27.02.2014 Views

The design of new grips for multiaxial fatigue tests

The design of new grips for multiaxial fatigue tests

The design of new grips for multiaxial fatigue tests

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

21 ème Congrès Français de Mécanique Bordeaux, 26 au 30 août 2013<br />

<strong>The</strong> <strong>design</strong> <strong>of</strong> <strong>new</strong> <strong>grips</strong> <strong>for</strong> <strong>multiaxial</strong> <strong>fatigue</strong> <strong>tests</strong><br />

V. I. PRISACARI a , P. GABORIT a , M. PONCELET a , B. RAKA a , J.M. VIRELY a<br />

a. Laboratoire de Mécanique et Technologie (LMT-Cachan),<br />

ENS de Cachan / UMR 8535 CNRS / Univ. Paris 6 / PRES UniverSud Paris,<br />

61 Avenue du Président Wilson, F-94235 Cachan Cedex<br />

Résumé:<br />

L’amélioration des simulations numériques de tenue à la <strong>fatigue</strong> nécessite des modèles multiaxiaux et leurs<br />

essais d’identification. Nombre d’entre eux se font sur des éprouvettes « en croix » depuis la démocratisation<br />

des machines de traction biaxiale [1]. La majorité des mors utilisés actuellement pour ce type d’essai a<br />

plusieurs avantages : robustesse vis-à-vis des charges radiales, montage simple et rapide des éprouvettes.<br />

Généralement, ils ont comme défauts communs une précision d’alignement insuffisante (induisant une<br />

flexion statique au montage) et potentiellement une flexion sous charge. Ces défauts sont évidemment de<br />

première importance lorsque des essais de <strong>fatigue</strong> sont réalisés. D’autres mors suppriment une partie de ces<br />

inconvénients, mais habituellement au prix d’un encombrement significatif. En outre, les techniques de<br />

mesure mises en place au sein des machines biaxiales ont beaucoup évolué depuis leur apparition [2-7]:<br />

Corrélation d’Images Numériques, Stéréo-corrélation, <strong>The</strong>rmographie, DRX. Il est donc nécessaire de<br />

prendre en compte ces nouvelles contraintes en plus des défauts des mors actuels. La solution proposée <strong>of</strong>fre<br />

une plus grande raideur en flexion tout en conservant un encombrement très réduit, permettant de<br />

nombreuses instrumentations. En outre, elle permet l’alignement du mors par rapport à son palier pour<br />

diminuer au maximum les mouvements hors plan (ce qui est vérifié avec une éprouvette étalon instrumentée).<br />

Plusieurs capteurs peuvent être embarqués sur le mors pour répondre à différentes fonctions, en particulier<br />

des capteurs de déplacement LASER <strong>of</strong>frent une mesure directe de l’écart entre mors et de leur inclinaison<br />

relative.<br />

Abstract:<br />

Most <strong>of</strong> the planar biaxial machine <strong>grips</strong> currently used <strong>for</strong> cross-shaped specimens have several advantages:<br />

stiffness to radial loads, fast and easy sample mounting. Generally, they lack a sufficient alignment precision<br />

(inducing a static bending during clamping) and might exhibit flexion during load. <strong>The</strong>se shortcomings are<br />

obviously <strong>of</strong> major importance when it comes to <strong>fatigue</strong> <strong>tests</strong>. Other types <strong>of</strong> <strong>grips</strong> avoid some <strong>of</strong> these<br />

problems, but usually at the cost <strong>of</strong> a significantly larger size. Moreover, measurement techniques used with<br />

biaxial testing machines have evolved considerably: Digital Image Correlation, Stereo Correlation,<br />

<strong>The</strong>rmography, X-Ray Diffraction. It is thus necessary to take into account these <strong>new</strong> requirements along<br />

with the defects <strong>of</strong> the existing <strong>grips</strong>. <strong>The</strong> considered solution <strong>of</strong>fers a higher bending stiffness without<br />

considerable increase in size, permitting significant additional instrumentation. Also, it allows the<br />

measurement and the adjustment <strong>of</strong> the alignment <strong>of</strong> the grip with respect to its hydrostatic bearing, thus<br />

reducing considerably the out <strong>of</strong> plane motions. Auxiliary devices can be attached to the <strong>grips</strong>, particularly<br />

LASER displacement sensors that provide direct measurement <strong>of</strong> the distance between the <strong>grips</strong> and <strong>of</strong> their<br />

relative tilt.<br />

Keywords: <strong>multiaxial</strong>, <strong>fatigue</strong>, testing machine, <strong>grips</strong><br />

1 Introduction<br />

Due to the increasing complexity <strong>of</strong> systems used in aeronautics, spatial, automobile industries, etc. the need<br />

<strong>for</strong> more sophisticated <strong>fatigue</strong> models to describe their behavior has grown in the past years [1]. This implies<br />

that experiments manage to get as close as possible to the <strong>multiaxial</strong> loading states encountered in service. In<br />

this sense, studies per<strong>for</strong>med on biaxial samples have grown in number and in complexity.<br />

1


21 ème Congrès Français de Mécanique Bordeaux, 26 au 30 août 2013<br />

<strong>The</strong> LMT Cachan laboratory owns a tri-tension/compression testing machine named ASTREE. It has three<br />

orthogonal axes, each <strong>of</strong> them containing two hydraulic actuators working in pairs. This machine is<br />

particularly interesting because <strong>of</strong> its mobile architecture, allowing the mounting <strong>of</strong> considerably large<br />

samples and permitting experiments where the center <strong>of</strong> the specimen is kept motionless.<br />

Most <strong>of</strong> the experiments per<strong>for</strong>med in ASTREE are biaxial tension-compression <strong>tests</strong> on planar cross-shaped<br />

samples [2-7]. Many sets <strong>of</strong> <strong>grips</strong> have been used <strong>for</strong> these experiments [3-7] but the majority fails in terms<br />

either <strong>of</strong> stiffness, size, or ultimate load. <strong>The</strong> <strong>design</strong> <strong>of</strong> <strong>new</strong> <strong>grips</strong> <strong>for</strong> ASTREE aims to fulfill the following<br />

needs: improving the stiffness <strong>of</strong> the <strong>grips</strong>, controlling unexpected solicitations, evaluating the quality <strong>of</strong> the<br />

alignment <strong>of</strong> the actuator lines, mounting <strong>of</strong> smaller and thicker samples. Moreover, the instrumentations and<br />

types <strong>of</strong> <strong>tests</strong> per<strong>for</strong>med in the laboratory have evolved considerably making it necessary to adapt the <strong>grips</strong> to<br />

these complex conditions.<br />

This study will first provide a brief description <strong>of</strong> the existing <strong>grips</strong> and the reasons why they fail the<br />

a<strong>for</strong>ementioned criteria. <strong>The</strong> architecture <strong>of</strong> the <strong>new</strong> <strong>grips</strong> will be then presented along with their innovations.<br />

Finally, the validation procedure and conclusions are stated.<br />

2 Existing <strong>grips</strong><br />

<strong>The</strong> current configuration contains four <strong>grips</strong> (fixed and mobile part), each one connected to the hydraulic<br />

actuator by means <strong>of</strong> a spacer (Fig 1a). <strong>The</strong>se <strong>grips</strong> (Fig 2a), <strong>design</strong>ed by J. Lemaitre and J.M. Virely (and<br />

called in the following L&V <strong>grips</strong>), have been used <strong>for</strong> most biaxial <strong>tests</strong> on ASTREE because <strong>of</strong> their clear<br />

advantages: sufficient ef<strong>for</strong>t transmission, small size and easy sample mounting. Nevertheless, some <strong>tests</strong><br />

have shown non-negligible bending values (Fig. 1b). Moreover, the alignment <strong>of</strong> the actuators is not properly<br />

evaluated (inducing static bending during mounting) and the compatibility with other types <strong>of</strong> samples is<br />

quite limited both in size and thickness.<br />

On the <strong>grips</strong> developed by Instron and used at the Centre des Matériaux laboratory an extra tightening,<br />

perpendicular to the load axis, has been added in order to ensure the contact between the fixed and the<br />

mobile part. <strong>The</strong> MTS standard <strong>grips</strong> used in uniaxial tensile machines have a more complex <strong>design</strong>: two<br />

wedges supported on the fixed part, with a mobile shaft which clamps these wedges. <strong>The</strong>y are available in<br />

“hydraulic” or "mechanical" versions. As shown in Fig. 1b, these types <strong>of</strong> <strong>grips</strong> endure high loads and induce<br />

almost no bending. However, their large diameter prevents the use <strong>of</strong> a stereo-DIC setup or a mobile X-ray<br />

diffraction (XRD) machine that was developed in LMT Cachan.<br />

L&V<br />

LMT<br />

Instron<br />

CdM<br />

MTS<br />

Hydr.<br />

MTS<br />

Mech.<br />

Diameter (mm) 150 225 204 170<br />

Axial ef<strong>for</strong>ts (kN)<br />

250 100 100 100<br />

- 250 - 100 - 100 - 100<br />

Radial ef<strong>for</strong>ts (kN) 200 110 - -<br />

Bending (mm) > 0,1 ≈ 0 ≈ 0 ≈ 0<br />

(a)<br />

(b)<br />

Fig. 1 – (a) Current configuration used in ASTREE, (b) Basic specifications <strong>of</strong> existing <strong>grips</strong><br />

2


21 ème Congrès Français de Mécanique Bordeaux, 26 au 30 août 2013<br />

3 Chosen <strong>design</strong><br />

Different architectures have been investigated. <strong>The</strong> chosen one is a combination between the L&V <strong>grips</strong> (Fig.<br />

2a) and to the Instron ones. It is simple (only two main parts) and stiff (because <strong>of</strong> its nearly symmetric<br />

geometry (Fig 2b), compactness and large ribs) despite its small size. Its rough shape and size derive directly<br />

from the collision limits with the neighboring devices or their size.<br />

(a)<br />

Fig. 2 – (a) L&V <strong>grips</strong> (b) Diagram <strong>of</strong> the <strong>new</strong> <strong>grips</strong><br />

(b)<br />

<strong>The</strong> upper mobile part will be attached during <strong>tests</strong> to the lower fixed one directly, using 8 M12 screws, and<br />

through sample contact using 8 M8 screws or less depending on the specimens. <strong>The</strong> height <strong>of</strong> the <strong>grips</strong><br />

allows the XRD machine to be used. This machine will be mounted above the horizontal plane <strong>of</strong> the <strong>grips</strong> in<br />

order to measure the stress state and phase changes while per<strong>for</strong>ming biaxial tension <strong>tests</strong>. <strong>The</strong> seemingly<br />

trivial aspect <strong>of</strong> the “steep” front end makes the region <strong>of</strong> interest more accessible in terms <strong>of</strong> visibility<br />

(lighting, XRD) and allows the use <strong>of</strong> shorter samples (no collision with neighboring <strong>grips</strong>).<br />

Another aspect that had to be taken into account was the cooling <strong>of</strong> the <strong>grips</strong> (controlling the heat flux from<br />

the <strong>grips</strong> to the samples or in the opposite way in case <strong>of</strong> high temperature <strong>tests</strong>). <strong>The</strong> four cooling pipes that<br />

go right through the <strong>grips</strong> allow maintaining their temperature at 20°C (the temperature <strong>of</strong> the actuator being<br />

about 60°C) without noticeably reducing their stiffness.<br />

Another clear improvement is the wider range <strong>of</strong> possible sample thicknesses (from 3 to 10mm) that allows<br />

coherence between the sample mid-plane and the actuators plane. One need only use the corresponding jaws.<br />

<strong>The</strong>se jaws, which complement the tightening M8 screws in terms <strong>of</strong> adherence, can be <strong>design</strong>ed with<br />

different contact surfaces (smooth, spiked, striated, etc.) depending on the material and on the loading type.<br />

4.1 Alignment solution<br />

<strong>The</strong> <strong>design</strong> <strong>of</strong> the <strong>grips</strong> aims to accomplish two purposes: on the one hand to allow the alignment <strong>of</strong> each<br />

grip with respect to its bearing in order to reduce as much as possible the out <strong>of</strong> plane movements that come<br />

from bearing rotation and on the other hand to allow the alignment <strong>of</strong> the grip assembly to avoid sample<br />

bending. Because <strong>of</strong> the inability to rigorously position the actuators vertically, these two conditions cannot<br />

be stricto sensu satisfied. An alternative is to verify the relative position <strong>of</strong> the actuators and then to modify<br />

the position <strong>of</strong> the <strong>grips</strong> with respect to the spacers. For the relative position verifications, one <strong>of</strong> the<br />

proposed solutions is to use a cross-shaped gauge with cylindrical ends placed on specific surfaces <strong>of</strong> the<br />

<strong>grips</strong> (Fig 2). Using this method, the measure is independent with respect to the rotation <strong>of</strong> each bearing and<br />

simplified (no iterations between each possible combination <strong>of</strong> <strong>grips</strong>).<br />

3


21 ème Congrès Français de Mécanique Bordeaux, 26 au 30 août 2013<br />

(a)<br />

(b)<br />

Fig. 2 – Alignment measuring system: (a) Support <strong>for</strong> the alignment cross which is coaxial with respect to<br />

the axis <strong>of</strong> the grip (b) Alignment cross gauge<br />

4.2 Out <strong>of</strong> plane motion measurement<br />

In order to measure the relative position <strong>of</strong> each grip with respect to its opposite neighbor, LASER sensors<br />

(Fig 3a) have been chosen. <strong>The</strong>y <strong>of</strong>fer high precision (e.g. 2 microns <strong>of</strong> measure incertitude at 10Hz, <strong>for</strong> a<br />

working range <strong>of</strong> 300-500 mm) without interfering with the other measurements. Because <strong>of</strong> their<br />

considerable size they could only be placed on top and on the bottom <strong>of</strong> the <strong>grips</strong>. Such a pair <strong>of</strong> sensors<br />

along with their corresponding “screen” on the opposing grip would allow the monitoring <strong>of</strong> the distance<br />

between the <strong>grips</strong> and the system’s alignment during loading (Fig 3).<br />

(a) (b) (c)<br />

Fig. 3 – Distance measuring using LASER: (a) Principle <strong>of</strong> LASER triangulation (b) Side view (c) Top view<br />

4.3 Numerical validation <strong>of</strong> the <strong>design</strong><br />

In order to quantify the gain in stiffness with respect to the L&V LMT <strong>grips</strong>, numerical simulations <strong>of</strong> the<br />

two systems were per<strong>for</strong>med. <strong>The</strong> purpose was to find the critical areas and to compare the way each set <strong>of</strong><br />

<strong>grips</strong> distributes ef<strong>for</strong>ts. Thus, the two systems are elastically loaded up to the maximum service values. To<br />

come closer to the real behavior, the ef<strong>for</strong>ts weren’t directly applied onto the <strong>grips</strong> but through a quarter <strong>of</strong> a<br />

typical cross-shaped sample. For the L&V <strong>grips</strong> an assembly with the fixed part, the mobile part and a<br />

sample in between was created (perfect contacts were used). For the <strong>new</strong> <strong>grips</strong>, in order to simulate the worst<br />

4


21 ème Congrès Français de Mécanique Bordeaux, 26 au 30 août 2013<br />

clamping scenario, a clearance was introduced between the upper grip and the lower one so that the ef<strong>for</strong>ts<br />

are only transmitted through the 5 M12 screws.<br />

<strong>The</strong> first simulation was simple tension, i.e. the most commonly encountered solicitation during <strong>grips</strong> usage.<br />

Thus, a <strong>for</strong>ce <strong>of</strong> 100 kN (the maximum load that can be induced by one actuator) was applied on the tip <strong>of</strong><br />

the quarter sample (Fig. 4). <strong>The</strong> upper limit <strong>of</strong> the represented stress range was set to 150MPa, i.e. far lower<br />

than the <strong>fatigue</strong> limit <strong>of</strong> the material (about 600 MPa). <strong>The</strong> <strong>new</strong> <strong>grips</strong> are below this value once optimization<br />

was done, whereas the L&V ones are considerably above in the connecting area and the lower ribs.<br />

Moreover, the deflection <strong>of</strong> the <strong>new</strong> <strong>grips</strong> is approximately 15 times smaller than <strong>of</strong> the L&V ones (Tab. 1).<br />

(a)<br />

Fig. 4 – Simple tension numerical simulations (100kN): (a) L&V <strong>grips</strong> (b) New <strong>grips</strong><br />

Secondly, accidental radial loads were simulated, where a <strong>for</strong>ce <strong>of</strong> 200kN was applied in the loading plane<br />

but perpendicular to the axis <strong>of</strong> the grip (Tab. 1 - Radial <strong>for</strong>ce 1). Although the found ef<strong>for</strong>ts are unrealistic,<br />

they validate qualitatively the gain in stiffness. Finally, system behavior was tested in the case <strong>of</strong> an<br />

accidental load perpendicular to the loading plane (Tab. 1 - Radial <strong>for</strong>ce 2). <strong>The</strong> considered load was 500kN,<br />

representing the maximum impact <strong>of</strong> an actuator from the vertical axis. As be<strong>for</strong>e, it can be seen that the<br />

compact <strong>design</strong> and the contribution <strong>of</strong> the mobile part are responsible <strong>for</strong> the considerably better response<br />

<strong>of</strong> the <strong>new</strong> system. In both cases, it is clear that the weakest link will be the M12 screws in such accidental<br />

loadings.<br />

(b)<br />

Measured value<br />

Axial <strong>for</strong>ce<br />

(100 kN)<br />

Radial <strong>for</strong>ce 1<br />

(200 kN)<br />

Radial <strong>for</strong>ce 2<br />

(500 kN)<br />

L&V <strong>grips</strong><br />

New <strong>grips</strong><br />

Deflection (mm) 0.3 - -<br />

Maximum stress (MPa) 350 2000 10000<br />

Deflection (mm) 0.02 - -<br />

Maximum stress (MPa) 140 600 4000<br />

Tab. 1 – Numerical simulations results <strong>for</strong> the L&V <strong>grips</strong> and the <strong>new</strong> <strong>grips</strong><br />

5 Conclusions and perspectives<br />

<strong>The</strong> study has presented the <strong>design</strong> <strong>of</strong> <strong>new</strong> <strong>grips</strong> <strong>for</strong> biaxial tension <strong>tests</strong>. <strong>The</strong> main focus was to resolve the<br />

issues found in most existing <strong>grips</strong>: improving the stiffness <strong>of</strong> the <strong>grips</strong>, controlling unwanted solicitations,<br />

evaluating the quality <strong>of</strong> the alignment <strong>of</strong> the actuator lines, mounting <strong>of</strong> smaller and thicker samples.<br />

Moreover, the complex techniques used while per<strong>for</strong>ming biaxial <strong>tests</strong> (Digital Image Correlation, Stereo<br />

Correlation, <strong>The</strong>rmography, XRD) created further needs that were partially covered (providing cooling pipes,<br />

easy LASER sensor attachment, mobile XRD machine accessibility, etc.).<br />

At the writing time <strong>of</strong> this proceeding, a validation procedure <strong>for</strong> the <strong>new</strong> <strong>grips</strong> was <strong>for</strong>eseen in a few weeks.<br />

<strong>The</strong> main purpose <strong>of</strong> the validation is to quantify the bending due to loading <strong>for</strong> the L&V and the <strong>new</strong> <strong>grips</strong>.<br />

5


21 ème Congrès Français de Mécanique Bordeaux, 26 au 30 août 2013<br />

This comparison will be per<strong>for</strong>med using different measuring techniques (strain gauge rosettes, Digital<br />

image correlation, LVDT) during cyclic proportional loadings with a stress ratio <strong>of</strong> -1 using two loading<br />

frequencies (20 Hz and 1 Hz). In order to per<strong>for</strong>m these <strong>tests</strong>, two reference test samples [8] were machined.<br />

One <strong>of</strong> them will be instrumented with 10 strain gauge rosettes (5 on each side) thus obtaining the averaged<br />

plane strain field in these areas <strong>of</strong> interest (Fig. 5a). <strong>The</strong> surface <strong>of</strong> the second one will be painted with a<br />

black and white speckle in order to obtain the full displacement field using Digital Image Correlation (Fig.<br />

5b).<br />

(a)<br />

Fig. 5 – Reference test samples used <strong>for</strong> grip validation using different measuring techniques: (a) strain<br />

gauge rosettes (b) Digital Image Correlation. <strong>The</strong> positioning system using centering pins is represented:<br />

primary contact (red) and secondary contact (blue).<br />

(b)<br />

References<br />

[1] Hannon, A. & Tiernan, P. A review <strong>of</strong> planar biaxial tensile test systems <strong>for</strong> sheet metal. Journal <strong>of</strong><br />

Materials Processing Technology 198, 1-13 (2008).<br />

[2] Wu, X.-D., Wan, M., & Zhou, X.-B. Biaxial tensile testing <strong>of</strong> cruci<strong>for</strong>m specimen under complex loading.<br />

Journal <strong>of</strong> Materials Processing Technology, 168(1), 181–183. doi:10.1016/j.jmatprotec.2004.11.003 (2005).<br />

[3] Bonnand, V., Chaboche, J.L., Gomez, P., Kanouté, P. & Pacou, D. Investigation <strong>of</strong> <strong>multiaxial</strong> <strong>fatigue</strong> in<br />

the context <strong>of</strong> turboengine disc applications. International Journal <strong>of</strong> Fatigue 33, 1006-1016 (2011).<br />

[4] Bedkowski, W., Determination <strong>of</strong> Critical Plane and Ef<strong>for</strong>t Criterion in Fatigue Life. Evaluation <strong>for</strong><br />

Materials under Multiaxial Random Loading. Experimental Verification Based on Fatigue Tests <strong>of</strong><br />

Cruci<strong>for</strong>m Specimens. Proc. 4th ICB/MF, May 31 - June 3, Paris, 435 – 447 (1994).<br />

[5] Sermage, J.-P., Lemaitre, J., Desmorat, R., 2000. Multiaxial creep <strong>fatigue</strong> under anisothermal conditions,<br />

Fatigue and Fracture <strong>of</strong> Engineering Materials & Structures. 23(3), 241–252.<br />

[6] Doudard, C., Poncelet, M., Calloch, S., Boue, C., Hild, F., & Galtier, A. Determination <strong>of</strong> an HCF<br />

criterion by thermal measurements under biaxial cyclic loading. International Journal <strong>of</strong> Fatigue, 29(4),<br />

748–757. (2007).<br />

[7] Poncelet, M., Barbier, G., Raka, B., Courtin, S., Desmorat, R., Le-Roux, J. C., & Vincent, L. Biaxial<br />

High Cycle Fatigue <strong>of</strong> a type 304L stainless steel: Cyclic strains and crack initiation detection by digital<br />

image correlation. European Journal <strong>of</strong> Mechanics - A/Solids, 29(5), (2010).<br />

[8] J. Lemaitre , J.-L. Chaboche , A. Benallal , R. Desmorat, Mécanique des matériaux solides (3e édition),<br />

(2008)<br />

6

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!