Post Weld Overlay Residual Stresses in Dissimilar Metal Welds Considering Various Butt Weld and Weld Repair Fabrication Sequences

Abstract

Weld overlays have been used to repair or mitigate stress corrosion cracking (SCC) in both boiling water reactor (BWR) and pressurized water reactor (PWR) nozzle-to-pipe dissimilar metal welds (DMW). One of the contributing factors to SCC is the high tensile residual stresses produced during the fabrication of the original butt weld, especially when local weld repairs were present during the welding process. In analytical simulations to determine the post weld overlay residual stresses, complete simulation of the original butt weld, weld repair and the overlay is desired. However, to reduce the computational effort, it is commonly assumed that the weld repair stresses overwhelm the original butt weld residual stresses such that the original butt weld need not be simulated and only the weld repair is simulated before the application of the overlay. Questions have also been raised as to why the butt weld and/or the weld repair need to be simulated since it is assumed that both of these fabrication processes would be overcome by the weld overlay process. This paper investigates three fabrication sequences in order to determine their effect on the post weld overlay residual stresses: (1) the butt weld is simulated followed by a weld repair and then the weld overlay is applied; (2) the butt weld is simulated followed by the weld overlay with no consideration of a weld repair; (3) the butt weld is not simulated but a weld repair is assumed and the weld overlay is applied. Five different nozzle-to-pipe size configurations were used in the study to determine the effect of pipe size on the three fabrication sequences described above. The investigation indicates that the post weld overlay residual stresses for Cases 1 and 3 are similar and hence simulation of the weld repair alone (without the butt weld simulation) prior to simulating the weld overlay is a reasonable assumption. However, not simulating the weld repair (corresponding to Case 2) may provide different residual stress distribution.