Abstract
The sensitivity of residual stress distributions in bainitic–martensitic steel welds to the transformation strains that arise when austenite decomposes on cooling has been assessed by examining the predictions of three models for a simple bead-on-plate configuration. These cover the following scenarios: case I, no phase transformations; case II, transformations with volume change effects only; case III, transformations with volume change effects and associated Greenwood–Johnson transformation plasticity. Austenite decomposition was predicted by implementing Kirkaldy's reaction rate equations as a subroutine in the finite element code Sysweld, eliminating the need for a continuous cooling transformation diagram. Predicted residual stresses were then compared and rationalised alongside measurements obtained by neutron diffraction and the contour method. It was found that serious errors in predicting the location and magnitude of the peak stresses occurred if transformations were not included, while cases II and III gave similar results generally in agreement with the stress maps. Indeed, the trends in the experimental results were intermediate between cases II and III. Differences between the models and the potential for further improvements are discussed.
Keywords
Get full access to this article
View all access options for this article.
