Analysis of Welding Residual Stresses and Prediction of Delayed Crack Initiation Location in Flat Butt Welds

Delayed cracks are particularly difficult to detect and can be highly hazardous. Their location is often difficult to predict, and if not identified in a timely manner, the cracks may expand during bridge operation, which usually leads to brittle fracture of the structure and poses a significant threat to the safety of the welded structure. To investigate the location of delayed cracks, a slant Y bevel welding test was designed while considering residual stress. The welding process was monitored using the DIC test system to measure the residual stress generated. The results were compared with the finite element ABAQUS calculated results of welded residual stress to validate and improve the finite element analysis method. Based on the residual stress results after welding, the distribution of diffusible hydrogen was simulated to predict the location of crack emergence. The stress field and diffusible hydrogen enrichment were taken into account for this prediction. In the test to verify the feasibility of the prediction method, the location of delayed crack generation was also observed. This was compared with the finite element prediction results.
The data indicated that: 1) The residual stress distribution in flat plate butt welding structures was inconsistent, with relatively uniform stress in the middle. Longitudinal and transverse stresses were the main residual stress. The stress showed a trend of tensile-pressure distribution, with the longitudinal residual stress in the center of the weld as a large tensile stress, reaching a maximum value of 320 MPa. As the distance between the weld center and the center of the plate decreased, the stress was converted into compressive stress. The residual stress distribution exhibited an 'M’ shape, with a maximum value of 336 MPa. The weld centerline experienced compressive stress. The transverse and longitudinal residual stresses did not exceed the initial yield strength, indicating a self-balancing state within the welded structure. In general, the simulated values of the numerical model and the measured values of residual strain DIC were in good agreement. 2) The distribution of hydrogen concentration in the heat-affected zone was affected by the residual stress gradient, which was higher in the weld zone than in the heat-affected and base metal zones. Additionally, there was a sudden change in residual stress in the transition region from the weld to the heat-affected zone. The distribution of hydrogen concentration in the weld and heat-affected zone was significantly higher than in the surrounding base metal. The root of the weld heat-affected zone exhibited an obvious hydrogen enrichment phenomenon, with the highest concentration of hydrogen in this region. The concentration of hydrogen was the second lowest in the weld area and the lowest in the base metal. As the distance increases, the concentration of hydrogen gradually converged to the initial set concentration. 3) Weld cracks were initiated in the heat-affected zone at the root of the weld. After initiation, the cracks expanded upward in the weld metal and then stopped cracking in the weld. This ensured that the hydrogen-rich and stress-concentrated locations in the finite-element simulation aligned with the location of the test weld crack initiation. The location of crack initiation was determined by the concentration of hydrogen, and changed in the hydrogen-rich area could alter the direction of the cracks.