Code characterisation of weld residual stress levels and the problem of innate scatter

Economic and safe management of operating nuclear power plant is increasingly dependent upon structural integrity assessments for pressure vessels and piping. The residual stress distribution assumed in defect assessments for welded joints often have a deciding influence on the analysis outcome. Guidance on characterising conservative levels of weld residual stress can be found in structural integrity codes and procedures such as R6, API 579 and BS7910. There is an increasing need to develop more realistic and reliable residual stress distributions that will deliver more accurate integrity assessments. However, future development of such distributions will have to deal convincingly with what is often termed the “innate scatter” of weld residual stresses. This paper first identifies and illustrates some of the origins of apparent innate scatter. The stability of the welding process is examined. The importance of transient weld bead starts/stops and the lay-up of passes in multi-pass welds are demonstrated. Uncertainties associated with two commonly used residual stress measurement techniques are reviewed and simple quantitative studies used to reveal the role of measurement gauge length and, more significantly, errors in spatial location on the level of measured residual stress. The final part of this paper surveys how structural integrity codes and procedures currently characterise welding residual stresses for defect-assessment purposes and discusses the development of more realistic residual stress profiles based on statistical treatment of scatter and uncertainties.