PSI - Issue 38

Hamza Abbad El Andaloussi et al. / Procedia Structural Integrity 38 (2022) 238–250 249 Hamza Abbad El Andaloussi, Luc Mouton, Firas Sayed Ahmad, Xabier Errotabehere, Stéphanie Mahérault-Mougin, Stéphane Paboeuf/ Structural Integrity Procedia 00 (2021) 000 – 000 12

fluorosilicone seal. Then, the aging the adhesive of the tested specimens is representative or even harsher than the one installed offshore. This last point represents an additional conservatism included in the interfacial S-N curve. Also, it seems that the results at load levels 62.5% and 56.5% fall in the same population. Whereas, the result at load level 50% falls in a distinct population. This could be explained by the increase of the dispersion of the results when the load level is decreased. Finally, one should note that all the 25 tested specimens present a similar failure mode: failure at the bondline. This result shows that the failure mode of the studied bonded assemblies is repeatable. 6.2. Benefit and limitation of stress-based approach As already explained, this new reinforcement solution presents a finite stress concentration on the edge, which allows using a stress-based strength criteria Erreur ! Source du renvoi introuvable. for design without mesh constraints in the FE model (but simply to have a converge stress gradient), or point-stress method. This is why the establishment of a S-N curve in peak stress has been possible and constitutes a first of a kind as per present literature review. Indeed, the results of the fatigue tests campaign combined with numerical and statistical analyses allow to determine this interfacial design S-N curve for the bonded reinforcement (called “ColdShield - Metal”). A first interesting point is that the mechanical behaviors implemented for polymeric materials correspond to quasi-static behaviors, whereas the experimental tests are performed at high speed. The strain-rate dependency of the mechanical behavior of polymeric materials is well-known. However, for the sake of simplicity and conservatively, quasi-static behaviors have been implemented instead of dynamic behavior that could be determined through additional tests. The consequence of this choice is that the interfacial stresses calculated numerically are lower than the experimental acting stresses. This means that for a given number of cycles at failure, the considered stress at failure is reduced and therefore the derived S-N curve is more conservative. Also, it is known Erreur ! Source du renvoi introuvable. that adhesive joints strength is governed by out-of plane interfacial stresses, namely out-of- plane shear stress and normal stress (also called “peel stress”). Thus, we have decided in this work to calculate a quadratic stress based on both out-of-plane stress components. This operation was mandatory in order to define a S-N curve. Without this operation, two stress values would be linked to each number of cycles, and the fatigue life assessment of the reinforcement would be more complex. As the strength of adhesive joints is dependent on the loading mode mixity, it is to be highlighted that the S-N curve built by calculating a quadratic stress is strictly correct for all loading modes that give a similar mode mixity. Basically, this fatigue design tool is correct for uniaxial loading modes (like tension-compression which constitutes the dominant loading mode for a FPSO deck) but cannot be generalized for all loading modes (e.g. out-of-plane loadings). 7. Nomenclature

FPSO Floating, Production, Storage and Offloading platforms. HAZ Heat affected zone WBM Wave Bending Moment SWBM Still Water Bending Moment S-N Stress-Number of cycles IDL Intermediate Deformation Layer RH Relative Humidity COV Coefficient of variation