A statistical framework for analysing weld residual stresses for structural integrity assessment

Thermal and mechanical processes during welding introduce complex three-dimensional distributions of residual stress. Management of residual stresses represents a major challenge for engineers in order to achieve safe and reliable operation of existing engineering plants. Consideration of how such stresses vary through the wall within welded components is critical in structural integrity assessments. Development of more accurate and realistic weld residual stress profiles through statistical analysis of high quality measured data is highly desirable. This can not be achieved without adequate interpretation of measured weld residual stress data through the development of an appropriate framework. This paper proposes a framework for analysing measured data at four levels of complexity depending on its spatial distribution. The framework provides clear guidance on how to perform appropriate statistical analyses of residual stress data and proposes how to deal with some practical issues that may arise from various stress measurements; for example the gauge size, spatial distribution, measurement uncertainties and data analysis assumptions. Parts of the proposed framework are then applied to measured stress data in a simple welded laboratory test specimen.Copyright © 2008 by ASME