Residual stress estimation in defect assessment procedures at weld toe and away locations on girth welds: Review of key parameters

The distribution of residual stresses in welded joints plays an important role within the fracture evaluation guidelines recommended in structural integrity assessment codes such as BS7910, API 579 RP-1/ASME FFS-1 and R6. The residual stress profile recommendations in these standards are based on extensive experimental results and finite element modelling (FEM) based parametric residual stress evaluations at the weld centerline and weld toe positions. The upper bound residual stresses’ profiles based on these recommendations vary significantly from one type of welding process to another for a given weld configuration with identical welding conditions. These fitness-for-service codes (FFS) depict great variability in estimating residual stress profiles during defect assessment, as BS 7910 & R6 recommends a constant profile at a distance away from welds and API 579 provides a single curve for all locations in the axial direction. Thus, conservatism is widely associated with these recommended profiles in fracture potential evaluation and assessments, leading to suboptimal recommendations. In this manuscript, a detailed review is undertaken of residual stress estimation in various FFS codes, showing vast variability among them for locations away from the weld toe on girth welds. Key distinct parameter characteristics, pipe radius to thickness ratio and heat input are detailed and found to have a significant effect on residual stress profiles in structural integrity assessment, using a stress decomposition technique. These recommendations establish an overall analysis of the interrelationship between key parameters, considering a generalized broad range of applications. A framework is proposed, based on the current review, for conducting detailed investigation by employing thermomechanical numerical modelling, coupled with measurement results (nondestructive and semi-destructive) from an experimental study, as input to machine-learning algorithms for application guidance to engineers.publishedVersio