Application of 3D Finite Element Modelling to Repair Weld Simulation in Industrial Applications
The Finite Element Method (FEM) has been implemented in 3D to predict welding residual stresses in repair welds. The analysis has been used to achieve more accurate residual stress predictions for the weld at the cost of long computation times. The use of this CPU intensive approach has been facilitated by the advent of ever-faster computer processors being made more accessible to the engineering community. The same technique has also been used with coarser meshes involving simplified welding sequences where a number of weld passes are “lumped” together to reduce the simulation time. The authors argue that this latter approach can be very useful in predicting the more global component response — in cases where 2D model symmetries are not applicable — and for rapid identification of problem areas where finer simulations would be prohibitive. The authors show an example of a residual stress prediction for a letterbox repair obtained using the FEM. Good agreement between this prediction and experimental measurements is shown. The FEM simulation technique has been used to predict residual stress formation during the welding process and subsequent service loading of the component. This analysis shows the residual stress field relaxation following “shakedown”. The component under service conditions is subjected to pressure loading and a small amount of bending stress. Based on recent residual stress experimental programmes conducted at Mitsui Babcock Energy Limited (MBEL), the authors provide a brief discussion on the ways in which various experimental techniques have been used to verify welding residual stress predictions from FE. The authors argue that just as there has been an interest in the field to measure residual stresses in the highly stressed regions of a weld, it is equally important to measure stresses in areas of relatively low stress to confirm that stresses do indeed die out away from welds. It is in the latter case where some experimental techniques cannot perform as well as other simple, well proven, strain measurement techniques.