 

method had to be set up first. As 

already discussed, the existing methods 

like the LCA 5 were not very helpful. 

The footbridge appeared to be too 

complex and too vaguely determined 

at this stage. Making detailed bridge 

designs and life-cycle inventories for 

all material options was, obviously, not 

the intention. Therefore, a simplified, 

but workable, two-way evaluation was 

chosen: 

– energy consumption analysis— 

taking also account of the energy 

“stored” in materials and products 

(the so-called “exergy” method 6 ); 

– analysis of loads (pollutions) to water 

and air as a result of material winning, 

processing, fabrication of the 

final product and its maintenance. 

In current views, the first approach 

can be seen as a measure of not only 

energy consumption as such (i.e. 

decrease of global energy resources) 

but also the processes resulting from 

fossil fuel combustion, like the greenhouse 

effect, rise in ocean level, global 

climatic changes, and so on. The second 

approach (loads to air and water 

apart) produced global pollution 

data of the bridge options under consideration. 

Loads to soil appeared 

to be insignificant, but they can be 

analysed in the same way, whenever 

relevant. 

Conceptual Designs 

As the materials in question represented 

in fact five groups of materials, 

the material grades had to be chosen. In 

accordance with the existing practice, 

the following grades were selected: 

– structural steel: S235J0 or S355J0, 

according to the European norm 

EN 10025. An arc-sprayed aluminium 

coating was considered as an 

alternative to the conventional paint 

system. 

– stainless steel: X2CrNi18-11 or 

X2CrNiMo18-14-3 according to the 

European norms (AISI 304L or 316L 

according to the US standards). 

– composite: fibreglass-reinforced 

polyester resin (FGRP) in pultruded 

sections. 

– aluminium: AlMgSi1,0 F31 according 

to the DIN 1748 code (or 6061 

and 6063 alloys according to the 

ASTM B221). 

– concrete: B35 according to the 

European norm EN 1992-1, 150 kg 

of reinforcement per 1 m 3 ; 100 kg 

of other steel accessories (e.g. handrails) 

per 1 m 3 . 

The next step was to complete five 

rough conceptual bridge designs, one 

in each optional material. It soon 

became clear that each option required 

a different form, system, manufacturing 

approach, and so on. In structural 

steel and concrete, for example, conventional 

girders with separate handrails 

were an evident choice, whereas 

in the other, more expensive materials 

the handrails were integrated in truss 

or truss-like girders. Major differences 

appeared also in section shapes, deck 

systems, and so on. In Fig. 3, one span 

of the bridge in each of the five materials 

is shown. The structural analysis was 

very brief in all cases. Nevertheless, it is 

fair to say that the bridge spans shown 

in Fig. 3 are representative for the 

considered materials, and comparable 

with each other in terms of strength 

and durability. 

The material mass estimations are 

based on a brief analysis and data 

from similar projects. These masses 

form the data for estimating both 

total costs and environmental impact. 

Remarkably large mass differences are 

seen between the material options. This 

requires a few comments. The mass of 

structural steel span would have been 

lower (2200–2500 kg) if truss girders 


Girder h = 360 

of HE240A 

Lower chord 

φ168, 3 × 6,3 

Upper chord/Handrail 

Lower chord 

U240 × 72 × 8 

Diagonals 

φ 70 × 60 × 5 

Chords 

φ 200 × 100 × 10 

Girder 

φ 450 × 200 

Option 1: Structural steel 

One span mass: 3000 kg 

Cross girders 

HE160B 

Option 2: Stainless steel 


Cross girders 

φ139, 7 × 6,3 

Option 3: Composite 


Cross girders 

H200 × 100 × 10 

Option 4: Aluminium 


Option 5: Concrete 

One span mass: 14 000 kg 

Deck plate 150 

FSC timber 50 mm 

Cross girder 

HE240B 

FSC timber 50 mm 

Cross girder 

φ168, 3 × 6,3 

Cross girder 

13 500 1600 

Fig. 3: Bridge span in five material options (length units: mm) 

integrated with handrails were chosen 

instead of beams. This has deliberately 

not been done to justify neglecting 

the impact of steel coating. In any 

case, however, the composite and aluminium 

bridges appear to be the lightest. 

The concrete bridge is 5–10 times 

heavier than the other bridges. The 

dead weight was of minor importance 

here, as long as it did not cause pontoon 

overloading. A smaller weight is, 

however, desirable in large bridges. It 

allows for higher traffic loads, lighter 

foundations, pillars, transport and 

assembly equipment. 

Global Assessment 

1100 

The bridge conceptual designs were 

employed to collect more data for the 

evaluation—not only the total material 

masses. The drawings in the form 

of outlines prompted specific questions 

and enabled collection of relevant 

data on the market. The desired 

data covered, in general, the following 

subjects: 

– quantities and unit prices of the 

materials involved; 

– available manufacturing technologies, 

their costs, conditions, quality 

assurances and risks; 

– transport and assembly requirements, 

like access, time, heavy equipment, 

specific provisions; 

– inspection and maintenance frequencies 

during the service life; 

– environmental impact of all processes 

involved. 

The accuracy of these data was not 

always high because of the preliminary 

nature of bridge design. In some cases, 

rough estimations had to be made. The 

concerned specialists agreed, nonetheless, 

that a sufficient, well balanced base 

was provided to evaluate the bridge 

options. The concise results of this 

evaluation are shown in Table 1. The 

general conclusions are as follows: 

– In terms of construction costs, the 

structural steel and the concrete 

bridges are favourable. The stainless 

steel bridge is too expensive; the 

composite and aluminium bridges 

score in the middle. 

– In terms of maintenance costs, the 

scores are opposite. The stainless 

steel bridge is the cheapest, followed 

by the concrete bridge. The structural 

steel bridge (conventionally 

painted) is the most expensive. The 

scores of the composite and aluminium 

bridges lie in between. 

– Adding construction and maintenance 

costs (whether or not capitalized) 

puts the concrete bridge in the 

first place and the structural steel 

bridge in the second. The composite 

bridge takes a good third place, 

closely followed by aluminium. The 

stainless steel bridge is evidently the 

most expensive. 

– Analysis of the energy consumption 

makes the composite bridge a 

winner. Every other option results 

in energy consumption that is more 

than two times as high. Energy consumption 

is seen as an important 

indicator of the contribution to the 

global warming effect. 

– The composite bridge is also the best 

in terms of the resulting water and 

air pollution levels. The structural 

steel bridge is the second, concrete 

bridge is the third and aluminium 

bridge is the fourth. 

The customer was advised as follows: 

if construction cost was the primary 

concern, the choice of a structural 

steel bridge was the best. But if a little 

extra cost was acceptable in the interest 

of the environment, the composite 

bridge of pultruded profiles was the 

best choice. An additional argument 

in support of the composite bridge 

was the innovative character of such 

a project. It was to be the first composite 

bridge of pultruded profiles in 

The Netherlands. The customer was 

indeed in a position, and willing, to 

choose the second, pro-environmental 

option. The composite bridge was constructed 

in October 2001. It has been 

closely monitored since then, confirming 

the results of the analysis. 


Bridge material 

Construction costs 

(EUR) 

Maintenance costs 

(EUR) 

Criterion 

Environment: Energy 

consumption (MJ) 

Environment: Critical volume 

of polluted air and water 

Structural steel Painted: 40 000 Painted: 30 000 “Exergy” method: 294 000 Water: 697,4 m 3 

Aluminium coated: 50 000 Aluminium coated: 6 000 Air: 7,09 × 10 6 m 3 

Stainless steel Steel AISI 316L: 110 000 

Steel AISI 304L: 96 000 

Steel AISI 316L: 6000 

AISI 304L more, life 

cycle shorter 

Rough estimation: 

“Exergy” method: 329 600 Not investigated but certainly 

more pollution than for structural 

steel 

“Exergy” method: 120 000 Water: 85,8 m 3 

Composite Pultruded sections of 

FGRP: 70 000 

17 000 

Air: 7,92 × 10 6 m 3 

Aluminium Quality AlMgSi1 Rough estimation: 19 000 “Exergy” method: 268 700 Water: 565,3 m 3 

acc. to DIN 1748: 77 000 Air: 41,10 × 10 6 m 3 

Concrete Reinforced concrete B35, 

handrails etc: 30 000 

Rough estimation: 

10 000 

Table 1: Performances of the five material options for the bridge 

“Exergy” method: 277 200 Water: 341,9 m 3 

Air: 31,04 × 10 6 m 3 

Eco-Analysis in Terms 

of Energy Consumption 

Ecological performances of a particular 

material option cannot be expressed in 

a single indicator, although it is advisable 

to keep the number of indicators 

small. Energy consumption, therefore, 

does not reveal everything about the 

ecological performances, but it is an 

important indicator in this field. It 

requires no argument today that energy 

consumption is a global environmental 

issue in both direct and indirect senses. 

In the first sense, it decreases the global 

energy resources which are—for the 

biggest part—not renewable. In the 

second sense, it harms the environment 

in many ways, including its contribution 

to the emission of CO 2 , other 

“greenhouse gases” and the resulting 

climatic changes. However, if the latter 

is seen as the main or only issue of ecoanalysis 

(which is not the author’s view), 

a direct analysis of greenhouse gas 

emission, 7 will be more appropriate. 

The required data is that of the energy 

use for the processing and manufacturing—from 

obtaining the raw materials 

to the final product—of one mass 

unit of the product in question (in 

MJ/kg). These data vary because the 

same materials and products can be 

obtained using different technologies. 

As eco-analyses are quite new, there is 

still much arbitrariness in defining the 

data. Therefore, it is always advisable 

to check which processes are covered 

by the received data. During this study, 

for example, the following energy consumption 

rates for structural steel products 

were found in various sources: 

– Source 1 (The Netherlands): 6 

46 MJ/kg; 


31 MJ/kg; 


18 MJ/kg; 

– Source 4 (USA): 10 6 MJ/kg. 

Such differences may be surprising to 

engineers who are used to approved 

specifications, standard codes and reliable 

and well tested data. However, 

the databases held by various institutes 

appear to be usable. When high figures, 

for example, for structural steel are 

quoted, they usually include energy 

input for rolling, surface treatment, 

transport, welding, fabrication, delivery 

and assembly of the structure. Low 

figures comprise smaller numbers of 

those processes. Data on other materials 

are collected in a similar way so that 

every database is usually consistent. It 

is, therefore, recommended to use data 

from the same source throughout the 

entire analysis. The lack of standards 

should temporarily be accepted. In 

the interest of the environment, one 

should rather critically apply the existing 

data than wait until they become 

better. 

The so-called “exergy” method was 

used to quantify the energy use for the 

five bridge options. In this method, the 

total energy consumption is a sum of 

energy value decreases for the materials 

in the processes involved. The 

analysis was limited to basic materials; 

wooden bridge decks in both structural 

and stainless steel bridges, stainless 

steel connectors in aluminium and 

FRP bridges and so on were ignored. 

The energy consumption rates per 

material unit were collected from the 

first 6 database except for composites 

(second 8 data base). Although both 

companies were involved in the official 

“Eco-indicator” project, 11 no uniform 

energy database for all materials 

was available at that time. The review 

resulted in some adjustments to the 

data for the purpose of this analysis 

(Table 2). According to recent views, 

the data for composites might still 

require a minor increase. These data 

should, however, not be confused with 

the much higher energy rates for plastics. 

Polyester resin usually makes up 

less than 50% in volume (about 30% 

in weight) of pultruded profiles. The 

rest is fibreglass. 

In the following example, energy consumption 

is estimated for a structural 

steel bridge: 

Total mass of two spans: 6000 kg. 

Assumed: 80% of the primary and 

20% of the secondary (recycled) 

material. Energy consumption on 

delivery: 

Ex 0 = 6000 × [0,8 × (46–7) + 0,2 

(1) 

× (36–7)] = 222 000 MJ 

The energy used during maintenance 

(2 × blast cleaning and painting) was 

approximated by subtracting the figure 

for unpainted structure (31 MJ/kg) 

obtained from another database. 9 To 

take account of the time delay (about 

20 and 35 years), a factor of 0,75 was 

introduced: 

Ex 1 = 6000 × 2 × 0,75 × (46 − 7 − 31) 

= 72 000 MJ (2) 

This gives the total energy 

consumption: 

Ex = Ex + Ex = 222 000 + 72 000 

0 1 

= 294 000 MJ 

(3) 

The energy consumptions for the other 

material options were estimated in a 

similar manner. This gave the energy 

impact graph for all the five bridge 

options (Fig. 4). 


Material 

Structural steel 

(e.g. S235J0) 

Stainless steel 

(e.g. AISI 316L) 

Composites 

(FGRP) 

Aluminium 

(e.g. AlMgSi1) 

Reinforced concrete 

(B35, handrails) 

Condition 

Primary 

Secondary 

Primary 

Secondary 

Primary 

Secondary 

These results are not as “hard” as, for 

example, those from structural analyses. 

One may wonder why the delay factor 

of 0,75 is used for the maintenance 

of the structural steel bridge—and if 

so, then why it is not applied to deck 

replacements in other bridge options. 

In this case, the engineers felt that spare 

decks of “unusual” materials (composite, 

aluminium) should be secured, that 

is, delivered together with the bridges. 

This assumption is, however, arbitrary. 

Another simplification is that the 

energy for dismantling after the service 

life has been neglected. Including 

it would probably point to the concrete 

bridge as the most energy-consuming 

option. Concrete demolition 

and utilization requires much energy. 

As mentioned, there are also differences 

in energy rating between various 

institutions and countries, especially in 

regard to composites. German data, 12 

often result in higher energy rates 

and American data 10 in lower energy 

rates. However, it is undisputable that 

the composite bridge had the lowest 

energy consumption. 

Loads to the Environment 

Energy analyses do not indicate how 

“clean” or “dirty” the considered 

Energy consumption 

value (MJ/kg) 

46 

36 

69 

54 

33 

— 

Remaining “stored” 

energy (MJ/kg) 

options are, that is, they provide no 

comparison in terms of environmental 

pollution. The problem with such a 

comparison is that each material option 

gives a spectrum of qualitatively different 

pollutions, which cannot simply 

be added up. The solution is found by 

taking account of the so-called “legal 

thresholds” of the particular pollutants. 

This was, to the author’s best knowledge, 

the first time that this approach 

was used in an infrastructure project. 

The applied method is derived from 

the so-called critical load method, 10 

and is based on the following two data 

records: 

– B m,i (kg/m 3 ), emitted masses of the 

pollutants i due to production and 

processing of 1 m 3 of the material m. 

Such emissions are usually recorded 

as loads to air, water and (exceptionally) 

soil. 

– B 0,i (kg/m 3 ), legal thresholds of the 

pollutants i in 1 m 3 of air, water and 

(exceptionally) soil. 

When these two data records are 

known along with the total mass G m 

and density γ m of the material m, the 

total critical volume of polluted air 

V a m or water V w m (m 3 ) can be computed 

as follows: 

7 

7 

11 

11 

9 

— 

Primary 137 33 

Secondary 45 33 

Primary 

Secondary 

Table 2: Energy consumption data for the five material options for the bridge 

350 000 

300 000 

250 000 

200 000 

150 000 

100 000 

50 000 

0 

Structural 

steel 

On delivery 

Stainless 

steel 

11 

— 

Maintenance 

Composite Aluminium Reinforced 

concrete 

Fig. 4: Energy impact of the bridge for the five material options 

2 

— 

V 

G 

γ 

m 

m 

= ×∑ 

m i 

B 

B 

mi , 

0 , i 

(4) 

Tables 3 and 4 present the emissions 

B m,i and their legal thresholds B 0,i for 

the four final material options: structural 

steel, composite, aluminium and 

concrete. The stainless steel option was 

not given up at that stage. The data for 

structural steel and aluminium bridges 

were collected from Refs. [10, 13]. The 

emission data for polyester resin were 

provided by the world market leader 

in this branch, and combined with the 

data for glass to give the aggregated 

emissions for FRP. The data for reinforced 

concrete (including steel accessories 

like handrails) were obtained 

by combining the records for concrete 

and steel. 

Apart from the global results (see 

Table 1), it is interesting to compare 

the pollutions to water and air qualitatively. 

For example, for the composite 

bridge, Eq. (4) and the data in 

Tables 3 and 4 give the following critical 

volumes of polluted air, V a cp and 

water V w cp: 

V 

V 

a 

cp 

w 

cp 

G B 

cp 

cp, 

i 

= × ∑ = 

γ i B 

cp 

0, i 

3 

−1 

4000 ⎛ 103 , × 10 12 , × 10 ⎞ 

= × 

+ ⋅⋅⋅+ 

3 

7 

1700 ⎝ 

⎜ 

− 

− 

90 , × 10 80 , × 10 ⎠ 

⎟ 

= 235 , × 3, 37 × 10 6 = 7, 92 × 10 

6 m 3 

(5) 

G B 

cp 

cp, 

i 

= × ∑ = 

γ B 

cp 

i 

0, i 

−6 

−2 

4000 ⎛ 20 , × 10 30 , × 10 ⎞ 

= × 

−5 

−3 

1700 

⎜ + ⋅⋅⋅+ 

⎝ 50 , × 10 10 , × 10 

⎟ 

⎠ 

= 2, 35 × 36, 5 = 85, 8 m 3 (6) 

The components of these sums, multiplied 

by the ratio G cp /γ cp , are represented 

in diagrams (left) in Fig. 5, along 

with the results for the other material 

options. The total critical volumes of 

polluted air and water are compared 

in pie charts (right) in Fig. 5. Also, the 

composite bridge appears to be more 

favourable than the other considered 

options. 

The analysis in this paper was deliberately 

kept simple. The bridge options 

were approached as single-material 

cases. Although there usually exists 

a single dominant material in all 

bridge projects, it may be advisable 

to consider other component materials 

as well. Examples are concrete 


Polluter Structural steel B st,i Composite B cp,i Aluminium B al,i Concrete B cr,i Threshold B 0,i 

CO 2 2,56 × 10 +3 1,03 × 10 +3 2,1 × 10 +4 4,95 × 10 +2 9 × 10 −3 

CO 9,58 × 10 +1 1,32 5,15 × 10 +1 3,48 4 × 10 −5 

CH 4 5,95 1,21 5,39 × 10 +1 9,89 × 10 −1 6,7 × 10 −3 

N 2 O 3,7 × 10 −2 4,8 × 10 −3 2,94 × 10 −1 1,51 × 10 −2 1 × 10 −7 

PM Fe/Al-oxi. * 2,2 × 10 −1 1,05 × 10 −1 1,65 6 × 10 −2 1 × 10 −7 

PM Si/Ca-oxi. * 4,2 × 10 −2 5,05 × 10 −1 2,7 × 10 −1 4,7 × 10 −1 3 × 10 −7 

SO 2 3,28 2,51 × 10 −3 1,27 × 10 +1 2,8 × 10 −1 1,2 × 10 −6 

NO x 3,08 2,83 2,45 × 10 +1 1,27 1 × 10 −5 

Styrene — 1,2 × 10 −1 — — 8 × 10 −7 

*PM = particulate matter (dust), here predominately Fe/Al or Si/Ca oxides. 

Table 3: Emissions to air for structural steel, composite, aluminium and reinforced concrete 

Polluter Structural steel B st,i Composite B cp,i Aluminium B al,i Concrete B cr,i Threshold B 0,i 

Aluminium 3,33 × 10 −6 2 × 10 −6 3,09 × 10 −5 1,65 × 10 −7 5 × 10 −5 

Ammonia 4,58 × 10 −3 1,1 × 10 −3 4,23 × 10 −2 2,38 × 10 −4 2,2 × 10 −3 

Cadmium 4,57 × 10 −5 2,1 × 10 −6 4,28 × 10 −4 2,18 × 10 −6 3,5 × 10 −6 

Copper 1,96 × 10 −8 7,9 × 10 −4 1,82 × 10 −7 0,99 × 10 −9 2 × 10 −4 

Cyanide 3,08 × 10 −4 7,4 × 10 −5 2,85 × 10 −3 1,6 × 10 −5 1 × 10 −4 

Fluoride 1,03 × 10 −1 2 × 10 −4 6,49 × 10 −3 3,51 × 10 −3 1,5 × 10 −3 

Manganese 6,07 × 10 −6 3,6 × 10 −6 5,64 × 10 −5 3,03 × 10 −7 5 × 10 −5 

Mercury 1,57 × 10 −4 7 × 10 −7 1,45 × 10 −3 7,53 × 10 −6 5 × 10 −6 

Zinc 3,97 1,4 × 10 −3 5,44 × 10 −2 1,35 × 10 −1 5 × 10 −3 

Cobalt — 3 × 10 −2 — — 1 × 10 −3 

Table 4: Emissions to water for structural steel, composite, aluminium and reinforced concrete 

Styrene 

NO x 

SO 2 

PM Si/Ca 

PM Fe/Al 

N 2 O 

Struct. steel 

CH 4 

Composite 

CO 

CO 2 

in 10 3 m 3 of air Aluminium 

Reinf. concrete 

0 5000 10 000 15 000 20 000 25 000 

Cobalt 

Zinc 

Mercury 

Manganese 

Fluoride 

Cyanide 

Copper 

Cadmium 

Ammonia 

Aluminium 

in 10 3 of water 

Struct. steel 

Composite 

Aluminium 

Reinf. concrete 

0 100 200 300 400 500 600 700 

31 040 

Loads to air 

341,9 

565,3 

7090 

Loads to water 

85,8 

7920 

41 100 

Fig. 5: Polluted air and water as a result of bridge construction with four material options 

and steel in cable-stayed bridges or 

steel and composite in steel bridges 

with composite decks. The discussed 

697, 4 

method can be applied in such cases 

too. Equation (4) then takes the following 

form: 

G B 

V = ⎛ 

j 

× 

⎞ 

ji , 

complex ∑⎜ 

⎟ 

j ⎝ γ 

∑ 

B 

(7) 

j i 0, i ⎠ 

where V complex is the critical volume 

of air or water polluted up to the 

respective legal threshold (m 3 ); G j 

is the total mass of material j in the 

considered complex material bridge 

option (kg); g j is the specific mass 

of material j (kg/m 3 ); B j,i is the mass 

of pollutant i emitted by production 

+ processing of 1 m 3 of material 

j (kg/m 3 ); B 0,i is the respective legal 

threshold of pollutant i in air or water 

(kg/m 3 ). 

This may look complex here, but once 

we have the databases B j,i and B 0,i in 

a PC, this sum presents no problem. 

In fact, it can easily be generated in a 

simple spread sheet, along with proper 

graphs. 

Conclusion and Future Outlook 

The considered case proves that synthetic 

composites (FRPs) constitute 

a very interesting material option for 

bridges in terms of environmental 

impact. A composite bridge project 


Fig. 6: Closed ‘traffic ducts’ concept 

requires less than half of the energy 

input that is required for an equivalent 

project constructed using steel, 

stainless steel, aluminium or concrete. 

In terms of loads to air, the composite 

bridge is the second “cleanest” option 

after the steel bridge. In terms of loads 

to water, the composite bridge is the 

undisputable winner. This makes composites 

an advantageous material for 

bridges, despite the slightly higher construction 

costs. 

The main reasons for the good performance 

of FRP are: 

– good mechanical properties, particularly 

the tensile strength, resulting in 

small quantities required; 

– very good chemical stability, resulting 

in low maintenance and long service 

life; 

– well-controlled processes, resulting 

in small error margins and low environment 

impact. 

The presented case should be seen as 

an indication, but not necessarily as 

evidence, for other bridge projects. 

Individual requirements and local 

conditions often play a decisive role 

in material selection. In the considered 

Noordland Bridge, for example, 

high corrosion resistance was particularly 

valued because of the surrounding 

environment (sea water). For road 

bridges, the relatively low elasticity 

modulus of composites may limit their 

applications or require other forms and 

structural systems, for example, “ribbon 

bridge”, 14 membrane deck, 15 high truss 

girders or closed traffic ducts. 16 The latter 

also (Fig. 6) offer other advantages 

for the environment. Yet, as the significance 

of environmental performances 

steadily grows, the synthetic composites 

will likely gain a stronger position 

in the construction market in the 

future. 

It is also predictable that the methods 

of environmental analyses will develop 

fast and that their results will enjoy a 

growing significance. It is important 

to develop objective, soundly based 

and well balanced tools enabling us to 

comparatively assess the environmental 

impacts on engineering choices. 

Only such tools can replace emotions, 

manipulations and free lobbying, 

which very often control these choices 

at present. Such tools should be rooted 

in official regulations, rather than in 

individual judgements. This is the main 

reason why the presented assessment 

method makes use of “legal thresholds”. 

Even if those thresholds are not 

perfect yet, they must be endorsed. 

The idea behind it is the same as for 

referring to the existing databases: it is 

better to use them and complain about 

their shortcomings than wait until they 

improve. 

References 

[1] Daniel RA. Environmental considerations 

to structural material selection for a bridge. 

Proceedings of the Bridge Engineering Conference. 

COBRAE, Rotterdam, March 2003. 

[2] Daniel RA. Construction material for a bridge 

with regard to the environment. Bautechnik 

2003; 80(1): 32–42. 

[3] Daniel RA. Ecological analysis of material 

selection for a bridge. Proceedings of 

33rd IABSE Symposium. IABSE, Bangkok, 

September 2009. 

[4] Daniel RA, Brekoo A, Mulder AJ. New 

materials for an old navigation lock. Land Water 

2001; 41(11): 36–38 (in Dutch). 

[5] Schmmelpfeng L, Lück P. Ökologische 

Produktgestaltung, Stoffstromanalysen und 

Ökobilanzen. Springer-Verlag: Berlin–Heidelberg, 

1999. 

[6] Elferink H. Energy analysis also useful for 

product improving. Energie- en Milieuspectrum 

1998; 11: 22–25 (in Dutch). 

[7] Anderson JE, Silman R. A life cycle inventory 

of structural engineering design strategies 

for greenhouse gas reduction. Struct. Eng. Int. 

2009; 19(3): 283–288. 

[8] Pré Consultants. Environmental Comparison 

of Harbour Approach Structures. Research 

report 2092. Amersfoort, August 1994 (in Dutch) 

(unpublished). 

[9] Intron Institute. Energy Analysis for the 

Motorway 16/13 Adaptation Options. Research 

report M715240/R980548. Houten, November 

1994 (in Dutch) (unpublished). 

[10] Mahadvi A, Ries R. Towards computational 

eco-analysis of building designs. Comput. Struct. 

1998; 67: 375–387. 

[11] Ministry of Housing, Urban planning 

and the environment. The Eco-indicator 99— 

Manual for Designers. The Hague, October 2000 

(in Dutch). 

[12] Hegger M, Fuchs M, Zeumer M. Integration 

vergleichender Nachhaltigkeitskennwerte von 

Baumaterialien. TU Darmstadt/Deutsche 

Bundesstiffung Umwelt, November 2005. 

[13] Sittig M. World-wide Limits for Toxic and 

Hazardous Chemicals in Air, Water and Soil. 

Noyes Publications: Park Ridge, NJ, 1994. 

[14] Schlaich M, Bleicher A. Carbon fibre 

stress-ribbon bridge. Proceedings of the 

COBRAE Conference: Benefits of Composites 

in Civil Engineering. COBRAE, Stuttgart, 

March 2007. 

[15] Daniel RA. Search of Optimal Shapes 

for Composite Bridges. Proceedings of the 

COBRAE Conference: Benefits of Composites 

in Civil Engineering. COBRAE, Stuttgart, 

March 2007. 

[16] Daniel RA. Shaping composite bridges 

for traffic and the environment. Proceedings of 

the 4th Int. Conference on Bridge Maintenance, 

Safety and Management. IABMAS, Seoul, July 

2008. CRC Press, Taylor & Francis Group: 

London–New York–Leiden, 2008. 


Experimental Assessment of Bond Behaviour 

of Fibre-Reinforced Polymers on Brick Masonry 

Enrico Garbin, Dr; Matteo Panizza, Dr; Maria Valluzzi, Prof. Dr; University of Padova, Dept. of Structural and Transportation 

Engineering, Padova, Italy. Contact: garbin@dic.unipd.it 

Abstract 

Existing masonry structures represent a significant amount of the architectural 

heritage. Many of these buildings are vulnerable to earthquakes. Consequently, 

they need structural improvements in order to meet the seismic requirements of 

recent building guidelines. In the last decade, there has been a growing interest in 

the application of Externally Bonded-Fibre Reinforced Polymers (EB-FRP) as 

strengthening and repair materials because of their high-performance mechanical 

characteristics, feasibility of application in civil structures, resistance to chemical 

attacks and other potentials. Brick masonry components are the most suitable 

substrates susceptible to improvements because of their more regular surface 

in comparison with stonework or rubble masonry. The bond behaviour of FRP, 

applied on a masonry substrate, is a critical issue for the effectiveness of the 

technique. In this paper, the results of an experimental assessment of the local 

behaviour of EB-FRP applied on clay bricks are presented. Experimental failure 

load results were compared with predictive bond strength models proposed in 

literature for concrete substrates. On the basis of measured strengths and local 

deformations, interface fracture energies were calibrated and an analytical function 

was proposed as bond stress-slip law. Finally, a bilinear law was calibrated 

for practical design applications. 

Keywords: FRP; glass; carbon; bond; masonry; brick. 

Introduction 

The application of Externally Bonded- 

Fibre Reinforced Polymers (EB-FRP) 

is a developing technique for the 

strengthening and repair of masonry 

structures. The bond behaviour between 

these products and the substrate is a 

crucial aspect that needs to be clarified, 

as it strongly influences the effectiveness 

of the intervention. In the last 

decade, the bonding of composite laminates 

on concrete substrates has been 

extensively investigated and characterized 

using different test set-ups. The 

most commonly used are the single-lap 

shear test, 1–4 double-lap pull–pull shear 

test, 5,6 double-lap push–pull shear test 7 

and beam-type test. 8 On the other 

hand, few investigations concerning 

the debonding from different masonry 

and masonry units are available. For 

instance refer to investigations on 

bond to natural stones, 9 comparison of 

natural stones and clay bricks, 10 testing 

the bond on solid clay bricks, 11,12 






usage of hollow clay blocks, 7 testing 

the bond on brick masonry prisms 13,14 

and the research 15 where historic clay 

bricks were used as substrate. All of 

the aforementioned studies adopted 

either the single-lap shear test or the 

double-lap push–pull shear test; the 

latter has also been referred to in literature 

as the double-shear push test 

or near-end supported double-shear 

test. 16 It involves applying tensile loads 

to two reinforcement strips symmetrically 

connected to the substrate mainly 

to create shear stresses at the interface 

while the brittle substrate is subjected 

to compressive stresses. The double-lap 

push–pull shear test set-up is conventionally 

based on the assumption that 

the applied load is equally distributed 

to the two strips. This test set-up is 

quite simple and suitable for commonly 

available universal testing machines. 

Several predictive models have been 

developed to evaluate the debonding 

load of composite-to-concrete 

joints. Reviews of available strength or 

bond-slip models were given by many 

researchers 17-20 . The models of Tanaka, 

Hiroyuki and Wu, Maeda, Khalifa (as 

The models of Tanaka, Hiroyuki and 

Wu, Maeda, Khalifa (as reported in 

Ref. [17]), Yang, Sato, Iso (as reported 

in Ref. [18]), express failure load as the 

product of an area and a nominal average 

tangential stress. The models of 

Izumo, 18 Neubauer and Rostàsy, 17 Chen 

and Teng 17 present other expressions 

for the failure load. Finally, 11 models 

provide an estimation of the fracture 

energy value that is correlated with 

the failure load. In particular, the models 

of Monti (as reported in Ref. [18]), 

Lu et al. 18 , herein labelled as Lu Bilinear, 

Brosens and Van Gemert 19 and Italian 

Research Council Guidelines 21 use a 

bilinear bond-slip law; the models of 

Nakaba et al. 6 and Savoia et al. 22 adopt 

a Popovic’s curve as bond-slip law; the 

models of Neubauer and Rostàsy, 17 Dai 

& Ueda, 23 Dai et al. 24 and Lu et al. 25 

(precise and simplified) resort to other 

types of bond-slip function. 

Several types of bond-slip laws have 

been used in the literature to describe 

the behaviour of externally bonded 

FRP, namely: (a) a cut-off function 

(Neubauer and Rostàsy, as reported 

in Ref. [17]); (b) a bilinear function, 

presented by some guidelines 26,21 and 

by researchers; 18,19,25 (c) a rigid function 

with linear softening behaviour; 17 

(d) a single function, as the Popovics 

curve 22,6 or an exponential curve; 24 

(e) two different non-linear functions 

for ascending and descending 

branches. 25,23 Therefore, it is generally 

assumed that bond behaviour of 

composite laminates exhibits softening, 

with an ascending branch followed 

by a descending one, and presents no 

residual stress for large slip. 

In the present work, the double-lap 

push–pull shear procedure was adopted. 

Results of five samples for highstrength 

carbon reinforcement and five 

samples for alkali-resistant glass reinforcement 

are presented and discussed. 

The predictions of 21 bond strength 

models, available in the reported literature 

for concrete as the parent material, 

were compared with the experimental 

debonding loads. The fracture energy 

of the composite-to-clay interface was 

evaluated using experimental failure 

loads. Furthermore, a simplified bondslip 

law was proposed on the basis of 

the data obtained from the load and 

strain recorded during the tests. Finally, 

a bilinear function was also calibrated. 


Experimental Tests 

Solid clay bricks, with nominal dimensions 

of 250 × 120 × 55 mm, were used 

as the substrate. High-strength Carbon 

Fibres Polymer (CFRP) on five specimens 

and alkali-resistant Glass Fibres 

Polymer (GFRP) on five other specimens 

were applied as reinforcement 

using a commercially available wet 

lay-up system. The mechanical properties 

of the bricks used are listed in 

Table 1 and the reinforcement system 

properties, obtained from manufacturers’ 

datasheets, are listed in Table 2. 

Each specimen was made by a single 

clay brick with two strips of reinforcement 

symmetrically bonded on the 

opposite wider surfaces. The thickness 

of the epoxy adhesive was approximately 

2 mm and the fibres were placed 

as accurately as possible on the centre 

line of the polymeric layer. Each strip 

was 50 mm wide and bonded to the 

brick over a length of L b = 200 mm 

(Fig. 1). An unbonded region of 30 mm, 

from the loaded end of the brick, was 

maintained in order to minimize edge 

effects. 

The testing machine was a universal 

mechanical press (Fig. 2). Each strip 

of reinforcement was bonded at the 

loaded end to a steel support connected 

to the machine. The brick was 

connected to the testing apparatus 

through a steel frame made by two 

steel plates linked by bolts (Fig. 2). 

The steel frame and support were connected 

to the machine by means of ball 

joints, to allow for even loading of the 

two composite strips. Samples were 

axially loaded in displacement control 

at a rate of 0,2 mm/min. The tensile 

load was monitored with a 100 kN load 

cell. For each specimen, seven strain 

gauges were applied to one of the 

two reinforcement strips and distributed 

as follows: one on the unbonded 

region next to the loaded end, and six 

on the bonded length. To optimize the 

Adhesive Saturant 

Characteristic compressive strength >80 N/mm 2 

Characteristic direct tensile strength >50 N/mm 2 

Maximum tensile strain 2,5% 

Tensile modulus of elasticity >3000 N/mm 2 

High-strength carbon fibre 

Equivalent thickness of one-ply fabric 0,165 N/mm 2 

Characteristic direct tensile strength 3430 N/mm 2 


Tensile elastic modulus 230 000 N/mm 2 

Alkali-resistant glass fibre 

Equivalent thickness of one-ply fabric 0,230 N/mm 2 

Characteristic direct tensile strength 1700 N/mm 2 


Tensile modulus of elasticity 65 000 N/mm 2 

Table 2: Properties of reinforcement components 

Reinforcement strips 

Steel plates 

Clay brick 

Clay brick 

50 mm 

Reinforcement 

Unbonded zone 

200 mm 30 mm 

Load direction 

Fig. 1: Geometry of specimens 

Mean cubic 

compressive strength 

Mean direct tensile 

strength 

Mean splitting tensile 

strength 

Mean flexural tensile 

strength 

Secant modulus 

of elasticity 

50,94 N/mm 2 

2,37 N/mm 2 

3,99 N/mm 2 

5,46 N/mm 2 

16 100 N/mm 2 

Table 1: Mechanical properties of clay 

bricks 

Fig. 2: Test machine (left) and a specimen ready for testing (right) 


–15 mm 

Loaded 

end 

SG1 

Fig. 4: Composite and brick surfaces after test 

SG2 SG3 SG4 SG5 SG6 SG7 

20 mm 

40 mm 

65 mm 

Clay brick 

Fig. 3: Distribution of the strain gauges 

95 mm 

130 mm 

170 mm 

Free end 

0 200 mm 

x axis 

and stress were uniformly distributed 

along the cross section and that it 

was possible to refer to the mechanical 

properties of one-ply dry fabric. 

This approach is accepted by many 

authors 17 and included in some guidelines 

21,26 . Therefore, it was possible to 

calculate the nominal tensile stress 

for each sample as the load divided 

by the cross-sectional area, A f =b' f·t f 

(where t f is the equivalent thickness of 

fibres) and also to calculate the modulus 

of elasticity as the linear best fit 

of the ratio between the nominal tensile 

stress and the experimental strain 

data recorded in the unbonded region, 

within 10 and 40% of the ultimate load. 

The average experimental modulus of 

elasticity was found to be higher than 

the manufactures’ values (22% for carbon 

reinforcement and 25% for glass 

reinforcement). Their large dispersion 

may be due to the wet lay-up application 

of the composites and to minor 

misalignments of the fibres. The average 

tensile stress reached by the reinforcement 

at the debonding load was 

64% of the maximum tensile strength 

of the carbon fibres and 68% of the 

glass fibres. 

As the reinforcement axial stiffness 

per unit width, E f·t f , was calculated for 

each specimen, failure loads per unit 

width, P u /2b' f , were tabulated against 

the axial stiffness. Trend lines were 

fitted, with respect to all data or to 

each set (carbon and glass fibres). The 

number of instruments and to monitor 

the whole bonded region, the strain 

gauges were closely spaced near the 

loaded end (Fig. 3). 

Experimental Test Results 

and Prediction of Strength 

At the end of the test, all the specimens 

showed complete detachment 

of the reinforcement from the support. 

Failure involved the brick surface 

(Fig. 4), where curved cracks 

and detachment of clay pieces were 

observed. Failure loads, P u , are listed 

in Table 3 for carbon (ShC) and glass 

(ShG) fibre reinforcements. Specimens 

strengthened with CFRP showed 36% 

higher failure loads than GFRP specimens. 

Table 3 also reports composite 

modulus of elasticity values, E f , 

nominal tensile stresses at debonding, 

s u , and failure loads per unit width, 

P u /2b' f (where b' f is the single strip 

width). To compute the modulus of 

elasticity, it was assumed that the strain 

Specimen E f (N/mm 2 ) P u (N) P u /2b' f (N/mm) r u (N/mm 2 ) 

High-strength carbon fibre 

ShC1 164 419 31 884 318,8 1932 

ShC2 336 439 34 233 342,3 2075 

ShC3 284 991 35 325 353,3 2141 

ShC4 277 511 39 210 392,1 2376 

ShC5 338 456 40 301 403 2442 

Mean value 280 363 36 191 361,9 2193 

Stand. dev. 70 696 3505 

COV (%) 25,2 9,7 

Alkali-resistant glass fibre 

ShG1 50 934 23 380 233,8 1017 

ShG2 87 014 27 940 279,4 1215 

ShG3 80 545 27 300 273 1187 

ShG4 102 598 26 400 264 1148 

ShG5 84 842 28 360 283,6 1233 

Mean value 81 035 26 676 266,9 1160 

Stand. dev. 18 817 1985 

COV (%) 23,2 7,4 

Table 3: Experimental results for carbon and glass fibre reinforcements 


Set of data c 1 c 2 R 2 

All data 12,876 0,310 0,876 

Glass data only 27,197 0,234 0,622 

Carbon data only 35,063 0,218 0,439 

All data (square root) 1,759 0,5 n.a. 

Glass data (square root) 1,953 0,5 n.a. 

Carbon data (square root) 1,681 0,5 n.a. 

Table 4: Regression constants of load versus axial stiffness trend lines (both per unit width) 

expression adopted for the trend lines 

is given in Eq. (1): 

P 

u 

c2 

c1 

Eftf 

 

2b' 

 

= ( ) (1) 

f 

where c 1 and c 2 are regression constants 

(values reported in Table 4). It 

can be observed that the fit of all data 

shows a better correlation than the fit 

of each single set. Moreover, adopting 

the relationship between the failure 

load and the square root (c 2 equals 

0,5) of the axial stiffness per unit width 

[Eq. (5)], from the literature 24,27,28 and 

Failure load per unit width (N/mm) 

600 

500 

400 

300 

200 

In(y) = 0,21758 In(x) + 3,5572 

In(y) = 0,23276 In(x) + 3,30118 

In(y) = 0,310003 In(x) + 2,55545 

100 

Carbon reinforcement 

Glass reinforcement 

0 

0 20 × 10 3 40 × 10 3 60 × 10 3 80 × 10 3 

Longitudinal axial stiffness (N/mm) 

Fig. 5: Failure loads per unit width versus reinforcement axial stiffness per unit width: 

experimental data and trend lines 

Failure load per unit width (N/mm) 

600 

500 

400 

300 

200 

100 

0 

0 

from guidelines 21 , it was possible to 

reduce the number of free parameters 

in Eq. (1). In this case, the regression 

coefficient for the GFRP was slightly 

higher than that for CFRP (around 

16%) and this could be significant for 

the fracture energy evaluation [see 

Eq. (6)]. Table 4 gives the values of 

carbon data set, glass data set and all 

data set, whereas Figs. 5 and 6 compare 

trend lines with experimental data. 

Experimental results were compared 

to estimations of strength given by 

the different models presented in the 

Introduction section (see Table 5 and 



20 × 10 3 40 × 10 3 60 × 10 3 80 × 10 3 

Longitudinal axial stiffness (N/mm) 

Fig. 6: Failure loads per unit width versus reinforcement axial stiffness per unit width: 

experimental data and trend lines based on the axial stiffness square root 

Fig. 7). All the predictions, except 

for that of Sato 18 and Izumo 18 in the 

in the case of carbon reinforcement, 

underestimated the average experimental 

failure load. Moreover, all formulations 

provided a prediction closer 

to test results in the case of CFRP 

than in the case of GFRP, except the 

models of Tanaka 17 and Hiroyuki 17 . 

However, results showed large differences 

from model to model. They 

varied between 44 and 154% of average 

experimental failure load for 

carbon reinforcement and between 

43 and 85% for glass reinforcement. 

Fracture Energy Calibration 

The interface mode II fracture energy, 

G f , is defined by Eq. (2) as a definite 

integral of the tangential stress, t, 

expressed as a function of the mutual 

slip between the composite and the 

substrate, s: 

 

f 

= ( ) 

G s d s 

0 

(2) 

One of the first analytical models 

of the composite-to-concrete bond 

strength was derived by Täljsten 29 , 

starting both from a linear approach 

based on the beam theory and from a 

non-linear approach related to fracture 

mechanics. For commonly used epoxy 

adhesives, a simplified formulation 29 

was obtained [Eq. (3)]: 

P 

u 

= b 

f 

2EtG 

f f f 

Et 

f f 

; 

T 

= (3) 

1+ 

 

Et 

T 

c c 

where b f is the reinforcement width, G f 

is the interface fracture energy, a T is 

a constant value and E c·t c is the axial 

stiffness per unit width of the concrete 

substrate. 

Yuan (as reported in Ref. [17]) proposed 

a modified constant value [Eq. (4)] that 

takes into account the width (b f and b c ) 

ratio of bonded materials: 

P 

u 

= b 

f 

2EtG 

f f f 

bEt 

f f f 

; 

W 

= (4) 

1+ 

 

bEt 

W 

c 

c c 

In most cases, the constant value a T , or 

a W , has a slight or negligible influence 

on the calculation. Several authors 22,24 

report the formulation in Eq. (5) without 

introducing any constant: 

Pu = bf 2 EftfG 

(5) 

f 

By applying Eqs. (3)–(5) to the experimental 

data of the present work, it was 


Model 

P u /2b' f 

(N/mm) 

CFRP 

Error 

(%) 

P u /2b' f 

(N/mm) 

GFRP 

Error 

(%) 

Tanaka 17 166 −54 166 −37,6 

Hiroyuki and Wu 17 158 −56,2 158 −40,6 

Maeda 17 254 −29,7 174 −34,9 

Khalifa 17 248 −31,4 170 −36,2 

Yang 18 192 −47 143 −46,3 

Sato 18 415 +14,6 116 −56,5 

Iso 18 271 −25,1 161 −39,5 

Izumo 18 557 +53,9 224 −15,9 

Neubauer and Rostàsy 17 283 −21,8 179 −32,7 

Chen and Teng 17 245 −32,2 156 −41,6 

Monti et al. 18 321 −11,3 204 −23,6 

Lu et al. Bilinear 25 220 −39,3 139 −47,7 

Brosens and 

Van Gemert 19 359 −0,9 228 −14,7 

CNR-DT 200 21 263 −27,4 167 −37,5 

Nakaba et al. 6 350 −3,3 222 −16,7 

Savoia et al. 23 328 −9,4 208 −22 

Neubauer and Rostàsy 18 266 −26,4 169 −36,6 

Dai and Ueda (1) 23 326 −9,9 207 −22,5 

Dai and Ueda (2) 24 322 −11 202 −24,2 

Lu et al. Precise 25 220 −39,3 139 −47,7 

Lu et al. Simplified 25 220 −39,3 139 −47,7 

Mean experimental 362 — 267 — 

Table 5: Predictions of failure load 

Predicted vs Exp. load ratio 

160% 

140% 

120% 

100% 

80% 

60% 

40% 

20% 

0% 

Tanaka 

HiroyukiI & Wu 

Maeda 

Khalifa 

Yang 

Sato 

Iso 

Izumo 

Neubrauer & Rostásy (1) 

Chen & Teng 

found that taking the parameters a T or 

a W into consideration only leads to a 

difference of less than 2%. It is worth 

noting that Eq. (5), demonstrated in 

some cases, 24,27 can be assumed in every 

case of regular interface law as pointed 

out by Savoia et al. 28 The formulation 

of Eq. (5), derived for concrete 

substrates, has been considered valid 

also for the clay substrate adopted in 

the present work, since both are quasibrittle 

substrates. Accordingly, it was 

possible to calibrate the fracture energy 

G f through Eq. (5), using mean values 

Monti et al. 

Lu et al. Bilinear 

Brosens & Van Gemert 



CNR DT–200/2004 

Nakaba 

Savoia 

Neubauer & Rostásy (2) 

Dai & Ueda (1) 

Dai & Ueda (2) 

Lu et al. Precise 

Lu et al. Simplified 

Fig. 7: Ratio of predicted failure loads versus average experimental value for carbon and 

glass reinforcement 

of failure load and elastic modulus, and 

the results are given in Table 6. The estimated 

value for glass reinforcement is 

approximately 35% higher than that 

for carbon reinforcement. 

Moreover, the fitting parameter c 1 

given in Table 4, when c 2 is 0,5 (square 

root based fit), allowed evaluation of 

G f as shown in Eq. (6). The results in 

Table 6 presented no significant difference 

from the values obtained by 

means of Eq. (5) for carbon and glass 

fibres, while using all data the value 

provided by Eq. (5) was 11% higher 

than the value provided by Eq. (6): 

2 

2EtG f f f= c1 

Et 

f f⇒ Gf= 0, 5c 

(6) 

1 

Calibration of a Bond-Slip Law 

Calibration of the bond-slip law on the 

basis of the experimental results was 

performed by adopting a combined 

approach where the tangential stress 

and interface slip points (t –s) were 

obtained from strain gauge recordings 

and the fracture energy value, G f , was 

calculated from failure loads by means 

of Eq. (5). The fracture energy represents 

an analytical restraint for the 

bond-slip function [Eq. (2)] and allows 

reduction of the number of free parameters 

involved in the calibration process. 

Equations (7)–(9), which briefly report 

the main relations among reinforcement 

strain e, interface tangential 

stress t and slip s, were obtained from 

simple equilibrium and compatibility 

considerations, disregarding the slip 

component due to the substrate, which 

is generally stiffer than the composite. 

The notation x indicates the coordinate 

along the central axis of the 

bonded region. 

d 

x 

dx 

( ) = ( ) 

1 

x 

E t 

f f 

dsx 

( x)= 

dx 

2 

d sx 

dx 

x 

( ) ( )= ( ) 

( ) ( )= 

f f 

 

sx xdx 

0 

(7) 

(8) 

1 x 0 

(9) 

E t 

To calculate the tangential stress and slip 

values from the corresponding strain 

recorded in discrete positions along the 

reinforcement, Eqs. (7)–(9) were modified 

and input into the discrete formulas 

given in Eqs. (10) and (11), as was done 


Reinforcement type G f from Eq. (5) (N/mm) G f from S.R. fit (N/mm) 

Carbon fibres 1,42 1,41 

Glass fibres 1,91 1,91 

All data 1,72 1,55 

Table 6: Evaluation of fracture energy 

in Ref. [30]. This allows for the manipulation 

of non-uniformly spaced data: 

1 

 

i 

= ( xi)= 

E t 

2 

f f 

i 

i 

i 

 

+ 1 1 

i 

+ 

 

x x x x 

 

i+ 

1 

i 

i 

i1 

(10) 

1 

si = s( xi)= si+ 1+ ( i 

+ 

i+ 1) ( xi+ 

1 

xi) 

2 (11) 

where the notation i − n indicates the 

strain gauge position. The x axis is 

oriented such that i increases from 

the loaded end (x = 0) to the free end 

(x = 200 mm). 

As explained in the Introduction section, 

it is assumed that the bond-slip law 

should show an ascending segment and 

a softening behaviour. Instead of using 

two different mathematical expressions 

for the ascending and the descending 

branches, a single function was selected 

on purpose. Although there could be a 

slight loss of adherence to experimental 

data, the single function reduces the 

required parameters, making the fitting 

easier. The proposed law, easy to integrate 

and derive, is given in Eq. (12): 

Bs 

( s)= Ase (12) 

where A and B are regression constants, 

t is the interface tangential stress and 

s is the slip. By applying the calibrated 

fracture energy value, it was possible 

to fit a function that depended on just 

one parameter, as shown in Eq. (13): 

After the optimization of the UniPd 

curves for carbon and glass reinforcements, 

it was also possible to calibrate 

a bilinear bond-slip law, which is commonly 

adopted by some guidelines 

(FIB Bulletin 26 ; CNR-DT 200 21 ). The 

analytical form of the bilinear law is 

reported in Eq. (15): 

( ) < 

 

max 

ss0 0 s s 0 

 

( s)= 

 

max (( s s) ( s s0 

)) s s< 

s 

 

 

0 s s f 

f f 0 f 

(15) 

where s f is the ultimate slip related to 

null tangential bond stress. 

Bond strees (N/mm 2 ) 

8,0 

6,0 

4,0 

2,0 

0,0 

0,000 0,200 0,400 

Slip (mm) 

Fig. 8: Calibrated bond-slip laws for CFRP reinforcement 

As the bilinear function depends on 

more parameters than the UniPd 

curve, the maximum tangential stress 

value, obtained from the fitted UniPd 

curve, and the calibrated fracture 

energy were imposed during the optimization 

process. Figures 8 and 9 show 

the optimized curves and the experimental 

stress-slip data. It can be noted 

that carbon reinforcement interface 

seems to be slightly stiffer than the 

glass reinforcement. 

Tables 7 and 8 report the significant 

values (fracture energy, peak tangential 

stress with related slip and ultimate 

slip) calculated by fitting the 

experimental data. They were compared 

with the values estimated using 

the 11 models reported in the literature 

and based on the fracture energy 

prediction. Estimated values varied 

across a wide range. Most models did 

not provide any difference for CFRP 

and GFRP reinforcements. Compared 

to the experimental results, they tend 

to underestimate the fracture energy 

(from 2 to 72%) and often do not 

CFRP 

UniPD curve 

Bilinear 

0,600 0,800 

∞ 

A 

2 

∫ τds 

= ⇒ A= 

B Gf 

; 

2 

B 

0 

2 

− Bs 

τ s B G s e 

()= ⋅ ⋅ 

f 

(13) 

The bond-slip law [Eq. (12)], herein 

labelled UniPd curve, can be rewritten 

in a normalized form as generally used 

in guidelines [Eq. (14)]: 

( s)= 

 

max 

 

s 

s 1 

s 

 

0 

 

s 

e (14) 

0 

where s 0 = 1/B and t max = t (s 0 ) are the 

coordinates of the maximum tangential 

stress point. 

Bond strees (N/mm 2 ) 

8,0 

6,0 

4,0 

2,0 

0,0 

0,000 0,200 0,400 

Slip (mm) 

Fig. 9: Calibrated bond-slip laws for GFRP reinforcement 

CFRP 

UniPD curve 

Bilinear 

0,600 0,800 


Curve G f (N/mm) s max (N/mm 2 ) s 0 (mm) s f (mm) 

UniPd fitting 1,42 7,22 0,072 — 

Bilinear fitting 1,42 7,22 0,034 0,392 

Monti et al. 18 1,11 5,37 0,046 0,415 

Lu et al. Bilinear 25 0,52 3,73 0,048 0,28 

Brosens and 

Van Gemert 19 1,39 2,71 0,012 1,025 

CNR-DT 200 21 0,75 7,46 0,056 0,2 

Nakaba et al. 6 1,32 7,08 0,065 — 

Savoia et al. 23 1,16 7,08 0,051 — 

Neubauer and 

Rostàsy 18 0,77 5,69 0,27 — 

Dai and Ueda (1) 23 1,15 8,58 0,103 — 

Dai and Ueda (2) 24 1,12 6,41 0,061 — 

Lu et al. Precise 25 0,52 3,73 0,054 — 

Lu et al. Simplified 25 0,52 3,73 0,048 — 

Table 7: Significant values for local bond of CFRP 

Curve G f (N/mm) s max (N/mm 2 ) s 0 (mm) s f (mm) 

UniPd fitting 1,91 6,33 0,111 — 

Bilinear fitting 1,91 6,33 0,048 0,603 

Monti et al. 18 1,11 5,37 0,046 0,415 

Lu et al. Bilinear 25 0,52 3,73 0,048 0,28 

Brosens and 

1,39 2,71 0,012 1,025 

Van Gemert 19 

CNR-DT 200 21 0,75 7,46 0,056 0,2 

Nakaba et al. 6 1,32 7,08 0,065 — 

Savoia et al. 23 1,16 7,08 0,051 — 

Neubauer and 

0,77 5,69 0,27 — 

Rostàsy 18 

Dai and Ueda (1) 23 1,15 7,1 0,107 — 

Dai and Ueda (2) 24 1,1 5,69 0,067 — 

Lu et al. Precise 25 0,52 3,73 0,054 — 

Lu et al. Simplified 25 0,52 3,73 0,048 — 

Table 8: Significant values for local bond of GFRP 

correctly estimate the maximum tangential 

stress and ultimate slip. 

Conclusion 

The bond behaviour of composite-to-clay 

brick interface was investigated using 

double-lap push–pull shear tests, for 

both high-strength carbon (CFRP) and 

alkali-resistant glass (GFRP) reinforcements. 

Far from being exhaustive, the 

experimental work was mainly focused 

on setting a procedure to design, perform 

and analyse this local phenomenon. 

Strength results showed a better performance 

of carbon reinforcement than 

glass, around 36% higher in the first 

case. 

Experimental strengths were compared 

with those obtained from 21 predictive 

models developed for concrete 

substrate. All predictions, except two in 

the case of CFRP, underestimated the 

test results. Models, except two in case 

of GFRP, appeared to provide better 

estimations for carbon reinforcement. 

However, the range of strength predictions 

was rather wide (between 44 and 

154% of average experimental failure 

load for CFRP and between 43 and 

85% for GFRP). 

From measured failure loads, different 

fracture energy values were obtained, 

around 35% higher in the case of glass 

reinforcement than carbon reinforcement. 

A mathematical function, easy to 

integrate and derive, was proposed for 

a bond-slip law and fitted for both carbon 

and glass reinforcement. Finally, 

two bilinear functions were also calibrated 

for design purposes according 

to strengthening guidelines. The optimized 

functions seem to show a bond 

behaviour for CFRP that is slightly 

stiffer than for GFRP. 

Further investigations are ongoing 

within the framework of the 

Rilem Technical Committee 223- 

MSC “Masonry Strengthening with 

Composite materials”, aimed at deepening 

the knowledge on the present 

topic and defining specific standardized 

procedures for testing the adhesion 

of composite materials applied to 

masonry. 


The authors wish to thank BASF CC Italia 

of Treviso, Italy, for the technical collaboration 

and for supplying fibres and the adhesion 

system. The research activity has been 

also partially supported by the National 

Italian Project ReLUIS. The authors would 

like to thank Eng. A. Cartolaro and the staff 

of the Laboratory of Material Testing of the 

Department of Structural and Transportation 

Engineering of the University of Padova, 

where tests were carried out. 

References 

[1] Chajes MJ, Finch WW, Januszka TF, Thomson 

TA. Bond and force transfer of composite material 

plates bonded to concrete. ACI Struct. J. 

1996; 93(2): 295–303. 

[2] Täljsten B. Defining anchor lengths of steel 

and CFRP plates bonded to concrete. Int. J. 

Adhes. Adhes. 1997; 19: 319–327. 

[3] Bizindavyi L, Neale KW. Transfer lengths and 

bond strengths for composites bonded to concrete. 

J. Compos. Const. 1999; 3(4): 153–160. 

[4] Mazzotti C, Savoia M, Ferracuti B. A new single-shear 

set-up for stable debonding of FRP – 

concrete joints. Const. Build. Mater. 2009; 23: 

1529–1537. 

[5] Lee YJ, Boothby TE, Bakis CE, Nanni A. 

Slip modulus of FRP sheets bonded to concrete. 

ASCE J. Compos. Constr. 1999; 3(4): 161–167. 

[6] Nakaba K, Kanakubo T, Furuta T, Yoshizawa 

H. Bond behavior between fiber-reinforced 

polymer laminates and concrete. ACI Struct. J. 

2001; 98(3): 359–367. 

[7] Camli US, Binici B. Strength of carbon 

fiber reinforced polymers bonded to concrete 

and masonry. Const. Build. Mater. 2007; 21: 

1431–1446. 

[8] De Lorenzis L, Miller B, Nanni A. Bond of 

FRP laminates to concrete. ACI Mater. J. 2001; 

98(3): 256–264. 

[9] Aiello MA, Sciolti MS. Bond analysis of 

masonry structures strengthened with CFRP 

sheets. Const. Build. Mater. 2006; 20: 90–100. 


[10] Faella C, Martinelli E, Paciello S, Perri F. 

Composite materials for masonry structures: 

the adhesion issue. In MuRiCo 3: Proceedings 

of the 3rd Nat. Conf. on Mechanics of Masonry 

Structures Strengthened with Composite 

Materials: Modeling, Testing, Design, Control; 

2009 April 1–3, Venice, Italy. Pitagora, Bologna, 

Di Tommaso A. (ed.), 2009; 266–273. 

[11] Briccoli Bati S, Rovero L, Tonietti U. 

Adesione fra blocchi in laterizio e rinforzo in 

CFRP. In Proceedings of the National Workshop 

on Materiali ed Approcci Innovativi per il 

Progetto in Zona Sismica e la Mitigazione della 

Vulnerabilità delle Strutture, Faella C, Manfredi 

G, Piluso V, Realfonzo R (eds), Salerno, Italy. 

Monza: Polimetrica, 2007; 213–220. 

[12] Panizza M. FRP Strengthening or Masonry 

Arches: Analysis of Local Mechanisms and 

Global Behaviour. Ph.D. [dissertation]. Padova 

(Italy): University of Padova; 2010. 

[13] Basilio I. Strengthening of Arched Masonry 

Structures with Composite Materials. Ph.D. [dissertation]. 

Guimarães (Portugal): University of 

Minho; 2007. 

[14] Subramanian K, Focacci F, Carloni C. 

An investigation on the interface fracture propagation 

between FRP and masonry. In MuRiCo 3: 

Proceedings of the 3 rd Nat. Conf. on Mechanics of 

masonry structures strengthened with composite 

materials: modeling, testing, design, control; 2009 

April 1-3, Venice, Italy. Pitagora, Bologna, Di 

Tommaso A (ed.), Bologna, 2009; 423–430. 

[15] Capozucca R. Experimental FRP/SRPhistoric 

masonry delamination. Compos Struct. 

2010; 92(4): 891-903. 

[16] Yao J, Teng JG, Chen JF. Experimental study 

on FRP-to-concrete bonded joints. Compos: B 

2005; 36: 99–113. 

[17] Chen JF, Teng JG. Anchorage strength models 

for FRP and steel plates bonded to concrete. 

ASCE J. Struct. Eng. 2001; 127(7): 784–791. 

[18] Lu XZ, Teng JG, Ye LP, Jiang JJ. Bond-slip 

models for FRP sheets-plates bonded to concrete. 

Eng. Struct. 2005; 27: 920–937. 

[19] Karbhari VM, Niu H, Sikorsky C. Review 

and comparison of fracture mechanics-based 

bond strength models for FRP-strengthened 

structures. J. Reinforced Plastics Compos. 2006; 

25(17): 1757–1794. 

[20] Ferracuti B, Savoia M, Mazzotti C. Interface 

law for FRP—concrete delamination. Composite 

structures. 2007; 80(4), 523–531. 

[21] CNR-DT 200/2004. Guide for the Design 

and Construction of Externally Bonded FRP 

Systems for Strengthening Existing Structures. 

Italian National Research Council: Rome, 2005. 

[22] Savoia M, Ferracuti B, Mazzotti C. Non linear 

bond-slip law for FRP-concrete interface. In: 

FRPRCS-6: Proceedings of the 6th International 

Symposium on Fibre-Reinforced Polymer 

Reinforcement for Concrete Structures; 2003 July 

8–10; Singapore, Tan KH (ed.). World Scientific: 

Singapore, 2003; 1–10. 

[23] Dai JG, Ueda T. Local Bond stress Slip relations 

for FRP Sheets-Concrete Interfaces. In: 

FRPRCS-6: Proceedings of the 6th International 

Symposium on Fibre-Reinforced Polymer 

Reinforcement for Concrete Structures; 2003 July 

8–10; Singapore, Tan KH (ed.), World Scientific: 

Singapore, 2003; 143–152. 

[24] Dai JG, Ueda T, Sato Y. Development of the 

nonlinear bond stress-slip model of fiber reinforced 

plastics sheet–concrete interfaces with a 

simple method. ASCE J Compos Const. 2005; 

9(1): 52–62. 

[25] Lu XZ, Teng JG, Ye LP, Jiang JJ. Bondslip 

models for FRP sheet/plate-to-concrete 

interfaces. In: ACIC 2004: 2nd International 

Conference on Advanced Polymer Composites 

for Structural Applications in Construction; 2004 

April 20–22; Guildford, Great Britain, Hollaway 

LC, Chryssanthopoulos MK, Moy SSJ (eds). 

Woodhead: Cambridge, 2004; 152–161. 

[26] FIB Bulletin 14. Externally Bonded FRP 

Reinforcement for RC Structures. Fédération 

Internationale du Béton: Lausanne, 2001. 

[27] Wu Z, Yuan H, Niu H. Stress transfer and 

fracture propagation in different kinds of adhesive 

joints. ASCE J. Eng. Mech. 2002; 128(5): 

562–573. 

[28] Savoia M, Ferracuti B, Mazzotti C. Una 

legge di interfaccia non lineare per placcaggi 

con lamine in FRP. In: AIMETA’03: Proceedings 

of the 16th Congress of Theoretical and Applied 

Mechanics; Sep 9–12; Ferrara, Italy, 2003. 

[29] Täljsten B. Strengthening of concrete prisms 

using the plate bonding technique. Int. J. Fract. 

1996; 82: 253–266. 

[30] Valluzzi MR, Tinazzi D, Garbin E, 

Modena C. FEM modelling of CFRP strips 

bond behaviour for bed joints reinforcement 

techniques. In: STRUMAS VI: Proceeding of 

the 6th International Conference on Computer 

Methods in Structural Masonry; Sep 22–24; Rome, 

Italy, 2003. 


http://www.rilem.net/tcDetails.php?tc=223-MSC 

http://www.cnr.it/sitocnr/Englishversion/CNR/ 

Activities/RegulationCertification.html 

http://www.reluis.it/ 

List of Symbols 

A Regression constant 

A f Cross-section area of a single strip 

B Regression constant 

b c Width of the concrete 

b f Reinforcement width 

b' f Width of a single strip 

c 1 Regression constant 

c 2 Regression constant 

e Euler’s Constant 

E c Modulus of elasticity of concrete 

E f Longitudinal modulus of elasticity of the 

fibres 

G f Interface fracture energy in mode II 

L b Bonded length 

P u Failure load 

s Interface relative slip 

s 0 Interface slip at the maximum tangential 

bond stress 

s f Ultimate interface slip at null tangential 

bond stress 

t c Thickness of the concrete 

t f Thickness of the fibres 

x Longitudinal abscissa along the bonded 

length 

a T Constant value 

a W Constant value 

e Reinforcement strain 

s u Nominal tensile stresses at debonding 

t Tangential bond stress 

t max Maximum tangential bond stress 


Bridges with Glass Fibre–Reinforced Polymer Decks: 

The Road Bridge in Friedberg, Germany 

Jan Knippers, Head of Institute, Universität Stuttgart, ITKE, Stuttgart, Germany; Eberhard Pelke, Head of Dept., Road and Traffic 

Administrative Dept. Hessen, Wiesbaden, Germany; Markus Gabler, Research Assistant, Universität Stuttgart, ITKE, Stuttgart, 

Germany; Dieter Berger, Development Eng., Road and Traffic Administrative Department Hessen, Wiesbaden, Germany. 

Contact: info@itke.uni-stuttgart.de 

Abstract 

In July 2008, the first road bridge in 

Germany using glass fibre-reinforced 

polymers (GFRP) was completed in 

Friedberg/Hessen. The structure has 

a span of 27 m and acts as a flyover 

across the federal road B3. The high 

durability of the new construction 

material and the fast assembly of the 

bridge were decisive factors in favour 

of GFRP. 

During the preceding years, several 

lightweight bridges using FRP had 

been constructed in the USA, Japan 

and also in Europe. Through these 

projects, valuable experience was gathered 

regarding construction, and the 

use and performance of composites 

could be demonstrated. 

The bridge in Friedberg extended this 

experience by taking into account the 

composite action between the FRP 

deck and the steel girders. It also 

followed, consequently, the approach 

of durable bridge construction by omitting 

any bearings or expansion joints 

and making the innovative material 

visible to passers-by. 

Keywords: glass fibre-reinforced polymers; 

composites; road bridge; installation; 

material tests; approval. 

Introduction 

The road and traffic administrative 

department in the German state of 

Hessen sees itself as a modern mobility 

service provider. It consequentially 

supports the development of new 

building methods for civil engineering 

structures, providing sustainability and 

reduction of life cycle costs as well as 

minimal interference with traffic during 

the construction. This includes the 

application of new materials in building 

methods. 






As the “hub of Germany” due to its 

central position, the state of Hessen 

has the highest average volume of 

traffic in Germany. Durability and 

low maintenance of civil engineering 

structures are hence prerequisites in 

terms of maintaining the economy’s 

efficiency and reducing costs of citizens 

by keeping traffic congestions to 

a minimum. The first approaches were 

conducted by the administration by 

observing traditional bridge building. 1 

Today, project calculations of road 

construction include approximate life 

cycle costs. In order to implement 

the objectives mentioned above, different 

new building techniques were 

examined and analysed. Lightweight 

constructions using fibre-reinforced 

polymer (FRP), which is very durable 

and fast to assemble, provide a potential 

solution. Such a bridge system was 

recently developed and realised in the 

context of the project “Congestion- 

Free Hessen”. 

Bridges with FRP Decks 

Realised Constructions 

The costs of FRP products are higher 

than that of conventional materials, 

therefore it is sensible to restrict their 

use to members that are susceptible to 

corrosion. For bridge superstructures, 

this means that decks can be built out of 

FRP, while the main girders are made 

of conventional materials such as steel 

and reinforced concrete (RC). This 

means not only lower costs but also less 

deflection of the superstructure due to 

the higher Young’s modulus of steel 

and RC. Steel FRP-composites have, 

especially in the USA, been extensively 

applied since the mid 1990s. The 

US Federal Highway Agency presently 

counts approximately 70 bridges 

built with FRP decks. 2 FRP decks are 

used very often in the refurbishment 

of existing road bridges. Damaged 

decks have been replaced on numerous 

RC-composite or steel bridges by 

FRP panels, whereas the abutment 

and the steel beams were preserved. 

The possibility of prefabricating large 

lightweight panels and transporting 

them to the building site, allows for 

extremely rapid refurbishment and 

reopening of the bridges. The curing 

time necessary for concrete decks is 

a fundamental reason for choosing 

the application of FRP. Several road 

bridges with FRP decks have also 

been built in Korea in the last few 

years, the 300 m long and 35 m wide 

“Noolcha Bridge” in Busan, for example. 

3 Positive experience was thereby 

consistently gathered—no major damages 

due to long term load have been 

observed in the structures built so far. 4 

Types of FRP Decks 

Available FRP decks can generally 

be classified into two categories: pultruded 

hollow sections and hand layup 

sandwich panels (Fig. 1). Pultruded 

decks consist of a row of prismatic 

bars that are manufactured through an 

automatic process. These hollow sections 

have a wall thickness that varies 

from 5 to 15 mm and an overall dimension 

of approximately 200 mm × 400 

mm. Owing to the continuous manufacturing 

process, the fibre direction 

is mostly longitudinal. The panels are 

formed by adhesively bonding the bars 

together. The bars are connected to the 

slab during assembly or previously at 

the factory in transportable sizes. The 

pultruded deck sections generally have 

a span of 2 to 3 m between the main 

girders. 

On the other hand, sandwich panels 

bear loads in two directions with the 

same stiffness and are therefore better 

suited to carrying concentrated 

loads. However, bigger tolerances 

occur through the manual manufacturing 

process than in pultruded profiles. 

Furthermore, research has shown that, 

owing to minimal heat transfer in the 

sandwich panels, very high temperature 

gradients can occur. Lastly, the 

intricate construction of connections 

and the connecting details are more 

challenging, which has led to sandwich 

panels not being introduced in Europe 

to date. 


Fig. 1: Comparison of different systems for FRP bridge decks; (left): pultruded hollow sections; 

(right): manually laid-up sandwich panels 

Dimensioning and Building 

Draft Principles 

Unlike the case of previously built 

structures, the aim of the design for 

the new road bridge in Friedberg, 

Germany, was to fully apply the composite 

action between the steel girders 

and the FRP deck, and to verify this 

experimentally as well as through analysis. 

For this reason, an adhesive layer 

was chosen for the FRP–steel connection: 

a joining method so far rarely 

used for civil engineering structures, 

although the material is appropriate 

and enables smooth load transmission 

without weakening caused by drilling. 

A frame structure was chosen with the 

purpose of long life and low upkeep of 

the construction, and therefore, track 

transition and bridge bearings were 

omitted. The structure was designed 

according to the ENV 1991–3 5 taking 

into account the government regulations 

of Germany. Due to lack of 

appropriate data, the fatigue-proof 

test of FRP could not be carried out 

through calculation. It was substituted 

by test data from the manufacturer of 

the FRP deck, which certified advantageous 

fatigue behaviour. 

Structural System and Composition 

The bridge structure (Fig. 2) has an 

integral design with a span of 27 m 

over the two-lane federal road B3a 

near Friedberg, in central Germany. 

The superstructure was made of two 

slightly curved and haunched steel 

girder with a structural height of 

625 to 900 mm. The overall 5 m wide 

FRP deck was glued onto both main 

girders (Fig. 3). A two-component 

epoxy adhesive was used to attach 

the FRP to the steel. The deck consisted 

of glued hollow sections and 

had a structural height of 225 mm. It 

was covered by a 45 mm thick layer 

of polymer concrete with applied 

polymer surfacing, which was sprayed 

with silica to provide surface roughness. 

The tensile strength of the polymer 

concrete was high enough to 

be considered in the calculation as 

strengthening for the top flange of the 

FRP deck. To obtain sufficient loadbearing 

safety, such composite action 

was necessary for withstanding concentrated 

wheel loads. The edge casings 

were made with a second layer of 

the same FRP sections. These casings 

together with the steel railing guarantee 

the required safety for the cars on 

the bridge. The design velocity of the 

road passing over the bridge is V ≤ 50 

km/h. The railing itself was connected 

to the front faces with anchor plates 

that were glued into the hollow sections 

of the FRP profiles of the deck 

and edge casings. Thus, it is possible 

to replace these parts in case of damage. 

The front face of the FRP decks 

was closed with thin-walled plates in 

order to prevent vermin from entering 

the hollow sections (Fig. 4). 

After assembling, the entire superstructure 

was rigidly connected onto 

the concrete abutments by grouting 

the attached fixing plates with 

welded head stud bolts. Making the 

FRP parts visible for passers-by was 

of utmost importance; therefore the 

front faces of the FRP deck remained 

uncovered, the hollow parts being 

recognisable as shadow gaps. The 

frame appearance of the bridge was 

underlined by the choice and orientation 

of formwork. 

625 

7000 20000 7000 

Fig. 2: FRP bridge in Friedberg (Hessen—Germany); (top): elevation; (bottom): longitudinal 

section (Units: mm) 

900 

8290 

Finite Element Method (FEM)— 

Calculations and Laminate Analysis 

The bridge structure was designed as a 

restrained frame with linear structural 

members. The composite cross section 

of the superstructure was therefore 

represented as an ideal steel cross section. 

It was already known from available 

test results 6 that full composite 

action can be assumed for the adhesive 

layer and the deck as the cross 

section virtually stayed even. With the 

actual dimensions of the Friedberg 

superstructure, the entire width of the 

FRP deck is under compression in 

the centre span. Owing to the direction 

of the fibre reinforcement, the 


East 

1300 

+245 

+45 

2,5% 

Steel girder (h = 625 ... 900) 

1200 1200 

5000 

slab mainly acts in uniaxial direction 

perpendicular to the bridge span. The 

load transfer presumptions were verified 

through tests. 

Specific values of material properties 

are necessary for the calculation of 

FRP structures as, on the one hand, 

different laminate set-ups are used for 

top flanges and webs of the deck and, 

on the other hand, the axial stiffness 

differs from the values in the lateral 

direction. A laminate calculation was 

therefore performed as a preparation 

to the actual analysis, determining the 

existing Young’s modulus and ultimate 

strength of the components based on 

the background of the respective fibre 

architecture (Table 1). 

For the input values of actions on 

bridges, in addition to ENV 1991–3, 

assumptions regarding the following 

issues had to be made: 

– For the additional ultimate limit 

state of creep rupture, 100% of dead 

load, 20% of imposed load and 50% 

of temperature load were taken into 

account and opposed to a reduced 

material strength. 

7 wearing course 

45 polymer concrete 

+155 

– 45 

Fig. 3: Standard cross section of composite superstructure (Units: mm) 

Fig. 4: Front face of superstructure (photo 

credit: Wilfried Dechau) 

– Rupture of an adhesively bonded 

joint was taken into account as an 

accidental action. 

– Analysis of fatigue loads was omitted 

due to existing verification 

through tests of the fatigue strength 

by the manufacturer. 

– Climatic temperature changes on 

steel–FRP composite superstructure 

were assumed to be T e = −20 to 41 K 

(overall cross section) and ΔT M = 

−18/+25 K (temperature moment). 

On the materials side, there are still gaps 

in building codes for FRP structures. 

In Germany, the recommendations of 

BUV (“BUV-Empfehlungen” 7 ) provide 

valuable design criteria. Structural 

Design of Polymer Composites— 

EUROCOMP Design Code and 

Handbook 8 is a helpful handbook for 

engineers. Furthermore, additional 

assumptions and alterations had to 

be considered for the analysis of the 

Friedberg Bridge: 

– superimposition of stress resultants 

according to the Puck/Knaust relationship 

of interaction; 

– relation of interaction at adhesively 

bonded connections; 

402 Scientific Paper Structural Engineering International 4/2010 

1300 

4% 

1000 

FRP-deck 

West 

N/mm 2 E xx E yy G xy f x,u,k f y,u,k 

Flange 28,000 19,000 5,000 250 175 

Outer web 17,000 24,000 4,500 160 125 

Inner web 17,000 26,000 3,000 210 195 

x: pultrusion direction y: Lateral to pultrusion direction 

Table 1: Properties of utilised FRP bridge deck 

– local buckling of the deck under 

stress vertical to the pultrusion 

direction. 

Tests for Technical Approval 

Overview 

The FRP deck ASSET used for the 

Frieberg Bridge was developed within 

the scope of an EU research project 

and had already been utilised for a 

smaller bridge in 2002. 9 From that project, 

material properties for the uniaxial 

load-bearing behaviour in the pultrusion 

direction were already available 

and applicable. Several additional specific 

values such as the shear strength 

of the FRP members and, moreover, 

the ultimate strength of the glue joint 

in tension and shear, the characteristics 

of the polymer concrete, the composite 

action between steel and FRP 

and the load-bearing behaviour due 

to concentrated wheel loads had to be 

determined. 

Composite Action between Steel 

and FRP 

Two main values are decisive in ensuring 

composite action: the strength of 

the adhesive joints between the steel 

girders and FRP deck panel, and the 

strength of the deck panel in compression 

vertical to the pultrusion direction. 

For the adhesive joints, tension 

and shear tests were preformed on 

small specimens. Rupture occurred 

within the FRP elements, and not 

within the adhesive or the interlayer. 

All tested specimens showed delamination 

within the top flange of the 

FRP deck. Therefore, it was of no 

importance to the ultimate strength, 

whether the surfaces were ground and 

degreased or not. The determined ultimate 

strength was far higher than the 

requirements identified in the analysis. 

The load-bearing behaviour of the 

deck under compression lateral to the 

Flange 

Inner web 

Outer web

pultrusion direction has already been 

described within the context of other 

research work. 6 Analogue experiments 

were conducted in order to confirm the 

values and secure a wider data basis. 

Therefore a 750 × 600 mm section of 

the deck was tested in an upright position 

under central load up to the point 

of rupture (Fig. 5). Because of local 

buckling, the strength was considerably 

lower than what could have been 

expected from the laminate architecture. 

The buckling appeared within the 

reinforcing layers of the overlap joint 

700 

between the individual elements. The 

failure mode became decisive for the 

design of the bridge. An optimization 

of the cross section, especially around 

the joining area could increase the 

compression strength perpendicular to 

the direction of pultrusion. 

Concentrated Wheel Loads 

Stress concentrations due to concentrated 

wheel loads are critical for the 

thin walls of the FRP deck. The polymer 

concrete used for the surfacing 

can bear tension and shear stresses. 

The surfacing system and the top 

flange of the FRP deck are assumed 

to be in composite action. Thus, the 

stress is distributed over a larger area. 

In addition, the polymer concrete surfacing 

reinforces the FRP deck flange 

(Fig. 6). This approach is similar to the 

design concept for orthotropic steel 

superstructures. The system is stable 

for wheel loads according to ENV 

1991–3. 

Construction and Assembly 

of the Superstructure 

Pre-Assembly in Shop 

Normal force on a 600 mm wide specimen (kN) 

600 

500 

400 

300 

200 

100 

The entire superstructure was preassembled 

in a storehouse located 20 

km from the building site. This allowed 

for protection against weathering and 

guaranteed optimal working conditions. 

The steel beams had already 

been welded when delivered and furnished 

with corrosion protection in the 

hall. The FRP deck had already been 

partially glued by the manufacturer 

and delivered in 5,00 × 1,50 m courses. 

The bonding of the deck panels onto 

the steel beams could be performed 

within 1 week. The FRP was ground 

and cleaned directly before the bonding; 

the adhesively bonded joints of 

the steel beams had already been 

sandblasted. 

0 

0 1 2 3 4 5 6 7 

Deformation (mm) 

Fig. 5: Test set-up and results for determination of compressive strength of FRP—bridge 

deck perpendicular to the direction of pultrusion 

Load on a area of 400 × 200 mm (kN) 

–175 

–150 

–125 

–100 

–75 

–50 

–25 

0 

0 

–1 

–2 

–3 –4 –5 

Deflection (mm) 

With surfacing 

Without surfacing 

Fig. 6: Test set-up and results for concentrated wheel loads, with and without surfacing 

layer 

–6 

–7 

–8 

On completion of the deck, the edge 

casings were affixed with a second 

layer of FRP sections and the front 

face was closed. After that a 45 mm 

thick layer of polymer concrete was 

applied onto the FRP. In the wet condition, 

the chosen coating was very 

stiff and debris-like and had to be 

levelled with appropriate tools to a 

nominal thickness. The treatment of 

the surfacing in one working sequence 

without joints proved to give the best 

results. The final jobs were the mounting 

of the railings, the edge sections at 

the transition between superstructure 

and abutment and the casting of 5 mm 

thick polymer surfacing on the edge 

casings. 

Transportation, Lifting and 

Positioning of the Superstructure 

Transporting the bridge to the building 

site had to be done in one night, 

as the public road next to the assembly 

hall could only be blocked for that 

long. The bridge was transported to the 

building site on a low platform trailer 

at a maximum speed of 50 km/h. The 

total weight of the superstructure was 


Fig. 7: Mounting of superstructure on construction 

site (photo credit: Fiberline Composites 

A/S, Denmark) 

approximately 60 tons. The total width 

of the carriage was only 5 m; support 

vehicles were not necessary and the 

motorway was not closed. The superstructure 

was lifted the next morning 

within 2 h onto the abutments (Fig. 7) 

by two truck-mounted cranes, after 

which the prepared supporting pockets 

were grouted (Fig. 8). With this 

completed, only the finishing jobs on 

the abutments were pending before 

the bridge could be made operational. 

Conclusion 

The use of fibre composites has great 

potential in bridge building. The possibility 

of pre-assembling in the hall 

allows for rapid installation on site. In 

this project, the critical working stages 

of gluing the FRP panels, as well as 

the lifting and positioning of the 

superstructure were performed without 

any problems. The handling of the 

prefabricated products is relatively 

non-sensitive. Detailed recording 

and tracking the quality of important 

components and working stages in 

the tender documents proved to be 

worthwhile. 

Monitoring the bridge over the next 

few years should contribute to gathering 

experience and optimising building 

methods for FRP bridge decks. Paying 

Fig. 8: Superstructure being lowered shortly before settling down in the grouting—pocket 

(photo credit: ASV Gelnhausen, Germany) 

close attention to detailed specifications 

of technical details in the tender documents 

and during construction proved 

to be beneficial. Thus, the necessary 

execution excellence could be obtained 

and risks for the principal could be 

minimised. With good collaboration 

between the consultant and the public 

authority, and a high commitment from 

all parties involved, this project could 

be completed successfully. 

References 

[1] Kuhlmann U, Pelke E, et al. Ganzheitliche 

Wirtschaftlichkeitsbetrachtungen bei Verbundbrücken. 

Stahlbau 76 (Heft 2), Seite 105ff, 

Germany. 

[2] Webpage of Federal Highway Administration 

(USA). www.fhwa.dot.gov/Bridge/FRP (as on 

Aug.16, 2010). 

[3] Lee SW, Kee-Jeung H. Experiencing More 

Composite-Deck Bridges and Developing 

Innovative Profile of Snap-Fit Connection. 

COBRAE Conference, Stuttgart, 2007. 

[4] Keller T, et al. Long-term performance of 

a glass fiber-reinforced polymer truss bridge. 

J. Compos. Constr. (ASCE) 01/2007, Vol. 11, 

pp.99–108.pp.99. 

[5] ENV 1991–3 Traffic Loads on Bridges. 

[6] Gürtler H. Composite Action of FRP Decks 

adhesively bonded to Steel Main Girders. PhD 

thesis No. 3135, Prof. Thomas Keller, EPFL 

Lausanne (CH), CCLab, 2004. 

[7] BÜV-Empfehlung. Tragende Kunststoffbauteile 

im Bauwesen (TKB). Bau- 

Überwachungsverein e.V., 2002 (Entwurf). 

[8] Clarke J, et al. Structural Design of Polymer 

Composites—EUROCOMP Design Code and 

Handbook. E&FN SPON: London, 1996. 

[9] Luke S, et al. Advanced composite bridge 

decking system—project ASSET. Struct. Eng. 

Int. 2002; 12: 76–79. 

SEI Data Block 

Owner: 

State of Hessen, Germany in the 

representation of Federal Republic 

of Germany 

Designer: 

Knippers Helbig Advanced 

Engineering, Stuttgart – New York 

Material tests and monitoring: 

Universität Stuttgart, Institute for 

Building Structures and Structural 

design (itke), Stuttgart (D) 

Design Construction Phase: 

KHP Planungsgesellschaft 

Main contractor: 

LS Bau GmbH & Co. KG, Giessen (D) 

FRP fabricator: 

Fiberline Composites A/S, 

Middelfart (DK) 

FRP (t): 13,5 

Span lengths (m): 27,0 

Construction cost (EUR million): 0,55 

Service date: August 2008 


Current and Future Applications of Glass-Fibre-Reinforced 

Polymer Decks in Korea 

Sung Woo Lee, President, Prof., Kookmin University, Seoul, Korea; Kee Jeung Hong, Prof., School of Civil Eng. Kookmin 

University, Seoul, Korea; Sinzeon Park, Associate Manager, Kookmin Composite Infrastructure Inc., Kyungki-do, Korea. 

Contact: kjhong@kookmin.ac.kr 

Abstract 

In comparison to concrete or steel 

bridge decks, glass-fibre-reinforced 

polymer (GFRP) compos ite bridge 

deck is highly economical, since its 

lightweight property reduces initial 

construction cost for the foundation, 

and the high durability decreases 

the life-cycle cost for the bridges. 

Furthermore, the duration of construction 

is reduced significantly because of 

the short installation time of the lightweight 

GFRP composite decks. 

Korea is one of the leading countries 

in the construction of composite-deck 

bridges in recent years. In Korea, 13 

road bridges (deck area = 15 917 m 2 ) 

have been constructed and six more 

road bridges (deck area = 9747 m 2 ) 

will be constructed in the near future. 

Furthermore, 19 footbridges (deck area 

= 14 921 m 2 ) have been constructed to 

date and six more footbridges (deck 

area = 14 204 m 2 ) will be constructed in 

the near future. Among them, there are 

two remarkable projects: the Noolcha 

Bridge in Busan Newport which was 

constructed with the world’s largest 

composite-deck panel, which is 300 m 

long and 35 m wide, and the existing 

1,7 km-long walkway of the Hangang 

Bridge in Seoul which was expanded 

from a width of 2,5 to 5 m by replacing 

the existing concrete decks with composite 

decks. 

Keywords: GFRP; composite deck; 

road bridge; footbridge; walkway 

expansion; snap-fit. 

Introduction 

To cope with the problems of deterioration 

and corrosion in conventional steel 

and concrete materials, highly durable 

and lightweight fibre-reinforced polymer 

(FRP) composites are considered 

to be one of the most promising alternative 

materials for civil infrastructures. 

Among many applications of 

these materials, glass-fibre-reinforced 

polymer (GFRP) composite decks for 

bridges are particularly notable, since 

they have been used successfully to 

replace conventional concrete and 

steel decks on a number of bridges. The 

GFRP composite decks for bridges 

have significant advantages compared 

to conventional concrete and steel 

decks as they are highly durable and 

corrosion free. Much longer service 

lives and lower maintenance costs are 

expected for GFRP composite decks 

than for conventional concrete and 

steel decks, which will, in most cases, 

result in much lower life-cycle cost 

(LCC). The lightweight of GFRP composite 

decks reduces the dead load by 

as much as 80% compared to conventional 

concrete decks. Much slimmer 

substructures are possible for bridges 

on account of the use of lightweight 

GFRP composite decks. When a GFRP 

composite deck is used for re-decking 

of a bridge, the capacity for live load on 

the bridge is consequently upgraded 

without strengthening its substructures. 

Furthermore, since GFRP composite 

decks can be installed easily and 

quickly, the duration of construction 

reduces significantly and the amount of 

disruption to traffic lessens. This allows 

considerable savings in direct and indirect 

costs to the urban community. 

In view of these notable advantages 

of GFRP composite decks, numerous 

studies have been carried out and an 

increasing number of field applications 

have been reported. 1,2 Moreover, 

many profiles of GFRP composite 

decks have been developed and put 

into practice since the 1990s. 2 In recent 

times, Korea has become one of the 

leading countries in the construction 

of composite-deck bridges. In Korea, 

more than 30 bridges (road and footbridges) 

with GFRP composite decks 

have been constructed. The “Noolcha 

Bridge” in the Busan Newport area 

was constructed with the world’s largest 

(300 m long and 35 m wide) composite-deck 

panel, and the 1,7 km-long 

walkway of the “Hangang Bridge” in 

Seoul was expanded from a width of 

2,5 to 5 m by replacing the existing 

concrete decks with the composite 

decks. Many more bridges with GFRP 

composite decks are scheduled to be 

constructed. 

The GFRP composite decks with 

tongue-and-groove connections were 

developed and commercialised by the 

authors’ university and a private company. 

3 Subsequently, the snap-fit connections 

of GFRP composite decks 

have been invented to significantly 

improve the construction quality and 

reduce the installation time compared 

to tongue-and-groove connections of 

GFRP composite decks. 4–11 

Developed GFRP Composite 

Decks 

Tongue-and-Groove Composite 

Decks 

Through extensive studies such as flexural 

tests, compressive fatigue tests, 

flexural fatigue tests, shear connection 

tests, traffic barrier tests, pavement 

bonding test, accelerated chemical 

tests and field load tests, a compositedeck 

profile that uses a tongue-andgroove 

connection was developed. As 

shown in Fig. 1a, it has three trapezoidal 

cells 200 mm in height (TG200), 

and is fabricated by pultrusion. To 

fabricate the laminate of the deck, 

8800 Tex E-glass rovings are used in 

the longitudinal direction in conjunction 

with multi-axial stitched fabrics 

(90°, ±45°), and unsaturated polyester 

is used as a resin base. The deck is 

designed for typical girders spaced 

at 2,5 to 3,0 m under the DB24 truck 

load (front and rear wheel loads are 

24 and 96 kN, respectively) according 

to the Korean Highway Bridge Design 

Specifications. 12 The pultruded deck 

tubes are horizontally assembled by 

epoxy bonding to build deck panels for 

bridges as shown in Fig. 2a. 

Snap-Fit Composite Decks 

Use of the tongue-and-groove connection 

method is prevalent in the assembly 

of composite decks, but it causes 

two problems when used for bridge 

decks. For conventional construction 

Structural Engineering International 4/2010 Technical Report 405

composite decks are rather limited 

and predictable. If the snap-fit composite 

decks are assembled without 

adhesive bonding, they can be easily 

disassembled and reused. The snap-fit 

connections also help to reduce the 

installation time and associated working 

costs. 

(a) Tongue-and-groove connection (TG200) 

Fig. 1: Pultrusion of composite decks 

Fig. 2: Connections for composite decks 

with concrete decks, shear connectors 

are installed on top of the girder prior 

to the placement of the concrete decks. 

For construction with the tongue-andgroove 

composite decks, in contrast, 

shear connectors cannot be installed 

before the placement and assembly of 

the composite decks on girders, since 

the composite decks are assembled by 

horizontal sliding on the top surface of 

the girder. Small holes at the locations 

of the shear studs must therefore be 

made on the composite decks prior to 

placing the composite decks on girders. 

After the placement and assembly 

of the composite decks, the shear studs 

are welded to the steel girders (or steel 

plates attached on the concrete girders) 

through small holes. Welding shear studs 

to girders through small holes is difficult 

and time-consuming and can result 

in poor welding quality. Accumulated 

horizontal gaps between the assembled 

composite decks become larger for a 

longer bridge. If the accumulated gaps 

are significantly large, it can cause mismatches 

between the locations of shear 

studs and the locations of the associated 

holes. This is the second problem that 

may produce construction errors and 

an increase in the construction time. 

To resolve the aforementioned 

problems related to conventional 

(a) Tongue-and-groove connection (TG200) 

(b) Snap-fit connection (SF200) 

(b) Snap-fit connection (SF200) 

tongue-and-groove connections, new 

profiles of composite decks with snapfit 

connections have been developed. 

The developed composite decks are 

assembled by mechanical snap-fitting 

in the vertical direction (Fig. 2b). 

Figure 1b shows the pultruded composite 

decks (SF200) with snap-fit 

connections for road bridges, which 

are not used yet in the field and are 

still under verification. Other snap-fit 

profiles have also been developed such 

as SF75L, SF75H, SF100 and SF125, 

which are not discussed in this paper 

explicitly. However, they are successfully 

applied in 19 footbridges in Korea 

in view of the easy assembly of the 

composite decks on the girder. The differences 

between these profiles lie in 

the height of the profile and the thickness 

of the flanges and webs. 

The newly developed snap-fit composite 

decks significantly improve 

workability and quality in welding 

shear studs to girders since the shear 

studs are welded with enough working 

space before placement of the composite 

decks. Furthermore, the developed 

decks help to avoid mismatches 

between the locations of shear studs 

and the locations of associated holes 

on the composite decks, since the 

horizontal gaps between the snap-fit 

Projects Examples 

Road Bridges 

The developed composite decks with 

tongue-and-groove connections at a 

height of 200 mm (Fig. 1a) have been 

applied for the construction of road 

bridges in Korea, with different girder 

types such as steel-plate girders, steelbox 

girders, pre-stressed concrete girders 

and reinforced concrete girders. 13 

road bridges (deck area = 15 917 m 2 ) 

have been constructed and six more 

road bridges (deck area = 9747 m 2 ) 

are planned to be constructed. Among 

these applications, the Bongsan Third 

Bridge and the Noolcha Bridge are 

shown in Figs. 3 and 4, respectively. 

The 36 m long and 7 m wide composite-deck 

panel was assembled on the 

steel-plate girders for the Bongsan 

Third Bridge as shown in Fig. 3a, construction 

on the bridge was completed 

in 2007. The Noolcha Bridge constructed 

in 2006 at the Busan Newport 

is considered a milestone since it has 

the world’s largest composite deck. 

Figure 4a shows the placing of the 

composite-deck panel from the storage 

site onto the RC girders using a crane 

for the Noolcha Bridge. Placing a single 

composite-deck panel (2 × 17,5 m) 

takes approximately 2 minutes, which 

is a significant reduction in construction 

time over that required for the 

placing of conventional concrete- or 

steel-deck panels. This quick installation 

of decks is a significant advantage 

in using the composite decks for 

bridge construction. Furthermore, the 

construction costs for the marine piles 

were significantly reduced because of 

the lightweight of the composite decks. 

Asphalt was used for the pavement of 

the Noolcha Bridge. 

Footbridges and Walkway Expansion 

The recently developed composite 

decks TG200, SF75L, SF75H, SF100 

and SF125 have been applied extensively 

for the construction of footbridges 

in Korea. These footbridges 

are constructed in short time periods 

as compared with conventional concrete- 

or wood-deck footbridges, since 

406 Technical Report Structural Engineering International 4/2010

(a) Placing GFRP composite decks (TG200) 

Fig. 3: Bongsan Third Bridge (steel-plate girder bridge) 

(b) Completed bridge 

of the Hangang Bridge in Seoul was 

upgraded with ease by replacing the 

existing concrete decks with the lightweight 

composite decks (SF75L). The 

existing walkway, which was 2,5 m in 

width, was expanded to 5 m through 

this upgrade without strengthening 

the substructures. After the existing 

concrete decks had been removed, 

Fig. 6a shows placing of the composite 

decks for the expanded walkway on 

both sides of the Hangang Bridge. The 

red lane for bikes and the grey lane 

for pedestrians are shown in Fig. 6b. 

Based on these successful applications 

of composite decks, more projects for 

new bridges and walkway expansions 

have been scheduled. 

Conclusion 

(a) Placing GFRP composite decks (TG200) 

Fig. 4: Noolcha Bridge (RC girder bridge) 

(a) Placing GFRP composite decks (SF100) 

Fig. 5: Wolchul Mountain Bridge (suspension bridge) 

(a) Placing GFRP composite decks (SF75L) 

the developed composite decks are 

much easier to install. 19 footbridges 

(deck area = 14 921 m 2 ) have been 

constructed and six more footbridges 

(deck area = 14 204 m 2 ) are planned 

for construction. 



(b) Completed walkway (four side-lanes) 

Fig. 6: Walkway expansion of Hangang Bridge (arch bridge and steel-plate girder bridge) 

In 2006, the developed composite 

decks (SF100) were quickly assembled 

at high elevation for a suspension footbridge, 

called the Wolchul Mountain 

Bridge. Shown in Fig. 5, it is 53 m long 

and 1 m wide. The 1,7 km-long walkway 

GFRP composite decks have been 

widely applied for many road bridges 

and footbridges in Korea. Among 

these GFRP deck bridges, there are 

two remarkable projects: the Noolcha 

Bridge in Busan Newport has been 

constructed with the world’s largest 

composite-deck panel and the 

Hangang Bridge in Seoul, has been 

expanded with ease from a width of 2,5 

to 5 m by replacing the existing concrete 

decks with the composite decks. 

The newly developed snap-fit connection 

of the composite decks resolves two 

problems encountered during construction 

with the tongue-and-groove composite 

decks: poor welding conditions 

for the connection of the shear studs and 

the possible mismatch between the locations 

of shear studs and the associated 

locations of pre-drilled holes on the 

composite decks. By snap-fitting composite 

decks without adhesive bonding, 

simplified assembly and disassembly of 

the composite decks are ensured. This 

enables their application in temporary 

bridges or temporary roads. It is anticipated 

that the newly developed snapfit 

connection will pave the way for far 

wider applications of GFRP composite 

decks in future, and that GFRP composite 

decks will be promising alternatives 

to c onventional concrete, steel or 

wood decks for bridges. 

References 

[1] DARPA. Advanced Composites for Bridge 

Infrastructure Renewal-Phase II Tasks 16- 

Modular Composite Bridge. Defense Advanced 

Research Projects Agency. Technical Report Vol. 

IV. USA, 2000. 

[2] Keller T. Use of Fiber Reinforced Polymers 

in Bridge Construction. Structural Engineering 

Documents 7. IABSE (International Association 


for Bridge and Structural Engineering). 

Switzerland, 2003. 

[3] Lee SW. Development of High Durable, Light 

Weight and Fast Installable Composite Deck. 

MOCT R&D Report. Ministry of Construction 

and Transportation, Korea, 2004. 

[4] Lee SW. Fiber Reinforced Polymer Composite 

Bridge Deck of Tubular Profile Having Vertical 

Snap-Fit Connection. US Patent No. US 7, 131, 

161 B2, USA; 2006. 

[5] Lee SW, Hong KJ. Development of Composite 

Deck Connection for Pedestrian Bridge Using 

Korean Traditional Wooden Joint Method. 

KOSEF Research Report. Korean Science and 

Engineering Foundation, Korea, 2007. 

[6] Lee SW, Hong KJ. Experiencing more composite-deck 

bridge and developing innovative 

profile of snap-fit connections. Proceedings of 

COBRAE Conference. Stuttgart, Germany, 2007. 

[7] Lee SW, Hong KJ. Constructing bridges 

with glass-fiber reinforced composite decks. 

Proceedings of 4th International Structural 

Engineering and Construction. RMIT University, 

Melbourne, Australia, 2007. 

[8] Lee SW, Hong KJ. Evolution of innovative 

snap-fit connection for pultruded ‘Delta Deck’. 

Proceedings of European Pultrusion Technology 

Association. Rome, Italy, 2008. 

[9] Lee SW, Hong KJ. Experiencing more 

GFRP composite bridge decks for vehicular 

and pedestrian bridges. Proceedings of IABSE. 

Chicago, USA, 2008. 

[10] Lee SW, Hong KJ, Ketel J. Composite ‘Delta 

Deck’ of innovative snap-fit connection for new 

and rehabilitated footbridges. Proceedings of 

Footbridge 2008. Porto, Portugal, 2008. 

[11] Lee SW, Hong KJ. Development of 

Light-Weight Composite Deck with Snap-Fit 

Connection for Rigmat and Bridge Deck. MOCT 

R&D Report. Ministry of Construction and 

Transportation. Korea, 2009. 

[12] Korea Road and Transportation Association. 

Korean Highway Bridge Design Specifications. 

Ministry of Construction and Transportation, 

Korea, 2005. 

SED 7 - Structural Engineering Document on FRP 

IABSE 

This Structural Engineering Document reviews the progress made 

worldwide in the use of fibre reinforced polymers as structural 

components in bridges until the end of the year 2000. It includes 

application and research recommendations with particular reference 

to Switzerland and is aimed at both students and practising engineers, 

working in the field of fibre reinforced polymers, bridge design, 

construction, repair and strengthening. 

Topics: 

Overview and Classification, Fibres and Matrices, Tensile Elements, 

Structural Components and Systems, FRP-Reinforced Concrete – Stateof-the-Art, 

Fibre Reinforced Polymers – State of the Art in Repair and 

Strengthening, Fibre Reinforced Polymers – State-of-the-Art in Hybrid 

New Structures, Fibre Reinforced Polymers – State-of-the-Art in All- 

Composite New Structures, Design, Codes and Guidelines, Application 

Recommendations, Research, Requirements and Recommendations. 

Price: CHF 40 for Members, CHF 70 for Non-Members 

www.iabse.org/publications/onlinshop 


Field Issues Associated with the Use of Fiber-Reinforced 

Polymer Composite Bridge Decks and Superstructures 

in Harsh Environments 

Louis N. Triandafilou, P.E., M.ASCE, Senior Structural Eng., Federal Highway Administration Resource Center, Baltimore, MD, 

USA and Jerome S. O’Connor, P.E., F. ASCE, Manager, Bridge Engineering Program, Dept. of Civil, Structural, Environmental 

Engineering, University at Buffalo, The State University of New York, Buffalo, NY, USA. Contact: lou.triandafilou@dot.gov 

Abstract 

There has been ample time for the 

United States to evaluate the use of 

fiber-reinforced polymer (FRP) composites 

to serve as bridge decks and 

superstructures under real-world environmental 

and operating conditions. 

This provides a great opportunity to 

weigh in on the decision to move forward 

with these materials. By studying 

the successes and failures, materials 

and techniques used for the design and 

construction of these bridges can be 

improved so that we can move closer to 

the goals of maintenance-free bridges 

and service lives exceeding 100 years. 

Though only a few of these structures 

are instrumented for structural health 

monitoring, there are a sufficient number 

of lessons that can be gleaned from 

the experiences of various owners during 

the fabrication, installation, operation, 

and inspection of these structures. 

By addressing these issues head-on, 

advocates will be better equipped to 

have direct dialogue with those in the 

profession who remain skeptical about 

the hidden potential for a construction 

material that is strong, light, and corrosion 

resistant. 

Keywords: FRP; deck; composite; 

bridge deck; superstructure; bridge. 

Introduction 

In the United States, public traffic has 

been carried on fiber-reinforced polymer 

(FRP) bridges since the installation 

of the first superstructure in 1996. 

A summary of these installations is 

available in the literature. 1 Bridge owners 

and researchers can benefit from 

these installations by considering their 

ongoing use as a real-world laboratory. 

Although the vast majority of the 

bridges constructed or rehabilitated 

with composites are considered a success, 

there have been some undesirable 

performance issues, such as debonding 

of wearing surfaces. Investigating 

these issues and openly sharing the 

experiences and findings with others 

should be encouraged. By making this 

knowledge available to a wide range 

of owners, engineers, and researchers, 

cooperation can be brought about in 

the development of the modifications 

to design and construction practices 

that will help advance the state of 

the art and the use of composites in 

practice. 

The scope of the discussion herein is 

restricted to 117 identified structures 

that have utilized FRP for the bridge 

deck or the superstructure. Although 

a few of these structures are too short 

to meet the federal definition of a 

bridge (at least 6,1 m), they are being 

considered here, since the lessons are 

the same, regardless of span length. 

Also included are the decks that have 

been taken out of service because of 

operational problems, and FRP superstructures 

that use concrete, timber, or 

asphalt for the deck or wearing surface. 

Experience from these bridges 

has provided lessons pertaining to the 

fabrication of the FRP, application of 

the wearing surface, and their longterm 

performance under harsh environmental 

conditions. Excluded from 

the discussion are pedestrian bridges, 

timber glue-laminated bridges that use 

FRP as a laminate, and FRP stay-inplace 

(SIP) forms for a concrete deck. 

Performance Summary 

Of the installations considered in this 

study, 95% are still in service and being 

monitored visually as part of regularly 

scheduled bridge inspections. Six have 

been taken out of service because of 

structural defects or failures of the 

deck or superstructure panels. An 

undisclosed number have also exhibited 

problems such as cracking and 

spalling of an adhesively bonded wearing 

surface. 

Structural Integrity 

It is often difficult to state conclusively 

what caused a particular structural 

failure. The six products taken out of 

service were from various manufacturers 

and manufacturing techniques. 

– Two types of deck were removed 

from the Salem Avenue bridge in 

Ohio (one vacuum-assisted resin 

transfer molded (VARTM), and 

one hand-laid sandwich section), 

because the FRP flange had separated 

from the FRP web core. There 

may have been contributing factors 

that led to failure of the decks after 

the bridge was placed in service, but 

an initiator of the problems seems to 

have been a manufacturing defect. 

The connection between the decks 

and the steel beams may have also 

played a role by allowing too much 

movement under live loads. 

– A detour bridge in Iowa was taken 

out of service and is not being used 

because it is not considered structurally 

sound. It is a VARTM sandwich 

section that separated between the 

FRP flange and web (top face skin 

and core). This could have been 

caused by a manufacturing defect 

or possibly by pounding from truck 

traffic along the leading edge after 

it was placed in service. The impact 

was apparently due to a slight height 

difference between the deck surface 

and the approach pavement. 

– In West Virginia, two corrugated 

core decks from the same manufacturer 

have been taken out of service 

because of structural integrity issues 

with the hand lay-up sandwich section. 

On one, clips between the deck 

and steel floor beam failed, allowing 

excessive movement that apparently 

led to a crushing of the web core of 

the deck. On the other, there were 

leaking joints between panels, and 

apparent debonding of the top face 

skin and crushing of the FRP core 

that resulted in “potholes”. 

– A pultruded FRP deck was removed 

from a bridge in Oregon owing to 

unsatisfactory performance of the 

support and connection details. The 

connection, which was designed to 

be rigid between the FRP deck and 

the steel structure, did not serve as 


intended and led to a failure of the 

grout haunch and shear stud pockets, 

excessive flexure of the deck, 

and cracking of the wearing surface. 

In addition, there are two bridges (in 

California and New York) that are 

being monitored closely. These were 

constructed by different manufacturers 

but both have been repaired after 

debonding of the sandwich sections was 

discovered. Figure 1 shows a void under 

an edge of the top face skin of one 

bridge superstructure and Fig. 2 shows 

a core of the top face skin of the other. 

Thermal Behavior 

Although composites used in other 

industries have been thoroughly investigated 

and used in service at extreme 

temperatures, there has been limited 

research done on bridges at extreme 

temperatures. 2 Thus, the observation 

of in-service bridges has been useful 

in demonstrating performance under 

these harsh conditions. 

FRP behaves slightly differently from 

traditional materials at cold temperatures 

and during fluctuations in temperature. 

For instance, not only is the 

coefficient of thermal expansion different 

from steel or concrete, within the 

same FRP unit, it may be different longitudinally 

as compared to transversely. 

In addition to the extent of the rmal 

expansion/contraction being different, 

the rate at which the changes occur is 

also different. In general, this has not 

created a problem. The exception is 

that the interface between the FRP and 

the applied wearing surface has sometimes 

failed; thermal behavior may be 

involved, though conclusive studies have 

not been made. See below for further 

discussion on bonded wearing surfaces. 

Drivers in cold climates are aware that 

the surface of a bridge can become 

icy before the roadway does. This is 

because the surface temperature of the 

roadway is moderated by that of the 

earth below it. Bridges do not benefit 

from the heat that is stored in that thermal 

mass, and respond to drops in air 

temperature more quickly. FRP decks 

react the same way, but the changes are 

even more dramatic because they have 

little thermal mass when compared to 

concrete decks. While this is not known 

to have caused any problems to date, it 

is important to be aware that the rate 

of temperature change is not what drivers 

may be expecting. Figure 3 shows 

frost on the deck while the approaches 

remain clear and dry. 

top surface while the bottom surface 

remains relatively cool. This typically 

occurs during autumn or spring when 

there are cool nights, but daytime temperatures 

rise rapidly and the dark 

roadway warms up by absorbing solar 

radiation. The difference in temperature 

between the top and the bottom 

has been observed to be as much as 

25°C (77°F). Analysis has shown that 

thermally induced stresses can be higher 

than stresses caused by live loads from 

traffic. Several FRP structures exhibit 

this phenomenon, but none have suffered 

damage as a direct result. 

Wearing Surface 

The serviceability issue most often 

reported as a problem concerns the 

wearing surface. Although no problems 

have resulted from the use of a 

concrete or timber deck used with 

FRP beams, or with asphalt used as a 

wearing surface on FRP decks (except 

in a case where there was a structural 

integrity issue with the deck itself), 

there have been numerous reports of 

cracking and debonding of polymer 

concrete wearing surfaces that were 

applied to FRP decks. The figures 

below highlight some of the relevant 

concerns. In general, wearing surface 

problems can be categorized as follows: 

– material selection; 

– application; 

– structural integrity of the FRP 

substrate; 

– movement or flexibility of the FRP 

substrate. 

Material Selection 

Fig. 1: Top face skin of the sandwich 

superstructure debonded from the core 

(California) 

Fig. 2: A sample drilled from the top face 

skin of a sandwich panel exhibits dry 

fibers. This manufacturing flaw may have 

led to debonding of the face skin from the 

core (New York) 

Fig. 3: Frosting of the deck surface while the 

roadway remains relatively warm and dry 

The use of snow plows on bridges that 

have FRP decks has not been cited as 

a problem. As is typical with conventional 

decks, high spots in the deck 

surface may exhibit polishing and wear 

sooner than the rest of the deck. 

High temperatures observed in the 

field frequently result in the “hogging” 

of thick FRP superstructure panels. The 

resulting camber (arching) does not 

seem to create a problem as long as the 

wearing surface and its bond with the 

FRP can handle the shear and tensile 

stresses. Hogging is a thermal distortion 

that results from a temperature gradient 

through the depth of a section that 

results in an expansion of the warm 

Fig. 4: Cracking along a construction joint 

created between two FRP superstructure 

panels. Use of a flexible material such as 

two-part silicone may have been a better 

material choice 

Figure 4 shows a construction joint 

between FRP superstructure panels 

that was filled with adhesive during 

installation. The wearing surface over 

the joint has broken up, making the 

joint susceptible to moisture intake. 


Figure 8 shows uneven wear of a 

thin epoxy concrete wearing surface. 

Although snow plows can be expected 

to polish high spots on any deck, 

these areas are most likely due to 

improper proportioning of aggregate 

and resin, or inadequate mixing during 

application. 

Fig. 5: Precautions must be taken when 

using asphalt for movable bridges because 

the weight of the material can cause a failure 

of the bond when the bridge is raised 

Figure 5 shows a movable bridge that 

is typically an attractive structure on 

which to use a lightweight deck. It 

serves as a reminder that the wearing 

surface sometimes needs to be able to 

withstand gravity loads in addition to 

being able to fulfill its other functions. 

When the bridge is in the up position, 

shear stresses between the deck and 

the wearing surface material need to 

be accommodated so that the material 

does not slide off the structure. 

On particularly hot days, the shear 

capacity of the bond and/or the material 

may become reduced. One agency 

has had this experience. 

Fig. 6: The debonding of this 7 mm wearing 

surface may have been due to improper 

material selection 

Figure 6 shows a partial debonding of 

a thin (6 mm) wearing surface. In this 

instance, a trial application was made 

using an acrylic-modified Portland 

cement based material. Most of the 

bonded surface survived but areas 

where it was thinly applied peeled off 

within a few years. 

Application 

Fig. 7: Debonding, possibly from 

inadequate surface preparation 

Figure 7 illustrates a thin polymer 

overlay that spalled due to poor surface 

preparation and installation. 

Although this illustration is that of a 

concrete deck, the result is typical of 

what can result on an FRP deck. 

Fig. 8: Polishing due to improper mixing 

and application. Durability may have also 

been reduced due to selection and use of an 

aggregate that was too small 

Fig. 9: Polymer concrete applied over a 

field joint between FRP panels. Insufficient 

mixing may be contributing to premature 

deterioration 

Figure 9 shows deterioration of a 

polymer concrete over a joint. Field 

applications such as this are difficult 

because of the additional variables 

involved (human and environmental 

factors). 

Fig. 10: An attempt to create an integral 

polymer concrete wearing surface during 

deck fabrication still resulted in surface 

degradation 

Figure 10 shows a superstructure that 

had the wearing surface made integrally 

with the FRP panel as part of 

the fabrication process. This was done 

by adding aggregate to the resin while 

it was applied by hand to the top of 

the panel. This integral surface was 

intended to circumvent the possibility 

of debonding of the wearing surface. 

Although this strategy was partially 

successful, it appears that the surface 

raveled because there was a partial 

cure in place before the top materials 

were applied. 


Figure 13 shows an example of a failing 

FRP panel that has been patched 

over. In this case, an epoxy resin was 

used; however, asphalt has also been 

used on decks where the depression 

(or pothole) was deep enough. 

Fig. 11: Polymer concrete spalling due to 

deviance from installation procedures 

Figure 11 illustrates a typical debond 

that exposes the FRP surface to ultraviolet 

(UV) light and wheel loads. The 

edges are also susceptible to water intrusion 

that will cause the hole to propagate. 

Fig. 12: Debonding of brittle epoxy wearing 

surface in cold temperatures 

Figure 12 is a picture of a 10 mm epoxy 

wearing surface that debonded during 

its first winter in service. The bond failure 

was attributed to improper surface 

preparation during installation but, as 

can be seen, the material broke up in a 

brittle fashion when it was debonded. 

The material’s lack of flexibility at low 

temperatures probably caused this. 

Water intrusion and its subsequent 

expansion upon freezing also accelerated 

the failure of the surface. 

Structural integrity of the FRP 

substrate 

Fig. 14: Random cracking over a 

failing sandwich superstructure panel. 

(Fig. 2 is a close-up of the core that 

was taken) 

Figure 14 shows another type of surface 

failure. This depression resulted 

from a softening or crushing of the 

FRP web core where there was water 

intrusion. The wearing surface is not 

intended to bridge defects such as this 

and exhibits alligator cracking. Bridge 

inspectors need to be aware of this 

condition since maintenance workers 

are likely to treat it as a pothole and 

just patch over it. 

Movement or flexibility of the FRP 

substrate. 

Fig. 15: Cracking has occurred over 

bond lines of pultruded sections that 

were joined in the factory. Selection 

of the adhesive may have played 

a role 

Figure 15 illustrates reflective cracking 

that has occurred over factory bond 

lines between pultrusions. 

Fig. 16: Cracking over a transverse steel 

floor beam, probably from stresses caused 

by flexure of the FRP panel over the top 

flange. Water intrusion was also observed 

Figure 16 illustrates tensile cracking 

that can occur over the edges of a floor 

beam as the more flexible FRP deforms 

under wheel loads and thermal stressing. 

In this case, the deck is secured to 

the top flange of the floor beam below 

by tightly bolting them together. 

Examples of successful wearing 

surfaces 

Fig. 17: Although asphalt wearing surfaces 

are not lightweight, they have worked well 

in many instances. No special bonding 

layer was applied to the FRP 

Figure 17 shows an FRP superstructure 

that was installed and then overlaid with 

asphalt. This approach has been used 

repeatedly with success. This adds a substantial 

amount of weight and detracts 

one of the benefits of using FRP (its 

lightweight nature). The additional dead 

load will also need to be addressed with 

respect to its contribution toward creep 

and the potential for brittle rupture. 

Although asphalt may not be desirable 

in all situations, there are some situations 

where it may be the best alternative. 

It also has the benefit of being a 

well-known material that maintenance 

crews are comfortable with. 

Fig. 13: This deficiency (which has been 

patched) is due to a failing sandwich 

superstructure panel rather than the 

material or methods used for the wearing 

surface. 


Fig. 18: Thin polymer concrete surfaces 

provide good skid resistance and protection 

with little extra weight 

Figure 18 illustrates a successful installation 

of a polymer concrete overlay. 

After 5 years of service, this deck surface 

is still in “like-new” condition. 

New York State and Oregon have 

undertaken research projects to 

explore some of the wearing surface 

issues presented above. 3–5 The New 

York report by Aboutaha recommends 

applying the surface as two courses, one 

specifically designed to insure a good 

bond and the upper course to provide 

the requisite durability and skid resistance. 

3 Reports of any such installations 

are yet to be made available. 

Conclusion 

The vast majority of FRP deck and 

superstructure installations have performed 

well structurally, but success 

in the field has been diminished by the 

poor performance of several wearing 

surfaces, especially the ones made of 

thinly applied bonded materials that 

were used for weight saving. Better 

performance in the future can be 

obtained by proper selection of material, 

meticulous adherence to wellplanned 

installation procedures, use of 

pull-off tests to check for proper surface 

preparation and bond, and use of 

a two-course system, where one is used 

to ensure bonding and a top layer is 

used for skid resistance and durability. 

References 

[1] Triandafilou L, O’Connor J. FRP Composites 

for Bridge Decks and Superstructures: State of 

the Practice in the U.S. Proceedings of International 

Conference on Fiber Reinforced Polymer 

(FRP) Composites for Infrastructure Applications, 

University of the Pacific, Stockton, CA, 

2009. 

[2] Dutta P, Kwon S. Fatigue Performance 

Evaluation of FRP Composite Bridge Deck 

Prototypes Under High and Low Temperatures. 

82nd Annual Meeting, Compendium of Papers 

on CD ROM, Transportation Research Board, 

Washington, D.C., 2003. 

[3] Aboutaha R. Investigation of Durability 

of Wearing Surfaces for FRP Bridge Decks, 

Syracuse University, 2008. 

[4] Wattanadechachan P, Aboutaha RS, Hag- 

Elsafi O, and Alampalli S. Thermal compatibility 

and durability of wearing surfaces on FRP bridge 

decks. ASCE Journal of Bridge Engineering 

2006; 11(4), pp. 465-473; July–August. 

[5] Barquist G, Lovejoy S, Nelson S, Soltesz S. 

Evaluation of Wearing Surface Materials for 

FRP Bridge Decks OR-DF-06-02, Oregon 

Department of Transportation, 2005. 


Examples of Applications of Fibre Reinforced Plastic 

Materials in Infrastructure in Spain 

Anurag Bansal, Research Engineer; John F. Monsalve Cano, Research Engineer; Bladimir O. Osorio Muñoz, Research 

Engineer; Carlo Paulotto, Dr., Research Engineer; Acciona Infraestructuras, Alcobendas, Spain. Contact: carlo.paulotto@acciona.es 

Abstract 

The introduction of new materials 

such as fibre reinforced plastics (FRPs) 

into construction practices has been 

discussed in this paper. Hereafter, 

some of the most significant examples 

of the applications of these new materials 

are presented in some detail. 

Keywords: carbon; glass; fibre; matrix; 

reinforcement. 

Introduction 

For the sake of clarity, the examples 

concerning the application of FRP 

materials in civil infrastructure in this 

paper are divided into two groups: 

structures and structural members 

entirely made with FRP materials and 

concrete structures or structural members, 

where FRP materials are used as 

external or internal reinforcement. The 

FRP beams employed to build up the 

girders of a road overpass along the A8 

freeway close to the Asturias Airport 

in the north of Spain and those used 

for the girders of two bridges along the 

M-111 freeway in Madrid belong to 

the former group. To the same group 

belong two FRP slabs, used as a channel 

bed during the channelling work of 

the Rio Mesoiro River in Galicia and 

the spiral stairs designed and manufactured 

for placement in a pumping well 

along one of the metro lines in Madrid. 

The FRP sheets used to reinforce both 

in shear and in bending a “reshaped” 

concrete beam in a building in Aragón 

and the FRP reinforcing bars used in 

some parts of the concrete substructure 

of the new tramway in Granada belong 

to the latter group. 

Airport in the north of Spain [1]. This 

is a four-span bridge of total length 

of 46 m (see Fig. 1a). The bridge girders 

are three continuous longitudinal 

carbon fibre beams on three intermediate 

supports connected to the 

reinforced concrete deck through 

alkali-resistant glass fibre shear connectors 

(see Fig. 1b). The beams are 

characterized by trapezoidal cross 

sections. They were manufactured in 

two parts and transported to the work 

site where they were joined by means 

of adhesive step joints. Each step joint 

was manufactured by placing the two 

parts of a beam side-by-side, wrapping 

them with pre-impregnated carbon 

fabric which was finally consolidated 

using the vacuum bag system. Using a 

crane, it took only half a day to place 

the three 46 m long beams on their 

supports. This was possible thanks 

to the minimal weight of the beams 

(a) 

at 1 kN/m. Glass fibre stay-in-place 

formworks were used to cast the concrete 

deck. This bridge was the result 

of an international research project 

involving Spain, Colombia and Cuba, 

and it was aimed at demonstrating the 

possibility of building a road overpass 

using FRP materials. 

The other two bridges are located 

on the outskirts of Madrid along the 

M-111 freeway [2]. These two bridges 

are identical, each made up of three 

simply supported spans (10, 14 and 

10 m) with a 20,40 m wide box-girder 

deck (see Fig. 2a). The four hybrid 

carbon–glass fibre deck beams have 

reverse “Ω”-shaped cross sections 

(see Fig. 2b). Each beam was closed by 

connecting its top flange to a sandwich 

panel along its entire length. This panel 

initially acts both as the top flange of 

the beam and as formwork for the 

Fig. 1: (a) General view of the bridge in Asturias; (b) View of the carbon fibre beams and 

the glass fibre stay-in place formworks 

(b) 

Structures and Structural 

Members Made Entirely 

of FRP Materials 

Asturias and M-111 Bridges 

Since 2003, three vehicular bridges 

have been designed, manufactured 

and erected whose girder beams are 

completely made of FRP materials. 

The first bridge is located along 

the highway leading to the Asturias 

(a) 

(b) 

Fig. 2: (a) General view of one of the twin bridges along the M-111 freeway; (b) One of 

the FRP beams during positioning 


uncured concrete deck slab. Once the 

concrete has strengthened, the slab 

contributes to the deck strength, as it 

is connected to the top flanges of the 

beams through steel studs. The minimal 

weight of the beams (3 kN/m) 

meant that they could be placed on the 

supports using a simple crane truck. 

12 beams were positioned during 

one working day. These two bridges 

were developed as demonstrators in 

the framework of a research project 

funded by the European Commission. 

The overall objective of the project 

was the development of a new highperformance 

and cost-effective construction 

concept for bridges based on 

the application of rapid-renewal and 

long-life service infrastructures in the 

countries that joined the European 

Union in 2004. 

lay-up and consolidated using the 

vacuum bag technique. The slabs were 

2,50 m wide and 14,50 m long, and had 

a thickness of 0,05 m (see Fig. 3). They 

were manufactured in a workshop in 

Madrid and moved to the work site by 

truck. After the two FRP bottom slabs 

were put in place, the lateral walls 

and top slabs of reinforced concrete 

were cast (see Fig. 4). The structural 

connection between the FRP bottom 

slabs and the lateral walls was ensured 

using steel studs with one of their ends 

embedded in the FRP slabs. 

Spiral Stairs for a Pumping Well 

in Madrid 

The stairs of one of the pumping 

wells along one of the metro lines 

of Madrid were exposed to a highly 

aggressive environment owing to 

water and gasoline infiltrations. The 

well has a depth of 21,15 m and a 

Rio Mesorio River Channelling 

During the channelling work of the 

Rio Mesorio River in the Spanish 

region of Galicia, the channel had 

to cross a pre-existing oil pipeline at 

one point [3]. According to the original 

design, the channel consisted of 

a reinforced concrete hollow section 

composed of two identical cells. The 

section was 2,40 m high and 5,40 m 

wide, with all the walls having the 

same thickness of 0,30 m, except for 

the central vertical diaphragm that 

had a thickness of 0,50 m. During the 

excavation phase, it was found that 

the thickness of the soil layer covering 

the pipeline was less than expected. 

The level of the top face of the channel’s 

bottom slabs however could not 

be varied, but executing the original 

plan would have meant interference 

between the bottom slabs and the oil 

pipes. To cope with this problem, the 

only feasible solution appeared to 

be reduction of the thickness of the 

slabs constituting the channel bed by 

replacing them with a material with 

better mechanical properties as compared 

with reinforced concrete. The 

use of steel was judged inappropriate 

due to possible corrosion problems. 

On the other hand, FRP materials, 

having both better mechanical properties 

than reinforced concrete and not 

suffering from corrosion phenomena, 

appeared to be the best alternative. To 

optimize the cost of the FRP slabs, it 

was decided to manufacture them with 

a mix of carbon and glass fibre fabrics 

that were impregnated with epoxy 

resin to overcome the constraint 

imposed by the carbon fibre sizing. 

The slabs were manufactured by wet 

Fig. 3: Positioning of the FRP channel bed slabs 

Fig. 4: View of the completed channel section 


3,98 m side square cross section, 

whereas the access to the well from 

the ground level is 1,00 × 0,80 m only. 

Originally, these stairs which were 

placed in the well to reach the pumps 

for their routine maintenance were 

made of steel, but intensive corrosion 

soon made it dangerous for the 

workers and the decision to replace 

them was made in November 2008. 

Initially, the plan was to substitute 

the steel stairs with aluminium or 

stainless steel stairs, but the owner 

judged both solutions too expensive 

and expressed concerns regarding 

possible problems in the construction 

of the new stairs because of the 

limited room available around the 

well entrance. In fact, the entrance is 

located in front of one of the Madrid 

airport terminals between a ramp and 

a belt conveyor, partially covered by 

a shelter. A solution of spiral stairs 

with the mezzanine floors completely 

made of FRP materials was then proposed 

(see Fig. 5). This solution had 

the advantage that the FRP material 

used was lightweight and corrosion 

free, thanks to its inherent properties. 

Glass fibre was chosen as the 

reinforcing fibre in order to bring 

down the cost of the stairs as much 

as possible, and phenolic resin was 

selected as the matrix due to its fireretardant 

characteristic. Moreover, it 

was proposed that a modular solution 

with the different stair steps 

progressively mounted one on top of 

the other to form the stairs, be used. 

This allowed the different parts of 

the stairs to be moved into the well 

Fig. 5: View of the FRP spiral stairs for the 

metro pumping well 

from ground level by simply using a 

pulley. In fact, a single stair step and a 

mezzanine floor have weights of 0,30 

and 1,47 kN, respectively. The different 

parts of the stairs were manufactured 

using different techniques. The 

57 steps characterized by a sandwich 

structure with a fire- resistant core 

were manufactured through resin 

transfer moulding (RTM). The five 

landings have the same sandwich 

structure of the steps and were manufactured 

by resin infusion. The three 

mezzanine floors were manufactured 

by resin infusion, while their supports, 

which were finally connected by steel 

dowels to the concrete walls of the 

well, were manufactured by hand 

lay-up. The design, manufacturing 

and construction of the stairs took 2 

months. 

FRP Materials as 

Reinforcement for Concrete 

Structures 

Beam Reshaping 

Because of a mistake during the 

design phase, a 11m span beam 

belonging to the reinforced concrete 

supporting structure of a building 

in the Spanish region of Aragon 

had an incorrect depth of 1,37 m 

instead of 1,07 m. This beam was 

positioned along the border of the 

building and its extra depth affected 

the height of a window beneath the 

beam, running along its whole length. 

The beam was initially reinforced in 

tension with eight steel rebars having 

a diameter of 20 mm and placed 

in two layers near the bottom of the 

beam cross section. The shear reinforcement 

consisted of 10 mm diameter 

steel stirrups spaced at 0,30 m. It 

was initially thought the beam depth 

could be reduced by cutting off 0,30 

m of the bottom of the beam and bolting 

steel plates to the reshaped beam. 

The heavy weight of the steel plates, 

however, would have represented a 

serious problem during the strengthening 

phase, because of the necessity 

of lifting them to the beam level and 

keeping them in position during the 

bolting phase. All these operations 

would have been further complicated 

by the provisional supports that 

needed to be used to sustain the beam 

after removal of the tensile reinforcement. 

For this reason, the solution of 

reinforcing the reshaped beam with 

carbon fibre plates was considered 

a better option. After cutting the 

beam, the cut surface was cleaned by 

blowing it with compressed air and a 

primer was applied on it. The cut, dry 

carbon fabrics were applied to the 

bottom and sides of the beam and 

impregnated with epoxy resin. At the 

bottom of the beam, the carbon fibre 

filaments were mainly oriented along 

the beam’s axis to resist the tensile 

stresses induced by bending, while at 

the beam-sides the carbon fibre filaments 

were oriented at ±45° with 

respect to the beam’s axis to sustain 

the external shear forces. 

New Tramway in Granada 

During the construction of a new 

tramway in Granada in 2009, the 

owner requested that steel rebars 

in some portions of the reinforced 

concrete slab supporting the rails be 

removed. The reason for this request 

was that normal steel rebars interfered 

with the electromagnetic system 

that detected the position of 

the tramcars along the line. To solve 

this problem, it was decided to take 

advantage of the electromagnetic 

transparency exhibited by glass fibre 

reinforcing bars. Owing to the permanent 

character of this application, 

vinyl ester resin was employed 

as matrix for the rebars to protect 

the glass filaments from the alkaline 

environment in the Portland concrete. 

A total of 13 560 m of 16 mm straight 

rebars were manufactured by pultrusion 

and 26 448 m of 12 mm diameter 

stirrups were produced through a 

process similar to pultrusion. A large 

experimental test campaign was carried 

out to first evaluate the thermomechanical 

properties of the rebar to 

be used in the design phase and then 

to ensure that these properties were 

maintained over all the rebar lots that 

were delivered. 

Conclusion 

From the applications presented in 

this article, it is evident that it is possible 

to take advantage of the inherent 

properties of composite materials, 

such as corrosion resistance, minimal 

weight, high tensile strength and 

electromagnetic transparency, and to 

cope with scenarios in which the use 

of the traditional construction materials 

would be impossible or imply 

higher costs in terms of money or 

time. 


References 

[1] Gutierrez E, Primi S, Mieres JM, Calvo I. 

Structural testing of a vehicular carbon fiber 

bridge: quasi-static, and short-term behavior. J. 

Bridge Eng. ASCE 2008; 13 (3): 271–281. 

[2] Primi S, Areiza M, Bansal A, Gonzalez A. 

New design and construction of a road bridge 

in composites materials in Spain: sustainability 

applied to civil works. In Proceedings of the 9th 

International Symposium On Fiber-Reinforced 

Polymer Reinforcement for Concrete Structures, 

2009. Oehlers DJ, Griffith MC, Seracino R 

(eds). Sydney, New South Wales, 2009, 62. ISBN: 

9780980675504. 

[3] Paulotto C, Primi S, Bansal A. A composite 

bed for the Rio Mesorio river. Proceedings of 

the Composites UK 10th Annual Conference, 

Birmingham, England, 2010. 


Owner: 

Ministerio de Fomento 

(for the Asturias bridge); 

Comunidad de Madrid 

(for the M111 bridges) 

Designer: 

Acciona Infraestructuras 

(for all the 3 bridges) 







FRP (t): 15 (for the Asturias bridge); 

22,5 (for each one the M111 bridges) 

Span lengths (m): 46 (for the Asturias 

bridge); 34 (for each one of the M111 

bridges) 

Construction cost (USD million): 0,5 

(for the Asturias bridge); 0,5 (for each 

one of the M111 bridges) 

Service date: 2004 (for the Asturias 

bridge); 2008 (for both the M111 

bridges) 

Sign up for an Institutional Subscription 

2 Print Copies per issue (Feb, May, Aug., Nov) 

Online Access to the Journal for the entire 

Institution via Site License 

Online Access to Current + Back issues (1991- ) 

at no extra cost (1500 and more SEI articles) 

only CHF 630 a year. 

www.iabse.org/SEI 


Fiber-Reinforced Polymer Decks for Movable Bridges 

Raymond D. Bottenberg, P.E., Oregon Department of Transportation, Bridge Preservation Engineering, Oregon, USA. 

Contact: raymond.d.bottenberg@odot.state.or.us 

Abstract 

Fiber-reinforced polymer (FRP) 

bridge decks are ideal for re-decking 

movable bridges. They can match the 

weight of existing steel grid decking 

to minimize bridge balance adjustment. 

They are bicycle- and motorcycle-friendly 

and quieter than steel 

grids. They protect the structure 

from precipitation and are corrosion 

resistant. 

In 2002, the Oregon Department of 

Transportation (ODOT) installed 

FRP decks on two movable bridges 

near Astoria. Cementitious grout 

pads supported the deck on the steel 

stringers, and the deck was secured 

to the stringers by welded steel studs 

in cementitious grout pockets. The 

wearing surface was to have asphalt 

concrete, but was changed to epoxy 

polymer concrete after the asphalt 

slid off a bascule leaf during a prolonged 

test lift. The grout details had 

failed on one of these bridges, resulting 

in deck replacement, and both 

bridges had experienced distress in 

the wearing surface. The attachment 

details and wearing surfaces used on 

these bridges did not accommodate 

the higher deflections associated with 

FRP decks, and were regarded as 

unacceptable. 

In 2005, the ODOT installed a FRP 

deck on a movable bridge in Florence. 

Neoprene sheets supported the deck 

on the steel stringers, and the deck 

was secured to the stringers with large 

structural blind fasteners installed 

from below. The wearing surface was 

urethane polymer concrete, placed 

by the broom and seed method. This 

installation showed none of the distress 

experienced on the two bridges at 

Astoria. The strong but flexible attachment 

details and wearing surface used 

on this bridge accommodate deflections 

well and were recommended for 

future installations. 

Keywords: bascule; movable bridge; 

pultruded FRP decking; steel 

grid decking; polymer concrete; 

structural blind fastener; staged 

construction. 

Introduction 

Historically, movable bridges have 

carried many kinds of decks including 

timber, reinforced concrete, welded 

steel grid, and orthotropic steel. One 

of the key properties of a movable 

bridge deck is its weight, which on 

both bascule-type bridges and vertical 

lift bridges directly affects the amount 

of counterweight needed and the span 

drive machinery power requirements. 

Often corrosion resistance is also 

important, as many movable bridges 

exist in corrosive marine environments. 

In 1997, the first pultruded fiber-reinforced 

polymer (FRP) bridge deck 

was produced by bonding tubular 

sections into panels. FRP decks have 

been installed on approximately 112 

bridges in the United States. In many 

cases, FRP bridge decking can closely 

match the weight and thickness of 

steel grid deck products. For movable 

bridges, this makes it an ideal deck retrofit 

material, as little adjustment of 

counterweights is required and little 

or no adjustment of roadway height 

is needed. Additional benefits include 

the inherent corrosion resistance of 

FRP, protection of the bridge structure 

from precipitation and road run-off, 

quiet ambient noise levels, and a more 

drivable surface for all vehicles, in particular 

motorcycles and bicycles. 

The ODOT rehabilitated two basculetype 

movable bridges in 2000 through 

2002, replacing deteriorated timber 

decks with FRP deck panels. Both 

bridges exist in a corrosive marine environment 

near Astoria, Oregon, and the 

FRP deck was selected to help prevent 

corrosion. Similarly in 2005, ODOT 

replaced a failing steel grid deck of 

a bascule-type movable bridge with 

an FRP deck. This bridge also exists 

in a corrosive marine environment in 

Florence, Oregon, and the FRP deck 

was selected to help prevent corrosion. 

ODOT Lewis and Clark 

River Bridge 

The Lewis and Clark River Bridge 

is a single-leaf bascule-type movable 

bridge built in 1924, carrying 

two traffic lanes on a 6,096 m wide 

roadway, which is supported by steel 

stringers on 0,851 m center-to-center 

spacing. Rehabilitation work undertaken 

in 2000 through 2002 included 

deck replacement, new operating 

machinery, machinery house restoration, 

bridge rail retrofit and some 

repairs to the timber approach spans. 

The FRP deck was a 178 mm thick 

product, which was delivered in 2,134 

× 6,199 m sections. Holes with 102 mm 

diameter were sawed through the top 

and bottom skins of the FRP decking 

on 610 mm spacing above the stringers, 

and foam dams were placed in the 

FRP panels to create 203 × 161 mm 

voids around each future shear stud 

location (refer to Figs. 1 and 2). The 

existing timber deck was removed, 

and the top flange of the stringers was 

cleaned by power tool cleaning according 

to Society for Protective Coatings 

(SSPC) SP11 standards. 1 Narrow neoprene 

strips were placed to support 

the FRP deck panels atop the stringers 

Welded shear stud 

Portland cement grout 

FRP panel 

Neoprene strip & 

Light gauge steel angle 

Steel stringer 

Portland cement 

grout pad 

Section view 

looking parallel to roadway centerline 

Fig. 1: Longitudinal detail view of FRP 

deck installation at shear stud location 

Welded shear stud 

Portland cement grout 

FRP panel 

Steel stringer 

Foam dam 

Portland cement 

grout pad 

Section view 

looking transverse to roadway centerline 

Fig. 2: Transverse detail view of FRP deck 

installation at shear stud location 


until the 25 mm cementitious grout 

pads were poured. Each FRP deck 

panel was placed on the stringers by 

a crane as shown in Fig. 3, and then 

jacked to seat the tongue and groove 

connection between panels, which was 

coated with adhesive. Wet-layup FRP 

doublers were added over each field 

splice. Light gauge steel angles were fastened 

to the bottom of the FRP panels 

along the sides of each stringer. Steel 

shear studs with 22 mm diameter were 

welded to the stringers through the 

holes in the deck panels. Cementitious 

grout filled the gap between the stringers 

and the FRP, and filled the pockets 

around each shear stud. Cells of the 

FRP panels near the joint armor at 

each end of the deck were also filled 

with cementitious grout. 

Following installation of the FRP 

decking, a 50 mm thick layer of asphalt 

concrete wearing surface was placed 

on the bridge. On 25 June 2001 during 

installation of the new operating 

machinery, the bascule leaf was left in 

the “open” position for approximately 

5 h. The result was cracking of the 

wearing surface, followed by large sections 

of asphalt sliding to the bottom 

of the open span. 2 Figure 4 shows the 

bridge with the bascule leaf in “open” 

position after the asphalt had slid-off. 

Fig. 3: FRP deck panel being placed on Lewis and Clark River Bridge 

Engineering studies and some laboratory 

tests 3 were conducted following 

this failure. The testing measured 

the bond shear properties of asphalt 

and identified the fact that asphalt is 

a viscoelastic material. When the bascule 

leaf was lifted, gravitational force 

produced a constant shear force acting 

on the asphalt and its bond, producing 

viscoelastic strains and eventual 

failure. Due to this behavior, it was 

concluded that asphalt is not suitable 

for an FRP deck on a bascule-type 

movable bridge. Recommendations 

were made to remove the asphalt and 

the “tack” coat, and replace it with a 

50 mm thick layer of epoxy polymer 

concrete. The old wearing surface was 

removed by the use of an excavator, 

solvent cleaning, and pressure washing. 

The exposed surface was dried by 

the use of “weed burner” torches. A 

primer was applied to the deck using 

paint rollers, and the epoxy binder was 

pre-mixed with aggregate, placed on 

the deck, and screeded. A topping coat 

was applied by the broom and seed 

method, where the deck was flooded 

with the polymer and the aggregate 

was broadcast into the curing polymer 

to provide skid resistance. 

Cracks began to appear in the epoxy 

polymer concrete wearing surface by 

May 2002, typically at the locations of 

shop and field splices in the FRP deck. 4 

Later ODOT tensile tests of polymer 

overlay materials 5 demonstrated that 

the epoxy polymer concrete exhibits 

brittle behavior at temperatures up 

to 38°C, making it unable to accommodate 

the strains concentrated at 

the splice locations when exposed to 

normal operating temperatures. These 

cracks have continued to grow at a 

moderate rate and replacement of the 

wearing surface will be required in the 

future. 

Old Young’s Bay Bridge 

Fig. 4: Asphalt concrete wearing surface failure on Lewis and Clark River Bridge 

The Old Young’s Bay Bridge, shown 

in Fig. 5, is a double-leaf bascule-type 

movable bridge built in 1921 that carries 

two traffic lanes on a 6,325 m wide 

roadway supported by steel stringers 

on 1,524 m center-to-center spacing. 

Rehabilitation work undertaken 

in 2000 through 2002 included deck 

replacement, operator’s house restoration, 

bridge rail retrofit, and some 

repairs to the timber approach spans. 

The FRP deck was again a 178 mm 

thick product installed using the same 

details as on the Lewis and Clark River 

Bridge. Following installation of the 


Siuslaw River Bridge 

Fig. 5: Historic view of Old Young’s Bay Bridge in the “open” position 

FRP decking, a 19 mm thick epoxy 

polymer concrete wearing surface was 

placed by the broom and seed method. 

Cracks began to appear on the epoxy 

polymer concrete wearing surface 

by May 2002 similar to the cracking 

observed on the Lewis and Clark 

River Bridge. These cracks continued 

growing at an aggressive rate, 4 probably 

exacerbated by increased deflections 

due to defective shop splices. An 

advanced example of this cracking 

is shown in Fig. 6. The cementitious 

grout around the shear studs failed 

as shown in Fig. 7. The cementitious 

grout between the FRP panels and 

the stringers also failed, starting at 

the joint armor and propagating along 

the stringers. Due to these failures 

and consequent FRP panel damage, 

this deck was replaced in June 2010 

with a steel grid deck at the request 

of ODOT district management. These 

failures could have been avoided by 

using ductile materials for the wearing 

surface and using strong but flexible 

materials for the attachment details 

to accommodate the deformations 

experienced between the flexible steel 

floor system and the more flexible 

FRP deck panels. 

Fig. 6: Cracking and spalling of epoxy polymer concrete wearing surface on the Old 

Young’s Bay Bridge 

The Siuslaw River Bridge is a doubleleaf 

bascule-type movable bridge built 

in 1936, carrying two traffic lanes on 

an 8,230 m wide roadway, which is 

supported by steel stringers on 0,914 m 

center-to-center spacing. Retrofit work 

undertaken in 2005 included deck 

replacement, recycled plastic lumber 

sidewalk decking, and new reinforced 

plastic walers on the fender system. 

The steel grid deck had been supporting 

the sidewalks, and the FRP deck 

retrofit required installation of new 

supports made from rectangular steel 

tubing, spanning between sidewalk 

brackets to keep the loads off the FRP 

decking. Stringer splices were modified 

by moving the splice plates to the bottom 

of the stringer flange and installing 

with countersunk fasteners. The 

FRP deck was a 127 mm thick product 

which was delivered in 2,438 × 4,089 m 

sections. The existing steel grid deck 

was removed, and the top flange of the 

stringers was repaired and cleaned by 

needle guns to SSPC SP11 standards. 1 

Neoprene pads with 3 mm thickness 

were placed on each stringer. Each 

FRP deck panel was placed on the 

stringers by a crane, and then jacked to 

seat the tongue and groove connection 

between panels, which was coated with 

adhesive. Wet-layup FRP doublers 

were added over each field splice. The 

FRP deck panels were secured with 16 

mm diameter structural blind fasteners 

through existing holes in the stringer 

flanges and newly drilled matching 

holes in the FRP. This installation is 

shown in Fig. 8. Cells of the FRP panels 

near the joint armor at each end of 

each deck were filled with urethane. 

The deck was installed in two stages 

to meet traffic needs, with the FRP 

panels spliced at roadway centerline. 

The bottom skins of the FRP panels 

were fastened securely to the flange 

of a stringer located at the centerline. 

The top skins of the FRP panels 

were fastened together using a 2,7 

mm thick 304 stainless steel plates, 

abrasive blasted and installed using a 

polyurethane adhesive sealant and 6,4 

mm diameter stainless steel structural 

blind fasteners. 

Owing to the problems with the polymer 

concrete wearing surfaces on the 

Astoria bridges, further laboratory 

tension testing 5 was performed with 

a variety of polymer concrete wearing 

surface materials cast into dogbone-shaped 

specimens. This testing 

revealed that the epoxy polymer 


e prevented by saw-cutting to break 

the bond and filling with sealant. There 

has been some loss of aggregate from 

the wearing surface, which may be 

attributed in some part to the use of 

unclean, dusty aggregate that did not 

adhere properly to the urethane. 

Conclusion 

Fig. 7: Grout pocket distress on Old Young’s Bay Bridge, following loss of grout pad 

Fig. 8: Attachment details of FRP deck installation on the Siuslaw River Bridge 

concrete wearing surface is ductile 

only at temperatures over 38°C, but 

urethane polymer concrete remains 

ductile at temperatures as low as 

−9°C. Urethane polymer concrete was 

selected for the Siuslaw River Bridge 

on the basis of this testing. The FRP 

deck was abrasive blasted and then reabrasive 

blasted at the request of the 

urethane supplier. The deck was then 

cleaned using acetone solvent and 

a 13 mm thick wearing surface was 

placed by the broom and seed method 

in two lifts. The completed installation 

is shown in Fig. 9. 

To date, this installation made with a 

ductile wearing surface and strong but 

flexible attachment details shows none 

of the distress experienced on the two 

bridges at Astoria. There are no cracks 

in the wearing surface over the shop 

and field splices in the FRP deck. At 

the ends of the deck, the edges of the 

FRP panels deflect and the joint armor 

does not, causing a crack that can 

Through these experiences, ODOT has 

learned that FRP decks are ideal for 

replacing decks on movable bridges, 

but only if appropriate details are 

used for attachment and for wearing 

surfaces. The weight of FRP decking 

and wearing surface can minimize the 

need for rebalancing the bridge, and 

the polymer concrete wearing surface 

is safer for vehicles including motorcycles 

and bicycles. This wearing surface 

is also much quieter than steel grid 

decking. 

It is important to note that the two 

bridges at Astoria did not experience 

FRP deck failure. They experienced 

failure of attachment details and 

wearing surfaces that were unable to 

accommodate the significant deflection 

of FRP. 

In addition, FRP decks have potential 

for rapid construction and staging 

applications as demonstrated by the 

two-stage installation accomplished on 

the Siuslaw River Bridge. 

The following guidelines are recommended 

for FRP deck retrofit installations 

for moveable bridges: 

1. Strong but flexible attachment 

details should be used and brittle 

materials avoided. ODOT has had 

good experience with structural 

blind fasteners with neoprene pads 

between the stringers and the FRP 

panels. 

2. The number of holes cut into the 

top and bottom skins of the FRP 

panels should be minimized. 

3. Polymer concrete wearing surface 

having high degree of ductility at 

expected operating temperatures 

should be used. 

4. Clean, dry aggregate should be 

used in any polymer concrete 

wearing surface. 

5. Asphalt concrete on bascule-type 

movable bridges should not be 

used. 

6. Adequate drain holes should be 

provided to prevent accumulation 

of water in FRP panels. 

7. When replacing steel grid decking, 

the likelihood that the stringers 


quality control programs should be 

specified. 

10. A saw-cut and flexible sealant 

between the FRP deck panels 

and rigid joint armor should be 

provided. 

References 

[1] Surface Preparation Standard No. 11, 

Power Tool Cleaning to Bare Metal. SSPC: The 

Society for Protective Coatings. Pittsburgh, 

PA, September 1, 2000 and November 1, 2004 

revisions. 

[2] Nelson J. Asphalt Crashes Down From 

Bridge. The Oregonian. Portland, Oregon, June 

27, 2001. 

[3] Lovejoy SC, Nelson SD, Patterson BM. 

Causes and Solutions to the Failure of the Wear 

Surface on the Lewis and Clark River Bridge. 

Unpublished report. Oregon Department of 

Transportation, November 21, 2001. 

Fig. 9: View of completed FRP deck installation on the Siuslaw River Bridge 

will have wear damage and may 

require replacement should be 

considered. 

8. Stringer splices with plates on top 

of the top flange should be modified 

to a configuration with countersunk 

fasteners and splice plates 

on the underside of the top flange. 

9. FRP deck products made by 

manu facturers that have effective 

[4] Lovejoy SC. Summary of Cracking in the 

Wearing Surface of the FRP Decks on Bridges 

0711 (Lewis and Clark River) and 0330 (Old 

Young’s Bay). Unpublished report. Oregon 

Department of Transportation, October 20, 2004. 

[5] Barquist G, Lovejoy SC, Nelson SD, Soltesz 

S. Evaluation of Wearing Surface Materials 

for FRP Bridge Decks. Oregon Department 

of Transportation and Federal Highway 

Administration, July, 2005. 

Make Your Client 

Remember You 

through out the Year.. 

Gift a SEI Subscription 

Contact: iabse@iabse.org, Subject: Subscription Gift 

CHF 220/Year for 4 issues. 


Glass Fiber Reinforced Polymer Strengthening 

and Evaluation of Railroad Bridge Members 

GangaRao V.S. Hota, Prof., Ph.D., P.E., Director; P.V. Vijay, Research Prof., PhD, PE; Dept. of Civil and Environmental 

Engineering West Virginia University, Morgantown, USA and Reza S. Abhari, MSCE, PE, Structural Eng., Alpha Corporation; USA. 

Contact: p.vijay@mail.wvu.edu 

Abstract 

In this work, damaged timber railroad 

bridge stringers and piles were 

rehabilitated with glass fiber reinforced 

polymer (GFRP) composites, 

and tested. Four timber stringers (152 

× 203 × 3560 mm) removed from the 

field were rehabilitated with GFRP 

spray lay-up and GFRP wrap vacuum 

bagging methods. GFRP strengthening 

increased the shear moduli of 

the two stringers by 41 and 267%. 

Rehabilitation and load testing were 

also performed on an open-deck-timber 

railroad bridge built during the 

early 1900s on the South Branch Valley 

Railroad (SBVR) owned by the West 

Virginia Department of Transportation 

(WVDOT) in Moorefield, WV, 

USA. Specifically, field rehabilitation 

involved repairing piles using GFRP 

composite wraps and phenolic formaldehyde 

adhesives. Static and dynamic 

tests using a 80 ton locomotive showed 

that the rehabilitated piles and pile cap 

showed a 43 and 46% strain reduction, 

respectively. Dynamic load amplification 

factor was noted to be almost 

close to a speed of 24 km/h. 

Keywords: strengthening; GFRP; composites; 

rehabilitation; timber piles; 

timber stringers. 

Introduction 

In the USA, timber was the primary 

railroad bridge construction material 

until the early 20th century. The use 

of lumber began to slowly decline in 

the early 20th century as other materials 

such as steel and concrete replaced 

wood for bridge construction. The US 

railroads have more than 2400 km of 

timber bridges and trestles in service 

today, which typically provide more 

than 50 years of acceptable service 

when properly designed, constructed, 

and maintained. 1 With the recent 

increases in railway car axle loads 

and long-term exposure to the environment, 

many of these bridges have 

reached the end of their service life or 

their load carrying capacity has lowered 

significantly because of material 

aging. This work describes the use of 

GFRP for rehabilitating in-service 

superstructure and substructure railroad 

timber bridge members. 

Objectives and Scope 

The work presented herein focuses on 

developing safer, faster, and most practical 

methods that allow in situ rehabilitation 

of timber railroad bridges 

without interrupting rail traffic. The 

objective of this research work is to 

evaluate the application of GFRP 

composite spray and wraps as a viable 

rehabilitation alternative for stringers 

and piles on in-service timber 

bridges. In this work, damaged timber 

bridge stringers and piles selected 

by the South Branch Valley Railroad 

(SBVR) in Moorefield, WV, USA were 

repaired and rehabilitated with the use 

of GFRP composites. The work consisted 

of: (a) determining bending and 

shear properties of damaged timber 

stringers removed from field-service, 

and performing analytical comparisons 

prior to and after repair through 

sprayed GFRP (SGFRP) and vacuum 

bagging methods, and (b) rehabilitating 

and evaluating timber railroad 

bridge piles under static and dynamic 

loads (up to 24 km/h). 

Stringer Rehabilitation Scheme 

A total of four 50+ year-old creosotetreated 

Douglas-fir stringers were 

acquired from the SBVR-WVDOT. 

The four stringers were deemed deficient 

due to significant checking and 

splitting along their length (Fig. 1). All 

four specimens (152 × 203 × 3560 mm) 

were load tested up to 26,7 kN before 

repair to determine flexural rigidity 

and shear modulus. Stringers were surface 

prepared by removing dirt and 

debris, by filling of void or splits with 

fillers (resin and saw dust mix), and 

edge smoothening to reduce stress 

concentrations. After sanding, primer 

coat (phenolic resin) was applied and 

cured. Retrofitting was done using the 

most compatible primer/resin combination 

using phenolic formaldehyde. 

Vacuum Bagging and 

GFRP Spray Methods 

Vacuum bagging creates clamping 

forces (127–152 mm Hg vacuum) 

to hold the resin-coated fiber/fabric 

components in place and removes the 

trapped air. Fabric pieces were cut 

to cover three sides of the specimen. 

The vacuum bagging method used a 

sealed plastic bag covering the lay-up 

area including a breather layer, and a 

vacuum pump to remove the air and 

apply uniform surface pressure during 

resin curing (Fig. 1). For the SGFRP 

method, two separate compressed air 

spray guns were used, in lieu of an 

Vacuum bag 

covering the 

wrapped area 

Vacuum pump 

(a) (b) (c) 

Fig. 1: Rehabilitation of timber stringers: (a) field-removed stringers; (b) vacuum bagging of wrapped beam; (c) spraying fiber on the 

wet surface 


integrated sprayer, one for spraying 

the resin and another for spraying the 

fiber. Chopped glass fibers, 12,7 mm 

in length, were sprayed on the primer 

coated wood surface in six cycles with 

a final SGFRP layer thickness measuring 

15 to 18 mm. Unlike the wrapping 

method that requires a smooth surface, 

spraying technique can easily be 

applied to any type of surface. 

Stringer Tests/Analysis 

After rehabilitation, stringers were 

tested under four-point bending with a 

shear span (a/h) ratio of 3,4. Stringers 

were instrumented with strain gages on 

the surface of the GFRP wraps at midspan 

on the tension side and rosette 

gages at the neutral axis (to evaluate 

shear and principal strains). Load versus 

deflection and shear strain were plotted 

and the modes of failure were identified. 

Based on the rule of mixtures and 

burn-off tests, the average elastic moduli 

for the GFRP wraps and SGFRP specimens 

were found to be 18 340 and 8963 

N/mm 2 , respectively. The lower modulus 

of the SGFRP is attributed to the lower 

density of the sprayed wraps. 

The results in Table 1 indicate that 

repairing damaged timber stringers 

using GFRP or SGFRP is effective in 

improving their flexural rigidity and 

shear moduli. Repairing shear zones of 

a stringer with GFRP composites had a 

significant impact on the shear modulus 

and deflection response. As the tested 

stringers were in poor condition with 

large voids, they were filled with strong 

filler material. Specimen #2 had wide 

longitudinal splitting, therefore the initial 

EI value was low, but it had more 

than tripled after repair. Specimens 

#3 and #4 containing less voids had a 

19,5 and 15% increase in the EI value, 

respectively. For specimen #3 repaired 

with four layers of GFRP wrap with 

vacuum bagging, the maximum strain 

measured at the extreme tension fibers 

at mid-span was 2488 µε compared to 

the maximum shear strain measured at 

the neutral axis of the specimen, which 

was 4446 µε. 

Equations (1) and (2) were used to 

calculate the parameters shown in 

Table 1 for beams prior to and after 

rehabilitation with FRP using SGFRP 

(beams #2 and #4) and vacuum bagging 

(beam #3) techniques. 2 The failure 

of a wooden beam in shear occurs 

when the maximum shear stress, s max , 

becomes equal to the shear strength 

of wood, s d , as given by Eq. (1). 2 The 

flexural rigidity (EI exp ) of the repaired 

specimens is computed from deflection 

equation (2) with bending and 

shear terms (Δ = Δ bending + Δ shear ). 

v 

bh 

EI 

d 

exp 

 

n h h 

+ 

 

h 

+ 2 

 

 

1 

frp 

1 

h 

frp 

 

2 

frp 

 

frp 

 

 

 

= 

 

3 frp 

+ 

frp 

 

1 n h = 2 

h 3 r (1) 

 

 

 

 

 

a P 2 2 

( L – a ) 

 

 

 

3 4 

 

= 

a P 

 

48 1 – 

GA 

 

 

 

2 

 

 

(2) 

Where V = shear force; s d = shear 

strength of wood in horizontal shear; 

b = width of the beam; I = moment of 

inertia of the section; h frp = height of 

the FRP section; h = height of the beam 

section; n = modular ratio = E SGFRP/ 

E wood ; r frp = FRP area fraction = 2th frp / 

bh; t = thickness of the FRP composite; 

r = ratio of shear capacity of FRPreinforced 

section to shear capacity of 

unreinforced section (effectiveness); 

L = span; k = shape factor; a = shear 

span; and (P/D) is the initial slope of 

load versus deflection curve. 

In Situ Bridge Pile 

Rehabilitation 

Overview 

In situ pile rehabilitation was carried 

out on SBVR-WVDOT track containing 

several timber bridges constructed 

in the early 1900s. Bridge No. 574 was 

approximately 14 m long with four piles 

that were highly deteriorated and/or 

damaged due to flooding (Fig. 2). Pile 

#1 was heart rotted (i.e. damaged core) 

and only about 20% of the cross sectional 

area was functional, and piles #2, 

#3, and #4 were heart rotted with 45% 

effectiveness. Instead of replacement, 

the badly damaged piles were rehabilitated 

to enhance their durability and 

load-bearing capacity. Phenolic-based 

adhesive with E-glass FRP composite 

wrap (0,4 kg/m 2 density with 0,2 kg/m 2 

in both 0 and 90 directions) was used 

owing to its cost advantage, chemical 

compatibility, excellent bonding, and 

electrical non-conductivity 3 . 

Rehabilitation Details 

Cofferdams were built around each 

submerged pile using metal sheets. 

Piles were pressure washed, dried, 

and sanded to obtain a smooth surface. 

Phenolic-based adhesive filler 

with 20% sawdust was used to fill the 

voids, cracks, and partially missing core 

of the core-damaged piles. The wrapping 

areas were pre-coated and cured 

for 6 h at ambient temperature (27°C) 

with phenolic-based primer. The precoated 

piles were wrapped to a length 

of 888 mm with two 482 mm resin wet 

GFRP fabrics having 76 mm overlap. 

A UV protection layer of resin was 

applied to the wrapped areas after 

rehabilitation. 

Field Testing, Results, Analysis, 

and Discussions 

Field testing was carried out using an 

80 ton locomotive prior to and after 

rehabilitation with different axle positions 

on the bridge for assessing the 

strains and stresses in pile caps and 

Specimen/ 

(repair scheme) 

Peak 

load (kN) 

s max , Shear 

strength 

(N/mm 2 ) 

Max. shear 

strain to failure 

(×10 −6 ) 

EI, Flexural rigidity (N/mm 2 ) G, Shear modulus (N/mm 2 ) 

Control 

(×10 6 ) 

Repaired 

(×10 6 ) 

Increase 

(%) 

Control Repaired Increase 

(%) 

#2 (SGFRP) 161,5 0,53 2937 2,85 8,7 305 — 216,2 — 

#3 (wrapping) 206,6 1,55 4446 11,1 13,3 19,5 295,3 793,6 267 

#4 (SGFRP) 209,3 1,28 3379 10,1 11,6 15 270,7 383,6 41 

Specimen #1 is control specimen and not repaired. Shear strains on specimen #2 without wrap are not considered. 

Table 1: Stringer properties (EI and G) before and after repair 


Wrapped 

height 

888 mm 

1219 

1981 

1067 

1524 

1676 

All dimensions 

in mm 

Average 

water 

level 

Pile 3 

Pile 2 

Pile 4 Pile 1 

305 

610 

508 

1575 

457 

3658 

Strain 

gage 

D = 305 D = 610 D = 610 D = 305 

(a) (b) (c) 

Fig. 2: Pile rehabilitation: (a) water submerged piles; (b) pile repaired with GFRP; (c) pile configuration and strain gage locations 

Loading 

Micro-strains 

Center of rear axle Before rehabilitation — −243 −144 152 230 

on pile bent 

After rehabilitation −185 −138 −127 108 129 

Reduction (%) 43 12 29 44 

24 km/h Before rehabilitation — −364 −150 278 253 

After rehabilitation −376 −229 −134 119 146 

Reduction (%) 37 11 57 42 

Negative (−) strain–compression; positive (+) strain–tension. 

Pile 

4 

Pile 

3 

Table 2: Results for different load cases before and after rehabilitation 

piles. The piles were not symmetrically 

placed when the bridge was built 

and this can be seen from the dimensions 

in Fig. 2. After rehabilitation, the 

strains induced in the timber piles and 

pile caps under different load conditions 

were reduced up to 43 and 46%, 

respectively (Table 2), with improved 

load distribution. Pile #2 was not rehabilitated 

and was considered as a control 

pile. It was also noted that strain 

values for pile #1 were in tension, 

instead of being in compression, which 

Pile 

2 

Pile 

1 

Pile 

Cap 

is attributed to local stress patterns and 

pile buckling at the top section and the 

displacement of the whole pile from 

its original axis. Dynamic load testing 

up to the SBVR-DOT recommended 

maximum speed on the bridge showed 

that the strains were similar to static 

values with a dynamic load amplification 

(impact) factor of 1 up to a speed 

of 24 km/h (Fig. 3). Dynamic load testing 

also showed a strain reduction in 

the FRP rehabilitated piles #1 and #3, 

and showed no change in the strain 

value of control pile #2, which was not 

rehabilitated (Fig. 3). 

Conclusion 

Damaged stringers strengthened with 

GFRP and tested under controlled 

laboratory conditions showed a 267 

and 41% improvement in the shear 

modulus when repaired with wrapping 

method and SGFRP method, respectively. 

The 50+ year-old in situ rehabilitated 

timber railroad bridge subjected 

to static and dynamic loads using an 

80 ton locomotive, revealed a reduction 

in pile cap strains by approximately 

44% and in the piles by approximately 

57%. The dynamic effect was negligible 

at the speed ranges up to 24 km/h 

in most of the piles. GFRP composite 

fabric in combination with phenolic 

formaldehyde adhesives was found to 

perform well with excellent bonding 

under harsh environments for GFRP 

wrapped timber piles. 

Acknowledgement 

Funding provided by Federal Railroad 

Authority (FRA) and West Virginia 

300 

200 

Pile 1: before rehab 

After rehab 

100 

Micro strains 

−100 

−200 

0 

10 11 12 13 14 15 


After rehab 

(a) 

−400 

(b) 


After rehab 

Fig. 3: Dynamic load testing: (a) 80 ton locomotive; (b) time versus strain data for piles at 24 km/h 

Structural Engineering International 4/2010 Technical Report 425 

−300

Division of Highways (WVDOH), 

USA for this work is acknowledged. 

References 

[1] Ritter M. Timber Bridges. Construction, 

Inspection and Maintenance. United States 

Department of Agriculture Forest Service, 1990. 

[2] Triantafillou T. Shear reinforcement of wood 

using FRP materials. J. Mater. Civil Eng. ASCE, 

9(2), 1997. 

[3] Abhari RS. Rehabilitation of Timber Railroad 

Bridges using GFRP Composites, MS Thesis, 

West Virginia University, 2007. 


Owner: 

South Branch Valley Railroad, 

Authority, West Virgin ia (WV) 

Department of Transportation (USA) 

Designer: 

Constructed Facilities Center, West 

Virginia University, Morgantown, WV. 


Constructed Facilities Center, 

West Virginia University, 

Morgantown, WV. 


Constructed Facilities Center, 

West Virginia University, 

Morgantown, WV. 

FRP (t): 0, 1 

Span lengths (m): 3560 mm 

(stringers) and 14000 mm (bridges) 

Rehabilitation cost (USD million): 

Less than 10,000 

Service date: 2001/2004 (Different 

times of rehabilitation) 


Design of the St Austell Fibre-Reinforced Polymer 

Footbridge, UK 

Jonathan Shave, Principal Eng., Parsons Brinckerhoff, Queen Victoria House, Redland Hill, Bristol, UK; Steve Denton, Engineering 

Director, Parsons Brinckerhoff, UK; Ian Frostick, Territory Civil Eng., Network Rail, UK. Contact: shavej@pbworld.com 

Abstract 

The St Austell Footbridge is a lightweight 

glass fibre-reinforced polymer 

(FRP) structure comprising pultruded 

and moulded elements, which crosses 

a railway line in St Austell, UK. When 

it was installed in October 2007, it was 

the first structure on the UK rail network 

to be entirely constructed from 

FRP materials. The bridge structure 

was designed to satisfy the aesthetic 

and environmental requirements of 

the client. Through rapid installation 

and the minimisation of any maintenance 

requirements, it has delivered 

economic, operational and sustainability 

benefits. 

This paper presents some of the design 

issues encountered in developing the 

innovative structure. These include the 

development of techniques and design 

philosophies to enhance the structure’s 

robustness through careful consideration 

of the potential behaviour of 

the structure and its components and 

connections up to the ultimate limit 

state. It also describes how the design 

was developed to ensure that vibrations 

caused by trains or pedestrians 

would be adequately controlled at 

the serviceability limit state. The findings 

of the research regarding behaviour 

of lightweight structures subject 

to aerodynamic impulses associated 

with passing trains are presented, and 

the corresponding effect on the design 

development is described. 

Also described in this paper are the 

regimes of testing and monitoring 

that were developed to ensure that 

the structure and its components were 

behaving as expected. 

Keywords: fibre-reinforced polymer; 

advanced composites; footbridge; 

design; robustness; vibration; buffeting; 

innovation. 

the town of St Austell, Cornwall, UK. 

The footbridge over the railway track 

gives pedestrians access to various 

amenities including a doctors’ surgery 

and a leisure centre. 

The old footbridge comprised three 

simply supported spans of approximately 

5, 14 and 6 m, respectively, and 

was supported by masonry piers and 

abutments. It was proposed to replace 

the superstructure with a new fibrereinforced 

polymer (FRP) structure 

and adapt the existing substructure for 

the new bridge (Fig. 1). 

Design Philosophy 

Design Concept 

The design concept for the new footbridge 

was developed on the basis of 

the Advanced Composite Construction 

System (ACCS), a modular system 

of pultruded structural panels. The 

system comprises cellular panels and 

boxes that are connected using adhesive 

joints and additionally secured 

with inserted mechanical connectors. 

This arrangement allows an anisotropic 

thick shell to be built up from standard 

modules in a variety of possible configurations. 

The designers recognised 

that it would be possible to use this 

system innovatively and efficiently in 

a U-frame configuration, as illustrated 

in Fig. 2. 

This approach minimised the construction 

depth from footway level to soffit 

level, and allowed the parapet walls to 

be used as structural members, giving 

the bridge sufficient stiffness to satisfy 

deflection requirements. Transverse 

frames were provided at regular intervals 

to resist distortional buckling of 

the cross section. 

Robustness 

As with any structural design, it was 

important to consider the robustness of 

the structure and ensure that progressive 

non-ductile failure modes would 

be avoided. The method of achieving 

this objective in the design of the FRP 

structure differed from conventional 

structures, as it was not appropriate to 

rely on any ductile behaviour of the 

FRP material or the adhesive joints. 

The high strength-to-stiffness ratio and 

low mass of the FRP material had the 

effect that the design was governed 

by serviceability limit states (such as 

deflection and vibration) rather than 

ultimate limit states. The factor of 

Introduction 

The St Austell Footbridge was required 

to replace a corroded wrought iron 

and cast iron structure over the 

Paddington–Penzance railway line in 

Fig. 1: St Austell Footbridge in service 


Longitudinal elements comprising 

ACCS panels built up into U-shape 

Fig. 2: U-frame concept 

Exterior skin 

comprising 

moulded panels 

Transverse frames to provide 

lateral stiffness and stability 

safety on the ultimate strength of the 

footbridge was high, and the theoretical 

load required to cause structural 

failure would never be encountered 

(and even if it were, the structure 

would already have become unserviceable 

due to excessive deflection). 

Notwithstanding this high factor of 

safety, there remained the possibility 

that an abnormal stress state could be 

triggered, for example, due to the unexpected 

failure of a joint. This scenario 

was considered carefully in the design 

of St Austell Footbridge, to avoid the 

possibility of a non-ductile progressive 

collapse of the structure. 

The vulnerability of the joints was initially 

minimised in the design by avoiding 

any in-span transverse joints and 

by careful specification of the adhesive 

(with associated testing) to ensure 

a sufficiently high factor of safety. In 

addition, the joints were given further 

robustness through the additional 

mechanical connector that effectively 

clamps each bonded joint together, significantly 

reducing the risk of any Mode 

I fracture behaviour. Notwithstanding 

these features, it was recognised that 

the potential vulnerability of the joints 

should be a key factor that had to be 

taken into account in the design stage 

when considering robustness. 

The philosophy was that the structure 

should be able to withstand the loss 

of any joint without causing ultimate 

collapse of the structure. Hence, if the 

most heavily loaded joints at the base 

of each parapet were to unexpectedly 

fail, losing the composite action 

between the parapets and the deck, 

then the deck section should have sufficient 

ultimate strength acting on its 

own to resist nominal loading (including 

pedestrian loading) without collapse. 

In this situation, there would 

be an increase in deflection and other 

noticeable indicators of distress. This 

behaviour is characteristic of a robust 

structure, improving safety and allowing 

effective monitoring techniques to 

be used. 

Outer Skin 

A further degree of robustness was 

achieved by encasing the main pultruded 

structural elements in a tough 

outer skin, providing additional protection 

from damage due to vandalism 

or adverse environmental conditions 

(such as ultraviolet radiation). Moulded 

FRP panels were used on the outside 

of the structure. These have the added 

advantage of improving the aesthetics 

and providing smooth curves to the 

structure. The inner faces of the parapets 

are protected with tough, replaceable 

anti-vandal panels, which prevent 

surface damage to the structure and 

have an anti-graffiti “deep-crinkle” 

finish. The non-slip walking surface 

was provided by a tough gritted FRP 

plate. 

Design for Vibration Serviceability 

The use of very light FRP materials 

allows the structure to have a low mass 

of only 5 t for the central 14 m span. 

There are various benefits to having 

a very light structure, principally concerned 

with installation and transportation, 

but the low mass also had some 

interesting effects on the vibration serviceability 

design, particularly under 

the actions of passing trains. This aspect 

of the design became the governing criterion, 

and required some cutting-edge 

research to be carried out in order to 

provide sufficient assurance that the 

structure would be serviceable. 

Two aspects of vibration serviceability 

were considered: pedestrian-induced 

vibration and train-induced vibration. 

Owing to the low mass of the structure, 

the design natural frequencies 

were generally higher than would be 

normal for a conventional footbridge. 

The fundamental vertical mode had a 

frequency of 12 Hz, and the horizontal 

mode had a frequency of 7 Hz, both 

of which were high enough for the 

structure to generally be deemed adequate 

for pedestrian-induced vibration 

serviceability. Even allowing for the 

unusually low mass of the structure, 

the accelerations arising from pedestrian 

actions were not expected to be 

high, and this proved to be the case. 

Vibrations arising from the aerodynamic-buffeting 

action of trains passing 

under the footbridge were harder 

to quantify. It was important to be able 

to demonstrate that the design accelerations 

would not cause discomfort to 

the pedestrians using the structure. 

Existing design standards provided only 

very limited guidance on the appropriate 

loadings to be used to model 

train buffeting effects. EN1991-2 1 

recommends a loading model, but this 

is not appropriate for use in the United 

Kingdom, as stated in the accompanying 

UK National Annex. 2 

Accordingly, it was necessary to derive 

a new loading model. An extensive 

literature search suggested that there 

was very little useful data on train buffeting 

vibrations, and so it was necessary 

to carry out new research based 

on actual measurements to derive an 

appropriate loading model. 

The designers, worked with Sheffield 

University experts to take acceleration 

measurements of a temporary footbridge 

over the railway at Goring, UK, which 

was exposed to considerable vibrations 

due to buffeting. These measurements 

were then back analysed to derive a proposed 

loading model for train buffeting 

to be used for the St Austell Footbridge 

design. Even using this loading model, 

and allowing for a reasonably modest 

line speed of 105 km/h at St Austell, the 

design of the footbridge was still governed 

by the limitation of vibrations due 

to train buffeting effects at the serviceability 

limit state. 

The development of the train buffeting 

load model allowed the footbridge 

design to be completed in an optimised 

and lightweight form, without the need 

for additional ballast or dampers to 

reduce theoretical vibration levels. 


Construction 

Fabrication of each span was achieved 

off site in factory conditions by 

building up the cross section from 

the ACCS components using adhesive 

bonds and mechanical inserts along 

the entire length of each span. 

Transverse U-frames were incorporated 

at regular intervals, with the 

horizontal frame members passing 

through the holes in the middle layer 

of the ACCS box. The internal faces 

were clad with the anti-vandal plates 

and the gritted base plate. 

The curved moulded panels on the 

exterior faces were produced by incorporating 

internal stiffening ribs to 

provide sufficient resistance to wind 

loading, and fixed to the structure at 

the top and bottom of each parapet 

wall. Moulded capping pieces were 

fixed to the top of each parapet wall, 

placed over the anti-vandal plate and 

the exterior moulded panels. 

Each bridge span was transported to 

the site as a single prefabricated unit, 

and was rapidly installed in a single 

night, as illustrated in Fig. 3. The spans 

were placed onto elastomeric bearings 

and held down to the new concrete 

padstones using stainless steel fixings 

and brackets. After installation of the 

spans, final panels were installed at the 

pier positions, providing an unbroken 

elevation and allowing relative movement 

and rotation at the bearings. 

Testing and Monitoring 

A significant amount of testing was 

carried out for this project, at the following 

levels: 

– Pultrusion testing 

– Adhesive testing 

– Moulded component testing 

– Complete structure testing (static 

and dynamic). 

The project specification was developed 

to include a comprehensive 

Fig. 3: Installation of the main span 

series of tests to ensure that all the 

materials and components had the 

correct properties. These included 

short-term and long-term stiffness and 

strength tests on the ACCS pultrusions, 

fire tests, adhesive tests at ambient and 

elevated temperatures and testing of 

the moulded panels for stiffness and 

strength. 

In addition, due to the innovative 

nature of the structure it was appropriate 

to specify testing of the complete 

structure under both static and 

dynamic loading conditions. 

The main span of the structure was 

tested under static loading conditions 

in the factory by filling it with 10,1 t 

of water up to a maximum level of 

510 mm, representing a design serviceability 

loading of 5 kPa (Fig. 4). 

The testing confirmed the following: 

– The structure as fabricated was adequate 

to carry crowd loading. 

– Deflections under serviceability 

loading were not excessive and were 

consistent with predictions. 

– The load–deflection behaviour was 

linear. 

To confirm the vibration behaviour 

of the footbridge, vibration testing 

was carried out a week after installation. 

Vertical and horizontal shakers 

were used in combination with accelerometers 

to obtain a full dynamic 

blueprint of the structure, including 

modeshapes, frequencies, modal 

masses and damping coefficients. The 

natural frequencies and mode shapes 

for the fundamental lateral and vertical 

modes will be monitored over the 

life of the structure in order to identify 

any reductions in stiffness or integrity 

of the structure. 

The dynamic response of the main span 

to pedestrian excitation was confirmed 

to be minor, as predicted. The test subjects 

reported that the vibrations were 

comfortable at all speeds. 

Fig. 4: Static load testing with water 

The main span was also tested for 

vibrations related to train buffeting 

loading and were well within acceptable 

serviceability limits. The speed of 

the trains was measured using a speed 

gun, and the accelerations were back 

analysed providing additional data for 

further improvement and calibration 

of the buffeting load model. 

Conclusion 

The St Austell Footbridge is a highly 

innovative structure, and the first all- 

FRP structure to be installed on the 

UK rail network. 

As a successful “pioneer” project, it 

paves the way for the increased use of 

these lightweight materials across the 

transportation sector to provide robust 

and aesthetically pleasing structures 

that deliver considerable economic, 

operational and sustainability benefits. 

Its design included the development 

of methods to ensure that a sufficient 

degree of robustness would be 

achieved. 

Research was carried out to ensure 

that vibration serviceability would be 

satisfied, particularly regarding the 

development and refinement of loading 

models for train buffeting effects. 

A significant degree of testing was carried 

out on the structure and its components, 

with highly successful results. 

References 

[1] BS EN 1991–2. Eurocode 1: Actions on 

Structures. Part 2: Traffic Loads on Bridges, BSi, 

October 2003. 

[2] National Annex to BS EN 1991–2. Eurocode 1: 

Actions on Structures. Part 2: Traffic Loads on 

Bridges, BSi, May 2008. 


Owner: 

Network Rail, UK 

Designer: 

Parsons Brinckerhoff 


BAM Nuttall 


Pipex Structural Composites 

FRP (t): 8 

Span lengths (m): 5, 14, 6 

Construction cost 

(EUR million): 0,4 

Service date: October 2007 


Aluminium Structures in Building and Civil Engineering 

Applications 

Frans Soetens, Prof., Eindhoven University of Technology, Eindhoven, The Netherlands; TNO Built Environment and Geosciences, 

Delft, The Netherlands. Contact: fsoetens@tue.nl 

Abstract 

Structural applications of aluminium are considered in this paper. Although 

the discussion is mainly devoted to Europe, the paper also refers, where possible, 

to developments in other parts of the world. The problems faced by a 

designer in creating an optimum design are described, followed by a brief 

review of the research carried out in the past four decades on the structural 

behaviour of aluminium and a preview of topics still to be investigated. 

A historical overview of standards is then given, starting from the ECCS 

Recommendations up to the recently published Eurocode 9. Finally, a number 

of structural applications are dealt with, as well as some future directions and 

concluding remarks. 

Keywords: aluminium; structures; design; research; standards; applications. 

Introduction 

Successful design and calculation solutions 

for aluminium elements and 

structures are only possible when 

the favourable properties of aluminium—in 

particular, its light weight and 

corrosion resistance—are taken into 

account, and an adequate design solution 

is provided for its unfavourable 

properties. These properties are unfavourable 

in comparison to the behaviour 

of steel structures and include 

the low Young’s modulus of aluminium, 

resulting in higher deformations, 

and a higher sensitivity to stability 

issues, as well as its vulnerability to 

fatigue issues and its lower fire resistance. 

Furthermore, a designer should 

also be aware of the freedom in the 

arbitrary shaping of aluminium cross 

sections as enabled by the extrusion 

process (Fig. 1). 

This implies that with design and calculation 

in aluminium—more than 

with other materials such as steel, 

concrete, wood—it is necessary to 

“think” in aluminium. In other words, 

the possibilities with aluminium, 

in particular the freedom of shaping 

by extrusion, must be kept in 

mind from the very beginning of 

a design. 1–8 

Overview of Research 

Activities 

State of the Art 

In the past four decades, significant 

research has been carried out on aluminium 

technology as well as on its 

structural behaviour in Europe and 

elsewhere, such as in the United States. 

This has resulted in the development 

of new alloys, improvement of material 

properties, combination of materials, 

new joining techniques, improved 

strength, stability and fire resistance. 

Many joint industry projects have 

been undertaken and many research 

committees have been involved on 

national and international basis. In 

Europe, Technical Committee 2 of the 

ECCS, the European Convention for 

Constructional Steelwork, has been 

particularly active. 9 

In the early 1970s, research priorities 

were set, and joint industry research 

work commenced. Some of the first 

topics tackled were strength, plasticity 

and ductility in order to study 

the application of plastic analysis 

for aluminium alloys. 10–13 Top priority 

was given to stability, more precisely 

global and local buckling, and 

this subject still requires substantial 

attention. 14–26 

Another important topic in Europe 

and elsewhere was connections, where 

the structural behaviour of both preloaded 

and non-preloaded bolts was 

investigated. 27–31 In The Netherlands 

in particular and also elsewhere, the 

structural behaviour of welded connections 

was studied. 32–36 Subsequently, 

adhesive-bonded connections were 

investigated to look after their applicability 

in building and civil engineering. 

37,38 High priority was also given 

to the topic of fatigue because of the 

higher vulnerability of aluminium 

structures to fatigue as compared 

to steel structures. 39–51 This work is 

still in progress in many countries; 

in Europe it has been coordinated 

within the ECCS-TC2. As in the case 

of steel, for aluminium, most attention 

was given to the fatigue behaviour of 

welded details (Fig. 2). Research was 

also done on aluminium shell structures, 

which resulted in design rules 

as given in Eurocode 9, Part 1–5, Shell 

Structures. 52–54 

Forthcoming Activities 

The above-mentioned knowledge of 

structural behaviour must be completed 

and extended. New designs and 

applications demand improved design 

rules, which in turn demands further 

study of specific topics. 




Paper received: Sept. 15, 2009 


Fig. 1: Aluminium extruded sections 

Fig. 2: Fatigue loading of welded 

aluminium bridge 


Some topics—related to the structural 

behaviour of aluminium—that require 

further study are as follows: 

– Research to optimise the loadbearing 

resistance of aluminium 

cross-sections, 55 which will further 

improve the existing design rules in 

Eurocode 9. Important aspects in 

this research are local and distortional 

buckling of thin-walled cross 

sections. 

– Research on “new” joining methods, 

such as “Friction Stir” welding 56–59 

and adhesive bonding. 60 For the 

latter, in particular, reliability and 

long-term behaviour (durability) 

are important aspects. This research 

will facilitate design rules for these 

joining technologies in Eurocode 

9, as well as in other national and 

international standards. 

– Research on improved methods to 

accurately determine the fatigue 

behaviour of aluminium components; 

except for existing design 

methods like safe life design and 

damage-tolerant design, the latter 

based on fracture mechanics, a promising 

continuum damage mechanics 

approach is under development. 

– Research on the structural behaviour 

of aluminium at elevated 

temperatures such as fire conditions. 

61–67 Although the strength 

and stiffness of aluminium decrease 

very fast at elevated temperatures, 

some aspects appear to be less sensitive 

at elevated temperatures compared 

to room temperature, such as 

buckling. 

Applied research such as that indicated 

above will result in better design rules, 

as has been shown with the research 

carried out in the past decades. 

In addition to extending the knowledge 

on the structural behaviour of aluminium, 

the designer must be instructed 

on how to apply this knowledge, that 

is, a type of knowledge transfer which 

must occur. This can be done through 

design projects where research institutions 

and industrial partners cooperate 

in order to demonstrate how to arrive 

at an optimum design in aluminium. 

This is particularly important for aluminium 

because aluminium is a rather 

young structural material as compared 

to steel, concrete, or bricks. 

Development of Design 

Standards 

The significant amount of research 

performed on the structural behaviour 

of aluminium thus far has culminated 

in design rules for structural applications. 

Many countries in Europe have 

an aluminium standard for building 

and civil engineering applications. 

However, most of them are quite outdated. 

The most recently updated ones 

are the British Standard BS 8118 68 and 

the Dutch Standard NEN 6710, 69 but 

these standards date from 1991; the 

old German Standard DIN 4113 70 was 

updated in 2002. 

As mentioned above, significant 

research in Europe was carried out 

under the umbrella of ECCS, the 

European Convention for Constructional 

Steelwork, where the 

Technical Committee 2 dealt with 

aluminium research that led in 1978 

to the first edition of European 

Recommendations. 9 Within the ECCS- 

TC2, research has continued since then, 

marking the beginning of a European 

Standard for the design of aluminium 

structures, Eurocode 9. 71 Although 

design rules in other national standards 

have been updated, Eurocode 

9 is by far the most comprehensive 

and up to date structural aluminium 

standard. 72–75 

The standard is subdivided into five 

parts: Part 1–1: General Structural 

Rules, Part 1–2: Fire Design, Part 1–3: 

Fatigue Design, Part 1–4: Cold-Formed 

Sheets, and Part 1–5: Shell Structures. 

Also, outside of Europe several design 

standards have been established in 

the past decades, for example, in the 

United States, Japan and Australia. 

Structural Design Applications 

The most important structural applications 

of aluminium can be found 

in transport and in building as well 

as civil engineering. In this paper, 

only building and civil engineering 

applications will be dealt with. The 

applications can be distinguished 

between offshore and onshore structures. 

Offshore applications are, for 

instance, helidecks, living modules, 

gangways, stairs, etc. Onshore applications 

are, for example, long-span 

roofs (space frames and domes), 

bridges, bridge decks, traffic gantries 

and sewage plants. A number 

of recently built structural applications 

of aluminium are reviewed by 

Soetens, 76 while many more and also 

older structural applications are dealt 

with by Mazzolani. 77 Some examples 

of recently built aluminium structures 

in The Netherlands are dealt with in 

the following sections. 

Aluminium Office Building 

An all-aluminium office building was 

built in Houten, The Netherlands, 

in 2000. It is the new office of the 

Aluminium Centre, the branch 

organisation of the Dutch aluminium 

industry. 78 

The concept of an all-aluminium 

office building, supported by hundreds 

of slender aluminium columns and 

designed by a Dutch architect, was the 

winning design in a contest and was 

chosen out of 64 designs. The architect 

was inspired by the rural landscape 

and called his design “aluminium forest”. 

The design is exceptional in that 

it is a one-storey building of about 

1000 m 2 supported by 380 aluminium 

columns (Fig. 3). Moreover, the stability 

of the system had to be provided 

by the columns as such; no additional 

bracings were allowed by the architect. 

The columns had a length of 6 m and 

diameters varying from 90 to 210 mm. 

Numerical simulations were carried 

out to investigate strength, stability 

and deformations of various models of 

the structure to decide upon the final 

design of the load-bearing system. 

To investigate the structural behaviour 

of the building system, a threedimensional 

finite element model of 

the entire system was developed. In 

all, six different designs were investigated. 

With the final design, sufficient 

stiffness (reduced deformations combined 

with an eigenfrequency above 

1 Hz) as well as sufficient stability and 

strength were achieved, among others, 

by introducing a small inclination to 80 

columns with a diameter of 210 mm. 78 

Aluminium Bridges 

In The Netherlands, the first aluminium 

bridge was installed in Amsterdam 

in 1955, followed by a few more aluminium 

bridges and/or bridge decks 

around the same time period, in different 

locations. In addition to the 

favourable properties of aluminium, 

as mentioned in the “Introduction” 

section, the supply of recycled aluminium 

after the 1945 war was one of the 

main reasons for these applications. In 

Europe and in North America only a 

small number of aluminium bridges 

have been built between the 1950s 

and the 1990s, probably because of the 

limited knowledge of the structural 

behaviour of aluminium. 

However, recently new initiatives have 

been taken and a “renewed” interest 

in aluminium bridges has arisen. 79,80 


Fig. 5: Residential area traffic bridge 

Fig. 3: Aluminium office building 

In The Netherlands, four application 

areas of interest were selected: 

– movable bridges; 81 

– residential area bridges; traffic but 

also pedestrian bridges; 82 

– extension of existing bridges; 83 

– renovation of bridge decks. 84 

Attention was given to all four areas, 

but most attention was paid to the movable 

bridge. A specific type of movable 

bridge is shown here: a non-balanced, 

hydraulically driven traffic bridge, with 

18 m span and 12 m width. 

A key requirement for the above type 

of movable bridge is light weight. In 

the meantime, many bridge projects 

have been designed and built in all 

four application areas, that is, one 

movable traffic bridge in Amsterdam, 

heavily loaded by trucks which necessitated 

a thorough fatigue analysis 

(Fig. 4); several residential area traffic 

bridges (Fig. 5); many pedestrian 

bridges with architecturally pleasing 

designs (Fig. 6); an aluminium extension 

of a steel bridge whereby the 

2,5 m-wide concrete pedestrian lane 

was replaced by 4,8 m wide prefabricated 

aluminium sections on the 

existing steel consoles (Fig. 7); and 

finally a number of aluminium decks 

replacing bridge decks (steel, concrete, 

wood). Figure 8 shows 600 m 2 of the 

wooden bridge deck of the movable 

bridge in the A29 highway replaced 

by prefabricated aluminium panels 

mounted on the existing steel beams. 

This system was also employed in the 

Fig. 4: Hydraulically driven movable 

bridge in Amsterdam 

renovation of two steel bridges in 

Kentucky, USA. 

In order to provide an alternative route 

in the case of road works on a bridge 

or on a road along a canal, a floating 

roadway would be particularly useful. 

Therefore a single-lane, floating roadway 

for cars travelling up to 80 km/h 

has been designed, 85 built and tested 

to full scale (Fig. 9). The floating road 

consists of aluminium pontoons, built 

out of a box frame of welded extruded 

sections, aluminium side walls and bottom 

plates. The pontoons had typical 

dimensions of 5,3 × 3,5 × 1,5 m (length, 

Fig. 6: Movable pedestrian bridge near 

central railway station Amsterdam 

Fig. 7: Aluminium extension of a steel 

bridge 

width, height). These dimensions and 

its light weight (about 2500 kg) allowed 

easy transportation by trucks as well as 

easy assembly and disassembly. 

In addition to the low weight, corrosion 

resistance, whereby surface protection 

was not necessary, and low maintenance 

costs were important criteria for 

the choice of aluminium in the above 

bridge applications. 

An interesting 50-year-old aluminium 

traffic bridge in the Ruhr area of 

Germany deserves to be mentioned. 

Owing to the heavy industrial atmosphere 

in the Ruhr area, it is inspected 

regularly whereby the latest tests and 

inspections carried out in 2003 showed 

that the bridge is still in a good condition. 

86 Furthermore, in Northern 

Europe, in particular in Norway and 

Sweden, a number of aluminium 

bridges and many bridge decks have 


Fig. 8: Wooden bridge deck replaced by aluminium sections 

Fig. 9: Aluminium floating road, full-scale testing 

been built in the last two decades. 87 

Apart from Europe, the concept of 

replacing existing bridge decks with 

aluminium decks is also considered in 

the United States. 88 Other aluminium 

structural applications are also receiving 

renewed attention. 89,90 

Concluding Remarks and 

Outlook 

Research carried out in the area of aluminium 

structures in the past decades is 

now starting to pay off. The knowledge 

on the structural behaviour as well as 

the up to date design rules allows better 

design of aluminium structures, in 

addition to avoiding stability problems 

and, in case of cyclically loaded structures, 

fatigue problems. Aluminium is 

a relatively young construction material 

as compared with concrete, steel, 

wood and masonry. The number of aluminium 

applications in load- bearing 

structures is still limited but, as has 

been reviewed above, there is a growing 

interest in the use of aluminium in such 

structures. 

Owing to the tendency to use lightweight 

and sustainable structures, the 

outlook for aluminium applications 

seems very positive. Enormous quantities 

of aluminium are available in 

the earth’s crust, and the opportunities 

for recycling are also very high, 

so there appears to be no limit to its 

application. This, combined with the 

availability of up to date design rules, 

promises a bright future for aluminium 

structural applications, provided the 

designer takes advantage of the favourable 

properties of aluminium and finds 

proper solutions to overcome the less 

favourable properties. 

Apart from the future probable 

research activities that have been 

described earlier in the “Overview of 

Research Activities” section, further 

areas of development are as follows: 

– material technology (new alloys, 

fibre metal laminates, self-healing 

materials, etc.); 

– structural applications of aluminium 

combined with other structural 

materials; 

– architectural design for building and 

civil engineering applications, which 

will take on more significance. 

This also means that the state of the 

art knowledge must be transferred to 

practice, to designers active in the field. 

One way to do this is to carry out joint 

industry projects where both research 

institutions and industrial partners 

take part. The European Aluminium 

Association (EAA) has taken an 

important initiative to contribute to 

this knowledge transfer: an e-learning 

tool—AluMATTER, in particular the 

“Structural applications module”—has 

been developed with interactive design 

examples based on Eurocode 9, which 

is of importance not only to practice 

but also for educational purposes. 

References 

[1] Mazzolani FM. Aluminium Alloy Structures, 

(2nd edn) E & FN SPON, Chapman & Hall: 

London, 1994. 

[2] Bulson PS (ed.), Aluminium Structural 

Analysis, Recent European Advances. Elsevier 

Applied Science Publishers, Barking, Essex, 

Great Britain, 1992. 

[3] Mazzolani FM (ed.), Aluminium Structural 

Design. CISM, Springer-Verlag: Wien, New York, 

2003. 

[4] Sharp ML. Behaviour and Design of 

Aluminium Structures. McGraw-Hill, Inc.: 

New York, 1992. 


[5] Kissell JR, Ferry RL. Aluminium Structures, 

A Guide to Their Specifications and Design. John 

Wiley & Sons, Inc.: New York, 1995. 

[6] Beaulieu D. Design of Aluminium Structures. 

Les presses de l’aluminium: Chicoutimi, Canada, 

2006. 

[7] Mazzolani FM. Design principles for aluminium 

structures, STAHLBAU SPEZIAL: 

Aluminium in Practice. Ernst & Sohn, Berlin, 

1998. 

[8] Mazzolani FM. Structural use of aluminium 

alloys in civil engineering. Proc. 2nd International 

Conference on Structural Engineering, Mechanics 

and Computation, Cape Town, South Africa, 

2004. 

[9] ECCS. European Recommendations for 

Aluminium Alloy Structures, 1st edn. Brussels, 

1978. 

[10] Mandara A, Mazzolani FM. Behavioural 

aspects and ductility demand of aluminium alloy 

structures. Proc. ICSAS Conference, Istanbul, 

1995. 

[11] Mazzolani FM, Piluso V. Prediction of rotation 

capacity of aluminium alloy beams. Proc. 

ICSAS Conference, Istanbul, 1995. 

[12] de Matteis G, Moen LA, Langseth M, 

Landolfo R, Hopperstad OS, Mazzolani, FM. 

Cross-sectional classification for aluminium 

beams—parametric study. ASCE, J. Struct. Eng. 

2001; 127(3): 271–279. 

[13] Lademo O-G, Hopperstad OS, Langseth 

M. Evaluation of yield criteria and flow rules 

for aluminium alloys. Int. J. Plast. 1999; 15(2): 

191–208. 

[14] Langseth M, Hopperstad OS. Local buckling 

of square thin-walled aluminium structures. 

Thin-Walled Struct. 1997; 27(1): 117–126. 

[15] Hopperstad OS, Langseth M, Hanssen L. 

Ultimate compressive strength of plate elements 

in aluminium: correlation of finite element analyses 

and tests. Thin-Walled Struct. 1997; 29(1–4): 

31–46. 

[16] Hoglund T. Shear buckling of steel and 

aluminium plate girders. Thin-Walled Struct. 

1997; 29: 13–30. 

[17] Mazzolani FM, Faella C, Piluso V, Rizzano 

G. Experimental analysis of aluminium alloy 

SHS-members subjected to local buckling under 

uniform compression, 5 th Int. Colloquium on 

Structural Stability, SSRC, Rio de Janeiro, 1996. 

[18] Moen LA, Hopperstad OS, Langseth M, 

Landolfo R, Dematteis G, Mazzolani FM. 

Rotational capacity of aluminium beams under 

moment gradient. Experiments and numerical 

simulations. J. Struct. Eng. 1999; 125(8): 921–929. 

[19] Mazzolani FM, Faella C, Piluso V, Rizzano 

G. Local buckling of aluminium channels under 

uniform compression: experimental analysis, 

Proc. ICSAS Conference, Espoo, Finland, 1999. 

[20] Soetens F, Mennink J, Snijder HH, van Hove 

BWEM. Experimental work and FEM calculations 

on the interaction of local buckling modes 

in aluminium sections. Proc. SIMS Conference, 

Lissabon, September 2000. 

[21] Chapuis J, Galambos TV. Restrained 

crooked aluminium columns. J. Struct. Div. 

ASCE 1982; 108(ST3): 511–524. 

[22] Sharp ML. Longitudinal stiffeners for compression 

members. J. Struct. Div. ASCE 1966; 

92(ST5): 187–211. 

[23] Sharp ML. Strength of beams or columns 

with buckled elements J. Struct. Div. ASCE 1970; 

96(ST5): 1011–1015. 

[24] Hopperstad OS, Langseth M, Tryland T. 

Ultimate strength of aluminium alloy outstands 

in compression: experiments and simplified analysis. 


[25] Aalberg A, Langseth M. Larsen PK Stiffened 

aluminium panels subjected to axial compression. 


[26] Mennink J, Soetens F, Snijder HH, van Hove 

BWEM. Buckling experiments and FE-analyses 

on complex aluminium extrusions. ASSCCA 

Conference, Sydney, June 2003. 

[27] Soetens F. Connections in aluminium alloy 

structures. Proc. ICSAS Conference, Cardiff, 

1987. 

[28] Ramirez J. Aluminium structural connections, 

conventional slip factors in friction grip 

joints. Proc. ICSAS Conference, Cardiff, 1987. 

[29] Menzemer CC. Failure of bolted connections 

in an aluminium alloy. J. Mater. Eng. 

Perform. ASM 1999; 8(2). 

[30] May JE, Menzemer CC. Strength of bolted 

aluminium alloy tension members. J. Struct. Eng. 

ASCE 2005; 131(7): 1125–1134. 

[31] Mazzolani FM, de Matteis G, Mandara A. 

Classification system for aluminium alloy connections, 

IABSE Colloquium, Istanbul, 1996. 

[32] Soetens F. Welded connections in aluminium 

alloy structures. HERON 1987; 32: 1–48. 

[33] Soetens F. Welded connections according 

to Eurocode 9 Design of Aluminium Structures. 

Proc. 7 th INALCO Conference, Cambridge 

(UK), April 1998. 

[34] Soetens F. Design of welded connections, 

STAHLBAU 67, Sonderheft Aluminium, Ernst 

& Sohn: Berlin, 1998. 

[35] Menzemer CC, Iasconne R. Reestablishing 

the shear strength of aluminium alloy fillet welds 

Weld. J. 2002; 81(4): 29–31. 

[36] Wang T, Hopperstad OS, Lademo O-G, 

Larsen PK. Finite element modelling of welded 

aluminium members subjected to four-point 

bending. Thin-Walled Struct. 2007; 45(3): 307–320. 

[37] Soetens F. Adhesive joints in aluminium alloy 

structures. Int. J. Adhes. Adhes. 1990; 10: 146–152. 

[38] Soetens, van Straalen IJJ, Dijkstra OD. 

Structural adhesive joints in aluminium. Third 

International Conference on Structural Adhesives 

in Engineering III, Bristol, June 1992. 

[39] Jaccard R, Kosteas D, Ondra R. Background 

Document to Fatigue Design Curves for Welded 

Aluminium Components, IIW doc. No. XIII- 

1588-95, 1995. 

[40] Kosteas D. On the fatigue behaviour of aluminium. 

STAHLBAU 67, Sonderheft Aluminium, 

Ernst and Sohn: Berlin, 1998. 

[41] Atzori B, Lazzarin P, Quaresimin M. A 

re-analysis on fatigue data of aluminium alloy 

bolted joints. Int. J. Fatigue 1997; 19(7): 579–588. 

[42] Sharp ML, Nordmark GE, Menzemer CC. 

Fatigue Design of Aluminium Components and 

Structures. McGraw-Hill, Inc.: New York, NY, 

1996. 

[43] Kosteas D, Jaccard R, Rothe J. Contribution 

to fatigue design rules in prEN 1999-1-3, R-ratio 

dependence, fatigue test factor and Simplified 

geometric stress concept. Proc. 9 th INALCO 

Conference, Cleveland, Ohio, USA, 2004. 

[44] Maddox SJ. Scale effect in fatigue of fillet 

welded aluminium alloys. Proc. 6 th INALCO 


[45] Soetens F. European research on fatigue 

of aluminium structures. Proc. 6 th INALCO 


[46] Maljaars J, Soetens F, van Straalen IJJ, 

van Hove BWEM. Fatigue tests on aluminium 

bridges. Proc. 9 th INALCO Conference, 

Cleveland, Ohio, USA, 2004. 

[47] Maljaars J, Soetens F, van Straalen, IJJ. 

Fatigue tests on aluminium bridges, HERON, 

Delft, 2005. 

[48] Maljaars J, Soetens F, Straalen IJJ. Fatigue 

of aluminium bridge decks, Struct. Eng. Int. 2006; 

16(4): 305–311. 

[49] Atzori B, Meneghetti G, Susmel L. A new 

procedure for the evaluation of fatigue strength 

of notched components applied to the evaluation 

of scale effect in aluminium alloy weldments. 

Proc. 8 th INALCO Conference, Munich, 2001. 

[50] Haagensen P, Oma S. Fatigue life improvement 

of welded aluminium alloys by peening. 

Proc. 8 th INALCO Conference, Munich, 2001. 

[51] Orjasaeter O, Skallerud B. Fatigue of wide 

plate weldments. Proc. 8 th INALCO Conference, 

Munich, 2001. 

[52] Mandara A, Mazzolani FM. On the stability 

of aluminium alloy cylindrical shells under 

axial compression. Costruzioni Metalliche 1989; 

2–3. 

[53] Mandara A, Mazzolani FM. Testing results 

and design procedures for axially loaded aluminium 

alloy cylinders. Proc. International 

Colloquium on Stability of Steel Structures, 

Budapest, 1990. 

[54] Mazzolani FM, Mandara A. Buckling of 

aluminium shells: proposal for European curves. 

Proc. 4 th International Conference on Thin- 

Walled Structures, Loughborough (UK), 2004. 

[55] Kutanova N, Soetens F, Snijder HH. Crosssectional 

instabilities of aluminium structural 

elements. Proc. CIMS Conference, Sydney, June 

2008. 

[56] Thomas WM. Friction stir welding and 

related friction process characteristics. Proc. 7 th 

INALCO Conference, Cambridge (UK), 1998. 

[57] Okura I, Naruo M, Vigh LG, Hagisawa N, 

Toda H. Fatigue of aluminium decks fabricated 

by friction stir welding. Proc. 8 th INALCO 

Conference, Munich, 2001. 

[58] Hori H, Tanikawa H, Seo N, Namba K. 

Tensile properties of friction stir welded joints— 

studies on characteristics of friction stir welded 

joints in structural aluminium alloys (report 3). 

Proc. 10 th INALCO Conference, Osaka, 2007. 

[59] Thomas WM, Verhaeghe G, Martin J. Staines 

DG, Stanhope C. Friction stir welding-process 

variants—an update. Proc. 10 th INALCO 

Conference, Osaka, 2007. 


[60] van Hove BWEM, Soetens F. Analytical 

and experimental research on strength properties 

of adhesive bonded joints in aluminium 

structures. Proc. 10 th INALCO Conference, 

Osaka, 2007. 

[61] Maljaars J, Fellinger JHH, Soetens F. Fire 

exposed aluminium structures, HERON 2005; 

50(4): 261–271. 

[62] Maljaars J, Soetens F, Katgerman L. 

Constitutive model for aluminium alloys 

exposed to fire conditions Metall. Mater. Trans. 

A 2008; 39: 778–789. 

[63] Langhelle NK, Eberg E, Amdahl J, 

Lundberg S. Buckling tests of aluminium columns 

at elevated temperatures. Proc. 15 th 

Offshore Mechanics and Arctic Engineering 

Conference, Firenze, Italy, 1996. 

[64] Maljaars J, Twilt L, Soetens F. Flexural buckling 

of fire exposed aluminium columns. Fire 

Safety J. 2009; 44: 711–717. 

[65] Maljaars J, Soetens F. Strength of MIG 

welded connections in fire exposed aluminium 

structures. J. Adv. Steel Constr. 2009; 5: 136–150. 

[66] Maljaars J, Soetens F, Snijder HH, Local 

buckling of aluminium structures exposed 

to fire. Part 1: Tests and Part 2: Finite element 

models. Thin-Walled Struct 2009; 47: 

1404–1428. 

[67] Maljaars J, Soetens F, Snijder HH. Local 

buckling of fire exposed aluminium members— 

a new design model. J. Struct. Eng., 2010; 136: 

66–75. 

[68] BS 8118. Structural Use of Aluminium, Parts 

1 and 2. BSI: London, 1991. 

[69] NEN 6710. TGB 1990 Aluminium 

Constructies. NNI: Delft, 1991. 

[70] DIN 4113. Aluminiumkonstruktionen unter 

vorwiegend ruhender Belastung, Teil 1 und Teil 2. 

DIN: Berlin, 2002. 

[71] EN 1999. Eurocode 9: Design of aluminium 

structures, Parts 1-1 to 1-5. CEN: Brussels, 

2007. 

[72] Mazzolani FM, Mandara A, Langseth M. 

Plastic design of aluminium members according 

to EC9. Proc. International Conference on 

Steel and Aluminium Structures, Espoo, Finland, 

1999. 

[73] de Matteis G, Mandara A, Mazzolani FM. 

Calculation methods for aluminium T-stubs: a 

revision of EC9, Annex J. Proc. 8th INALCO 

Conference, Munich, 2001. 

[74] Mazzolani FM. Conception des structures 

en alliage d’ aluminium selon l’Eurocode 9(EN 

1999). Les Eurocodes—Conception des bâtiments 

et des ouvrages de genie civil, Editions Le 

Moniteur, Paris, 2005. 

[75] Soetens F. van Hove BWEM. Design 

of welded connections in aluminium structures. 

IABSE, Struct. Eng. Int. 2006; 16(4): 

326–331. 

[76] Soetens F, Structural applications of aluminium: 

an overview. Proc. 10 th INALCO 

Conference, Osaka, 2007. 

[77] Mazzolani FM. Structural applications of 

aluminium in civil engineering. IABSE, Struct. 

Eng. Int. 2006; 4: 280–286. 

[78] Soetens F, van Hove BWEM, Mennink J. 

Dutch aluminium technology office building 

in the Netherlands. Proc. Aluminium 2000 

Conference, Essen, Germany, 2000. 

[79] Soetens F, Maljaars J, de Kluyver D. 

Aluminium bridges and bridge decks—an 

overview of recent applications. Proc. 3rd 

International Conference in Building and 

Construction, Moscow, 2008. 

[80] Siwowski T. Aluminium bridges—past, 

present and future. Struct. Eng. Int. 2006; 16(4): 

319–326. 

[81] Soetens F, van Hove BWEM, Mennink J. 

Aluminium bridges in The Netherlands. Proc. 8 th 

INALCO Conference, Munich, 2001. 

[82] Soetens F, Mennink J, van Hove BWEM. 

Aluminium bridges, actual designs and prospects. 

Proc. ASSCCA Conference, Sydney, 2003. 

[83] Maljaars J, Soetens F, Burggraaf H, Design 

of an aluminium bicycle path integrated in a 

steel bridge. Proc. IABSE Symposium, Weimar, 

Germany, 2007. 

[84] Maljaars J, Soetens F, de Kluyver D. Structural 

design of aluminium bridge decks for existing 

bridges. Proc. IABSE Symposium, Chicago, 2008. 

[85] Soetens F, van Hove BWEM, Maljaars J, 

Janssen EGON, Mennink J. Floating aluminium 

roads. Proc. 9 th INALCO Conference, Cleveland, 

Ohio, USA, 2004. 

[86] Mader W, Pieper A. Schwansbell bridge 

celebrating 50 th birthday, Struct. Eng. Int. 2006; 

16(4): 356–360. 

[87] Hoglund T, Nilsson L. Aluminium in bridge 

decks and in a new military bridge in Sweden. 

Struct. Eng. Int. 2006; 16(4): 348–352. 

[88] Kissell JR, Das SK, Kaufman G. 

Opportunities for extruded aluminium alloy 

bridge decks for existing bridges. Light Metal 

Age 2008; 66(3): 12–18. 

[89] Kissell JR. Recent Developments in 

Aluminium Structural Design. Aluminium Now, 

Aluminium Association: Washington, DC, 2002. 

[90] Kissell JR, Skillingberg MH. Aluminium in 

construction applications. Aluminium Int. Today 

2006; 18(2): 25–26. 


1. http:// www.eaa.net/en/education/ 

AluMATTER. 


Glass Tensegrity Trusses 

Maurizio Froli, Prof.; Leonardo Lani, Dr-Ing.; Department of Civil Engineering, University of Pisa, 

Pisa, Italy. Contact: l.lani@ing.unipi.it 

Abstract 

High transparency and modularity, retarded first cracking, non-brittle collapse 

and fail-safe design were the basic requirements that inspired and guided the 

development of a new kind of glass beams. The two basic conceptual design 

goals were to avoid any cracking at service and to get a ductile behaviour at 

failure. These objectives were reached by a preliminary subdivision of the beam 

into many small triangular laminated panes and by assembling them together by 

means of prestressed steel cables. Two prototypes have been constructed at the 

University of Pisa, tested in the elastic domain under dynamics loads and successively 

brought to collapse under quasi-static, increasing load cycles. In order to 

investigate the decay process of residual mechanical resources, the second prototype 

has been repaired twice by substituting just the damaged triangular panes 

and then tested again each time up to failure. Experimental results resulted in a 

good agreement with non-linear numerical simulations performed by appropriate 

finite element modelling. 

Keywords: structural glass; prestressed glass structures; post-breakage behaviour; 

structural ductility; fail-safe design; chemical tempering; fracture mechanics. 

Introduction 

Ductility is usually associated with 

metallic materials that are capable of 

developing large plastic deformations. 

On the other hand, fragility is traditionally 

associated with glass materials 

or ceramics. 

Nevertheless, some outstanding and 

innovative glass structures, like the 

Haus Pavilion in Rheinbach the 

Yurakucho canopy in Tokyo, the 

glass stairs of the Apple Stores in San 

Francisco, have been built even in seismic 

areas where fragile failures must 

be definitely avoided. 

Indeed, although glass is fragile and 

weakly tension resistant, it has a very 

high compressive strength and, if conveniently 

connected with other ductile 

materials, for example, by means 

of gluing or by prestressing, it is able 

to form composite structures of high 

mechanical performances and also 

global ductility. 

It is well known that stress concentrations 

that occur at the apex of 

microscopic surface cracks, always 

present even in virgin specimens, are 

responsible for the intrinsic fragility 




Paper received: March 04, 2010 

Paper accepted: June 29, 2010 

of glass 1 and for its relative low tensile 

strength. An apparent higher tensile 

strength is obtained by thermal or 

chemical tempering treatments, which 

induce surface compression stresses 

that inhibit crack initiation and propagation 

but do not exert any influence 

on fragility. 2 

Prestressed Composite 

Glass Beams 

Basic Concepts 

The intrinsic fragility of glass may be 

overcome by organizing the whole 

structure in two or more hierarchic 

levels, each of them composed of a 

parallel, redundant assemblage of at 

least two structural components. 

The hierarchic organization of the 

components ensures that the sequence 

of progressive damage follows a preestablished 

order starting from the 

level where the weakest components 

are. Therefore, if we put ductile materials 

at the lowest level, we can be 

sure that the failure process will start 

from here accompanied by large plastic 

deformations, that is, with a global 

ductile behaviour. 

On the other hand, redundancy 

ensures at each level that, when a single 

component fails, the other partner 

components are still able to bear the 

load although with a reduced degree 

of safety. 3 In laminated glass panes, 

the application of this principle also 

ensures a pseudo-ductile behaviour: it 

is known that if a glass sheet breaks, 

the other sheets are still able to bear 

the load, and even if all the sheets 

break into large fragments, (only fully 

thermally tempered glass breaks into 

many small fragments), the redundant 

sandwich structure assures a postbreakage 

stiffness and bearing capacity 

of the component that is similar, to 

some extent, to material ductility. 4 

For these reasons, a suitable application 

of the two basic principles of hierarchy 

and redundancy can provide a 

structure with decisive properties of 

global ductility and fail-safe design 

even if mostly composed of glass components. 

Fig. 1 shows the structural 

organization of the present type of 

glass beams. 

Additionally, if the integrity of the 

structure is enhanced by prestressing, 

compression stresses superimpose in 

glass elements to those produced by 

tempering, thus increasing the apparent 

tensile strength of the material. 

Structural Conceptual Design 

of Trabes Vitreae Tensegrity Beams 

Experiments reveal that when a traditional 

glass beam is submitted to 

increasing flexural loads, it cracks at 

a certain load, developing characteristic 

crack patterns. To avoid an uncontrolled 

process of crack initiation and 

propagation, the idea considered was 

to govern it by regularly pre-cutting 

the glass surface into many equilateral 

triangle panes and connecting them 

together using a system of prestressed 

steel cables. 

The principle of tensegrity permeates 

this concept, therefore it was decided 

to call these beams Trabes Vitreae 

Tensegrity or TVT, mixing Latin and 

English words. 

Each triangular pane is composed of 

two 5 mm thick chemically tempered 

glass sheets 5 laminated by means of 

a 1,52 mm thick PolyVinylButyral 

(PVB) interlayer. 

The beam is composed of two parallel 

twin curtains 174 mm apart, braced 

on the upper side by a horizontal truss 


Redudancy, parallel assemblies 

Curtain A Curtain B 

Steel 

cable 

Steel 

cable 

Fig. 1: Hierarchy and redundancy organization 

Phase 0: Pure Prestress 

The structural behaviour of TVT 

beams is analogous to that of segmenand 

connected together at the lower 

edge nodes by means of hollow stainless 

steel structures (Figs. 2 and 3). 

Each curtain is made of a Warrenlike 

appearance of glass panes (Fig. 2) 

jointed at the apex by means of stainless 

steel nodes (Fig. 3). Mechanical 

bolting between the glass panes and 

steel nodes was avoided as dangerous 

local tensile peaks always occur in 

glass holes. 

Instead, the nodes are mutually 

connected by means of AISI 304 stainless 

steel cables tensioned by screw 

tighteners. Consequently, only contact 

pressure is exchanged between glass 

and steel nodes due to the prestress 

action. In order to attenuate local 

contact peaks, the vertices of the glass 

panels are round, and 1 mm thick 

AW-1050A grade aluminium alloy 

Glass sheet 

Glass sheet 

Glass sheet 

Glass sheet 

Laminated 

pane 

Laminated 

pane 

1st Level 2nd Level 3rd Level 

Hierarchy, increasing mechanical strength 

sheets are interposed between steel 

and glass. 

The redundancy principle is applied at 

two different levels: the first is that of 

the doubly laminated panes and the 

second that of the parallel arrangement 

of the two twin curtains of glass 

panes and steel cables as sketched in 

the scheme of Fig. 1. The relatively 

large spacing of the two curtains gives 

the beam an appreciable torsional 

stiffness and good lateral torsional 

buckling stability. 

Qualitative Structural Behaviour 

tal prestressed concrete beams. During 

the shop assemblage of a beam, the 

two twin curtains are placed on a 

horizontal plane and prestressed. The 

dead load of the curtains is entirely 

sustained by the surface, thus only 

prestress forces act, inducing a quasiisotropic 

distribution of compression 

stresses in the glass panes (Fig. 4). 

Phase 1: Glass Decompression 

When in service, under the flexural 

action of dead loads and increasing 

external loads, tension stresses in the 

lower parts of the glass panels gradually 

diminish prestress compressions 

until a limit state of decompression is 

reached in the central part of the beam. 

When the external loads are further 

increased, the decompression propagates 

from middle span towards the 

supports. This stage has been denoted 

as Phase 1—glass decompression. 

Since the steel nodes exert unilateral 

restraint only at the point of contact, 

the decompressed vertices of the glass 

panels detach and simply move a small 

distance from their supports without 

developing tension stresses. The static 

scheme of the beam changes thus into 

that sketched in Fig. 5 where flexural 

and shear tension forces are sustained 

respectively by the lower steel bars 

and one order of the diagonal steel 

bars. Compression stresses flux within 

the glass panels following typical 

“boomerang-shaped” patterns visible 

Fig. 2: The prototype TVTb beam 

Fig. 4: Phase 0 calculated compression isolines 

Fig. 3: Steel node 

Fig. 5: Phase 2 calculated compression isolines 


1000,00 N 

1000,00 N 

1000,00 N 

1000,00 N 

Y 

X 

Fig. 6: Global model of TVTb 

in the same graph. Only secondary 

tension stresses of lower intensity 

affects glass. 

Phase 2: Buckling of Upper Cables 

Compressed steel cables are gradually 

de-tensioned: when the prestress load 

is fully compensated, they buckle away. 

This limit state has been denoted as 

Phase 2—buckling phase. 

Phase 3: Collapse 

After Phase 2 has been reached, a further 

increase in load leads to an augmentation 

of stress compressions in 

the glass panes and tension stresses in 

the steel rods. The dimensioning of the 

different component parts of the beam 

can be performed so that the final 

Phase 3—collapse takes place due to 

the yielding of the steel cables and not 

because of glass rupture in compression, 

thus resulting in a ductile collapse 

accompanied by large displacements. 

Depending on the slenderness of the 

steel cables and on the prestress intensity, 

Phase 2 may even follow Phase 3, 

as indeed happened in the prototype 

beam, and illustrated in the following 

text. 

Numerical Modelling 

Four different finite element model 

(FEM) analyses have been performed 

to predict the various aspects of the 

structural behaviour of the beam and 

to better calibrate the design of the 

prototypes (Fig. 6): 

– 2D non-linear geometrical analysis 

to evaluate the effect of prestress on 

the flexural stiffness of the beam; 

– 2D local buckling analysis to evaluate 

instability effects of prestress for 

each single glass pane; 

– 3D local geometric non-linear analysis 

to evaluate the transversal stiffness 

of the different joints; 

– 3D geometric non-linear analysis 

to evaluate the torsional stiffness of 

the beam. 

The glass panels have been modelled 

by Shell elements (4-node, 24 degrees 

of freedom) while the steel cables 

have been reproduced by Bar elements 

(2-node, 6 degrees of freedom) 

that react only to tension stresses. The 

constitutive laws for the two materials 

have been deduced from European 

Code 6 , that is, glass has been schematized 

as a linear brittle elastic material 

and stainless steel as linear elastic– 

plastic material with a linear hardening 

branch. 

In order to model contacts between 

glass panels and steel nodes a set of 

point contact elements was introduced 

that were capable of reacting just 

to compression stresses. Aluminium 

sheets were not included in the model 

because of their relatively small thickness 

and, for practical reasons, owing to 

the coincidence of the Young’s moduli 

of aluminium and glass. The transversal 

stiffness of the joint was preliminarily 

investigated by a 3D local model with 

solid elements (8-node, 24 degree of 

freedom) (Fig. 7). 

Calculations have substantially confirmed 

the intuitive predictions synthetically 

described in the section 

Qualitative Structural Behaviour with 

the only exception that the buckling 

phase of upper steel cables (Phase 2) 

does not influence significantly the 

flexural response of the beam (Fig. 8). 

On the other hand, the decompression 

phase (Phase 1) and the yielding phase 

(Phase 3) of the lower steel cables can 

be clearly recognized. 

Figure 9 shows the load factor versus 

displacement of the middle span point 

for different prestress (from 2 to 12 

kN) N p load in the steel cables, assuming 

P = 1 kN, the force applied to each 

of the eight nodes of the beam, corresponds 

to a load factor one. The first 

stiffness reduction is associated with 

the decompression of the lower part 

of the beam (Phase 1). By increasing 

the prestress level, the intensity of the 

external load that induces the decompression 

phase increases. The second 

step of stiffness decay is associated 

with the yielding of the lower steel 

cable but, of course, the ultimate limit 

load results independent of N p . 

Figure 10 shows the influence of N p 

on the axial force in the lower cable of 

the beam, and how the ultimate load is 

independent on the value of prestress. 

The hardening properties of stainless 

steel allowed the analysis to progress 

beyond the yield initiation of the lower 

bars until the buckling of the upper 

bars and of the middle span glass 

panels. 


It can be observed that the theoretical 

mechanical response of the beam is 

substantially bilinear until yielding of 

the steel bars occurs, while the buckling 

of the upper cables does not seem 

to have a relevant influence on the 

overall residual stiffness. 

Furthermore, the static principle, 

which states that prestress controls 

only serviceability limit states but not 

the ultimate limit states, is confirmed. 

(a) 

Z 

Y 

X 

Experimental Tests 

Virgin Specimens 

Y 

After the construction and testing of 

a first prototype (TVTa), which suffered 

some assemblage problems concerning 

prestress operation, a second 

prototype beam (TVTb) was prepared 

having a length of 3300 mm and a 

height of 572 mm. This prototype has 

been submitted to dynamic and quasistatic 

cyclical laboratory tests to completely 

characterize the experimental 

structural behaviour of the specimen 

and to compare it with numerical 

predictions. 

(b) 

Fig. 7: Joint model (a) and (b) 

Plate stress 22 Mid plane (MPa) 

−4 

−8 

−13 

−17 

−21 

−26 

−30 

−34 

−35 

Fig. 8: Calculated principal compression stresses 

Z 

X 

Dynamic Test 

Dynamic tests have been performed 

by inducing sudden impulses both 

in the horizontal and in the vertical 

direction at middle span. To produce 

the impulses, a mass of 34 kg was 

slowly applied to the beam by means 

of a steel rod thus inducing a state of 

initial distortion. As the rod was suddenly 

cut, the beam was submitted to 

damped free oscillations. Vertical and 

horizontal acceleration histories were 

recorded at some representative points 

of the beam, which allowed to evaluate 

eigen vibration periods and to control 

the attitude of the structure to damp 

free oscillations. 

Load factor 

9 

8 

7 

6 

Np = 2 kN 

Np = 4 kN 

5 

4 

Np = 6 kN 

Np = 8 kN 

Np = 10 kN 

3 

Np = 12 kN 

2 

1 

0 

−10 0 10 20 30 40 50 60 70 80 90 

Displacement (mm) 

Fig. 9: Load factor versus vertical displacement 

The assessment of the structural frequencies 

was deduced from the trend 

of the accelerations, appraising the distance 

between two consecutive maxima. 

Table 1 compares the theoretical 

and the experimental results. 

Quasi-Static Cyclic Tests 

The static test of TVTb prototype 

has been performed in two different 

stages: during the first one, the specimen 

was submitted to a progressively and 

cyclically increasing loading condition. 

In the second stage, the load was 

increased monotonically until the 

collapse of the beam took place at a 


Load factor 

9 

Np = 2 kN 

8 

Np = 4 kN 

7 

Np = 6 kN 

6 

Np = 8 kN 

Np = 10 kN 

5 

Np = 12 kN 

4 

3 

2 

1 

0 

0 2000 4000 6000 8000 10 000 12 000 14 000 16 000 18 000 20 000 

Axial force (N) 

Fig. 10: Load factor versus axial force 

T est no. Form Measured frequency (Hz) Calculated frequency (Hz) 

1 In plane 19,1 76,9 

2 In plane 16,9 76,9 

1 Out of plane 13,4 12,1 

2 Out of plane 15 12,1 

Table 1: Results of the dynamics tests 

total applied load of 61,84 kN (load 

factor = 7,73). 

Before the application of external 

loads, vertical and horizontal displacements 

induced by self weight and prestress 

were measured over a few days. 

Such investigations led to the conclusion 

that tension and rigidity reductions 

that could occur with time due to 

the relaxation of the cables or to other 

viscosity phenomena in the PVB interlayer 

could be substantially neglected. 

The program of cyclic loading has 

been performed to check precedent 

intuitions and theoretical analyses and 

to evaluate the different structural 

resources of the prototype, namely, the 

ability to sustain repeated occurrence 

of increasing loads without damage 

or significant performance decay, the 

response to the influence of unavoidable 

geometrical imperfections and the 

eventual capacity to dissipate energy 

without damage. 

The results of the cyclic test program 

are represented in Fig. 11 in terms of 

total applied load versus middle span 

vertical displacement. The experimental 

results are compared in the same 

graph with the analytical monotonic 

response of the 2D FEM model. 

The comparison allows the conclusion 

that numerical and experimental 

9 72 kN 

8 

7 

6 

Numerical model 

Final pushover test 

64 kN 

56 kN 

48 kN 

results are rather close to each other 

with some limited discrepancies: 

– The first knee related to Phase 1 

(glass decompression) is recognizable 

although less marked as in 

numerical analysis; 

– The actual stiffness of the beam 

before the first decompression 

knee is lower than that obtained by 

numerical analysis; 

– At each new load cycle, the level of 

the decompression load decreases 

although the residual deformation is 

very limited. 

Each cycle of Fig. 11 encloses a finite 

area showing that the beam is also 

surprisingly able to dissipate energy 

before any damage occurs in the 

component materials. This can be 

attributed to the friction that develops 

due to relative slip movements at 

the interface between the glass and 

steel nodes and perhaps also to viscoelastic 

slip movements in the PVB 

interlayer. 

Very small transversal displacements 

were measured (not shown here for 

the sake of brevity) at each load cycle 

evidencing how good the torsional 

stiffness of the beam is and how it 

remains constant throughout the progression 

of the load cycles. 

After the completion of the cyclical 

load program, the beam has been submitted 

to a monotonic increasing load 

up to collapse. In the graph of Fig. 11, 

the load versus displacement curve of 

this stage (blue line) is compared with 

the theoretical curve. The experimental 

curve has almost no decompression 

knee but now the second knee is visible, 

corresponding to yielding of the 

lower steel bar, which occurred at a 

higher load level than predicted. First 

failure symptoms therefore manifested 

in the ductile component material of 

the composite structure (Fig. 12). 

Load factor 

5 

4 

3 

Cyclic tests 

40 kN 

32 kN 

24 kN 

2 

16 kN 

1 

8 kN 

0 0 kN 

0 10 20 30 40 50 60 70 


Fig. 11: Load of middle point versus vertical displacement compared with FEM 

(dotted line) 

Fig. 12: TVTβ at failure 


7 

56 kN 

Conclusion 

Load factor 

6 

5 

4 

3 

2 

1 

0 

−5 0 5 10 15 20 

TVTβ 

Fig. 13: Displacement origins have been shifted to the right in TVTb bis/tris 

Owing to the hardening properties 

of stainless steel, the load can be 

increased even beyond yielding: the 

final collapse of the specimen was 

reached when the upper parts of middle 

span glass panels buckled away. 

Repaired Specimens 

After prototype TVTb completely 

collapsed as a consequence of the 

breaking of middle span panels, it was 

repaired by substituting just the broken 

panels. Prestress was restored at 

the same levels of the virgin specimen 

TVTb. 

The first repaired prototype was 

labeled as TVTbbis and submitted 

to the same increasing load cycles of 

TVTb. 

Since the clam plates of the central 

panels in TVTb resulted in a little dam- 

TVTβbis 

TVTβtris 

48 kN 

40 kN 

32 kN 

24 kN 

16 kN 

8 kN 

0 kN 

25 30 35 40 45 50 55 60 65 


age after buckling, they could not offer 

the same degree of restraint as the virgin 

ones. 

Therefore, the central panels of 

TVTb bis buckled corresponding to 

a load factor of 4 instead of the 

previous 7,5. 

Next, in prototype TVTb bis, the collapsed 

central panels were substituted. 

The second repaired prototype was 

called TVTbtris and exhibited almost 

the same ultimate load factor as the 

precedent version. 

Figure 13 shows the load factor versus 

middle span vertical displacement 

cycles of the virgin specimen and of 

the two repaired versions. A rather 

good retention of the stiffness properties 

of the repaired specimens can be 

observed together with a progressive 

increase in the dissipated energy. 

The numerical results and the test 

experiments on virgin TVTb prototype 

of a prestressed composite glass–steel 

beam allow the conclusion that the 

constructional principle is valid and 

merits further technological improvement 

and research work. 

The experimental and numerical 

results have underlined that TVT composite 

glass–steel beams are able to 

develop a ductile break-up and that the 

serviceability limit state is governed by 

the level of prestress in the steel cables. 

The cyclic load programme also evidenced 

that these glass beams are able 

to dissipate energy through friction 

and viscoelasticity without damage. 

The segmental, modular features and 

the tensile integrity of these beams 

allow substitutions to be limited just to 

the collapsed or cracked panels, thus 

reducing repair costs. 

References 

[1] Menčik J. Strength and Fracture of Glass and 

Ceramics. Elsevier: London, 1992. 

[2] Sedlacek G. Ein Bemessungskonzept zur 

Festigkeit thermisch vorgespannter Gläser. 

Shaker Verlag: Aachen, 2000. 

[3] Rice P, Dutton H. Structural Glass, 2nd edn. 

Spon Press: London, 2004. 

[4] Kott A, Vogel T. Safety of laminated glass 

structures after initial failure. IABSE Struct. 

Eng. Int. 2004; 14(2): 134–138). 

[5] Macrelli G. Process control methods for chemical 

strengthening of glass on industrial scale. 

Proc. XIX Int. Cong. Glass, Edinburgh, 2001. 

[6] EN572 – Glass in Building, Basic Soda Lime 

Silicate Glass Products, Definitions and General 

Physical and Mechanical Properties, CEN/TC 

129 Glass in Building, 2004. 


A Simplified Serviceability Assessment of Footbridge 

Dynamic Behaviour Under Lateral Crowd Loading 

Luca Bruno, Associate Prof. and Fiammetta Venuti, Postdoc Fellow; Dept. of Structural and Geotechnical Engineering, 

Politecnico di Torino, Torino, Italy. Contact: fiammetta.venuti@polito.it 

Abstract 

In this paper a simplified approach to footbridge serviceability assessment under 

lateral crowd loading is proposed. The approach is based on the determination of 

a limit crowd density, which causes the lateral acceleration of the deck to reach 

the human perceptibility threshold. The approach relies on a lateral load model 

that describes the action of both uncorrelated and synchronised among-eachother 

pedestrians by proposing a law that links the percentage of synchronised 

pedestrians to the value of the crowd density, which is assumed to be uniformly 

distributed along the span. The approach is applied to three case studies and the 

results are compared with those obtained through the application of other serviceability 

and stability criteria. 

Keywords: serviceability criteria; footbridges; pedestrian load; synchronisation; 

lock-in. 

Introduction 

Over the last decade, the problem of 

footbridge serviceability under humaninduced 

excitation has been one of the 

important areas of research in the field 

of civil engineering. The interest of 

researchers and engineers is due to the 

recent trend towards the construction of 

more and more slender and lightweight 

footbridges, which are extremely prone 

to vibration. This interest is testified by 

an intense research activity (reviewed, 

e.g. in Refs. [1, 2]) and the publication of 

guidelines for the design of footbridges, 

especially focused on their dynamic 

behaviour under pedestrian loading. 3–6 

In particular, considerable attention 

has been directed towards the phenomenon 

of the so-called synchronous lateral 

excitation (SLE), first observed on 

the T-bridge in Japan 7 and brought to 

the world’s attention after the closure 

of the London Millennium Bridge in 

2000. 8 In recent years, SLE has been 

recognised to be affected by two 

kinds of synchronisation phenomena 

(e.g. Refs. [9, 10]): the pedestrian– 

structure (ps) one, also called lock-in, 

takes place when a pedestrian perceives 

the lateral motion of the walking surface 

and tends to synchronise his/her pacing 




Paper received: December 01, 2009 

Paper accepted: May 28, 2010 

rate to that of the surface in order to 

maintain body balance; the pedestrian– 

pedestrian (pp) synchronisation arises 

when the pedestrian’s walking is constrained 

because of high crowd density 

values so that people tend to walk with 

the same frequency and a null relative 

phase angle, that is, they synchronise 

with one another. The complex relation 

between ps and pp synchronisation 

remains a subject of debate in the scientific 

community, especially in relation to 

their mutual role in the SLE triggering 

process. 11 Nevertheless, what is widely 

accepted is that the ps synchronisation 

cannot occur if the deck vibrations 

are below the pedestrian perceptibility 

threshold, while the pp synchronisation 

cannot take place if the crowd 

density is below the threshold for 

unconstrained free walking. 

Several criteria have been suggested to 

avoid occurrence of lock-in and guarantee 

pedestrian comfort. According 

to the authors, these can be roughly 

classified into two approaches corresponding 

to different stages of the SLE 

process in which the criterion is derived. 

According to the first approach, lock-in 

can be avoided if the amplitude of the 

lateral acceleration of the deck does 

not exceed a limit value corresponding 

to the pedestrian perceptibility threshold—for 

instance, the Sétra/AFGC 4 

and Synpex 5 technical guides recommend 

a limit value ̇ż lim =0,1m/s . The 

2 

second approach (e.g. Refs. [8, 12, 13]) 

is derived once the lock-in has occurred 

(i.e. the lateral vibration is higher than 

the perceptibility threshold) by identifying 

two degrees of severity in the ps 

synchronisation process: a stable condition, 

where the lock-in has been triggered 

but the lateral response remains 

below a value of about 10 to 15 mm 13 

and an unstable condition, where the 

pedestrians exert larger lateral forces 

leading the structural response to an 

unstable amplification. The approach 

defines a limit number of pedestrians 

N lim as the one that sets off the unstable 

condition. For this reason, the criteria 

based on the second approach 

should be stability rather than serviceability 

criteria. 

Both approaches imply the need 

to define suitable load models that 

describe the load condition in the considered 

stage of the SLE process. The 

models related to the first approach 

consider the pre-lock-in stage and 

describe the pedestrian behaviour as 

random, that is, the pedestrians are 

assumed to walk with random phases 

(e.g. in Ref. [4]). On the contrary, the 

models related to the second approach 

rely on the assumption that lock-in has 

already occurred, that is, the pedestrian 

motion and force are synchronised 

and proportional to the lateral 

motion of the deck. 

According to the authors, the following 

main considerations can be 

outlined: 

– the pp synchronisation is not explicitly 

accounted for in none of the 

models, even though it could occur 

both in the pre- and post-lock-in 

stages, since it depends on the crowd 

density value; 

– it is not possible to establish which 

approach is the most conservative in 

absolute terms. In fact, the first one 

is the most conservative in terms of 

deck acceleration amplitude (the 

perceptibility threshold is certainly 

lower than the stability one) but this 

feature does not necessarily imply 

the same in terms of corresponding 

pedestrian number, because the 

lateral force of a single pedestrian is 

higher within the lock-in stage than 

in the pre-lock-in one. 


In this paper an alternative approach 

to footbridge serviceability assessment 

under lateral crowd loading is 

proposed. This approach is based on 

the consideration that the two concepts 

of limit values could be linked, 

that is, the limit number of pedestrians 

may be interpreted as the number 

of pedestrians that induce the 

limit value of the acceleration. The 

positive aspects of the two approaches 

are therefore retained: on one hand, 

the definition of a critical number of 

pedestrians could be useful for the 

designer and the footbridge owner in 

view of the development of control 

strategies to limit the flow of pedestrians; 

on the other hand, the analysis 

of the pre-lock-in stage is expected to 

be more relevant in order to predict 

and prevent the occurrence of SLE. 

The approach is accompanied by the 

proposal of a load model within the 

pre-lock-in stage that also accounts for 

the pp synchronisation due to crowd 

density. In the following, the proposed 

force model and comfort criterion are 

described and compared with other 

serviceability/stability criteria found 

in the literature through their application 

to some case studies. 

Proposed Approach 

The proposed approach is based on 

the derivation of a critical crowd density 

r lim , which causes the lateral acceleration 

of the deck to reach the limit 

value ̇ż lim =0,1m/s 2 . This limit value 

corresponds to the pedestrian perceptibility 

threshold recommended by 

Sétra/AFGC 4 and Butz et al. 5 , that is, 

to the upper bound of the pre-lock-in 

stage. The following assumptions have 

been adopted: 

1. The structural response can be estimated 

with sufficient accuracy using 

a single-degree-of-freedom (SDOF) 

modal equation for the mode of 

interest. 

2. The deck mass is constant along the 

footbridge span. 

3. The crowd is uniformly distributed 

along the span, so that both the 

crowd mass and the force exerted 

are constant in space. The force is 

applied on the deck so that its sign 

matches the sign of the deformed 

shape. 

4. The pre-lock-in stage is considered, 

that is, the pedestrians are not 

synchronised to the bridge motion, 

but they could be synchronised 

among each other owing to crowd 

density. 

It follows that the critical value of the 

crowd density r lim refers to pedestrians 

uniformly distributed along the 

footbridge walking path, that is, to a 

load condition widely adopted in the 

footbridge design and corresponding 

to a uniform crowd flow. 

The structural dynamics is described 

by the SDOF equation of motion as 

follows: 

Zt ̇̇ Zt ̇ 2 Ft () 

() + 2ξω 

s 

() + ω 

s 

Zt () = (1) 

M 

in which Z(t) is the generalized coordinate 

expressing the motion of the system, 

t, the time, being the independent 

variable; z is the damping ratio; w s = 

2πf s is the natural circular frequency of 

the mode of interest; F(t) and M are 

the generalised force and mass, respectively, 

expressed as follows: 

Ft () = ft () ∫ ϕ ( x) 

d x 

L 

0 

L 

(2) 

M= m∫ ϕ( x)2 

d x 

(3) 

0 

where L is the length of the footbridge 

deck, x is the space coordinate along 

the longitudinal axis of the footbridge, 

j (x) is the mode shape, f(t) and m are 

the force and mass per unit length, 

which are constant in space according 

to hypotheses 2 and 3. 

The mass m is intended to be the overall 

mass of the structure and crowd 

systems: 

m= m + m ( ρ) , withm ( ρ) 

= ρ BG (4) 

s p p 

where the subscripts s and p refer to 

the structure and pedestrians, respectively, 

r (ped/m 2 ) is the crowd density, 

G = 70 kg is the average mass of one 

pedestrian and B is the width of the 

walking path. 

The lateral force f(t) per unit length 

is expressed on the basis of the load 

model proposed by Venuti et al. 10 and 

Venuti and Bruno. 14 The original force 

model had been conceived to account 

for both pp and ps synchronisation 

and to describe the crowd force during 

the triggering, self-exciting and selflimiting 

SLE stages. According to the 

above-mentioned simplifying assumptions, 

the lateral force model reduces 

to the following: 

f() t = ρeq 

B⋅F0 sin( ωpt) 

(5) 

with 

ρ eq 

= N eq 

/( BL) (6) 

and 

N = N S + N( 1 −S 

) (7) 

eq pp pp 

where r eq is the equivalent crowd density; 

F 0 = 28 N is the amplitude of the 

lateral force exerted by one pedestrian; 

w p = 2πf p is the circular lateral walking 

frequency; N = rBL is the number 

of pedestrians on the deck; S pp is the 

coefficient of synchronisation among 

pedestrians. The square root term in 

Eq. (7) represents the contribution 

of uncorrelated pedestrians, according 

to the model of Matsumoto et al. 15 

The lateral walking frequency f p is 

expressed as a function of the walking 

velocity v, in turn dependent on 

the crowd density, according to the 

following laws: 

2 3 

fp = ( 293 , v − 159 , v + 035 , v )/ 

2 

⎧⎪ 

⎡ ⎛ 1 1 ⎞ ⎤⎫⎪ 

v= vM 

⎨1 

− exp ⎢−γ 

− 

⎝ 

⎜ 

⎠ 

⎟ ⎥⎬ 

⎩⎪ ⎣ ρ ρM 

⎦⎭⎪ 

(8) 

(9) 

where the values of the coefficient g, 

the free speed v M and the jam density 

r M are made sensitive to different travel 

purposes (L = leisure/shopping, C = commuters/events, 

R = rush hour/business) 

and geographic areas (E = Europe, 

U = USA, A = Asia) (Fig. 1). The laws 

in Eqs. (8) and (9), first proposed by 

Venuti and Bruno, 16 are fitted to the 

experimental data reported in Refs. 

[5, 17, 18]. The values of the empirical 

parameters g, n M , r M of interest for the 

case studies discussed in the following 

section are reported in Table 2. 

As far as the synchronisation coefficient 

S pp is concerned, in the last 2 

years some new studies 5,19,20 concerning 

pp synchronisation due to crowd 

density have been carried out by means 

of experimental tests performed within 

different ranges of crowd density. 

Araújo et al. 19 found that, in the crowd 

density range from 0,3 to 0,9 ped/m 2 , 

there is no evidence of synchronisation 

among pedestrians, since the standard 

deviation of the walking frequencies 

is almost constant for different densities 

and the phase angles are totally 

random. Ricciardelli and Pansera 20 

observed that, in the crowd density 

range 0,5 to 1,5 ped/m 2 , initially different 

walking frequencies and phases 

tend to get closer for increasing crowd 

densities, giving rise to synchronisation 

nuclei within the crowd. Finally, 

Butz et al., 5 who performed tests with 

the highest values of the crowd density 

(1,2–3,0 ped/m 2 ), observed a reduction 


v (m/s) 

(a) 

2 

1,5 

1 

0,5 

0 

0 

Venuti and Bruno (2007) - L 

-C 

2,5 

-R 

Butz et al. (2008) 

2 

Oeding (1963) - L 

-C 

1,5 

-R 

1 

Bertram and Ruina (2001) 

Venuti et al. (2007) 

0,5 



0 

1 2 3 4 5 6 0 0,5 1 1,5 2 2,5 

r (ped/m 2 ) 

(b) 

v (m/s) 

Fig. 1: v(r) (a) and f p (v) (b) laws and comparison with the ones proposed in Ref. [5] 

of the standard deviation of the walking 

frequencies for low walking speed, 

that is, for increasing density, but they 

did not report any data concerning 

the phase angles. Owing to the scarceness 

of quantitative results concerning 

the phase angles, in the following, 

the standard deviation of the walking 

frequency s f will be considered as a 

suitable indicator of synchronisation, 

that is, the smaller s f is, the higher the 

pedestrian tendency to walk at the 

same mean walking frequency (Fig. 2). 

On the basis of these experimental 

observations, the qualitative S pp (r) 

law proposed in Ref. [10] is replaced 

by a new law developed according to 

the scheme represented in Fig. 3a as 

complementary to s f /s f,0 , where s f,0 = 

0,12 Hz is the mean standard deviation 

in the case of unimpeded walking 

measured in the same experimental 

s f (Hz) 

0,12 

0,08 

0,04 


Araujo et al. (2009) 

0 0 0,5 1 1,5 2 

r (ped/m 2 ) 

Fig. 2: Values of the standard deviation s f as 

measured in Refs. [5, 19] 

1 

s f /s f,0 

S pp 

2f p (Hz) 

campaigns. 5,19 The fitting function 

is inspired to the trend of the coherence 

against the coupling strength in 

the Kuramoto model 21 : the coherence 

(i.e. S pp herein) is null until the coupling 

strength (i.e. r herein) reaches 

a threshold value (r c ), which corresponds 

to a phase transition; for r > rc 

the coherance grows towards perfect 

synchronisation. S pp (r) is, therefore, 

expressed as follows: 

S 

0 

c 

pp( ) 

, 

1 exp[ a( 

c)/ 

M 

] 

c 

(10) 

where a = 8,686 derives from the fitting 

to the set of points of coordinates 

(r, 1 − s f /s f,0 ) and r c is set equal to 

0,6 ped/m 2 (Fig. 3b). Figure 4 plots 

the trend of r eq /r M versus r/r M (the 

trend of F is proportional): the function 

is bounded between two limit 

curves, which represent the cases of 

all pedestrians synchronised to each 

other ( ρ eq,sync 

= NBL, / e.g. marching 

soldiers) and all pedestrians uncorrelated 

( ρ eq,rand 

= NBL / ), respectively. 

The proposed law matches the 

model of uncorrelated pedestrians for 

r < r c , while it tends to the full synchronisation 

as r approaches r M . It 

is worth pointing out that the density 

range in which the pp synchronisation 

develops and the law differs from the 

limit curves corresponds to the one of 

0,2 

Butz et al.’s data 

Araujo et al.’s data 

0 0 

Fitting 

0 r c /r M 

1 

0 r 0,2 0,4 0,6 0,8 

r/r c 

M 

(a) 

(b) r M 

r/r M 

Fig. 3: S pp (r) relation: scheme (a) and proposed law (b) 

S pp 

1 

0,8 

0,6 

0,4 

1−s f /s f,0 from: 

1 

r eq /r M 

10 0 

10 −1 

10 −2 

10 −3 0 r c 

r M 

practical interest in real world crowd 

events on footbridges. 

The amplitude of the steady-state lateral 

acceleration can be found referring 

to the following well-known 

expression: 

Ż̇ 

F M D ρeqBF 

= = 

m 

0,2 0,4 0,6 0,8 1 

∫ 

∫ 

L 

0 0 

L 

0 

r/rM 

Fig. 4: Diagram of r eq versus r 

r eq 

r eq,sync 

r eq,rand 

ϕ( x) 

dx 

D 

2 

ϕ( x) 

dx 

(11) 

where F is the amplitude of 

the generalised force and 

2 2 2 −05 

D= [( 1− f ) + ( 2 f ) ] . 

r 

ξ 

r 

is the 

dynamic amplification factor, f r = f p /f s 

being the frequency ratio. Substituting 

Eqs. (4)–(10) in Eq. (11), the amplitude 

of the steady-state acceleration 

can be expressed as a function of the 

crowd density r through the variables 

r eq , f r and m. For instance, Fig. 5 plots 

the trend of Z ̇̇ versus r for four different 

combinations of geographic 

area (Europe E and Asia A) and 

travel purpose (rush hour R and leisure 

L) and for given structural properties 

of the footbridge (L = 90 m, B 

= 4 m, m s = 2000 kg/m, x = 0,005, f s 

= 0,9 Hz). It should be stressed that 

the four curves are valid for Ż̇ 

≤ ż̇ 

lim 

, 

while the grey branches represent 

the ideal trend of Z ̇̇ in the absence 

of ps synchronisation. The blue dots 

indicate the limit condition (r lim , ̇ż lim 

). 

The travel purpose especially affects 

the value of the structural response, 

by modifying the walking frequency– 

density relation and the f r values in 

turn. The geographic area parameter 

has the main effect of varying the 

value of the maximum density r M , 

therefore shifting the maximum structural 

response. 

It is worth pointing out that Z ̇̇ = Z ̇̇ ( ρ ) 

is not a bijective function, as clearly 

shown in Fig. 5; therefore, Eq. (11) 

is not invertible and the value of 

r lim is herein determined through 

an iterative procedure based on 

the “Goal Seek” tool in Microsoft ® 

Excel ® , even if other algorithms can 


z lim (m/s 2 ) 

.. 

be programmed and employed (e.g. 

the ones based on dicothomic search). 

The obtained critical density value 

can be easily checked by substituting 

it in Eq. (11) and verifying that the 

corresponding acceleration amplitude 

is close to but lower than the 

perceptibility threshold ̇ż lim 

. 

Application and Comparison 

with Other Criteria 

The proposed approach is applied to 

three case studies and the results are 

compared to those obtained with other 

serviceability and stability criteria. 

The first comparison is performed by 

applying the serviceability criterion 

presented herein, but substituting the 

accompanying force model with the 

one proposed by Sétra/AFGC 4 : 

f() t = N F sin( ω t) 

N 

eq 

10 1 

10 0 

.. 

z lim 

10 −2 

eq 

0 

s 

2 

⎧⎪ 

10, 8 Nξ 

ρ< 

1 ped/m (12) 

= ⎨ 

⎩⎪ 185 , N ρ ≥ 1ped/m 2 

with F 0 = 35 N. The equivalent number 

of pedestrians N eq has been 

N lim = 

10 −3 10 −1 10 0 

Dallard 

et al. 8 Newland 12 Roberts 13 

8πξ fM s 

k 

s 

r (ped/m 2 ) 

2ξmL 

s 

αβG 

E-R 

E-L 

A-R 

A-L 

Fig. 5: Diagram of ̇̇ Z versus r for different 

combinations of geographic area and 

travel purpose 

mL 

s 

2 

Gf D 

Table 1: Stability criteria proposed in Refs. 

[8, 12, 13] 

r 

10 1 

obtained as the number of pedestrians 

—uniformly distributed and walking in 

phase with the same frequency as the 

footbridge—that produces the same 

effect as random pedestrians. 

Three stability criteria are applied and 

shortly summarised in the following 

(Table 1). In the equation proposed by 

Dallard et al., 8 M s is the modal mass 

of the bridge and k = 300 N s/m is a 

proportionality constant that has been 

tuned on the London Millennium 

Bridge experimental data. The a and 

b coefficients in Newland’s equation 

12 represent the ratio between the 

motion of the pedestrian centre of 

mass and the bridge motion (equal to 

2/3) and the percentage of pedestrians 

synchronised to the structure (equal to 

0,4), respectively. As for the product 

f 2 r D in the Roberts’ equation, 13 mean 

values are suggested by the author for 

different ranges of the frequency ratio 

f r . Differently from the two previous 

models, the last one does not necessarily 

imply that f p = f s . 

The case studies chosen for the applications 

are the T-bridge (T-br.) and 

M-bridge (M-br.) in Japan 7,22 and the 

south span of the London Millennium 

Bridge (LM br.). 8 The input data 

adopted for the calculations are summarised 

in Table 2, while the results 

in terms of r lim are compared in 

Fig. 6. For each case study, the results 

obtained with both the stability (black 

points) and the serviceability criteria 

(red points) are reported. In particular, 

the present results distinguish 

among different travel purposes and 

the filled points refer to the expected 

traffic condition in service. First, it can 

be observed that the values of r lim 

calculated with the serviceability criteria 

are generally higher than those 

derived through the stability criteria. 

This could be explained by considering 

that the force models adopted to 

derive stability criteria usually assume 

the force as proportional in amplitude 

and resonant to the deck motion, 

thus inducing a higher response. In 

particular, the model of Dallard et al. 

provides by far the lowest values of 

r lim ; moreover, this model has been 

specifically tuned on the LM br. data, 

therefore, its applicability to other 

case studies is questionable. It can be 

observed that the criteria of Sétra, 

Newland and Roberts result in comparable 

values of the limit density in 

all the case studies, even though the 

criteria rely on rather different modelling 

assumptions. The method proposed 

in this article leads to the less 

conservative results for the LM br. 

and M-br. cases, allowing a denser 

crowd on the bridge to reach the limit 

acceleration. It should be pointed out 

that, for each case study, the lowest 

r lim value is associated to the travel 

purpose, which determines the f r value 

closest to unit: in other words, when f r 

is near the unit, the proposed method 

provides results that are close to those 

obtained with the criteria that assume 

resonant conditions. Generally speaking, 

the lateral force is not applied in 

resonance with the deck motion, but 

the walking frequency is determined 

as a function of the crowd density 

and is affected by the travel purpose. 

Hence, the present model allows the 

designer and the owner to take into 

account different traffic scenarios and 

retain the most suitable one: the most 

recurrent scenario, if it can be set (e.g. 

a footbridge linking a bus terminal, as 

in the T-br. case) or the one that gives 

the most conservative result (i.e. the 

lowest r lim value). 

Conclusion 

In this paper, a simplified criterion for 

the assessment of footbridge serviceability 

under lateral crowd loading has 

been presented. It is based on the derivation 

of the limit crowd density that 

induces a lateral acceleration of the 

deck, corresponding to the threshold 

of human perceptibility. The approach 

relies on a load model that accounts 

not only for the action of random 

pedestrians but also for the possibility 

of pp synchronisation due to crowd 

density. Moreover, the proposed lateral 

load model describes the actual walking 

behaviour of pedestrians by referring 

to constitutive laws, namely, the 

speed–density and frequency–speed 

relations. 

L (m) B (m) m s (kg/m) f s (Hz) w M s (kg) f R–C–L q M (ped/m 2 ) v M (m/s) R–C–L f 2 r D 

T-br. (A) 179 5,25 4200 0,93 0,0113 214 010 2,1–1,65–1,89 7,7 1,48–1,37–1,04 14 

LM br. (E) 108 4 2000 0,8 0,007 160 000 1,64–1,28–1,47 6 1,69–1,56–1,12 16,2 

M-br. (A) 320 1,5 600 1,025 0,0027 97 200 2,1–1,65–1,89 7,7 1,48–1,37–1,04 28 

Table 2: Input data for the three case studies: T-bridge, London Millennium bridge (south span), M-bridge (central span) 


lim (ped/m 2 ) 

1,4 L 

1,2 

1 

R 

0,8 

C 

0,6 

0,4 

0,2 

0 

R 

C 

L 

T-br. LM br. M-br. 

Fig. 6: Comparison between the r lim (ped/m 2 ) 

obtained with the different approaches 

L 

C 

R 

R - C - L 

Sétra/AFGC 

Dallard et al. 

Newland 

Roberts 

With respect to stability criteria, 

serviceability criteria appear more 

appropriate for preventing the occurrence 

of SLE, since they are based on 

the description of pedestrian behaviour 

in the pre-lock-in stage. The application 

to real case studies has shown 

that stability criteria are generally 

more conservative and imply more 

severe restrictions to the footbridge 

service than the serviceability criteria. 

According to the authors, the scattered 

results coming from the applications 

of the criteria are mainly due to the 

rather different load models that the 

criteria rely on and they highlight the 

fact that the definition of a universally 

accepted criterion to prevent SLE is 

still an open issue. In particular, the 

mechanism of synchronisation among 

pedestrians due to crowd density 

requires further experimental tests to 

be properly measured and completely 

clarified. 

To the authors’ opinion, the proposed 

approach and load model offer 

a framework susceptible to be easily 

adapted in the case of vertical crowd 

load and resulting vibrations. 

References 

[1] Živanovic S, Pavic A, Reynolds P. Vibration 

serviceability of footbridges under humaninduced 

excitation: a literature review. J. Sound 

Vib. 2005; 279: 1–74. 

[2] Venuti F, Bruno L. Crowd-structure interaction 

in lively footbridges under synchronous 

lateral excitation: a literature review, Phys. Life 

Rev. 2009; 6: 176–206. 

[3] Federation International du Beton (FIB). 

Guidelines for the Design of Footbridges, Bulletin 

No. 32, Lausanne, 2006. 

[4] Service d’Études Techniques des Routes 

et Autoroutes (Sétra/AFGC). Passerelles 

piétonnes – Evaluation du comportement 

vibratoire sous l’action de piétons. Guide 

méthodologique. Paris, 2006. 

[5] Butz C, Feldmann M, Heinemeyer C, Sedlacek 

G, Chabrolin B, Lemaire A, et al. Advanced load 

models for synchronous pedestrian excitation and 

optimised design guidelines for steel footbridges 

(SYNPEX), Report RFS-CR 03019, Research 

Fund for Coal and Steel, 2008. 

[6] HIVOSS, Design of Footbridges: Guideline 

and Background Documents, Research Fund for 

Coal and Steel, 2008. 

[7] Fujino Y, Pacheco BM, Nakamura S, 

Warnitchai P. Synchronisation of human walking 

observed during lateral vibration of a congested 

pedestrian bridge. Earthquake Eng. Struct. Dyn. 

1993; 22: 741–758. 

[8] Dallard P, Fitzpatrick T, Le Bourva S, 

Low A, Ridsdill RM, Willford M. The London 

Millennium Footbridge. Struct. Eng. 2001; 79(22): 

17–33. 

[9] Ricciardelli F. Lateral loading of footbridges 

by walkers. Proceedings Footbridge 2005, Venice, 

2005. 

[10] Venuti F, Bruno L, Napoli P. Pedestrian 

lateral action on lively footbridges: a new load 

model. Struct. Eng. Int. 2007; 17(3): 236–241. 

[11] Brownjohn JMW, Živanovic S, Pavic A. 

Crowd dynamic loading on footbridges. 

Proceedings Footbridge 2008, Porto, 2008. 

[12] Newland DE. Pedestrian excitation of 

bridges. Proceedings of the institution of 

mechanical engineers. J. Mech. Eng. Sci. 2004; 

218c: 477–492. 

[13] Roberts TM. Lateral pedestrian excitation 

of footbridges. J. Bridge Eng. 2005; 10: 

107–112. 

[14] Bruno L, Venuti F. Crowd-structure interaction 

in footbridges: modelling, application to a 

real case-study and sensitivity analyses. J. Sound 

Vib. 2009; 323: 475–493. 

[15] Matsumoto Y, Nishioka T, Shiojiri H, 

Matsuzaki K. Dynamic design of footbridges. 

IABSE Proc. 1978; P17/78: 1–15. 

[16] Venuti F, Bruno L. An interpretative model 

of the pedestrian fundamental relation. C. R. Mec. 

2007; 335: 194–200. 

[17] Oeding D. Verkehrsbelastung und 

Dimensionierung von Gehwegen und anderen 

Anlagen des Fußgängerverkehrs. Straßenbau 

und Straßenverkehrstechnik 1963; 22. 

[18] Bertram JE, Ruina A. Multiple walking 

speed-frequency relations are predicted by constrained 

optimization. J. Theoret. Biol. 2001; 209: 

445–453. 

[19] Araújo MC, Brito HMBF, Pimentel RL. 

Experimental evaluation of synchronisation in 

footbridges due to crowd density. Struct. Eng. 

Int. 2009; 19(3): 298–303. 

[20] Ricciardelli F, Pansera A. An experimental 

investigation into the interaction among walkers 

in groups and crowds. 10th International 

Conference on Recent Advances in Structural 

Dynamics RASD 2010, Southampton. 

[21] Strogatz SH. From Kuramoto to Crawford: 

exploring the onset of synchronization in 

populations of coupled oscillators. Phys. D 2000; 

143: 1–20. 

[22] Nakamura S, Kawasaki T. A method for predicting 

the lateral girder response of footbridges 

induced by pedestrians. J Constr. Steel Res. 2009; 

65: 1705–1711. 


The Construction of the Main Bridge of the Yichang 

Yangtze River Railway Bridge in China 

Yiqiao Zhou, Civil Eng.; Lichao Zhang, Civil Eng.; China Railway Major Bridge Engineering Co. Ltd., Wuhan, China. 

Contact: yqzhou@ztmbec.com 

Abstract 

Railway construction technology in 

China, especially of high-speed railway 

long-span bridge construction, has been 

developing rapidly since the beginning 

of the twenty-first century. To maintain 

the dynamic characteristics, travel 

safety and passenger comfort when 

the train travels over a bridge at high 

speed, and, at the same time, to meet 

the requirement of economical and 

technical viability of construction of the 

bridge—was a challenging problem. For 

solving this problem, many new types 

of bridge structure have been developed 

for railway and high-speed railway 

long-span bridges. The arch beam 

hybrid structure is one of the structures 

that have been widely used in China. It 

is normally made up of a prestressed 

concrete (PC) beam, a rigid frame, or 

V-shaped piers and a steel arch or concrete-filled 

steel tube arch. 

Keywords: hybrid structure; long-span 

PC rigid frame; concrete-filled steel 

tube flexible arch; arch rib rotate– 

lifting synchronously; railway bridge. 

Introduction 

Recently, a long-span arch beam hybrid 

bridge structure has been successfully 

used in constructing the main structure 

of the Yichang Yangtze River Railway 

Bridge, which combined the PC rigidframe 

girder with a concrete-filled steel 

tube flexible arch. The bridge carries 

double rail lines designed to support 

a train velocity of 160 km/h. The span 

arrangement of the main bridge is 130 

+ 2 × 275 + 130 m. This is not only the 

first use of this kind of structure for a 

railway bridge in the world but also, 

with a main span length of 275 m, the 

bridge ranks among the longest of its 

type in the world at present. 

General Description 

This bridge across the Yangtze River 

at Yichang City is a vital link between 

the newly constructed railway line in 

Yichang, Hubei Province and the city 

of Wanzhou in the Chongqing municipality. 

The span arrangement of whole 

bridge from the southern to the northern 

bank is 10 × 49,2 m simply supported 

PC box girders + (130 + 2 × 275 

+ 130 m) continuous PC rigid-frame 

girder with concrete-filled steel tube 

flexible arch + 14 × 48,2 m simply supported 

PC box girder + (56 + 108 + 

56 m) continuous PC girder + 9 × 32 m 

simply supported PC beam; the total 

length is 2526,73 m. A three-directional 

pre-stress system and C60 highperformance 

concrete was adopted 

for the PC girder of the main bridge. 

The structure of arch rib is a parallel 

concrete-filled steel tube truss and the 

arch axis is a quadratic parabola. The 

calculated span of the arch is 264 m 

and the arch rise is 52,8 m. The rise– 

span ratio is 1/5,0. C50 grade microexpansion 

concrete has been filled in 

the rib tube, and the suspenders that 

connect the arch rib and PC girder are 

parallel steel strands (Fig. 1). 

Construction of Substructure 

The foundation of the main bridge 

comprises reinforced bored piles; all 

the pile tips were embedded into the 

solid bedrock beneath the riverbed. 

There are 12 piles (each of diameter 

Φ 3,0 m) for each of the three main 

piers and 9 piles (Φ 2,0 m) for each 

of the two side piers. The pile caps are 

17,0 m × 23,0 m × 5,0 m rectangular 

reinforced concrete structures with 

rounded corners in cross section. A 

rectangular single-cell reinforced concrete 

pier shaft of cross section 8,0 m × 

12,0 m has been used for Pier 12, and 

a twin-wall reinforced concrete pier 

shaft of cross section 3,0 × 12,0 m, with 

5,0 m central spacing between walls is 

adopted for Piers 11 and 13. The height 

of main pier shaft is 38,5 m. Each of 

the two side piers (Piers 10 and 14) 

has a solid rectangular shaft with cutcornered 

cross section and a tray-type 

pier cap. 

Piling works were carried out from 

the working platform over water by 

a boring machine using the air-lifting 

reversed-circulation method. When 

all the piles of a pier had been constructed, 

a steel cofferdam was sunk 

to the designed depth. Construction 

of the pile cap and the lower section 

of pier shaft commenced after the bottom 

was sealed by tremie concrete and 

the water was pumped out of the cofferdam. 

The remaining sections of the 

pier shaft were constructed by means 

of a climbing formwork. 

Construction of PC 

Rigid-Frame Girder 

The superstructure of the main bridge 

is a single-box double-cell PC girder 

130,8 

275,0 275,0 130,8 

Yichang 

Wanzhou 

37,0 

26,0 5,0 

45,0 5,0 

40,0 5,0 

4,0 

30,0 

38,5 

38,5 

38,5 

44,0 

14,5 

14,5 

14,5 

10 

9φ1,5 m bored piles 

11 


12 



13 


34,0 4,0 

14 

Fig. 1: Elevation for the main bridge (Units: m) 


with trapezoidal side webs. The width 

of the top flange is 14,4 m and that of 

the bottom flange varies from 9,2 m at 

the pier top to 12,7 m at mid span. The 

girder depth is 14,5 m at pier top and 

4,8 m at mid span. The thickness of the 

top flange is 500 and 400 mm for double- 

and single-layer tendon arrangement, 

respectively. The thickness of the 

web is changed gradually from 600 mm 

at the pier top to 300 mm at mid span; 

the thickness of the bottom flange varies 

from 1400 mm near the pier top to 

350 mm at mid span. 

The pier top segments were constructed 

on brackets that were installed 

on the pier heads. To ensure construction 

quality of the pier top segment, 

the concreting-works was divided 

into two pours horizontally. The first 

pouring height was 8,7 m, including 

the 1,2 m pier top shaft; the second 

one was 7,0 m up to the top flange. 

The balanced cantilever construction 

method was used for constructing the 

girder and there were 177 segments 

in all (Fig. 2). The longitudinal length 

of the segments was divided into 3,0, 

3,5, 4,0, 4,5 and 5,0 m, and the heaviest 

segment was 3583 kN. Travelling 

formwork was composed of the main 

truss, travelling system, front and rear 

anchorage system, formwork and 

hanging system. The end segments of 

the side spans were concreted on the 

side pier tops and brackets. 

Galvanized steel strands (Φ 15,24 mm) 

were adopted for the longitudinal and 

transverse prestressing systems. 128 

longitudinal tendons and 31 strands 

for each were arranged in the top 

flange at the main piers; they were 

used to meet the requirement for 

the cantilevering construction of the 

girder segments. 28 longitudinal tendons 

and 19 strands for each were 

installed in the bottom slab of the side 

spans—these were tensioned after 

closure of the girder. Considering 

that the girder should resist the thrust 

force from the arch, 12 and 40 tendons, 

each consisting of 19 strands, 

were arranged in the top and bottom 

flange, respectively, at midway across 

the main central spans. Transverse 

prestressing tendons were arranged in 

the top flange only, with five strands 

for each tendon and longitudinal spacing 

of 800 mm. A Prestressing Screw 

Bar (PSB) 930 (Φ 32 mm) threaded 

steel bar was embedded in the webs 

as the vertical pre-stressing system 

and the longitudinal spacing was 

400 mm. For ensuring grouting quality, 

the vacuum auxiliary grouting 

method was applied for all tendon 

and threaded steel bar ducts for elimination 

of the air content in the grout 

and to make the grout completely fill 

the ducts. 

The PC frame girder followed the closure 

sequence of the side spans first, 

followed by the central spans. The two 

side spans were allowed closure at different 

times, but the closure of the two 

central spans had to occur simultaneously. 

The practical closure elevation 

difference is 8 and 4 mm on the two 

side spans, 3 and 6 mm; for the two 

main spans respectively, these were 

much lower than the allowable error 

according to the industrial specification 

in China. 1 

To reduce the influence of concrete 

shrinkage and creep, and enhance the 

cracking resistance ability of the concrete, 

C60 high-performance concrete 

was used for the girder structure. A 

compound–mineral admixture (namely 

ground slag and fly ash mixture) was 

used for the concrete to enhance its 

compactness and crack resistance 

ability, and reduce hydration heat in 

the early stages. In addition, a polycarboxylic 

acid water-reducing agent 

was introduced to optimize the mixture 

ratio, reduce the slump loss, and 

enhance the fluidity of the concrete. 

Installation of Steel Tube Arch 

The main arch is composed of two arch 

ribs and the central spacing is 12,35 m. 

Each arch rib is composed of four steel 

tubes (Φ 750 mm) that are arranged in 

a double dumb-bell-shaped cross section. 

Steel tubes (Φ 450 mm) connect 

upper and lower Φ 750 mm tubes 

vertically to form a truss plan, and a 

steel plate connects the left and right 

Φ 750 mm tubes transversely. The central 

spacing of the four steel tubes is 

1,7 m in the transverse direction, 4,0 m 

at the arch toe and 3,0 m at its crown 

in the vertical direction. The two arch 

ribs are connected by 11 lateral braces 

on each span (Fig. 3). 

1700 

φ = 750 

φ = 450 

Fig. 2: Construction of girder 

Fig. 3: Typical cross section of an arch 

rib (Units: mm) 


The arch rib was manufactured in sections 

in a factory, and delivered at the 

site in barges. In order to assemble and 

install the main arch, arch rib assembling 

falsework on the bridge deck and 

three stay cable towers on the top of 

each main pier were installed. The sections 

of the arch rib were lifted from 

the barge and placed on the falsework 

by cranes (Fig. 4). 

After the four half-span arch ribs had 

been assembled, they were rotated 

vertically and closured by the lifting 

system, which includes stay cable towers, 

balance cables, stay cables, anchor 

cables, anchorages and hydraulic jacks. 

The magnitude and distribution of the 

stay forces were the critical factors for 

successful rotation and closure of the 

ribs. Each stay cable was controlled 

by a hydraulic jack with lifting capacity 

of 3200 kN. All the hydraulic jacks 

were operated by a computer that 

controlled the cable force and ensured 

Fig. 4: Assembling the arch rib 

Fig. 5: Lifting and rotating the arch rib 

synchronized working of the jacks during 

operation. The main arch between 

Piers 11 and 12 was rotated and closured 

first, and closure of the main 

arch of Piers 12 and 13 followed thereafter 

(Fig. 5). 

Once the arch ribs of two spans were 

closured and the lateral bracing members 

between the two main arches 

installed, C50 micro-expansion concrete 

was filled in the rib tubes. Since 

normal concrete has autogenous 

shrinkage and temperature shrinkage 

properties, the concrete in the tube 

could separate from the inner wall of 

the steel tube, thus decreasing the loadbearing 

capacity of the concrete-filled 

steel tube. This meant that the concrete 

filled into the rib tube should have selfcompacting 

and self-expanding properties. 

In order to meet the requirement, 

the concrete mix ratio was carefully 

selected and vacuum auxiliary concrete-pumping 

method was adopted. 

Concrete was filled in the Φ 750 mm 

steel tube of the arch ribs and the space 

between the upper and lower lateral 

connecting steel plates, the Φ 450 mm 

steel tube was kept empty. Concrete 

filling was carried out synchronously 

both up- and downstream of the arch 

rib tubes of the same span. Concrete 

was pumped up from the toes of the 

rib tubes; the grout let-out valves and 

vacuumizing vents were installed on 

the crown of the rib tubes. The vacuum 

level in the tubes was kept between 

−0,1 and −0,09 MPa. 

Every suspender connecting the arch 

rib and PC girder is composed of a 

pair of cables that are composed of 

Φ 15,24 mm galvanized steel strands. 

The longitudinal interval between 

suspenders is 10 m and there are 100 

suspenders (200 strand cables) in 

total. Each cable consists of 15 strands, 

except for the three pairs that are near 

each arch toe, which consist of nine 

strands. The upper end of suspender 

was the tensioning end and anchored 

in the rib tubes; the lower end was the 

fixed end and anchored in the stiffened 

cross-beam beneath the bridge deck. 

A dynamometer was installed on the 

fixed end of every cable in order 

to monitor cable force during the 

construction and operation stages 

to ensure safety of the structure. All 

the strands were wax coated and 

Polyethylene (PE) sheathed individually 

for corrosion protection. High 

Density Polyethylene (HDPE) duct 

was applied for the external sheath of 

all cables. 

Since the Yichang Yangtze River 

Railway Bridge is a very complex 

hybrid structure, the load borne by the 

structure in every construction stage 

is different, and hence structure alignment 

control and stress measurement 

were very important during the girder 

cantilever construction and arch rib 

installation; the structure alignment and 

stress monitoring and control works 

was therefore carried out from the initial 

till the final stages of construction. 

The measurement points were set out 

and strain gauges were embedded in 

the structure; the calculated stress and 

displacement data were used to guide 

construction work in all stages. 

On completion of construction of 

the Yichang Yangtze River Railway 

Bridge, the dead and dynamic load 

tests were carried out on the bridge. 

The test results show that all the design 

requirements of the bridge structure 

have been fulfilled (Fig. 6). 


the arch and the girder. The whole 

structure thus has a graceful profile in 

addition to the excellent economic and 

technical viability. 

References 

[1] Code for Construction on Bridge and Calvert 

of Railway TB10203-2002, China Railway 

Publishing House, Clause 9.5.7, 87. 

[2] Luo S, Yan A, Liu Z. The research of long 

span continuous rigid frame-flexible arch hybrid 

bridge structure. Journal of Railway Science and 

Engineering 2004; 2: 57–62. 

Fig. 6: The completed Yichang Yangtze River Railway Bridge 


Conclusion 

The PC rigid-frame, concrete-filled, 

steel tube arch hybrid structure 

adopted for the Yichang Yangtze River 

Bridge is a new application for construction 

of a double-tracked railway 

bridge. The PC rigid frame girder and 

the concrete-filled steel tube arch bear 

the load together; the dead load of 

the girder is mainly borne by the rigid 

frame itself; the secondary load and 

live load are borne jointly by the girder 

and the arch. The magnitude of the 

forces resisted by the girder and arch 

depends, respectively, on the rigidity of 

its individual structure and the area of 

the flexible suspenders. 2 The mechanical 

behaviour of the structure to resist 

bending is made up of resistance of the 

arch to the compressive stress and that 

of the girder to the tension stress. The 

horizontal thrust of the arch is balanced 

by the axial tension in the girder. Since 

the structure has excellent stiffness 

and stability, most external loads could 

not cause a horizontal thrust on the 

pier As a result, the bending moment 

resisted by girder is decreased; so the 

section dimensions of the girder could 

be reduced accordingly, maximizing 

the advantages of force resistance of 

Owner: 

Wuhan Railway Administration 

Bureau, Ministry of Railways, PRC 

Designer: 

China Railway Siyuan Survey and 

Design Group Co. Ltd., PRC 

Contractor: 

China Railway Major Bridge 

Engineering Group Co. Ltd., PRC 

Steel (t): 25 167 

Concrete (m 3 ): 13 672 

Total cost (USD million): 64,18 

Service date: 

Expected by 

December 2010 


Static and Dynamic Analysis of the “Piedra Movediza” 

Replica Rock, Argentina 

María Inés Montanaro; Civil Eng., Maria Haydee Peralta; Civil Eng., Norma Ercoli; Civil Eng., Maria Laura Godoy; 

Civil Eng., Irene Rivas; Prof., National University of the Centre of Buenos Aires Province, Civil Engineering, Buenos Aires, 

Argentina. Contact: mmontana@fio.unicen.edu.ar 

Abstract 

This paper describes the activities 

involved in the development of the project 

“Piedra Movediza” Replica, located 

on the La Movediza hill in the city of 

Tandil, Argentina, in May, 2007. The 

nature of the project warranted multidisciplinary 

works involving topographers, 

geologists and private enterprises 

for the design and realisation of the 

covering material, construction of the 

anchors, hoisting and the subsequent 

assembly. Attention was especially 

given to the structural aspect, for which 

a typology consisting of an internal 

steel structure formed by a framework 

consisting of four grids arranged in two 

orthogonal planes was adopted. These 

grid beams transferred the load to the 

grid column or mast with its vertical axis 

coinciding with the vertical axis of support. 

The structure was complemented 

by cross, longitudinal and horizontal 

frames, which played a dual role: first, 

to copy the rock’s external geometry 

and serve as a mould for the outer covering, 

and second, to comply with the 

resistance function of transmitting the 

external loads to the internal structure. 

A static and dynamic analysis of the 

resistant structure was performed. 

Keywords: steel; composite; static 

and dynamic analysis. 

History and Aim 

Tandil has always been famous for 

the imposing Piedra Movediza, which 

used to rest at an inconceivably steep 

angle on one of the hilltops of the city. 

This rock finally smashed on the valley 

floor on 29 February 1912. The place 

where this rock once stood is one of 

the most visited places in Tandil. 

The Piedra Movediza (or “moving 

stone”), a large boulder, stood seemingly 

miraculously balanced on the 

brink of a chasm. To demonstrate the 

slight movements of the boulder, it 

was common practice to place bottles 

or some other things on its base to 

see them break. 1 The moving stone 

toppled on 29 February 1912. In May 

2007, a replica was set up in the same 

place where the original rock stood. 

Work Team 

An agreement was signed between 

the Tandil Government and the 

Universidad Nacional del Centro de la 

Provincia de Buenos Aires (UNCPBA) 

to commence the studies that would 

embark on the project to create a replica 

of the moving rock, a team was 

formed with teachers and a graduate 

of the Civil Engineering Department, 

coordinated by the engineer, and 

external participants. The engineer, 

and a team from the Universidad 

Nacional de La Plata, together with 

the Aeronautics Department and a 

geologist, also took part. 

Plan of Action 

A plan of action was established following 

certain given guidelines: 

– The replica should have the same 

geometrical dimensions as the original 

rock. 

– It should be fixed in the same position 

as the original rock. 

– It should be situated at the same 

place on Cerro La Movediza. 

– It should consist of a steel structure 

and it should be coated with a composite 

material capable of reproducing 

the colour and texture of the 

original rock. 

Activities 

Topographic Studies 

The planialtimetric topographic surveys 

determined the position and location 

of the replica. The geometry of 

the replica was determined from an 

analysis of the available information 

and from the survey of the geometry 

of the fallen rock that lies at the foot 

of the Cerro La Movediza. Infrared 

rays, laser electronic tachometres and 

GPS receiver systems were used to this 

end. 1523 points among the three existing 

pieces of rock were surveyed in a 

planialtimetric way. The study showed 

that the rock had an approximate volume 

of 91 m 3 , an approximate weight 

of 248 tons and an external surface of 

approximately 133 m 2 . 

Geological and Geotechnical Surveys 

These studies included, on one hand, 

the evaluation of the hill discontinuities 

existing in the location. It was 

observed that the hill had a fissure, the 

geological and geotechnical characteristics 

of which were studied, leading to 

the conclusion that this did not, in any 

way, render the task of anchoring the 

replica to the hill impossible. 

On the other hand, the characteristics 

of the rock base, which determined the 

depth of anchors and their size, were 

studied. It was reported that the solid 

rock on the hill had a granite structure 

and that it presented very good 

geological characteristics 2 ideal for 

anchoring the replica. 

Wind Tunnel Study 

This study allowed the assessment 

of the impact of the topography on 

the location and the geometry of the 

structure in the distribution of the 

wind pressure. The static and dynamic 

forces at the anchoring for each direction 

of the wind were also measured. 

The range of standard frequencies 

where higher levels of non-stationary 

aerodynamic load can be found was 

also determined. 

A 1 : 40 scale prototype model of 

the replica of the “Piedra Movediza” 

and part of the summit of Cerro La 

Movediza were tested in the wind tunnel 

of the Aeronautics Department 

at the Faculty of Engineering at the 

UNLP. The model of the hill summit 

was modified to allow its rotation 

in the tunnel test section and to 

study the winds in eight directions (S, 

SW, W, NW, N, NE, E and SE). The S 

directions (direction of more frequent 

winds) – SW (direction of winds of 

higher intensity) and SE – follow the 


original geometry. On the basis of the 

observations at the location, it was 

estimated that the changes in the modelling 

of the hill were conservative, as 

they may induce higher wind loads in 

the test than those obtained in reality. 

Figure 1 shows an image of the tested 

model and the equipment used. 

For each wind direction, forces and 

distribution of pressure at different 

speeds were measured to verify the 

independence of the Reynolds number 

of dimensionless coefficients of force 

and pressure. 

The distribution of pressure in the replica 

model for the eight wind directions 

was measured in 56 nodes by means of 

sensors, presenting the information as 

a dimensionless coefficient of pressure 

C p . Analysis of the specified pressure 

coefficients reveals the strong influence 

of the irregular geometry and the 

topography (at location) in their distribution. 

The data reported by the tests 

allowed the corresponding calibration 

of the numerical models used and the 

comparison of results. 

Covering Material 

The covering material has a structural 

function, necessitating that certain 

requirements of strength, stiffness and 

durability be followed. Because of the 

nature of the project, the covering 

material should simulate the texture 

and colour of the original rock and 

achieve an optimum weight to facilitate 

the replica hoisting. 

From the resistance point of view, the 

replica should resist the pressure of 

the wind and transmit it to its internal 

resistant structure. 

In response to the request for simulating 

the texture and colour of the 

original rock, a large amount of previous 

works carried out with composite 

materials in this regard were referred 

to. Additionally, from a construction 

and assembly point of view, these kinds 

of materials have advantages because 

of their lower weight, and based on 

this, a composite material for the covering 

was adopted. 

Structural Project 

The project was designed considering 

that the replica would have the same 

geometrical dimensions, same appearance, 

texture and colour. It would also 

be located in the same position and 

location on the hill with respect to the 

original rock. Data from the previously 

indicated stages of the study made the 

structural project possible. 

Structural Typology 

The typology adopted, as shown in 

Fig. 2, consists of an internal steel structure 

built in a construction company 

and made of a framework consisting of 

four grids arranged in two orthogonal 

planes according to Fig. 3. These grids 

relieve in a column, with its vertical 

axis coinciding with that of the original 

rock. The structure is complemented 

by vertical and horizontal frames, made 

of stiffened steel plates, which play a 

dual role: first, copying the external 

geometry of the rock and serving as a 

mould for the outer covering, and second, 

transmitting the loads caused by 

the wind to the internal structure. The 

connection of the horizontal external 

frames with the grids is mainly through 

the beams at three planes of stiffness. 

Structural Analysis 

A software based on Finite Element 

Method, was used for the static and 

dynamic structural analysis. 

Fig. 2: Structural typology 

Static Behaviour 

In the first instance and for the purpose 

of making a pre-dimensional study 3 of 

the structure elements, simplified analyses 

in orthogonal planes coincident 

with those of the main grids of the 

frame were performed. These analyses, 

apart from enabling the initial predimensional 

study, allowed making 

the typologies of the framework grids 

consistent with the external shape and 

dimensions of the original rock. 

Simplified load combination related to 

adopted models, and load hypotheses 

arising from the overlapping of the 

dead load (weight structure + weight 

of the covering) and wind were considered. 

To this effect, a distributed 

covering weight of 3 tons and a wind 

pressure corresponding to a speed of 

50 m/s equivalent to 180 km/h was 

considered. The distribution of the 

wind pressures was obtained from 

the coefficients of pressure obtained 

through the tests in the wind tunnel for 

the eight tested directions and for the 

highest speed corresponding to 50 m/s. 

For the transverse grids, the overload 

corresponding to the north wind as the 

worst combination was considered. 

Following the pre-dimensional analysis 

of plane models, spatial static analyses 

were conducted considering nine load 

hypothesis corresponding to the dead 

loads and the combination of the eight 

states of independent wind loads with 

the dead load. The worst results for the 

central column corresponded with the 

column base plate. 

Dynamic Behaviour 

The unique characteristics of the structure 

were obtained by dynamic analysis 

for purposes of comparison with inputs 

from wind tunnel tests, with reference 

to the standard frequency range in 

which the highest level of energy from 

non-stationary aerodynamic loads is 

concentrated. The reported range of 

normalised frequencies corresponds 

to a 1,7 and 6,5 Hz dynamic load 

Fig. 1: Prototype model in wind tunnel 

Fig. 3: Frameworks 


frequency. The first dynamic analysis 

performed with the pre-dimensional 

study used in the static analyses yielded 

values of the fundamental frequency, 

which was in the range of the highest 

load energy measured in the wind tunnel. 

This contributed to the stiffness of 

the column of the structure and other 

areas of importance that led to a fundamental 

frequency of 8 Hz away from 

the range of excitement mentioned. 

Design of the Foundation 

The design of the foundation of the 

moving rock replica basically includes 

four structural anchors of 5 m length, 

embedded in the base rock held by a 

single anchor plate of 70 mm thickness, 

with stiffeners arranged orthogonally 

and inserted into the rock with a 

grout-type injection material between 

the rock and the plate to ensure proper 

adherence. Additionally, three vertical 

inserts 1,5 m deep and of construction 

nature were used. 

To facilitate the union of the structure 

with the foundation, the central column 

was welded to a 1200 × 1150 × 

50 mm base plate. This base plate was 

bolted, after the hoisting, to an ISO 8,8 

anchor plate with 20 bolts of 1 ½" ISO 

8,8 displayed on the anchor plate for 

that purpose. The base plate had corresponding 

holes that were perfectly 

aligned for the assembly. 

Construction Process 

The construction process was led by 

the Secretary of the Ministry of Public 

Works and Services for the city of 

Tandil. The structure was built in a 

construction company in Tandil. 

Figures 4 and 5 show the column, the 

grids and part of the frames, and the 

arrangement of the cross frames and 

the base plate, respectively. 

The outer covering was built in as 

specified, with some adjustments in the 

colour and texture. 

The drilling for the anchoring was 

done under the supervision of a geologist, 

taking appropriate precautions 

considering the particular site where 

the task had to be carried out. The final 

re-design for the arrangement of the 

anchors and the corresponding plate 

proved to be an arduous task. To this 

effect, a pattern of the dimensions of the 

anchor plate, which facilitated the rearrangement 

in response to the requirements 

of the geological survey with 

regard to the separation of the fissure 

Fig. 4: Central column 

Fig. 5: Framework construction 

plane, was built. The anchors were built 

by a company that specialised in rock 

anchoring; they also set the inserts and 

injected the grout according to the 

specifications stated. The anchor plate 

was hoisted by a crane already installed 

at the foot of the hill for mounting of 

the replica. The anchor plate with its 

20 bolts was fixed to the replica via the 

base plate with holes. Before the hoisting 

of the replica, a test of uprooting 

the anchors was done, with results that 

were within the established limits. The 

relocation of the replica, from the construction 

company to the foot of the 

hill, was accompanied by applauding 

and waving of flags by deeply touched 

people, which showed what the rock 

meant to them and to the city. 

The construction process lasted for 

approximately 4 months. The replica 

successfully crowned the hill on 13 

May 2007. 

A huge crane of about 9 tons hoisted 

the replica (Figs. 6 and 7). 

Conclusion 

The notable geometry of the replica 

and the site characteristics of the location 

made this project a real challenge. 

The development of the project and 

its subsequent execution showed the 

importance of working together for a 

project, especially when different disciplines 

are involved. It also made possible 

the sharing of knowledge between 

Fig. 6: The relocation of the replica 

Fig. 7: The replica crowned on the hill 

university and community, making the 

society highly appreciative of the university’s 

social role. 

References 

[1] “La Piedra Viva” Elías El Hage, Pomy Levy. 

Alfredo Bossio. Artes Gráficas. 1° Ed. Mayo, 

2007. 

[2] Informe petrográfico geofísico de los estudios 

realizados sobre la Piedra Movediza de Tandil. 

Inédito. Secretaría de Minería. Buenos Aires, 

1962. 

[3] Reglamento Argentino de Estructuras de 

Acero para Edificios. CIRSOC 301, Buenos 

Aires, Julio, 2005. 


Owner: 

Municipalidad de Tandil (Argentina) 

Contractor: 

Metalúrgica Marcelo Fernandez 

Designer: 

National University of the Centre of 

Buenos Aires Province 

Steel (t): 7 

Composite (t): 3 

Estimated cost (EUR million): 0,2 

Service Date: May 2007 


Footbridge Studenci over the Drava River in Maribor, 

Slovenia 

Viktor Markelj, Structural Eng., Manager, PONTING d.o.o. Maribor; Lecturer, Faculty of Civil Engineering, University of Maribor, 

Slovenia. Contact: viktor.markelj@ponting.si 

Abstract 

The paper presents the design and 

construction of the new footbridge 

Studenci over the Drava River in 

Maribor. In 2004, the City of Maribor 

issued an open, anonymous national 

call for proposals for the reconstruction 

of the old bridge. In spite of the 

fact that it was an existing bridge 

that was to be reconstructed, the new 

bridge appeared in a new, entirely different, 

contemporary image. 

The existing old bridge structure of 

two steel I-shaped girders with concrete 

deck was replaced by a new, 

steel, transparent, space truss structure, 

with a wooden deck and linear 

LED illumination. Only the supports 

from the old bridge were preserved 

and these also were reconstructed and 

strengthened. 

The ‘structural solution’ as presented 

in this paper, received a Footbridge 

Award in 2008. 

Keywords: footbridge; pedestrian 

bridge; reconstruction; design competition; 

steel structure. 

Introduction 

The new footbridge over the Drava 

River in Maribor is in fact a reconstruction 

of the old bridge where the 

existing supports were strengthened 

and the old superstructure was substituted 

with a new one. 

In spite of a relatively simple and clear 

structural system, the bridge is distinctly 

recognizable due to the interaction 

of its walking surface and structure 

elevation. Because of the transparency 

of its truss structure and symmetry, it 

is neutral to both the environment and 

the river landscape. In addition to the 

attractive configuration, the original 

design made way for an attractive and 

economic method of construction. 

History 

The history of the old footbridge is full 

of character. It was constructed in the 

year 1885 with three simply supported 

truss girders and wooden supports in 

the riverbed. After the flood in 1903, 

the bridge was provided with intermediate 

stone supports (Fig. 1). During 

World War II, it was destroyed twice 

and each time reconstructed by the 

army. In the year 1946, the bridge was 

swept away by floods; two years later, 

the superstructure was substituted by a 

new one—with two welded plated girders 

(Fig. 2). After the completion of the 

hydro power plant in 1968, the water 

level increased by about 5 m. In 2007, 

a new superstructure was constructed 

according to the winning solution of 

the design competition conducted in 

2004. 

Design of the New Footbridge 

To minimize construction costs, it was 

necessary to reuse the existing supports 

of the old footbridge, thus dictating 

the spans of 3 × 42 = 126 m, 

and, at the same time, it was necessary 

to increase the existing navigation 

clearance under the bridge from 3,0 to 

3,6 m of clear opening. Beside common 

requirements such as safety, durability 

and economy, the city of Maribor also 

sought an attractive, unique bridge 

that would be in harmony with the 

environment and would prove appealing 

to the town dwellers. 

The selected solution was a triangular 

steel truss as a primary longitudinal 

structure (spine) along which a 

secondary transversal structure was 

to be raised (ribs). As neither of the 

banks had sufficient height for the 

structure, the structure was raised 

upwards through the walking surface, 

thus being divided into two parts. 

With a gradual rise of the deck, the 

structure disappears under the floor, 

joining both walkways over the middle 

of the river and providing a common 

surface for an unimpeded view 

of all sides and a peaceful meeting 

place. 

Fig. 1: Footbridge Studenci from 1885 to 1946 Fig. 2: Footbridge Studenci from 1948 to 2007 


Longitudinal section / side view 

1 

5,25 

42,00 

2 

136,50 

42,00 

3 

42,00 

4 

5,25 

Studenci 

Taborsko nabrezje 

Koblarjev zaliv 

Lent 

Ruska cesta 

Strma ulica 

3,05 

3,60 

Drava river 

3,05 

Support axis 1 


Plan / view 



5,25 

42,00 

136,50 

42,00 42,00 

5,25 

3,20 

5,80 

Drava river 

Fig. 3: Longitudinal section and plan of the new footbridge (Units: m) 

Description of the Footbridge 

and the Structure 

The footbridge axis is in a straight 

line, the longitudinal alignment is 

in a convex vertical rounding with 

R = 1045 m and with a maximal slope 

of 5%. In the middle of the footbridge, 

the deck width is 3,20 m and at 

either end it is split into two parts of 

2,40 m each, divided by a visible main 

structure breaking through the deck 

(Fig. 3). 

The variation of the structure and deck 

alignment was solved by dividing the 

structure into two systems supporting 

the deck: 

– The spine—main steel structure is a 

centrally positioned triangular truss 

of a constant structural height and 

width. The geometry enabled a simple 

production in a workshop: after 

assembling on site by welding, it was 

erected in place simply by incrementally 

launching it over the existing 

structure. The main steel structure 

is made of structural steel S355 and 

protected with anticorrosive coats in 

light grey. 

– Ribs—secondary steel structure rises 

along the main structure and carries 

the walking surface. Erected 

by bolting after the erection of the 

main structure, it is protected by hot 

galvanizing. 

– Wooden deck structure made of 

transversal boards from hard exotic 

wood of 42 mm thickness, with a 

sawtooth profile on the top. 

The main structure is a space steel 

truss, consisting of three longitudinal 

Secondary structure 

cross bracket 

1,760 

Cross bracket width 

f298,5 mm 

d = 8–2 mm 

B = 3,200–5,800 

Bridge axis 

Main structure attachment 

plates (altering position) 

f298,5 mm 

d = 26–20 mm 

f114,3 mm 

d = 8–20 mm 

1,500 

0,720–3,140 

Main structure width 

1,750 

1,760 

Cross bracket width 

Fig. 4: Cross section of the footbridge is of constant shape—changing only at the ends 

of the secondary structure 

New pier cap 

Micropile 

f25 cm, L = 15,0 m 

Strengthened existing pier 

min. 7,000 m 

Jet-grouting piles 

0,25 

1,60 

Pier axis 

2,70 

3,50° 

1,57 

0,25 

1,75° 1,75° 

254,524 

New bearing block 

0,50 

253,200 

14,50 

15,00 

Water level 

River bed 

New pier cap 

Micropile 

f25 cm, L = 15,0 m 

Strengthened existing pier 

Fig 5: Reconstruction of the old river piers with jet-grouting piles (Units: m) 

min. 7,000 m 

Jet-grouting piles 

0,25 

7,50° 

os brvi 

5,45 

6,50 

0,25 

7,50° 7,50° 

3,17 3,17 

0,25 

7,50° 

New bearing block 

254,524 

0,50 

253,200 

14,50 

15,00 


30,0° 

30,0° 

Original (glued) wooden section 240/80 mm 

240,0 

30,0° 

30,0° 

LED linear diodes (L = 2400 mm, e = 40 mm) 

pipes—one pipe for the upper chord 

and two for the bottom chord. The 

triangular cross section is of constant 

form; the change is only in the 

pipe thickness and the position of the 

accessory piece for bolting the secondary 

structure (Fig. 4). The axial 

distance between the upper and bottom 

pipes is 1,75 m which gives a total 

structural height of 2,05 m; the axial 

distance between the lower flanges is 

1,50 m. The longitudinal pipes are of 

diameter ϕ = 298,5 mm, wall thickness 

varies from 8 to 20 mm and for 

the diagonal and cross pipes, ϕ = 114,3 

mm. The camber of the main structure 

is of a constant radius R = 4000 m, 

the upright connections between layers 

are not vertical but radial, so that 

the element lengths and mutual angles 

are equal along the total bridge length, 

making the construction cheaper and 

simpler. 

The old river piers were reinforced 

with six piles and the abutments— 

because of greater width—with nine 

piles. The total length of the jet-grouting 

piles is 15 m—8 m through the old 

pier structure and 7 m through the 

gravel base (Fig. 5). The tops of existing 

piers have been adapted for the 

needs of the new structure. 

80,0 

Polycarbonate protected 

strip d = 3 mm 

Fig 6: Footbridge lighting is concealed within the wooden top rail of the steel railing 

Fig. 7: Main truss of new footbridge positioned over the old structure 

Fig. 8: Dismantling of the old structure 


method of erection was possible. After 

erection, the new structure was supported 

by the remainder of the existing 

structure. Truss elements were erected 

above the intermediate supports and 

the bearings were grouted. Then the 

old footbridge was dismantled (Fig. 8). 

Some details of the new footbridge 

are shown in Fig. 9. The profile of the 

reconstructed footbridge Studenci can 

be seen in Fig. 10. 

Fig. 9: Some details of the new footbridge 

Conclusion 

Since the city of Maribor could not 

give up more than 120 years old 

pedestrian route, it decided to reconstruct 

the old destroyed footbridge - 

on the existing supports, but with 

the new superstructure. The original 

design of the bridge’s reconstruction 

provides an attractive form and at 

the same time enables a very costeffective 

way of building. The design 

solution has been presented with the 

Footbridge Award 2008 in the category 

of Technical medium span, with 

the judges declaring it as »A pleasing 

technical solution to a difficult set of 

criteria« and noticing its »Interesting 

new technical ideas«. 


Fig. 10: Reconstructed footbridge Studenci over the Drava River in Maribor 

The secondary structure, supporting 

the wooden deck structure, is 

composed of transverse cantilever 

girders and longitudinal girders and 

is screwed to the main structure all 

along. The transverse cantilevers 

are at the same time railing balusters 

and are equal along the entire 

bridge length. The deck is made of 

transversely placed bangkirai wooden 

boards. The discreet lighting from the 

handrail is another special feature of 

the footbridge. The installed lighting 

power of LED diodes is only 350 W 

(Fig. 6). 

Construction 

The footbridge construction was innovative 

and was built at a low cost, as it 

used the existing structure as a support 

for erection. The structure was divided 

into two parts, the main longitudinal 

one and the secondary transverse one. 

The main structure is a triangular truss 

that was welded by segments and progressively 

positioned over the existing 

structure (Fig. 7). The separation 

into main and secondary structures 

reduced the transverse size of the 

structure to such an extent that this 

Owner: 

City of MARIBOR, Slovenia 

Structural design: 

PONTING d.o.o., Mari bor, Slovenia 

Contractors: 

SGP Pomgrad GNG d.o.o., Slovenia 

Steel structure: Meteorit d.o.o., Slovenia 

Bridge type: Steel space truss 

Bridge size: 

Length: 130 m, 

Width: 

3,2–5,8 m 

Deck: 

Bangkirai wood: A = 550 m 2 

Steel S355: 

93 000 kg 

Lighting: 

LED 350W 

Total cost (EUR million): 1,2 

Service date: December 2007 


Sanchaji Bridge: Three-Span Self-Anchored Suspension 

Bridge, China 

Gonglian Dai, Prof., Dr, Civil Eng.; Xuming Song, Lecturer, Dr, Civil Eng.; Nan Hu, Graduate Student; Central South University, 

Changsha, China. Contact: daigong@vip.sina.com 

Abstract 

Sanchaji Bridge across the Xiangjiang 

River is located at the northern section 

of the Second Ring Road Project 

in Changsha, Hunan Province, China. 

Its main span is the longest in the 

world among all self-anchored suspension 

bridges constructed using double 

towers and double cables till now. 

For a self-anchored bridge, structural 

behaviour and construction methods 

are totally different from those of a 

traditional suspension bridge. A main 

cable containing 37 prefabricated 

strands and streamlined steel box cross 

section was used as stiffened girder 

with a height of 3,6 m. A fully welded 

anchoring chamber was adopted to 

connect the main cable with the stiffened 

girder for the first time. The main 

tower is of reinforced concrete with a 

variable hollow box section. As for the 

construction method, launching methods 

were selected for the erection of 

steel box girder and “non-stress” 

method for installation of hangers. The 

construction of the bridge was started 

on 10 September 2004 and completed 

on 1 September 2006. The completion 

of the bridge effectively relieved the 

traffic pressure in the northern region 

of Changsha and played a vital role in 

improving local economy. This paper 

introduces several features of the 

Sanchaji Bridge, including type selection, 

structural behaviour and construction 

methods. 

Keywords: self-anchored suspension 

bridge; structural system; design 

parameters; construction method. 

a (5 × 65 m) PSC continuous beam. 

Before project bidding, there were 

five bridges across Xiangjiang River in 

Changsha, including two arch bridges, 

two beam bridges and a cable-stayed 

bridge. The self-anchored suspension 

bridge system was selected because 

of its logical structural principles and 

the aesthetic value addition to the 

landscape 1 . 

Geological Condition 

The Sanchaji Bridge is located on a quaternary 

stratum, which mainly contains 

soil in the upper and slate in the lower 

levels. There are not many ups and 

downs at the surface of the river bed, 

where the main ingredients are sandy 

slate and metamorphic sandstones. 

The uniaxial compressive strength is 

20,9 to 65,60 MPa. Rock forms the 

fourth level. The arrangement of the 

main span and the geological conditions 

are shown in Fig. 2. It can be seen 

that piers 11 and 12 that formed the 

foundation for the two towers adopted 

drilled pile groups of C30 class concrete 

with length between 16 and 24 m. 

The C30 concrete pile cap under each 

tower is a dumbbell-shaped structure 

including two caps with diameter of 

17 m and thickness of 5 m linked by 

7 × 4 m tie beam. Under each tower, 

there are 18 piles with a diameter of 

2,4 m. The width of the channel located 

in the west of the island is about 700 

m and its greatest depth is at about 

6,0 m. Flood peak of Xiangjiang River 

frequently appears from late April to 

June, which covers about 86% of total 

number of peaks occurring in 1 year. 

The basin is so large that the flooding 

is quite heavy. 

Design Criteria 

The main design criteria for Sanchaji 

Bridge are high-way-I with design 

speed of 60 km/h; three-lane traffic 

loading in each direction with pedestrian 

load of 4 kN/m; deck width of 35 m; 

lateral slope of bridge deck is 2,0% in 

two directions; longitudinal slope of 

bridge is 1,5%; temperature: design 

normal temperature is 20°C with mutative 

range of 0 to 40°C; normal design 

wind speed refers to average speed of 

25,9 m/s; designed to withstand upto 7 

grade intensity of magnitude for earthquake; 

anti-collision standard refers 

to third-level fairways with 400 kN 

Introduction 

Sanchaji Bridge shown in Fig. 1 is 

located at downstream Xiangjiang 

River in the northern region of 

Changsha, where the width between 

river banks is about 1442 m. The total 

length of the main bridge is 1577 m, 

with span arrangement of the whole 

bridge containing a (8 × 65 m) prestressed 

concrete (PSC) continuous 

beam, a (70 + 132 + 328 + 132 + 70 m) 

self-anchored suspension bridge and Fig. 1: The close view main span of Sanchaji Bridge 


K41 + 504,35 K42 + 236,35 

70 12 + 12 × 9 + 12 = 132 

732 

11 + 34 × 9 + 11 = 328 

12 + 12 × 9 + 12 = 132 70 

129,655 

129,655 

with a spacing of 600 mm. The thickness 

of the inclined web flange is 8 mm 

with longitudinal closed rib stiffeners 

at 400 mm spacing. 

55,770 

56,720 

57,964 

58,515 

57,964 

56,720 

55,770 

Main Tower and Pier 

23,496 

9 10 11 12 13 14 

Fig. 2: The main span arrangement of Sanchaji Bridge (Units: m) 

in longitudinal and 550 kN in lateral 

direction; channel clearance is 10 m 

while flood level is 36,78 m. 

Structural System 

Since no bridge of a similar type was 

constructed ever before, 1 : 28 overall 

model test and 1 : 5 steel anchoring 

chamber test were carried out to 

verify the accuracy of finite element 

analyses. Test results were used in conceptual 

design and the details of the 

design of the bridge are as follows: 

(a) a five-span continuous girder was 

used in a self-anchored suspension 

bridge for the first time, which eliminated 

tremendous uplift force created 

by the anchorage of main cable at #10 

and #13 piers. In this way, a negative 

reaction force at the bearings could 

be basically avoided under the dead 

load, and very little rotation at the 

beam end occurred during the service 

period; (b) the issue of stability is 

vital in this bridge because the cable is 

anchored at the girder of the suspension 

bridge. So, it was addressed by 

placing precast concrete blocks in the 

box girder near the section of the pier 

where the heaviest weight was 8000 kg 

at #10 and #13 piers and 6000 kg at #11 

and #12 piers. In this way, there was no 

negative reaction force at the bearings 

of these four piers mentioned under 

both dead load and live load, and 

the safety coefficient was maintained 

above 1,3; (c) the selection of restraint 

type would have a huge impact on 

structural behaviour. On the basis of 

analysis and comparison, four MSTU 

25,202 

dampers were installed between the 

stiffened girder and tower columns 

at both towers, where each damper 

required a force of 100 tons. Therefore, 

the stiffened girder would be a floating 

system along the bridge under 

some slow loads, such as temperature 

variation, while the girder would be 

controlled by dampers at columns of 

the tower and reduce the force on the 

foundation of the tower during special 

live loads, such as vehicle braking 

force and earthquakes. 

Stiffened Girder 

In a self-anchored system, the stiffened 

girder will be under a great axial force. 

Therefore, the cross section of girder 

adopted a streamlined steel box type 

with Q345d steel, after the wind tunnel 

test, as shown in Fig. 3. The main 

dimensions of the steel box girder 

cross section were as follows: the clear 

height of the box girder at the bridge 

axis was 3,60 m and overall width was 

35 m. So, the high-span ratio of the box 

girder is 1 : 91,1 and high-width ratio 

is 1 : 9,72. The bridge deck adopted 

an orthotropic plate with a thickness 

of 12 to 14 mm at the top plate, 12 to 

16 mm thickness at the web plate and 

10 mm at the bottom plate, except 

at some local section near the tower. 

There is a diaphragm every 3 m with 

a thickness of 10 mm at normal section 

and 16 mm at the position of 

bearing. The U-shaped closed rib deck 

stiffeners are 260 mm deep and 8 mm 

thick with a spacing of 600 mm. At 

the bottom flange, the dimensions are 

190 mm (depth) and 6 mm (thickness) 

The main tower adopted a C50 class 

RC structure with variable cross sections, 

as shown in Fig. 4. The height 

of the tower is 106,16 m in #11 and 

104,453 m in #12 piers (from the top 

of pile cap). Above the bridge deck, 

the height is about 71,752 m. The cross 

section of the tower is of a hollow box 

type with dimensions of 5 (along the 

bridge) × 3,5 m (along the section) at 

the top. The column width along the 

direction of the bridge expands from 

the top to bottom according to the 

ratio 1 : 100. The two columns of the 

tower are linked by crossbeams with a 

spacing of 25 m at the top. There are 

two PSC crossbeams between columns 

with a hollow rectangular cross section. 

The crossbeam below the deck 

is 4,0 × 6,0 m in cross section and 0,6 

m thick, equipped with 48,12-ϕj15,24 

strands. As for the upper crossbeam, 

the dimensions are 3,0 × 4,5 m, with 

a thickness of 0,5 m, equipped with 

eight 12-ϕj15,24 strands. The thickness 

of the column is variable, being 

700 mm above the lower crossbeams 

and 1 m below the same. Furthermore, 

all the joint parts between crossbeams 

and columns are strengthened, with a 

gradual variation in thickness. Because 

of saddle installation, the solid column 

area was adopted with a length of 4 m. 

1000 2900 2200 

35 000/2 

11 400 

2070 1200 

3600 

3693 

13 807 

Fig. 3: The semi cross section of the bridge after the wind tunnel test (Units: mm) 

Fig. 4: The main tower above the bridge 

deck 


Lightning protection devices were set 

up at the top of the column. Except 

for the two main towers, the remaining 

piers for approach are of twin-column 

type with a diameter of 3 m. 

Cable System 

On account of the larger rise–span 

ratio than that of the gravity-anchored 

suspension bridge or single tower one, 

it is vital to take into consideration 

the selection of hanger clamps, shape 

of cable design, installation method 

of hanger, and so on. The rise–span 

ratio of Sanchaji Bridge is 1/5 at midspan 

and 1/10,6245 at side span 2 . 37 

parallel wire strands, comprised of 

127 j 5,1 mm high-strength galvanised 

steel wires, formed one main cable, 

which was installed by Prefabricated 

Parallel Wire Strands (PPWS) method. 

The total length of the main cable was 

680,70 m with a spacing of 25,0 m 

between the wires. The amount of steel 

in the main cable was about 1026 t. The 

shape of the main cable was catenary 

at every construction stage before 

operation, because the shape between 

two tower saddles was just caused by 

the dead load along the cable. The 

material adopted for strand anchor 

tube and cover plate was ZG310-570 

and Q235-A, while the filled material 

in tube was zinc–copper alloy. 

The fully welded anchoring chamber 

was adopted for connecting the main 

cable with the stiffened girder because 

the height of girder was not required 

to increase in order to ensure superior 

appearance and convenient construction 

of the girder. Moreover, the 

structural behaviour of this anchor- 

ing chamber is superior to the concrete 

one because cracking caused by 

huge local compressive stress could 

be avoided. Two hanger ropes were 

installed at each cable band, as shown 

in Fig. 5. Except for the two pairs of 

rigid hangers near the #10 and #13 

piers, all hanger ropes were made of 

85j 5,1 mm galvanised steel wire. The 

prefabricated parallel wire strand 

hanger was pinned by a socket with 

cable and anchored at the girder. There 

are 122 pairs of hangers in the whole 

bridge with a spacing of 9 m. Screws 

with bolts were used for the connections 

between the hanger and girder, 

so that the flow of force would be 

clear and installation easy. The hanger 

clamp is a symmetrical steel casting 

component and the length of the clamp 

is controlled by declining cable force 

and clamp traction force. On the basis 

of an experiment on hanger clamp, 

the traction force is maintained at 12 

MPa instead of the regular 10 MPa to 

avoid the huge length of clamp. There 

are four saddles in the whole bridge 

with the self-balancing system so that 

blocks at the top of the tower are not 

needed. The major material of saddle 

is cast GZ25 with a weight of 40,5 t, 

including the upper component, lower 

plate, filled panel and so on. 

Construction Method 

The whole construction period was 

divided into five stages: foundation 

and tower erection; incremental 

launching of stiffened girder; erection 

of main cable; installation of hang- 

ers; deck system pavement. The major 

difficulties during the construction 

included four aspects as follows: (a) 

Incremental launching of the steel box 

girder. The launching platform was 

set between #9 and #10 piers where 

the water was deep and four temporary 

piers were placed for carrying 

the main girder at middle span. The 

standard prefabricated segments were 

divided into two types (12 and 9 m) 

and assembled in site by welding. The 

weight of standard segment was 195,8 t 

of 12 m and 152,6 t of 9 m. According 

to the requirement of waterway during 

the construction, temporary piers 

were placed with a maximum span of 

77 m, and the length of launching nose 

was 48 m. (b) The line shape control 

of the main girder and cable. Since the 

main girder was on a vertical curve 

with 24402,745 m radius, the segments 

were installed on the platform according 

to the line shape of the bridge 

girder and pre-camber consideration. 

Owing to a larger rise–span ratio, 

there was a huge difference in line 

shape before and after the installation 

of hangers. So, the installation of main 

cables should follow strict limits and 

enable the controlling plate to adjust 

the errors during cable fabrication. (c) 

“Non-stress” method in installation 

of hangers. To install the “non-stress” 

hangers, the girder was lifted by four 

temporary piers (740 mm raised in 

two piers near the tower and 1,48 m 

in piers of middle span), as shown in 

Fig. 6. After the installation of the 

hangers was completed, the temporary 

piers were removed and the girder 

system could find its final position, 

Fig. 5: The hanger rope between main cable 

and stiffened girder 

Fig. 6: Sanchaji Bridge after the installation of hanger (during construction) 


Fig. 7: The view of Sanchaji Bridge at night in service period 

the transformation of the structural 

system was completed. In this way, the 

time consuming hanger tension process 

could be avoided and construction 

speed could be accelerated. It took 

only a week to complete the installation 

of 244 hangers in Sanchaji Bridge. 

(d) Concrete pouring and pumping. 

The hydration problem of huge concrete 

structures such as pile caps and 

crossbeams was managed by adjusting 

the mix ratio and the water circulation 

system inside and heat insulation 

system outside the concrete. Concrete 

pumping to high places such as the top 

of the main tower was addressed by 

improving workability of concrete by 

adding silica to the pipeline. 

Conclusion 

During the design process of Sanchaji 

Bridge, a series of theoretical analyses 

and model test were carried out so that 

it provided a solid basis for the construction. 

Although sound structural 

performance and attractive appearance 

make self-anchored suspension 

bridge a competitive proposal within 

a certain span, such bridge types have 

not come so much into practice yet. 

Sanchaji Bridge (Fig. 7), as the longest 

self-anchored suspension bridge in the 

world, will provide a valuable experience 

and reference for the design and 

construction of such bridge types in 

the future. 

References 

[1] Song X, Dai G, Fang S. Conceptual design 

and construction method of Sanchaji Bridge. 

18th China Bridge Engineering Symposium, 

Tianjin, China, 2008 (in Chinese). 

[2] Song X, Dai G. The main cable shape control 

and design of Sanchaji bridge. IABSE Rep. 2007; 

93: 586–587. 


Owner: 

The Administration of 2nd Ring Road 

(China) 

Contractor: 

The 5th Co. Ltd of China Major Bridge 

Engineering Group 

Designer: 

Changsha Planning and Design 

Institute Co. Ltd/Central South 

University (China) 

Steel (t): 13 182 

Concrete (m 3 ): 18 970 

Estimated cost 

(USD million): 

approx.51,8 

Service date: September 2006 


The First Extradosed Bridge in Slovenia 

Viktor Markelj, Structural Eng., Manager, PONTING d.o.o. Maribor; Lecturer, Faculty of Civil Engineering, 

University of Maribor, Slovenia. Contact: viktor.markelj@ponting.si 

Abstract 

This paper presents the conceptual 

design and technical solutions for the 

bridge over the artificial lake of Ptuj 

(power plant reservoir on the Drava 

River) on the new south main road, 

which connects the city of Ptuj with the 

Maribor–Zagreb motorway. 

The horizontal axis of the 430,0 m long 

and 18,0 m wide bridge has a radius of 

curvature of R = 460,0 m. The bridge is 

designed as an “extrados bridge,” representing 

an intermediate form between 

girder bridges and cable-stayed bridges. 

Structural spans of the bridge are 65 + 

100 + 100 + 100 + 65 = 430 m. In the cross 

section the bridge is a mono-cellular, 

box-shaped, longitudinally prestressed 

concrete girder; H = 2,70 m, with classic 

prestressing tendons inside and additional 

exterior extrados cables over the 

low pylons. 

The conceptual design won the open 

design competition in 2004 and the final 

tender design was completed in March 

2005. The construction of the innovatively 

designed “extrados bridge” started 

in November 2005, and the bridge was 

opened for traffic in May 2007. 

Keywords: extrados bridge; prestressed 

concrete; stay cables; deviator; saddle. 

Introduction 

In May 2004, the municipality of 

Ptuj and Motorway Company in the 

Republic of Slovenia announced an 

anonymous competition to select the 

best design solution for a bridge over 

the Drava River near Ptuj, the oldest 

city in Slovenia. On the basis of the 

winning design, the tender design was 

prepared in 2005; later in September 

that year, and the construction contract 

for EUR 8,8 million was signed. 

The bridge, named “Puch Bridge” after 

the famous Slovenian inventor, was 

opened to traffic in May 2007. 

Environmental Conditions 

and Other Restrictions 

The Ptuj Lake is the largest artificial 

lake in Slovenia, with a length of 

over 5 km, width up to 1,2 km and 

depth up to 15 m. At the location of 

the bridge, the lake is 250 m wide and 

about 5 m deep. The surrounding terrain 

is very flat. The main problem was 

the lake itself, particularly the shores 

that house all the facilities and equipment 

for the operation of the reservoir 

(safety embankments, sealing curtain 

and drainage), the sewer pipes on both 

sides and the stream beyond the right 

bank of the lake. 

For specific locations, it was necessary 

to take into account the relative vicinity 

of the old town of Ptuj with its old 

castle. This being a site of historical 

heritage, very strict restrictions were 

imposed in order to preserve the views 

of the old city. This limited the height 

of structures (in this case, pylons) to a 

maximum of 10 m. 

The following were other significant 

restrictions and conditions that 

affected the technical design of the 

bridge: 

– severe restrictions on the support 

layout due to the slurry wall on the 

banks of the lake; 

– road geometry with a sharp radius 

of curvature of R = 460 m; 

– low elevation of the bridge and the 

required waterway and shipping 

clearance of 4,0 m underneath; 

– difficult foundation conditions in 

the lake because of the rocky bed 

located 20 to 30 m below the water 

level; 

– other obstacles (existing sewage and 

other municipal water ditches and 

streams, the existing discharge facility, 

sharp crossing of the road with 

the left bank of the river); 

– two-way carriageway of width 

8,10 m and a separate lane of width 2 

× 3,10 m for pedestrians and cyclists. 

Design Concept 

The design concept of the bridge was 

the result of a search for the optimal 

response to the very difficult conditions 

for spanning the lake. Owing to 

the many restrictions, the bridge concept 

required large spans, but at the 

same time, a relatively shallow structure 

was necessary. 

The longitudinal disposition with the 

spans of 65 + 100 + 100 + 100 + 65 = 

430 m proved to be the best solution 

for bridging the obstacles. There was 

no dilemma in arriving at the cross 

section: taking into consideration the 

road curvature and torsion, bridge 

installation and maintenance, a box 

girder of maximum feasible height of 

2,70 m was selected. The slenderness 

ratio L/H = 100 m/2,7 m = 37 was obviously 

too large for a normal concrete 

continuous girder structure of constant 

depth. The usual prestressed concrete 

girder with a variable height, constructed 

by the free cantilever method, 

would require a girder height of 5,0 to 

5,8 m (L/20–L/17) at the support. This 

would close the navigation channel 

and was not acceptable for this reason. 

It was therefore necessary to support 

the structure from above, for example, 

by cables through the pylon. However, 

taking into consideration the cultural 

heritage of the region and the need to 

protect the view of the city, a restriction 

of a maximum pylon height of 

10 m was imposed on structural elements 

located on the upper side of the 

carriageway. 

The solution to this problem was the 

use of a new system in bridge building, 

the so-called “extrados” bridge, which 

is a kind of intermediate step between 

the cable-stayed and the girder bridges. 

In this case, the bridge girder could 

be more slender than a normal continuous 

girder, and the pylon could be 

lower than the pylon for cable-stayed 

systems. 

Structure Description 

The bridge length between the expansion 

joints is 433 m and the total width 

is 18,70 m. The static system is a continuous 

externally prestressed box structure 

2,7 m high, constructed according 

to the so-called system of the “extrados” 

bridge with structural spans of 65 

+ 100 + 100 + 100 + 65 = 430 m. In its 

entire length, the bridge lies in the curvature 

R = 460 m (Fig. 1). 

The entire structure is continuous with 

expansion joints only on abutments. 

Two central supports have longitudinally 

fixed bearings and carry all horizontal 

loadings, while the transverse 

horizontal loadings are carried by all the 


430 

65 

100 

100 

100 

65 

1 2 

3 

4 

5 6 

Channel River Drava lake Stream 

1 

2 

100 

3 

430 

100 

4 

100 

5 

6 

65 

65 

3,1+01,80 

3,2+01,80 

3,0+01,80 

3,3+01,80 

2,9+36,80 

3,3+66,80 

Fig. 1: Spans 65 + 100 + 100 + 100 + 65 = 430 m in curvature R = 460 m 

supports. Three of the four intermediate 

supports are located in the reservoir. 

The Substructure and Foundation 

The substructure consists of two abutments 

and four intermediate piers, 

three of which are located in the reservoir 

and one on the shore. All the 

supports are on deep foundation with 

piles of diameter Φ 1,50 m. Owing to 

the low and relatively stiff supports, a 

rigid connection between the piers and 

the girder is not possible; however, all 

the supports have bearings at the top. 

Supports 3 and 4 have fixed pot bearings 

at the top, while supports 1, 2 and 

5, 6 have longitudinally movable and 

transversally fixed bearings. 

The bearings summary is given in 

Table 1. 

The intermediate piers are relatively 

low, 2,4 m thick and about 8 m wide, 

of a full cross section and are fixed 

into the pile head at the bottom 

(Fig. 2). 

At the foundation in the lake, the soil 

mechanics data had to be taken into 

account as well as the execution feasibility, 

structure maintenance and the 

hydraulic consequences. For piers, the 

foundation on eight piles of diameter 

Φ1,50 m with the oval-shaped “floating” 

pile head was foreseen in the 

design. As the superstructure lies on 

a curve, the piles are arranged asymmetrically 

so that they are uniformly 

loaded under the permanent load (the 

intermediate supports are more loaded 

on the inner side of the curvature—on 

the bearings, the ratio is 58–42%). At 

the pile head surface, it was simple 

to perform the temporary supporting 

during the cantilever construction. 

Support in axis 

Bearing (kN) 

Outside curve 

Inside curve 

Abutment1 PNe 6100 PNe 5400 

Pier 2 PNe 21 700 PNe 28 200 

Pier 3 PN 23 000 PN 28 200 

Pier 4 PN 23 000 PN 28 200 

Pier 5 PNe 21 700 PNe 28 200 

Abutment 6 PNe 6100 PNe 5400 

PNe—pot bearing movable in longitudinal direction. PN—pot bearing fixed. 

Table 1: Bridge bearings 

Superstructure 

The superstructure consists of three 

main elements: 

– girder roadway structure; 

– low pylons; 

– inclined cables. 

The roadway structure is a trapezoidal 

prestressed RC box of structural 

height 2,70 m (Fig. 3). The web thickness 

is 0,50 m, the upper slab is 18,16 m 


3,0% 

Drazenci 

Ptuj 

3,0% 

Fig. 2: Characteristic river pier 

wide and the bottom slab 9,10 m. The 

slab thickness varies from 0,22 to 0,50 

m. The cantilever is 4,24 m long, at the 

fixing point 0,50 m thick and rebar 

reinforced. Owing to the flow of compression 

stresses, the thickness of the 

lower slab is enlarged to 0,80 m at the 

supports. The structure is made of concrete 

of compressive strength C45/50. 

The superstructure is prestressed with 

internal bonded tendons (negative in 

the cantilever and positive in the span) 

3,0% 

Drazenci 

2 

Ptuj 

Drava Drava Water 

Gravel 

Sand/clay 

Gravel 

Sand/clay 

Marl base 

18,70 

0,50 3,10 1,70 

4,05 

4,05 1,70 3,10 0,50 

3,0% 

Drazenci 

Ptuj 

3,0% 3,0% 

Fig. 3: Stay cables anchored adjacent to concrete webs (Units: m) 

k.niv. 

2,70 

and additionally with 2 × 5 pairs of 

extrados cables, 31 strands of diameter 

15,7 mm. They support the box at 

a spacing of 5 m, adjacent to the webs, 

so that the force flows directly into the 

longitudinal load-carrying system and 

represents no problem. 

The superstructure also includes short 

pylons of total height 8,5 m (L/11,8), 

two on each support. The pylons with 

the cross section 1,20 m/2,80 m are 

inclined outwards with an inclination 

7,5 : 1, so that the cables, because of the 

route curvature, do not interfere with 

the clearance. The deviators for extrados 

cables in each pylon are designed 

to make possible the replacement of 

individual cables. The pylons are vertically 

prestressed on the tension side 

with dywidag bars with Φ = 40 WR 

(950/1050 MPa), with 11 bars on the 

inner side of the radius, and with 5 bars 

on the outer side. Thus the loading due 

to the structure curvature is compensated. 

The strength of the concrete for 

the pylons is C45/55. 

Inclined “Extrados” Cables 

In addition to the bonded tendons 

(negative and positive) in the girder 

and vertical prestressing bars in short 

pylons, there is also the third type of 

prestressing element, namely, low 

inclined stays, also known as extrados 

cables. To maintain uniformity, all the 

extrados cables are of the same bearing 

capacity and with equal number 

(31) of bearing strands with the section 

150 mm 2 . The cables are 46 to 88 

m long with the stressing force from 

3600 to 3800 kN. 

Each inclined cable consists of the 

following: 

– the free length; 

– two equal anchorages on the girder 

beam; 

– deviator or saddle located on the 

short pylon (Fig. 4). 

Free Length 

Extrados cables consist of 31 monostrands 

of 15,7 mm (cross section 

150 mm 2 ) in the outer PE protection 

pipe that has been grouted with the 

cement mortar. 

The composition of the free length of 

the inclined cable is as follows: 

– Bearing element: seven-wire strand 

with nominal diameter of 15,7 mm 

and the section 150 mm 2 and of 

quality 1570/1770 MPa with a very 

low relaxation (

φ193,7 

Fig. 4: Short pylons with saddles and vertical prestressing bars 

– static test on individual strands; 

– dynamic test on individual strands 

according to fib recommendations 

for extrados cables, which means the 

stress modification Δs = 140 MPa 

with s k = 0,55f u and in 2 × 10 6 loading 

cycles. 

The durability of the inclined stays 

anchorages is assured by the material 

selection and the selected corrosion 

protection system. The anchorage head 

and the winding nut are made of rustresistant 

alloy, all the other anchorage 

elements are hot galvanized. The inner 

and upper part of the anchorage and 

cable are additionally grouted with 

cement mortar; the head, wedges and 

strands, making possible subsequent 

prestressing, are grouted with greases 

in the protection cap. 

550 

550 

550/550 

After construction, the space between 

the monostrands and external 

PE cover was grouted with cement 

mortar to achieve an additional 

mechanical resistance and prevent 

water condensation in the pipe. 

Anchorage 

Each inclined stay has two anchorages 

into the girder structure. The typical 

anchorage VT 31-150 SK has been 

used with a minor modification at the 

external head protection (Fig. 5). 

6260 

Fig. 5: Stay-cable anchorage system (Units: mm) 

φ323,6 

PEHD φ180 

Fig. 6: The saddle consisting of two bent steel tubes (Units: mm) 

2200 

φ193,7 

φ323,6 

PEHD φ180 

The anchorage and system have been 

used and verified on numerous bridges 

with inclined stays, having more strict 

requirements than those with extrados 

cables. In addition, dynamic tests 

of the entire anchorage for s k = 0,45f u , 

Δs = 200 MPa and 2 × 10 6 cycles were 

carried out. 

As the system used had additional 

modifications (galvanized wires) especially 

for the bridge over the Drava 

River, two additional tests were performed, 

with positive results: 

φ355,6 

φ180 

Saddle 

For anchoring and the transition 

through the pylon respectively the 

so-called deviating saddle or deviator 

was used. This solution is being used at 

extrados bridges, as the cable-breaking 

angle is smaller than in cable-stayed 

bridges. In this case, the curvature 

radius of 4,60 m for a length of about 

2,2 m was used. 

The deviator consists of two bent steel 

pipes, namely, of the external pipe 

(Φ = 323,6/7,1 mm) and the internal 

deviator pipe (Φ = 193,7/5,6 mm). 

Within the deviating pipes, the course 

of the settled parallel strands was 

achieved with PE spacers, assuring distance 

between individual strands and 

the pipe. Inside the deviation pipe, the 

PE cover and the grease were removed 

from the monostrands. The prepared 

strands were thereafter grouted with a 

fast curing acrylic resin mortar, which 

is chemically neutral to the galvanized 

wires. 

The detail of the deviator is shown 

in Fig. 6. The external pipes for the 

deviators were inserted in a group 

into the pylon with the auxiliary steel 

structure, assuring a more exact placement 

into the inclined pylon, according 

to the shop drawings. Owing to the 

horizontal curvature of the road and 

the vertical curvature of the elevation, 

each deviator in the pylon has its own 

position. 

Construction 

The bridge erection, lasting from 

October 2005 till May 2007, was a very 


demanding task. The foundation in 

the lake was carried out with artificial 

islands that were built with the help of 

steel sheet piles. This was followed by 

deep foundation with piles of diameter 

Φ 1,50 m and of length 25 to 30 m, 

reaching the marl base. The transportation 

of equipment and material on 

the lake was carried out through heavy 

barges. 

Fig. 7: The balanced cantilever construction with the help of “extrados cables” 

Fig. 8: Bridge construction 

The bridge deck was built by the balanced 

cantilever method with the help 

of inclined extrados cables. Movable 

scaffolding with segment length of 5 m 

was used for erection. Since the superstructure 

is of a constant height, the 

scaffold was not modified as for the 

usual variable height girders. The setting 

and prestressing of stay cables was 

performed. Doing this saved time, consequently 

reducing the duration of a 

single construction stage to one week, 

which is normal. 

One of the particularities of the 

construction was the “extrados 

cables”, which were assembled near 

the bridge and then mounted as a 

whole through saddles in the pylon 

to anchorages (Fig. 7). Prestressing 

was performed with mono jacks from 

both sides, using the strand-by-strand 

method. 

The stressing protocol ensured that 

each strand in the stay cable was tensioned 

with the same force at the end 

of the tensioning procedure. Cables 

have an ultimate load capacity of 8200 

kN; in service stage being loaded with 

approximately 4000 kN. 

Fig. 9: Completed bridge 

Very complex was also the monitoring 

of deformations in construction 

stages, due to the flexible cantilevered 

deck of the bridge (Figs. 8-11). 

The construction of such a bridge 

would not have been possible without 

the most modern software solutions 

in the field of cable-stayed bridges 

and the most accurate surveying 

equipment. 

Conclusion 

Fig. 10: Bridge over the Drava River at Ptuj, Slovenia 

The solution employing the so 

called “extradosed bridge” system 

managed to ingeniously overcome 

numerous restrictions and difficulties 

imposed on the Drava River 

spanning. The sharp curvature, which 

confronted design and building with 

a very challenging task, was successfully 

carried out by using some new 

design solutions and original structural 

details. 



Owner: 

DARS, d.d., Motorway Company, 

Celje, Slovenia 

Structural design: 

PONTING, d.o.o., Maribor, Slovenia 

Contractors: 

SCT, d.d., Ljubljana, Slovenia 

Porr AG, Wien, Austria 

Fig. 11: Night view of the heritage town and its new bridge 

The bridge provided the city of Ptuj 

a solution for their traffic problems 

and a unique engineering structure. 

In addition to that, the bridge does 

not compete with the city’s urban 

cultural heritage; moreover, it has 

the potential to become a new architectural 

landmark and symbol for the 

modern city of Ptuj on its southern 

border. 

Bridge type: Concrete extrados bridge 

Bridge size: L = 433 m, 

W = 18,70 m, A = 8097 m 2 

Concrete (m 3 ): 9488 

Reinforced steel (t): 1300 

High quality steel 

for cables (kg): 189 900 

High quality steel 

for stays (kg): 99 060 

Total cost (EUR million): 8,8 

Service date: May 2007 

SED 12 - New Structural Engineering Document Published 

IABSE 

This SED book provides case studies of structural rehabilitation, repair, 

retrofitting, strengthening, and upgrading of structures. Selected studies 

are presented in this SED and cover a variety of structural types from 

different countries. 

This document is a summary of practices to help structural engineers. 

The reader will discover different approaches to put forward strengthening 

or rehabilitation projects. Even identical technical problems 

could have very different efficient solutions, as discussed in the papers, 

considering structural, environmental, economic factors, as well as 

contractor and designer experience, materials, etc. 

Price: 

CHF 40 for Members, CHF 70 for Non-Members 

www.iabse.org/publications/onlinshop 

Structural Engineering Documents 

12 

Case Studies of 

Rehabilitation, Repair, 

Retrofitting, and 

Strengthening 

of Structures 

International Association for Bridge and Structural Engineering 

Association Internationale des Ponts et Charpentes 

Internationale Vereinigung für Brückenbau und Hochbau 

IABSE 

AIPC 

IVBH 

Structural SED 12.indd Engineering 1 International 4/2010 Technical 28/10/10 Report 11:28 467 AM

Recent PhD Abstracts 

Structural Engineering International (SEI) would like to help you stay up-to-date with some of the exciting and cutting 

edge research being carried out at universities around the world. Our new rubric, “Recent PhD Abstracts,” will provide 

a window onto recent research activities, while giving recent PhDs a chance to disseminate their results quickly. Readers 

wishing to follow up on the information presented in the abstracts will be encouraged to contact the authors, thereby stimulating 

direct contact between researchers and those most interested in their results. The full-length manuscript will also 

be available using the DOI or the URL that will be given with most of the abstracts. Submission form can be downloaded 

from: www.iabse.org/journalsei/asanauthor 

Simple Models for Analysis of 

in-plane Loaded Masonry Walls 

Author: Dr. Alvaro Viviescas Jaimes, 

Columbia 

Email: aviviescasj@hotmail.com 

Supervisor: Dr. Pere Roca Fabregat, 

Universitat Politècnica de Catalunya, 

Barcelona, Spain 

URL: www.tesisenxarxa.net/TESIS_ 

UPC/AVAILABLE/TDX-1229109- 

130304//AVJ1de1.pdf 

Language of this document: Spanish 

A simplified method for the analysis 

of the ultimate capacity of walls subjected 

to in-plane forces is presented. 

The method is based on simple equilibrium 

models representing the combination 

of compression or tension 

stress fields mobilized at the ultimate 

condition. The thesis presents tentative 

rules for the construction of the 

models with some specific models proposed 

for walls subjected to different 

loading conditions. The performance 

of the proposed models is analyzed 

by comparison with numerical results 

generated by means of the well known 

micro-modeling approach specifically 

developed for the analysis of masonry 

structures. 

These simple models are based on the 

struts and ties method which uses the 

struts to represent the compressive 

stress fields and the ties for tension 

zones, both of them forming a resistant 

mechanism (Fig. 1). The objective 

was to predict satisfactorily, by means 

of the numerical micro-model, the ultimate 

loads and mechanisms of fracture 

observed in the experimental tests on 

masonry walls. Calibration and validation 

by means of experimental results 

and numerical simulation constitutes 

an essential and necessary feature of 

the method. 

The examples of application presented, 

corresponding to walls subjected to 

uniform vertical loading or confined 

walls subjected to uneven vertical loading, 

illustrate the ability of the models 

Numerical model 

Fig. 1: Numerical and Simple Model 

to satisfactorily predict the maximum 

horizontal forces that the walls can 

resist. More specifically, the models 

have been able to provide good estimations 

of reference strength values 

obtained by means of an up-to-date 

micro- model. The parametric studies 

carried out show that the simple models 

proposed take into account adequately 

the influence of the geometry 

(in particular, the width to height ratio) 

and the main material properties (the 

masonry compression strength and 

joint-unit frictional properties) on the 

in-plane strength of the wall. 

Railway Bridge Response to 

Passing Trains: Measurements 

and FE Model Updating 

Author: Dr. Johan Wiberg, Sweden 

Email: jw@kth.se 

Supervisor: Prof. Raid Karoumi, 

Prof. Håkan Sundquist; KTH Royal 

Institute of Technology, Dept. of Civil 

and Architectural Eng., Division of 

Structural Design and Bridges, 

Stockholm, Sweden 

URL: http://kth.diva-portal.org/smash/ 

get/diva2:241373/FULLTEXT02 

Language of this document: English 

Existing railway bridges are being 

analysed in detail for their response 

a 

r c 

b c 

x f 

Simple model 

to moving loads due to the increase in 

speeds and axle loads. These numerical 

analyses are time consuming as 

they involve many simulations using 

different train configurations at different 

speeds as well as many other 

considerations. Thus, simplified models 

are often chosen for practical and 

time efficient simulations. The New 

Årsta Railway Bridge in Stockholm 

was successfully instrumented during 

construction and a simplified 3D beam 

element FE model was prepared. The 

model was first manually tuned based 

on static load testing. The most extensive 

work was performed in a statistical 

identification of significantly 

influencing modelling parameters to 

be included in an optimised FE model 

updating. The amount of parameters 

included in the optimisation was in 

this way kept at an optimally low level. 

For verification, measurements from 

several static and dynamic field tests 

with a fully loaded macadam train and 

Swedish Rc6 locomotives were used. 

The implemented algorithms were 

shown to operate efficiently and the 

accuracy in static and dynamic load 

effect predictions was considerably 

improved. It was concluded that the 

complex bridge can be simplified by 

means of beam theory and an equivalent 

modulus of elasticity for simplified 

global analyses. That modulus was 

b p 

468 Recent PhD Abstracts Structural Engineering International 4/2010

in this case approximately 25% larger 

than the specified mean value for the 

concrete grade in question. The optimised 

FE model was used in moving 

load simulations with high speed train 

loads according to the design codes. 

Typically, the calculated vertical acceleration 

of the bridge deck was lower 

than the allowable code value. This 

indicates that multispan continuous 

concrete bridges are not so sensitive 

to train induced vibrations and may be 

suitable for high speed traffic. Finally, 

the relevant area of introducing the 

proposed FE model updating procedure 

in the early bridge design phase 

is outlined. 

Residual Sresses in Stainless 

Steel Box Sections 

Author: Dr. Michal Jandera, Czech 

Republic 

Email: michal.jandera@fsv.cvut.cz 

Supervisor: Prof. Ing. Josef Machacek, 

Dr. Sc, Czech Technical University, 

Prague 

URL: www.ocel-drevo.fsv.cvut.cz/ 

ODK/cz/docs/Disertace/Disertace- 

Jandera.pdf 

Language of this document: Czech 

The investigation is focused on stainless 

steel cold rolled SHS. In total, 14 

SHS stainless steel stub columns were 

tested for subsequent FEM validation. 

The material properties of the flat and 

corner areas of the sections as well as 

the initial deflections of all plates were 

investigated and compared with existing 

predictive formulae. The research 

embraces experimental investigation 

of residual stresses induced by the 

forming process of the sections. The 

residual stress pattern in the sections 

was determined according to the sectioning 

method for the longitudinal 

and as well as the transversal direction. 

In addition, trough-thickness 

measurements using an X-ray diffraction 

method were performed. Patterns 

of the membrane and bending stress 

distribution along the section were 

generalized and suitable predictive 

formulas for general use were developed. 

A high correlation between the 

suggested formulas and measured 

pattern is demonstrated. The patterns 

were finally used in FEM parametric 

studies to show the possible influences 

of each part of the residual stresses on 

the compressive strength of sections 

subjected to local and global buckling. 

The studies are assessed for all substantial 

parameters such as web plate 

slenderness, column slenderness and 

also for the nonlinearity parameter of 

the Ramberg-Osgood formula. Paradoxically, 

it was found that inclusion 

of residual stresses in stainless steels 

(unlike common carbon steels) generally 

led to an increase in the loadcarrying 

capacity. This was attributed 

principally to the influence of the 

bending residual stresses on the material 

stress–strain curve. It was found 

that despite the secant modulus being 

consistently reduced in the presence 

of the residual stresses, the tangent 

modulus was increased in some 

regions of the stress–strain curve. For 

cases where column failure strains 

coincided with these increased tangent 

modulus regions (which was over the 

majority of the practical slenderness 

range, except large column slenderness), 

higher buckling loads resulted. 

Design Oriented Constitutive 

Model for Steel Fiber 

Reinforced Concrete 

Author: Dr. Filipe Laranjeira de 

Oliveira, Spain 

Email: filipe.laranjeira@gmail.com 

Supervisor: Prof. Antonio Aguado, 

Prof. Climent Molins, Universitat 

Politècnica de Catalunya 

URL: www.tesisenxarxa.net/TDX- 

0602110-115910/ 

Language of this document: English 

In the last years, industry has been 

demanding the use of steel fiber reinforced 

concrete (SFRC) in structural 

applications (Fig. 2). Because the postcracking 

strength of this material is not 

negligible, the crack-bridging capacity 

provided by fibers may replace, 

partially or completely, conventional 

steel reinforcement. Therefore, an 

appropriate characterization of the 

SFRC uniaxial tensile behavior is of 

paramount interest. However, in spite 

Fig. 2: Uniaxial tensile test on concrete 

specimen with 60kg/m 3 of steel fiber 

reinforcement 

of the extensive research and recently 

advanced standards, there is no agreement 

on the constitutive model to 

be used for the design of SFRC. The 

crack-bridging capacity provided by 

steel fibers improves both the toughness 

and the durability of concrete. 

Conventional SFRC is a material that 

presents a softening response under 

uniaxial tension, but may develop 

hardening behavior in bending due to 

its ability to redistribute stresses within 

the cross-section. This evidence has 

contributed to an increasing interest 

and growing number of applications 

of this material. In this doctoral thesis, 

a direct and rationale approach to predict 

the tensile response of SFRC for 

structural design calculations is developed. 

The proposed design-oriented 

constitutive model differentiates itself 

from previous studies in multiple 

aspects and defines a new philosophy 

for the design of SFRC elements. This 

model provides a direct and practical 

procedure to obtain the material’s 

tensile behavior by means of parameters 

with physical meaning and based 

on clear concepts: fiber pullouts and 

orientations. One of the major contributions 

of this work is the ability to 

predict the stress-crack width curves 

that reflect the specific combination of 

the properties of the matrix and fibers 

applied. Furthermore, it introduces 

a novel philosophy for the material 

design by taking into account influences 

from the production process, 

fresh-state properties and the element 

to be built in order to define the constitutive 

diagram. 

Structural Engineering International 4/2010 Recent PhD Abstracts 469

Eminent Structural Engineer: Christian Menn—Bridge 

Designer and Builder 

Eugen Brühwiler, Prof., Dr. Civil Eng., ETH, Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland. 

Contact: eugen.bruehwiler@epfl.ch 

Brief CV 

1927 Born on 3 March in 

Meiringen, Switzerland. 

1950 Graduated from the 

Swiss Federal Institute 

of Technology (ETH) 

Zurich with a degree in 

Civil Engineering. 

1953 Accepted a position as 

assistant to Professor 

Pierre Lardy at ETH 

Zurich, where he 

completed his doctoral 

degree in Civil 

Engineering in 1956. 

1956 Worked as an engineer 

with the French 

construction contractor 

Dumez, on the 

construction of the 

UNESCO building in 

Paris. 

1957 Established his own 

engineering design 

office, specializing in 

reinforced concrete 

construction, in Chur, 

Switzerland. 

1959 Built his first bridges at 

Letziwald and Cröt in 

Switzerland. 

1971–1992 Was Professor of 

Structural Engineering 

at the Swiss Federal 

Institute of Technology 

(ETH) Zurich. 

Since 1992 Commenced private 

practice as a consulting 

engineer. 

1996 Received honorary 

doctorate from the 

University of Stuttgart. 

2008 Received honorary 

doctorate from the Ecole 

Polytechnique Fédérale 

de Lausanne (EPFL). 

2009 Received the 

International Award 

of Merit in Structural 

Engineering from IABSE. 

Fig. 1: Christian Menn 

Introduction 

Christian Menn (Fig. 1) ranks among 

the most important and creative bridge 

design engineers of the recent decades. 

His bridges are witnesses of his exclusive 

engagement in bridge engineering 

and more than 50 years of continuous 

experimentation in conceptual, 

structural and aesthetic design. They 

teach us that the true art of structural 

engineering is characterized by 

innovation and imagination with the 

objective to improve the environment 

through structural art. His work has 

been exhibited at various universities 

in Europe and at art museums 

across the United States, and attests to 

his extraordinary creativity and absolute 

mastery of bridge engineering. 

Fig. 3: Reichenau bridge at Tamins, Switzerland, 1963 

Bridges in Switzerland 

The first bridges designed by him 

(Fig. 2) were inspired by the innovative 

bridge forms—the three-hinged hollow 

box girder and the deck-stiffened 

arch—of his Swiss predecessor Robert 

Maillart. After 1960, Menn developed 

his own style of the arch bridge. This 

consists of a monolithic frame system 

composed of a thin polygonal 

arch supporting slender cross walls 

at a wide spacing, and a stiff partially 

prestressed box girder deck with a 

wide cantilevered roadway. The most 

prominent example is the 100 m span 

Reichenau Bridge (Fig. 3) over the 

Rhine River, which represents a new 

form for deck-stiffened arch bridges 

with long spans. Menn has used this 

Fig. 2: Letziwald bridge at Avers, Switzerland, 

1959 

470 Eminent Structural Engineer Structural Engineering International 4/2010

system for several other arch bridges, 

such as the Viamala Bridge and the 

Nanin and Cascella bridges on the 

Moesa River, built between 1966 and 

1968 in the canton of Graubünden. 

In 1970, in collaboration with an engineering 

firm, Menn won a competition 

to build the six-lane 1100 m long 

Felsenau highway bridge across the 

Aar Valley in Bern. He designed a 

structure in prestressed concrete with 

two central spans of 156 m each. The 

hollow box girder with inclined webs 

and variable depth is curved and supports 

a 7,6 m wide cantilevered deck 

slab. The bridge is characterized by 

its slender appearance and impressive 

aesthetic despite its large size (Fig. 4). 

In 1971, Menn accepted a position as 

professor at ETH Zurich. He left his 

engineering office but continued to 

design exceptional bridges throughout 

Switzerland in collaboration with 

different engineering firms. He regularly 

sat on juries for bridge design 

competitions. 

Christian Menn designed the Ganter 

Bridge built in 1980 on the Simplon 

Pass road, whose extraordinary silhouette 

presents a new aesthetic for 

Fig. 4: Felsenau bridge at Berne, Switzerland, 

1975 

Fig. 5: Ganter bridge at Eisten, Switzerland, 1980 

structures in the Swiss Alpine environment 

(Fig. 5). The structure consists of 

steel cable stays encased in a sheet of 

concrete and anchored in short pylons 

to form a rigid structure 150 m above 

the valley floor. The Ganter’s central 

span of 174 m is still the longest in 

Switzerland. The distinctive structural 

form arose from Menn’s imagination 

and aesthetic inspirations. 

In the same period, Menn collaborated 

with an engineering firm in Bellinzona 

to win a design competition for the 

Biaschina Viaduct, a highway bridge 

completed in 1983 in Switzerland’s 

Ticino valley. The bridge’s tall columns 

and roadway, built using a balanced 

cantilever construction method, result 

in a form of harmonious proportions. 

Towards the end of the 1980s, in collaboration 

with another engineering 

firm, Menn conceived the Chandoline 

Bridge, a cable-stayed structure over 

the Rhone River at Sion. 

At the end of his academic career, 

Menn designed the Sunniberg Bridge, 

built between 1996 and 1998 to divert 

highway traffic around the town of 

Klosters in the canton of Graubünden 

(Fig. 6). This five-span cable-stayed 

structure is 526 m long and crosses the 

valley at a height of 50 to 60 m. The 

roadway’s strong curvature in the plan 

required that the pylons lean away 

from the bridge deck so that the cables 

did not interfere with the road clearance. 

Because of its strong curvature, 

the bridge deck could be fixed to the 

abutments without dilation joints, and 

longitudinal deck length variation is 

taken by radial displacements of the 

structural system. This is probably the 

first time this solution is applied worldwide. 

The slender pylons, thin deck, and 

harped cables form an elegant ensemble 

and confer a striking aesthetic 

on the bridge, expressing boldness in 

and inviting admiration for the art of 

engineering. In 2001, the Sunniberg 

Fig. 6: Sunniberg bridge at Klosters, 1998 

Bridge was awarded the Outstanding 

Structure Award by the International 

Association for Bridge and Structural 

Engineering (IABSE). 

Bridges Worldwide 

After retiring from professorship in 

1992, Menn began a new vocation as 

consultant for the design of several 

significant bridges around the world. 

This pursuit has allowed him to further 

develop his passion for the art of 

bridge design. A series of innovative 

bridge projects arose from situations 

calling for structures of both functional 

and symbolic importance, notably 

in the United States. 

In 1991, Menn proposed a cable-stayed 

bridge to the civic leaders of Boston as 

a means to satisfy several complex project 

constraints. The bridge’s original 

character stems from the play in the spatial 

configuration of stay cables as experienced 

by automobilists crossing the 

bridge (Fig. 7). The cables on the 227 m 

central span are anchored to the outer 

edges of the bridge deck—at 60 m wide, 

it is the widest cable-stayed bridge in 

the world—while on the end spans, the 

cables are anchored along the median. 

The two “inverted Y-shaped” towers 

symbolize the entrance into downtown 

Boston. The stay cable arrangement 

allows for a significant reduction in 

transverse bending of the pylons due 

to the asymmetric cross section and 

eccentric traffic loads. The concept of 

the bridge structure is the product of 

a design process led by solely optimizing 

the flow of forces while respecting 

the stringent boundary conditions. 

The aesthetic appearance results thus 

simply from an optimized and refined 

structural form. The Leonard P. Zakim 

Bunker Hill Bridge, named after a civil 

rights activist, opened on 12 May 2002 

with more than 200 000 people crossing 

the bridge on foot. The bridge has 

received several distinctions, and in 

honour of the designer, 3 November 

2000 was proclaimed as “Christian 

Menn Day” in Massachusetts. 

Structural Engineering International 4/2010 Eminent Structural Engineer 471

Future Bridge Project 

Fig. 7: Leonard P. Zakim Bunker Hill Bridge in Boston, USA, 2002 

Menn developed the designs of several 

other major bridges in the United 

States. He designed the winning proposal 

for a bridge competition held in 

1998 for the 2 km long Woodrow Wilson 

Bridge crossing the Potomac River, 

south of Washington, DC, but only the 

basic structural system remained from 

Menn’s original design. The originally 

slender, elegant V-shaped piers were 

abandoned in favor of unmotivated 

curved massive elements. 

Since 2001, Menn has developed several 

designs for a new bridge crossing 

the Niagara River near its famous falls. 

These designs for the “Peace Bridge” 

between Buffalo and Fort Erie have 

won the support largely of the public 

and local leaders, but the official process 

of approvals is still ongoing. 

Fig. 8: Streicker Pedestrian Bridge over 

Washington Road on the Princeton 

University Campus, Princeton, USA, 2009 

Fig. 9: Grimsel lake bridge at Guttannen, Switzerland (Project) 

Menn’s conceptual designs for a bridge 

crossing the valley in front of Hoover 

Dam near Las Vegas, the east span of 

the San Francisco Oakland Bay Bridge, 

a double self-anchored suspension 

bridge for crossing the Ohio River at 

Louisville and a new Inner Belt Bridge 

in Cleveland were unfortunately all 

not considered, sometimes because of 

their seemingly too innovative nature, 

and more conservative approaches 

were preferred instead. 

On the Princeton University campus, 

Menn’s design of the slender pedestrian 

bridge passing over Washington 

Road took the form of a bold arch. The 

Streicker Bridge, completed in 2010 and 

named after its principal benefactor, 

links four buildings designed by internationally 

acclaimed architects (Fig. 8). 

In 2007, Christian Menn was asked 

to propose bridge designs to provide 

access to a large island to be developed 

in Abu Dhabi in the United Arab 

Emirates. For the two largest most 

em blematic bridges, he proposed an 

arch rising above the roadway surface 

from which the bridge deck is suspended, 

and a cable-stayed bridge with a single 

“spindle-shaped” pylon supporting a 

wide, softly curving roadway just a few 

meters above the water surface. These 

projects are currently in the design 

development phase, and construction is 

expected to begin in the near future. 

In Switzerland, near the birthplace 

of Menn, plans to raise the Grimsel 

dam and lake require redrawing of 

the road over the Grimsel Pass which 

has to cross more than 300 m across 

the artificial lake. In 2005, Christian 

Menn designed a cable-stayed bridge 

crossing the lake with a single span 

of 352 m. This will be the longest in 

Switzerland, supported by two vertical 

75-m high inversed Y-shaped pylons. 

The aesthetic expression is characterized 

by the slender deck and purely 

shaped pylons, as well as the semifan 

arrangement of the stay cables 

and their concrete anchorage blocks 

shaped like mountain crystals (Fig. 9). 

Lessons in Bridge Design 

Christian Menn has designed arch, box 

girder and cable-stayed bridge structures. 

He has developed new ideas for 

each of these types of structural systems 

in order to create unique forms 

and an original aesthetic. His bridges 

are the result of a design process reducing 

structural elements to meet the 

given functional and environmental 

requirements. They are characterized 

by harmonious proportioning of structural 

elements, a slender appearance 

and coherent integration of shapes, 

which continually led to new creations 

and expressions. Menn’s works are 

among the most beautiful bridges built 

in the last 50 years; they are symbols of 

modern technology, and they establish 

a technical aesthetic. 

Menn’s bridges express technical efficiency 

with an accent on slenderness 

and transparency. They emphasize to us 

the importance of understanding how 

structural systems function. A sound 

engineering concept is the solid basis 

for a far-reaching aesthetic quality and 

for finding simple yet elegant structures. 

Guided by the basics of structural 

mechanics and natural sciences, this 

approach continues to be very efficient 

and valuable, in particular nowadays, 

when (architect-led) bridge designs, 

often based on a spectacular metaphoric 

idea rather than on an efficient structural 

concept, have produced structures 

that are excessively expensive to build 

and maintain and thus are controversial. 

In his numerous lectures on bridge 

design, Christian Menn has always 

insisted that the art of structural engineering 

should be appreciated and be 

given much more importance, in particular, 

in bridge design competitions and in 

the education of structural engineers. 

472 Eminent Structural Engineer Structural Engineering International 4/2010

IABSE Annual Meetings 

Venice, Italy, September 19–21, 2010 

SEI Editorial Board and Nominees 

The 2010 Annual Meetings were held 

at the Palazzo del Casino, at the Lido 

in Venice, prior to the 34 th IABSE 

Symposium. There was a good attendance 

with 140 participants and 55 

accompanying persons. Accompanying 

persons discovered Venice and islands 

and enjoyed the social events together 

with committee members: on Sunday 

Ms Evelyne Stampfli, Deputy Consul 

General of Switzerland, gave a reception 

at the Excelsior Hotel. On Monday 

evening, Jacques Combault, President 

of IABSE welcomed all delegates, and 

Carlo Urbano, Chair of the Italian 

Group, in his turn welcomed all to 

Venice with drinks and delicious food 

on Tuesday evening. 

The Annual Meetings included the 

Administrative, Executive, Permanent 

Committee and Chair National Groups, 

Technical Committee, Editorial, Correspondents 

and E-Learning Boards, 

Working Commissions, Work ing 

Groups, Scientific Committees, Young 

Engineers Board, the Advisory Group 

to the Executive Committee and the 

IABSE Foundation Council. 

The Permanent Committee approved 

the annual statement of accounts 

2009 and the budget 2011. The annual 

accounts 2009 closed with total net 

revenues of CHF 989'066 and an 

excess of revenues of CHF 17'505. The 

Association funds amount to CHF 

467'658 as on December 31, 2009. 

For the year 2011 a budget with total 

net revenues of CHF 1'087'700 was 

approved. 

The Permanent Committee changed 

the article 14 of the By-Laws and made 

English the sole official language of 

IABSE as from January 1, 2011. 

Predrag (Pete) Popovic, USA, New President of IABSE 

Jacques Combault ended his term as 

President on October 31, 2010. At the 

Closing Ceremony of the 34 th IABSE 

Symposium he thanked his colleagues 

on the Administrative and Executive 

Committees, and all those who have 

made the Technical Committee more 

dynamic and efficient by encouraging 

the creation of new Working Groups 

and making IABSE E-Learning 

become a reality. During the three 

years of his presidency several successful 

international events were held: 

the Congress in Chicago, Symposia in 

Bangkok and Conferences in Helsinki 

and Dubrovnik, a Bridge Workshop 

and spectacular tour in China. Jacques 

Combault attended four Outstanding 

Structure Award plaque presentations 

and welcomed one new National 

Group to the Association. The 80 

years of IABSE were celebrated by 

making all IABSE publications from 

1929–1999 available for free online 

to the general public. A new more 

dynamic website is on its way to serve 

the Association for an even better 

exchange of structural engineering 

knowledge. 

Pete Popovic took the opportunity 

during the Closing Ceremony to thank 

Jacques Combault and his wife Danièle, 

who has assisted and supported her 

husband during his presidency. 

Pete Popovic from Wiss, Janney, 

Elstener Associates, Inc., USA, has 

taken office as President of IABSE 

on November 1, 2010, for a period of 

three years. He is the second IABSE 

President from USA. Pete Popovic 

is Member of IABSE since 1985 and 

knows the Association well. He has 

been Chair of Working Commission 8, 

Member of the Technical Committee, 

Member of the Outstanding Structure 

Pete Popovic, USA, President of IABSE 

Award Committee and Vice-President 

of IABSE. His contributions to IABSE 

Structural Engineering International 4/2010 Panorama 473

conferences are numerous and he was 

the Chair of the Organising Committee 

for the Chicago Congress in 2008. 

Pete Popovic’s fields of expertise are 

the design, assessment and repairs of 

bridges and buildings. He has in particular 

expertise in assessment and repair 

of concrete structures and of fatigue 

damage in steel bridges, and exterior 

facades of high-rise buildings. 

During the first ten years of practice, 

he participated in structural design 

of major steel bridges and rapid transit 

systems in Chicago, New York and 

Atlanta, USA. He was engaged in the 

design of post-tensioned box girder 

bridges in Kuwait. Over the last 30 

years, he has evaluated and designed 

repairs for over 1500 structures. Major 

projects included assessment of steel 

bridges for fatigue damage, investigation 

of collapses of bridges and buildings, 

assessment and design of repairs 

for exterior facades of high-rise buildings 

up to 60-stories tall, and assessment 

and repair of over 100 parking 

structures. 

Pete Popovic has published over 40 

technical papers on assessment, load 

testing, strengthening and repair of 

bridges, buildings and parking structures 

and is a contributing author to 

Jacques and Danièle Combault and Pete Popovic 

several books. He has received awards 

from the International Concrete 

Repair Institute for innovative repair 

projects and is an invited lecturer at 

the University of Wisconsin, USA and 

World of Concrete (USA and Mexico) 

on topics of concrete repairs, rehabilitation 

of parking structures, and prevention 

of structural failures. 

As President of IABSE Pete Popovic 

intends, to work in making IABSE 

more visible and increasing IABSE 

membership. His goal is to have 

National Groups play an increasing 

role in recruiting new members and 

retaining existing members. The goal is 

to increase IABSE membership by 500 

over the next three years. 

IABSE Awards 2010 

Jacques Combault, President of IABSE, 

presented the IABSE Awards at the 

Permanent Committee (Honorary 

Memberships) and at the Opening 

Ceremony of the 34 th IABSE Symposium 

in Venice on the 21 st and 22 nd of 

September. 

Honorary Membership 

Honorary Membership is presented to 

an Individual Member of IABSE, for 

exceptionally great services rendered 

to the Association. 

The Executive Committee of IABSE 

has awarded Honorary Membership 

to Prof. Aarne Jutila, Finland. The 

President of IABSE presented the 

Award at the Permanent Committee 

meeting on September 21, 2010, 

‘in recognition to his outstanding 

and dedicated services to the 

Association’. 

Aarne Jutila, Finland 

Born 1940 in Helsinki, Aarne Jutila 

received his Civil Engineering degree 

at Helsinki University of Technology 

(TKK) in 1966, with major subject 

“Bridge Engineering”. After 

graduation he studied a year at ETH, 

Zurich, as “Bundesstipendiat” under 

the guidance of Bruno Thürlimann. 

Later he worked as bridge designer 

at Kjessler and Mannerstråle AB 

in Stockholm and Tapiola, Finland, 

as Assistant Lecturer at Queen’s 

University of Belfast, Northern Ireland, 

and as section chief at the Finnish 

Road Administration’s bridge design 

office (TVH) in Helsinki before founding 

of and working for three consulting 

engineering companies. Besides that 

he also worked as assistant, laboratory 

engineer and, since 1984, as Professor 

of Bridge Engineering at TKK. He 

retired in August 2008 and continues 

his bridge engineering activity as 

Managing Director of Extraplan Oy, 

consulting engineers, that he founded 

in 1977. 

Aarne Jutila joined IABSE in 1967, 

and has since then held numerous 

functions within the Association: 

Secretary of the Finnish Group 

1972–88 and Chair since that, Vice- 

Chair of the Organising Committee 

of the 1988 Helsinki Congress, SEI 

474 Panorama Structural Engineering International 4/2010

Correspondent and Member of the 

Editorial Board 1991–2000, Member of 

several scientific committees (Malmö 

1999, New Delhi 2005, Dubrovnik 

2010), Chair of the Scientific 

Committee of the Lahti Conference 

in 2001, Member of the Executive 

Committee and Vice-President 

1999–2007. 

He continues his engagement for 

IABSE: Chair of the Finnish Group, 

Member of the Permanent Committee, 

Member of the Foundation Council 

Board, Member of the Advisory 

Group to the Executive Committee of 

IABSE. 

Honorary Membership 

Hai-Fan Xiang, China 


has awarded Honorary Membership 

to Prof. Hai-Fan Xiang, China. Jacques 

Combault gave a speech at the 

Permanent Committee meeting on 

September 21, 2010, and informed that 

Prof. Xiang, was not able to travel to 

Venice and that Pete Popovic, future 

President of IABSE, would present the 

Diploma to Prof. Xiang at a Ceremony 

at Tongji University, ‘in recognition of 

his outstanding and dedicated services 

to the Association’. 

Hai-Fan Xiang graduated from Tongji 

University in 1955 and acquired 

his master in 1958. Since then he 

has worked at Tongji University for 

more than 50 years. He gained the 

Research Fellowship of Alexander von 

Humboldt Foundation and worked as 

a visiting professor at Ruhr University, 

Bochum, Germany in 1981 and 1982. 

As a pioneer in bridge wind engineering 

in China, he devoted his research 

field to the wind-resistance of longspan 

bridges after returning to Tongji 

University in 1982. He was awarded 

the title of National Outstanding 

Expert in 1986. In 1995, he was elected 

as an Academician of the Chinese 

Academy of Engineering. In 2007, 

he became an Emeritus Professor of 

Tongji University. 

He has been the first Chairman of 

Department of Bridge Engineering, 

the founding Dean of College of Civil 

Engineering, and the Director of the 

State Key Laboratory for Disaster 

Reduction in Civil Engineering. 

He has published 12 books, numerous 

articles domestically and internationally. 

He has received more 

than 20 national awards and several 

international awards including 

the R.H. Robert Scanlan Medal 

of ASCE and the Anton Tedesko 

Medal of the IABSE Foundation 

for the Advancement of Structural 

Engineering. 

Hai-Fan Xiang is President of the 

Insti tution of Bridge and Structural 

Engineering of China, and Chairman 

or Co-Chairman for more than ten 

organisations. He joined IABSE in 

1992 and has dedicated his time to 

the Association on several Scientific 

Committees for conferences held in 

Seoul 2004, Shanghai 2004 (Chair) 

and New Delhi 2005. Former Vice- 

President of IABSE (2001-2009), 

he is currently a Delegate to the 

Permanent Committee and on the 


(SEI) Advisory Board and a Member 

of the Foundation Council of 

IABSE. 

Dagu Bridge, Tianjin, China 

International Award of Merit in 

Structural Engineering 

The International Award of Merit in 

Structural Engineering is conferred 

for outstanding contributions in the 

field of structural engineering, with 

special reference to their usefulness 

to society. Contributions may include 

various aspects in Planning, Design, 

Construction, Materials, Equipment, 

Education, Research, Government, 

and Management. The Executive 

Committee of IABSE has conferred 

the International Award of Merit in 

Structural Engineering to Man-Chun 

Tang, USA, “for blending art and 

engineering, and together successfully 

creating innovative concepts for signature 

bridges that are admired by 

both his peers and the general public 

alike”. 

Man-Chung Tang, USA 

Man-Chung Tang is the Technical 

Director and Chairman of the Board 


of T.Y. Lin International, a globally 

recognised consulting firm with headquarters 

in San Francisco, USA. 

Man-Chung Tang received his Doctor 

of Civil Engineering in 1965 from 

the Technical University Darmstadt, 

Germany. His career spans more than 

44 years, and encompasses designing 

and constructing over 100 bridges 

worldwide, including 32 cable-stayed 

bridges, four major suspension bridges, 

and numerous segmental bridges. A 

true leader and icon, Dr. Tang’s contributions 

to innovations in bridge 

design are demonstrated through 

teaching, writing over 100 technical 

papers, and offering numerous presentations. 

He is an honorary professor at 

ten universities, a member of the U.S. 

National Academy of Engineering, a 

foreign member of Chinese Academy 

of Engineering, and an honorary member 

of the American Society of Civil 

Engineers (ASCE). 

A world authority on cable-stayed 

bridges, Man-Chung Tang served as 

Chairman of the American Society of 

Civil Engineers (ASCE) committee on 

cable-suspended bridges and published 

the definitive guideline for the design 

of cable-stayed bridges, used today by 

engineers all over the world. Dr. Tang 

is also a founding member of the Post- 

Tensioning Institute (PTI) committee 

that published “Recommendations for 

the Design and Testing of Stay Cables,” 

also used worldwide. 

Man-Chung Tang’s bridges, besides 

being safe, functional and economical, 

are considered works of art–beautiful 

structures that blend seamlessly with 

their surroundings. It is often quoted 

that “the sun never sets on a Dr. Tang 

bridge,” as his designs can be found 

all around the globe. Man-Chung 

Tang continues to advance the field 

of bridge engineering as an innovator 

and an educator and he has been continually 

recognised by his peers for his 

dedication to the field. 

IABSE Prize 

The IABSE Prize was established in 

1982 to honour a Member early in 

his, or her career for an outstanding 

achievement in the field of structural 

engineering, in Research, Design or 

Construction. The Prize is presented to 

Individual Members of IABSE, forty 

years of age or younger. 


has presented the IABSE Award 2010 

to Roberto Revilla Angulo, Spain, ‘in 

Roberto Revilla Angulo, Spain 

recognition of his significant involvement 

in many major bridge projects, 

specially for his contribution in the 

design, project of Montabliz Viaduct’. 

Roberto Revilla Angulo, was born in 

Bilbao (Spain) in 1970. He studied at the 

Technical Civil Engineers University 

College of Santander, and graduated 

in 1995. He has since then worked 

with Apia XXI, where he has been the 

Head of the Structures Department 

since 2000. At the time he is finishing 

his Doctoral Thesis “Stability of 

great high piers of bridges built by 

cantilever method”, at the Structural 

and Mechanical Department at the 

University of Cantabria. Montabliz 

Viaduct has allowed him to participate 

in some research works such as special 

studies of earthquake, wind and fire; 

wind tunnel tests; terrain-structure 

interaction studies of the foundation 

of piers and monitoring both static and 

dynamic structural behaviour during 

its construction. 

Important projects he has developed 

include: Navas Viaduct, Caviedes 

Viaduct, Viaduct over Voltoya River, 

New Bridge over Ebro Reservoir, 

Santander’s Bay Ring Footbridge, 

Montabliz Viaduct and New Bridge 

over Llobregat River. He has won 

the Idea Tenders of the New Bridge 

over Guadaira River (Sevilla) and 

the New Bridge over Llobregat River 

(Barcelona). Roberto Revilla’s professional 

passion has always been to 

design bridges taking care of aesthetics 

and in full harmony with the surrounding, 

breaking civil engineering 

architecture barrier. 

Outstanding Paper Award 

The Outstanding Paper Award is 

remitted each year to the author(s) 

of a paper published in the preceding 

year’s issues of the IABSE Journal 


(SEI), encouraging and rewarding contributions 

of the highest quality. It was 

first launched in 1991. 

The Outstanding Paper Award Committee, 

chaired by Professor Akira 

Wada, Japan, has conferred the 

Outstanding Paper Award to Andreas 

Breum Ølgaard, Jens Henrik Nielsen, 

and John Forbes Olesen, Denmark, for 

their paper: 

“Design of Mechanically Reinforced 

Glass Beams: Modelling and 

Experiments” Published in Structural 

Engineering International (SEI) May 

2009. 

The paper is a study on how to obtain 

a ductile behaviour of a composite 

transparent structural element. The 

structural element is constructed 

by gluing a steel strip to the bottom 

face of a float glass beam using 

an epoxy adhesive. The composite 

beam is examined by four point 

bending tests, and the mechanisms 

of the beam are discussed. Analogies 

to reinforced concrete beam theory 

are made; thus, four different design 

criteria, depending on the reinforcement 

ratio, are investigated. Analytical 

expressions are derived that are capable 

of describing the behaviour in an 

uncracked stage, a linear cracked stage 

and a yield stage. A finite element 

model, capable of handling the cracking 

of the glass by killing elements, is 

presented. 

Both analytical and numerical simulations 

are in fairly good agreement 

with the experimental observations. It 

appears that the reinforcement ratio 

is limited by the risk of anchorage 

failure and must be adjusted accordingly 

to obtain safe failure behaviour 

in a normal reinforced mode. Analysis 

of anchorage failure is made through 

a modified Volkersen stress analysis. 

Furthermore, different aspects of the 

design philosophy of reinforced glass 

beams are presented. 

Outstanding Structure Award 

The Outstanding Structure Award 

(OStrA) was established in 1998. It is 

one of the highest distinctions awarded 

by IABSE and recognises, in different 

regions of the world, some of the 

most remarkable, innovative, creative 

or otherwise stimulating structures 

completed within the last few years. 

The Outstanding Structure Award 


Committee is chaired by Mr. William J. 

Nugent, USA. In 2010 the Outstanding 

Structure Award is awarded to. 

The National Aquatics Center, 

Beijing, China, 

for being, “a breathtaking interlocked 

soap bubble architecture of ETFE pillows 

within a polyhedral steel space 

frame resulting in outstanding aesthetic 

harmony of form function and structure 

which is energy efficient and pleasing 

to all”. 

Unusually in this era of architectural 

form making, the Beijing National 

Aquatics Centre, was generated as 

much by engineering intent as for its 

beauty. It is the result of an outstanding 

collaboration between Arup, PTW 

architects and CCDI. The primary 

purpose of the “box of bubbles” is to 

trap as much solar energy as possible 

and use it to both heat the swimming 

pools and light the internal spaces. This 

“insulated greenhouse” saves 30% of 

the required heating energy and 55% 

of the artificial lighting. This energy 

saving is equivalent to cladding the 

whole building with solar panels. It 

also provides a quality of internal light 

and space that needs to be physically 

experienced to be really appreciated. 

The structure is based on a solution 

to the century old mathematical 

“The Water Cube”, China 

conundrum posed by Lord Kelvin: 

“What is the most efficient way 

of subdividing three dimensional 

space?” This puzzle is now thought to 

have been solved by Professor Weaire 

and Dr Phelan, whose foam is also the 

geometry of a perfect array of soap 

bubbles. 

The geometry of Weaire-Phelan foam 

provided a unique structure that may 

be the most earthquake resistant 

building in the world. The building is 

not a pattern applied to a box, but a 

box carved from a theoretically perfect 

and repetitious array of bubbles. 

The final building is satisfying on 

many levels: as a beautiful object; 

a poetic expression of bubbles and 

water; the physical manifestation of 

an abstract theoretical geometry; a 

through thickness pattern; complementary 

to its neighbour, the Bird’s 

Nest, and uplifting to be in. But most 

importantly it is entirely sustainable 

and achieves pure engineering 

objectives. 

Outstanding Stucture Award 

Finalists 

Starting with the 2010 Outstanding 

Structure Award, IABSE is pleased to 

present the three Finalists selected by 

the OStrA Committee. 

Mode Gakuen Spiral Towers, Nagoya, 

Japan. Its design includes three towers 

interwined in a spiral form, suggesting 

the intertwined rising energy of the students 

of Mode Gakuen’s three schools: 

its fashion school (MODE), computer 

and animation school (HAL), and 

medical school (lSEN). 

The building has 36 floors above 

ground, three basement levels, and 

two penthouse levels. Its height is 170 

meters above ground and 21 meters 

underground. A central core having 

an oval cross-sectional shape consists 

of three wings having fan-shaped cross 

sections, radially arranged next to 

each other. The planar configuration 

changes with height. Three classrooms 

are arranged in the respective wings 

around the central core, which 

includes stairwells and elevator shafts. 

Ascending higher in the building in 

a spiral pattern, the rooms gradually 

become smaller in size. Displacement 

of the centers of rotation of the three 

wings produces an external appearance 

of organic curves. 

Twelve straight columns are arranged 

aro und this core, and braces are connected 

to these columns in a mesh 

network, forming the thick central 

trunk of the tubular structure (called 

an “inner truss tube”). This tubular 

structure is highly strong and rigid 

with regard to horizontal and twisting 

forces exerted on the building by 

earthquakes and high winds, providing 

the necessary structural performance. 

With no braces around the outside, a 

transparent appearance is achieved; 

and minimal, thin-diameter columns 

provide lower rigidity for a light frame 

that does not bear seismic forces. 

Mode Gakuen Sprial Towers, Japan 


Sutong Bridge, China. Sutong Bridge 

is located in the southeast of Jiangsu 

Province, China, which is in the lower 

reaches of the Yangtze River. The 

visionary project was motivated by 

the need for a highway route crossing 

the Yangtze River and linking 

Suzhou and Nantong at the opposite 

banks. The total length of the Bridge 

project is 32,4 km, consisting of three 

main parts, the viaducts on both banks 

of the river and the central part over 

the water, which is about 6 km long. 

The central part comprises of the main 

cable-stayed bridge with the world 

record 1088 m span as main navigational 

channel, a continuous rigid 

frame bridge with a main span of 268 m 

as secondary navigational channel, and 

approach bridges. 

The bridge substructure was designed 

and constructed with the emphasis on 

sustainable development or environmental 

protection in the mother river 

of China, Yangtze River. This aim 

was achieved through selecting group 

pile foundation instead of caisson to 

alleviate the impact on the river flow, 

installing various scour protection to 

minimise the erosion in the river bed, 

Sutong Bridge, China 

and adopting partially hydrolysed 

polyacrylamide (PHP) system for clay 

mud treatment to reduce the disposal 

of bored pile construction. One of 

the most significant challenges in the 

construction of this super-long cable 

stayed bridge was geometry control. 

The unique complexity of Sutong 

Bridge required specially developed 

methods and procedures to control 

the geometry profile and safety of the 

bridge during the construction period. 

Heathrow Terminal 5A, UK. The 

156 m clear span roof encloses three 

mio. sq. ft. floor space framed in steel 

over three storeys. The unconventional 

height of the building was in 

response to the challenge of having 

to build within the constraints of two 

runways and the greenbelt beyond 

them. 

The T5A roof is an awe inspiring 

structure that arches over the terminal 

building. The roof carries huge compression 

forces which are essential to 

prevent the buckling of its individual 

parts and of the structure as a whole. 

One of the pioneering analysis techniques 

employed on this project was 

modal buckling analysis. This calculated 

the effective reduction in lateral 

stiffness that is caused by compression 

forces within the structure and 

used eigenvector analysis to predict 

the most critical possible buckling 

modes. The mode shape data was 

then processed to give sets of design 

forces, ensuring a consistent reserve of 

strength against buckling, without providing 

extra strength where it was not 

needed. 

Heathrow T5A, UK 

The central arched section of the roof 

needed to be assembled, clad, and 

pre-stressed at ground level before 

being lifted into position using strand 

jacks. This creative approach was vital 

to ensure that the whole operation 

could be carried out below the airport 

radar ceiling and that the risk from 

working at height would be reduced. 

This idea became an integral part of 

the building design and construction 

planning of the whole Terminal. 


The IABSE Foundation Anton Tedesko Medal 

The Anton Tedesko Medal is awarded 

by the IABSE Foundation for 

the Advancement of Structural 

Engineering. The Award has two components: 

the first is a medal awarded 

to the Laureate in recognition of his 

contribution to the advancement of 

structural engineering. The second 

part is a sum of 25’000 Swiss Francs 

to be used by the Laureate in order 

to organize and finance a study leave 

abroad for a young promising engineer 

(Fellow) outside his/her home 

country with prestigious engineering 

firms. Klaus Ostenfeld, Chair 

of the IABSE Foundation Council, 

conferred the Anton Tedesko Medal 

to Prof. Koichi Takanashi, at the 

Symposium Opening Ceremony 

“in recognition of his dedication to 

excellence in structural engineering 

and his role as a mentor for young 

engineers” 

Koichi Takanashi has supervised many 

students and directed research projects 

at the Universities of Tokyo, Chiba 

and Kougakuin. His research has been 

focused on plastic design and seismic 

design. One of his outstanding accomplishments 

was the establishment of 

the overall testing method to combine 

numerical analysis of structural 

system in the computer and the test 

of structural frames in 1974. This 

method has advanced research to precisely 

understand the behaviour of 

structural frames. This method is now 

widely used in the world and developed 

as one of the standard methods 

in earthquake response analysis 

and structural testing. As his important 

role in the structural engineering 

society, he chaired the ‘Structural 

Design Appraisal Committee of Tall 

Buildings’ for eight years, where structural 

design works of all tall buildings 

in Japan were examined and appraised 

from the viewpoint of the structural 

performance against earthquake. 

Also, he directs big research projects. 

His latest project is “Development of 

a new structural system” which was 

reported in SEI Vol. 20 No. 1. Prof. 

Takanashi is currently President of 

the Japan Society of Structural Steel 

Construction (JSSC), Member of the 

Architectural Institute of Japan (AIJ) 

and the International Association for 

Bridge and Structural Engineering 

(IABSE). 

Koichi Takanashi, Japan 

Within IABSE Koichi Takanashi has 

contributed comprehensively since 

his first presentation of his paper in 

IABSE Symposium Lisbon in 1973. 

He has submitted numerous papers 

and given his support to an IABSE 

Congress and Symposium as an Invited 

and a Keynote Speaker. He participated 

in Working Commission 8 and 

5 as a member. He convened IABSE 

Symposium Davos, Rome and Kobe as 

a Member of the Scientific Committee. 

He further has extended his efforts 

as a Vice-President from 1997–2005, 

served as Chair of the Japanese Group 

from 1999–2005, he was also a Member 

of the IABSE Outstanding Structure 

Award Committee. 

IABSE Symposium Venice, September 22–24, 2010 

Large Structures and Infrastructures for Environmentally Constrained and Urbanised Areas 

Venice 2010 Symposium Banner 

The Italian Group of IABSE, chaired 

by Carlo Urbano, welcomed the 

world’s structural engineers and their 

accompanying persons to an important 

event that promoted science 

and practice in Bridge and Structural 

Engineering at its highest levels. For 

the IABSE Symposium it was a return 

to Venice, after it had been hosted in 

the Serenissima for the first time back 

in 1983. 

The Organising Committee was chaired 

by Enzo Siviero, Italy, the Scientific 

Committee by Massimo Majowiecki, 

Italy and the Advisory Committee by 

Anton Steffen, Switzerland. The excellent 

Secretaries were Bruno Briseghella 

for the Organising and Tobia Zordan 

for the Scientific Committee. The symposium 

topic had found the interest 

of a considerable number of Italian 

Universities, including the Istituto 


and on a CD. The book (899 

pages) and CD can be ordered at: 

www.iabse.org/publications/onlineshop/ 

BASAAR 

Several IABSE Working Commissions’ 

and Working Groups held 

their annual ‘BASAAR’ (Briefings 

About Structural Applications and 

Research) on Wednesday afternoon, 

September 22. The purpose of the 

BASAAR is to promote a lively discussion 

about a topic predetermined 

by each Working Commission or 

Group. Following topics were presented 

and discussed: 

Carlo Urbano, Chair, Italian Group of IABSE 

Universitario di Architettura di 

Venezia and the Polictecnico di Milano 

as co-organisers, and the Universities 

of Naples, Rome, Trento and Turin as 

supporters. 

The Palazzo del Casino and the 

Palazzo del Cinema, at the Lido in 

Venice served as venues. The attendance 

was high with almost 600 participants 

including accompanying 

persons. Out of 400 abstracts submitted 

to the theme ‘Large Structures and 

Infrastructures for Environmentally 

Constrained and Urbanised Areas’ 

the Scientific Committee selected 210 

papers for oral presentation and some 

150 for poster presentation, from 47 

countries. 

The Opening Ceremony 

Welcome and Introductory Speeches 

were made by Carlo Urbano, Chair 

of the Italian Group of IABSE; Enzo 

Siviero, Chair of the Organising 

Committee; Antonio Paruzzolo representing 

Giorgio Orsoni, Mayor of 

Venice, Don Alberto representing 

Angelo Scola, Cardinal of Venice, and 

Jacques Combault, President IABSE, 

who subsequently presented the 

IABSE Awards 2010. Klaus Ostenfeld, 

Chair of the IABSE Foundation 

Council then conferred the IABSE 

Anton Tedesko Medal. Alberto 

Scotti gave an invited lecture on ‘The 

Venice Mose Project: an Holistic 

and Interdisciplinary Approach for 

Innovative Interventions’ and completed 

the Ceremony. 

Symposium Contents 

Keynote Lectures, Presentations and 

Posters addressed the following main 

topics: Basis of Design; Infrastructure 

and Design as Meeting Point for 

Architecture and Engineering; Infrastructure 

Hazard and Safety Concepts; 

Management and Planning of Operation 

and Maintenance; Ethics and 

Social Responsibility. 

Keynote Lectures 

• Giorgio Diana, Italy 

The Messina Strait Bridge: Major 

Problems Affecting the Design 

• Klaus Ostenfeld, Denmark 

An Integrated Multidisciplinary Approach 

to Design of Major Fixed Links 

• Jiemin Ding, China 

Recent Applications and Practices of 

Large-Span Steel Structures in China. 

Invited Lectures were given on the 

following days of the Symposium: 

‘Traceability Systems and Quality 

Systems: two Sides of same Coin’, by 

Corrado Baldi; ‘The Raising of the 

Buildings of the City of Venice for the 

Safeguard from the High Water’, by 

Enzo Siviero. 

Symposium Report 

Three Keynote Lectures and 

375 contributions have been collected 

in the Symposium Report 

The Long-term Structural 

Performance (Life-Cycle Costs) 

by F. Biondini, Italy and M. Torkkelli 

Finland, 

(Co-ordinator M. SØderkvist, Finland, 

WC1) 

Monitoring – Does it make Sense? 

by R. Geier, Austria and S. Nakamura, 

Japan, 

(Co-ordinator: R. Geier, Austria, WC 2). 

Structural Safety Assessment of 

Concrete Structures 

by J. McGormley, USA; M.Matsumoto 

Japan and C. Bob, Romania, 

(Co-ordinator: J. Tortorella, USA, WC 4). 

The Long Term Issues Facing 

Structural Engineers in 

Environmentally Constrained Urban 

Areas 

by A. Boegle, Germany; A. Meyboom, 

Canada and F. Saad, Egypt, 

(Co-ordinator: W. Anderson, Canada, 

WC 5) 

Integrating Sustainability into 

Structures 

by J. Kanda, X. Ruan, China and 

J. Anderson, USA, 

(Co-ordinator: J. Kanda, Japan, WC 7) 

Seismic Resistance of Structures – 

Lessons from Devastating 

Earthquakes (Chile, Mexico, Turkey, 

Haiti, Greece) 

by L.F. Fargier Gabaldon, USA; 

C. Mendez, Switzerland; M. Gercek, 

Turkey and D. Sonda, Italy, 

(Co-ordinator: St. Dritsos, WG 7) 

In addition to the BASAAR a video 

presentation ‘Vibrations of the Volga 

Bridge in Volgograd, Russia’ was 

organised by S. Mozalev, Chair of the 

Russian Group of IABSE. 


Young Engineers Programme 

Benefits for young engineers at this 

Symposium were jointly offered by 

the IABSE Fellows and the Symposium 

Organising Committee: 

Young Engineers enjoyed reduced 

registration fee and were offered 

free IABSE membership for the 

year 2011. The Award Jury represented 

by A. Chen, China, P. Collin, 

Sweden and R. Zandonini, Italy, conferred 

two prizes each of 2000 EUR, 

sponsored by IABSE Fellows to two 

young authors for their outstanding 

contributions: 

Johan Berger, Austria: ‘New Ap proach 

for Bridges with Very High Durability’ 

and Xin Ruan, China: ‘Failure Analysis 

of a Long Span Pre-stressed Concrete 

Box Girder Bridge’. 

YEP Awardees Johan Berger, Austria 

and Xin Ruan, China 

Technical Visits 

On Friday a technical visit was organised 

to the moveable dams of ‘MOSE’, 

a defence system consisting of rows of 

concealed gates designed to stop high 

tides at the three lagoon inlets. On 

Saturday another technical excursion 

to three structures: the fourth bridge 

over the Grand Canal in Venice, 

designed by Santiago di Calatrava; the 

Ponte Strallato di Marghera, a curved 

cable-stayed bridge linking the city of 

Mestre to Porto Marghera and to the 

‘Laguna Palace’ glass roof: a complex 

constituted by a hotel building and a 

private dock. 

Special Public Session 

On Friday, September 24, after the 

Symposium a Special Session on 

‘Recent Large Earthquake and Seismic 

Risk Reduction with Rreference to 

a Sustainable Development’ was 

organised, and open for free to the 

Public. 

Social Events and Sightseeing 

On Wednesday, the Organising Committee 

welcomed all participants and 

accompanying persons to the beautiful 

Chiostro di San Nicoletto, dating to 

the origins of the independent Venice, 

in the early Middle Ages. Drinks and 

food and a wonderful soprano recital 

were enjoyed in a white decoration 

at candle-light. Later in the evening 

unexpected visitors interrupted the 

peaceful atmosphere and initiated 

what was later called the ‘mosquito’ 

dance, which many guests abandoned 

to continue with a nice dinner in 

Venice or to join the young engineers 

social event at the historical Nicelli 

Lido Airport for free drinks and music 

entertainment. 

Prior to the Symposium Dinner on 

Thursday evening, a visit to either 

San Marco’s Church with a magnificent 

organ concert or to the Scuola 

Grande di San Rocco were organised. 

The evening was continued with a 

fine six course menu at the beautiful 

Cà Giustinian Palace, the seat of the 

Biennale, located close to San Marco 

Square and overlooking the Grand 

Canal. 

Welcome Reception at Chiostro di San Nicoletto 

Social Tours 

Symposium delegates and accompanying 

persons were offered various 

guided tours in Venice by foot or 

Gondola, outings to Murano, Burano, 

or longer excursions to cities such 

as Padova, Verona, Florence and 

Rome. 

Members of the Organising Committee and Guests 


Dhaka Conference ‘Advances in Bridge Engineering-II’ 

Organised by the Bangladesh Group of IABSE and JSCE Steel Structures Committee, Japan 

The Dhaka Conference on ‘Advances 

in Bridge Engineering-II’ held from 

August 8–10, 2010, was a great success. 

It was jointly organised by the 

new Bangladesh Group of IABSE 

and JSCE Steel Structures Committee, 

in Association with the Roads and 

Highways Department Government 

of Bangladesh and the Institution 

of Engineers, Bangladesh and 

co-organisers from the Bangladesh 

Sections of the American Society 

of Civil Engineers and Institution 

of Civil Engineers, UK. One of the 

major aspects of this conference 

was bringing major professional 

societies and bodies working in the 

region together to achieve an effective 

collaboration and partnership in 

future. 

B.C. Roy, Vice-President of IABSE 

presented a message on behalf of 

IABSE President, Jacques Combault 

at the Opening Ceremony, and two 

Honorable Ministers responsible for 

the Ministries of Communication, 

Science and Information Technology, 

Government of Bangladesh, attended 

the Conference Dinner and Closing 

Ceremonies of the conference. The 

presence of Muhammad Yunus, Nobel 

Laureate for Peace in the Conference 

Dinner inspired participants greatly. 

Experts, professionals and academicians 

from Asia, Europe, Australia 

and the North America discussed the 

issues of Bridge Engineering and its 

further development for the benefit of 

the people and community. The total 

number of registration to the conference 

was 338. 

Six Keynote addresses were supported 

by 56 technical papers from 

11 countries and four continents. Bridge 

engineering themes were on: History 

and Planning Materials; Analysis; 

Design and Construction Design for 

Dynamic Forces; Geo-environmental 

Issues and Administration; Maintenance 

and Monitoring. Ref.: www. 

iabse-bd.org 

Conference participants perceived the 

technological challenges that exist in 

Bangladesh for bridge construction and 

maintenance, but also recognised the 

necessity of having world class bridges 

in Bangladesh for the effective transportation 

not only within the country, 

but also within the Asian region. They 

identified the necessity of technology 

transfer, assimilation and formulation 

of a unified bridge code for the Asian 

countries in the coming years. 

A volume with 594 pages has been 

published, for more information, 

please contact A.F.M. Saiful Amin: 

samin@ce.buet.ac.bd 

Md. Yunus (right), Nobel Laureate for Peace. From left: 

H. Mutsuyoshi, Saitama Univ.; M. Nagai, Nagoaka Univ. 

of Tech. Japan; K. Wheeler, Maunsell AECOM 

Left to right: M. A. Sobhan (Chair, Bangladesh IABSE Group); 

Md. Yunus (Nobel Laureate); A.F.M. Saiful Amin (General 

Secretary, Bangladesh IABSE Group) 


Maschinenfabrik GmbH & Co. KG 

A matter of confidence. 

Stressing of stay cables 

PAUL supplies 

• Prestressing installations incl. planning work 

• Anchor grips 

• Prestressing machinery 

(single-/multi-stressing jacks) 

• Strand pushing and cutting equipment 

• Automatic prestressing machines for railway sleepers 

• Prestressing equipment for bridge construction 

(prestressing cables and stay cables) 

The experts in 

Prestressed Concrete Technology. 

Max-Paul-Straße 1 

88525 Dürmentingen / Germany 

www.paul.eu 

Phone: +49 (0) 7371 / 500 - 0 

Fax: +49 (0) 7371 / 500 - 111 

Mail: stressing@paul.eu

Brilliant Slide 

MSA ® – New for 

spherical bearings! 

MAURER Sliding Alloy 

New alloy MSA ® for bearings in 

bridge and building structures. 

Shining – special treatment of 

the surface for reduced sliding 

resistance and abrasive wear 

Resistant to agressive 

industrial air – especially against 

fluorine and chlorine 

Superior corrosion resistance 

Better economic efficiency 

Longer lifetime 

Approved by the construction 

authorities (AbZ Z-16.4-436) 

Our Product Range 

Maurer Expansion Joints 

Maurer Structural Bearings 

Maurer Seismic Protection Systems 

Maurer Vibration Control Systems 

Maurer Monitoring of Structures 

Maurer Söhne GmbH & Co. KG 

Frankfurter Ring 193, 80807 Munich/Germany 

Phone +49 89 32394–0 

Fax +49 89 32394–306 

ba@maurer-soehne.de 

www.maurer-soehne.com

SEI-november-2010-journal

Create successful ePaper yourself

Delete template?

Save as template?