PSI - Issue 57

Lorenzo Bercelli et al. / Procedia Structural Integrity 57 (2024) 437–444 2 L. Bercelli, C. Guellec, B. Levieil, C. Doudard, F. Bridier, S. Calloch / Structural Integrity Procedia 00 (2019) 000 – 000 1. Introduction Welded joints constitute a critical component to the structural integrity of navalindustry applications. Indeed, the geometrical singularity of the weld toe leads to stress concentration makingit a preferential location for the initiation and propagation of fatigue cracks. In that regard, robust fatigue dimensioning tools must be provided to engineers. More specifically, the effect of welding and pre-loading residual stresses on crack propagation should be considered in predictive models as they are known to strongly influence the fatigue life, as illustrated in the work of Alam et al. (2009), Barsoum and Barsoum (2009), Sonsino (2009) and Williams et al. (1970). The residual stresses alter the loading ratio close to the weld toe as they add up with the mean stress, which can slow down the propagation of cracks when compression occurs as seen in the work of Smith and Smith (1983). In the study of Hensel (2020), the introduction of an effective stress ratio was proposed, taking into account residual stresses, allowing for the definition of a mean stress sensitivity factor , linking the stress range Δ at any stress ratio to a reference stress range at a stress ratio =−1 for a same fatigue life . Through monitoring of crack propagation via Thermoelastic Stress Analysis (TSA), Bercelli et al. (2023) obtained results consistent with this sensitivity factor. However, the definition of an effective stress ratio relies on the knowledge of the residual stresses near the weld toe, which can be difficult to estimate. Moreover, one has to consider the evolution of the residual stress field with the applied loading, because of the stress concentration at the weld toe where plasticity can occur. For these reasons, the availability of an unambiguous in-situ method for the determination of the stress state close to the fatigue crack should greatly facilitate the study of the residual stresses influence on fatigue life. In the present study, an experimental method based on TSA is proposed in order to acknowledge the effect of the stress state near the weld toe on crack propagation. Firstly, the material and test set-up are presented, as well as a short reminder of the theoretical background of TSA. Secondly, the experimental results are presented and the crack-closure phenomenon is detailed. Thirdly, a predictive fatigue model is introduced with consideration of the effect of the stress ratio in the definition of a crack-opening rate . Finally, a short conclusion is proposed. 2. Material and test set-up 2.1. Test configuration The welded joints consist of thick T-joints of 50 width which base metalis a high strength steel. The T-joints are tested in fatigue in four-point bending (Fig. 1), and four different test configurations are considered: - As-welded joints (AS-W) loaded at the machine loading ratio = 0.1 ; - As-welded joints (AS-W) loaded at = 10 ; - Pre-loaded in tension (PRE-T) at the yield stress and the loaded at = 0.1 ; - Pre-loaded in compression (PRE-C) at and the loaded at = 10 . Bending fatigue tests are performed until failure ( i.e. presence of a through-width crack) at a constant stress amplitude. The face on which the stiffener is welded is monitored using an infrared camera (Fig. 1). 3000-frame infrared films of acquisition frequency = 100 are recorded regularly throughout the test to allow for TSA. 2.2. Theoretical background of TSA The TSA technique relies on the field measurement of the temperature response of a sample linked to the thermoelastic coupling. The local heat equation states = − (− ⃗) + ℎ +Δ , (1) with the density, the specific heat, the temperature variation, ⃗ the local heat flux, ℎ the thermoelastic coupling heat source and Δ the dissipation heat source. Given the expression of ℎ proposed by Boulanger et al. (2004), and considering a typical steel material in its elastic domain, for small temperature variations, and under adiabatic conditions, it comes ℎ =− 0 1 , (2)

438