Local Stress Variations at Bead Interruptions in Austenitic Multi-Pass Groove Welds

This paper presents finite element analyses of residual stress in an austenitic multi-pass groove weld. The aim was to establish the effect upon the residual stress of stop-start interruptions during the deposition of weld beads. Comparison of measured residual stress profiles with the residual stress distributions predicted by finite element (FE) modelling aimed to validate the FE method for predicting residual stresses around stop-start features. This paper presents a comparison of measured and modelled residual stress distributions in a series of simple welded 304 stainless steel plates. The plates were machined with a v-groove designed to be filled using eight weld passes. Samples which included interrupted weld beads contained two stop-start features in the fifth pass. In the first feature the welding power was ramped down over 15 seconds; this represented normal welding good practice. The second feature investigated was an abrupt stop, where the welding power was removed instantaneously; this represented an extreme stop. Three welded plates were considered. One contained five weld passes, such that the final pass contained stop-start features and resulted in partially filling the weld groove. Two welds plates each containing eight passes have also been considered; one contained stop-start features in the fifth pass and the other contained no stop-start features. This allowed a comparison of the effect of stop-start features and the effect that subsequent beads have upon any perturbations in the residual stresses produced. Residual stress measurements have been performed using neutron diffraction. 3D weld modelling has been carried out using VFT and the Abaqus finite element package. Results from the welding FE analyses were compared with the neutron diffraction measurements. Good agreement between the modelled and measured residual stresses is achieved in the uninterrupted 8-pass sample and after deposition of the bead containing stop-start features in the 5-pass sample. Following deposition of subseqeunt beads perturbations in the residual stress profile are retained in the neutron diffraction measurements, but all perturbations are removed from the residual stress profiles predicted using both VFT and Sysweld. This work suggests that modelling welding stop-start features is only necessary in the final weld capping passes, if residual stresses over a short length scale are of interest.