Fatigue crack propagation and life assessment of welded joints of corrugated web I-beams considering residual stress

Corrugated web I-beams (CSWs) are widely used in bridge and construction engineering applications due to their lightweight, high strength, and superior buckling performance. Welding residual stress (WRS) has a significant impact on the fatigue performance of CSWs. However, its time-dependent behavior during crack propagation remains insufficiently studied. In this paper, the fatigue crack propagation mechanism and fatigue life evolution of CSWs welded joints are investigated through both experimental testing and numerical simulation. The specimens were fabricated using carbon dioxide gas shielded welding. Surface residual stresses on the flange plates were measured using the hole-drilling method, and a thermo-mechanical coupled finite element model was developed to validate the simulation accuracy. The results show that the simulated stresses agree well with the measured data. As the wave angle increases from 30° to 60°, the peak stress in the inclined flat region of the web increases from 159.9 MPa to 220.89 MPa, indicating more pronounced stress concentration. Crack propagation is dominated by the Mode I stress intensity factor (SIF), with the SIF at the deepest crack location being significantly higher than at the surface. A coupling relationship exists between crack geometry parameters, and an increase in the a/c ratio weakens the effect of twist angle on the SIF. The SIFs due to external loading and residual stress are independently calculated, and a residual stress relaxation mechanism is introduced to simulate stress redistribution during crack propagation. Residual stress significantly reduces fatigue life, and neglecting its influence leads to overestimation of fatigue life. Furthermore, ignoring the relaxation behavior during crack propagation amplifies prediction errors and causes substantial deviation from actual service conditions. The life assessment approach proposed in this study enhances prediction accuracy while maintaining computational efficiency and is applicable to the fatigue reliability analysis of complex welded structures.