ABSTRACT Defect tolerance assessment is a prerequisite for ensuring the safe service life of welded structures. In this study, a hybrid experimental–numerical framework was proposed to evaluate the influence of pores and inclusions on the mechanical properties in 316H stainless steel welds. Based on the GTN damage theory, the characteristic damage parameters of 316H steel at 550°C were calibrated through response surface methodology (RSM) and particle swarm optimization‐backpropagation (PSO‐BP) neural network. The PSO‐BP algorithm significantly enhanced the efficiency and accuracy of parameter calibration with a minimal prediction error of 3.2%. The upper limit of defect size in overmatched welded joint was analyzed to improve the safety assessment of welded structures. Numerical simulations systematically quantified the impact of defect characteristics on the static strength and fatigue life of welds, such as number, size, shape, location, type, and volume of pores. Defect size and shape were critical factors in static strength of welds, while fatigue life showed strong sensitivity to defect interaction. Pores played a dominant role in the decrease of load‐bearing capacity, with a porosity threshold of 5.03%.
Li et al. (Tue,) studied this question.