A number of previously published works have shown that the presence of residual stresses can significantly affect measurements of fracture toughness, unless they are properly accounted for when calculating parameters such as the crack driving force. This in turn requires accurate, quantitative residual stress data for the fracture specimens prior to loading to failure. It is known that material mechanical properties may change while components are in service, for example due to thermo-mechanical load cycles or neutron embrittlement. Fracture specimens are often extracted from large scale components in order to more accurately determine the current fracture resistance of components. In testing these fracture specimens it is generally assumed that any residual stresses present are reduced to a negligible level by the creation of free surfaces during extraction. If this is not the case, the value of toughness obtained from testing the extracted specimen is likely to be affected by the residual stress present and will not represent the true material property. In terms of structural integrity assessments, this can lead to ‘double accounting’ — including the residual stresses in both the material toughness and the crack driving force, which in turn can lead to unnecessary conservatism. This work describes the numerical modelling and measurement of stresses in fracture specimens extracted from two different welded parent components: one component considerably larger than the extracted specimens, where considerable relaxation would be expected as well as a smaller component where appreciable stresses were expected to remain. The results of finite element modelling, along with residual stress measurements obtained using the neutron diffraction technique, are presented and the likely implications of the results in terms of measured fracture toughness are examined.
