Numerical prediction of low-cycle fatigue crack propagation in cruciform welded joints

Abstract Low-cycle fatigue is a major cause of damage in steel structures due to the accumulation of large plastic deformations. In particular, welded joints are prone to low-cycle fatigue failure due to large strain concentration under cyclic loads, such as those caused by earthquakes. Therefore, assessing and predicting low-cycle fatigue cracking in welded joints is crucial for preventing significant damage. This study aimed to predict low-cycle fatigue crack propagation in cruciform welded joints using a cyclic J-integral (Δ J ), which is a criterion for fatigue crack growth based on elasto-plastic fracture mechanics. Calculation of Δ J was carried out by 3D finite-element (FE) analysis with a sub-modeling technique. Low-cycle fatigue tests were performed on fully and partially penetrated welded specimens, and low-cycle fatigue cracks were experimentally observed at the weld toe. To simulate the experimental conditions, 3D FE models that consider each weld profile were created, and Δ J was calculated using the contour integration method. The crack growth rate was evaluated using the modified Paris law, expressed as Δ J , and the predicted fatigue crack propagation behavior was compared with that in the experiments. The predicted crack propagation was highly consistent with the experimental results in all cases. These results demonstrate that the Δ J -based approach can effectively predict low-cycle fatigue crack growth in welded joints.