Statistical Analysis of Residual Stresses in a Stainless Steel Edge Welded Beam

A statistical approach based on Bayes’ theorem has been applied to determine the ‘best estimate’ means of two sets of measured residual stress profiles at various point locations through the depth of a stainless steel edge welded beam. Residual stresses introduced into an AISI Type 316H stainless steel beam (250mm × 50mm × 10mm) by welding along the 10mm edge have been measured using high energy synchrotron and neutron diffraction techniques. Mean values of measured residual stresses are compared with predicted stresses obtained from a two-dimensional finite element simulation employing isotropic, kinematic and mixed isotropic-kinematic hardening models. The accuracy of the prediction has been assessed using a root mean square error approach (RMS).