Abstract

In many ferritic steel nuclear pressure vessels, weld deposited austenitic stainless steel or nickel-based alloy is used as a protective cladding for corrosion resistance. In the UK, it is common practice to provide additional assurance of structural integrity through a flaw tolerance assessment of postulated defects carried out during the design process. These assessments are usually carried out using the R6 failure assessment diagram-based J estimation approach and are targeted at regions with the highest likelihood of structurally significant defects, such as welds. Defining an appropriate method for assessing clad regions is therefore important to provide appropriate assurance without the need for multiple, highly refined assessments of clad regions.

A previous paper [1] described studies into the impact of Cladding Residual Stresses (CRS) on crack driving force calculations using the R6 procedure. It was concluded that, in many cases, it may be appropriate to exclude CRS from the defect tolerance assessment, especially in cases where residual stresses have been relaxed by prior application of large primary loads, such as pressure tests. This paper further develops the studies to implement refinements that address two main limitations in the previous work: (1) the simplified residual stress profile (which did not include any tensile stress in the parent material adjacent to the cladding interface) and (2) the fact that only fully circumferential defects were considered.

A series of cracked body Finite Element (FE) models have been analysed to determine the crack tip loading under different combinations of pressure, residual and thermal stresses. Stress Intensity Factors (SIFs) derived from elastic-plastic FE models were compared to R6 results to determine the levels of conservatism when applying different assumptions. Three different residual stress fields were considered: (1) no CRS, (2) an idealised version of the profile obtained from deep hole drilling measurements and (3) 80% of the idealised profile. Three different R6 methods for modelling the interaction between primary and secondary loads were also considered. Both fully extended and semi-elliptical defects were considered. The modelling was carried out in Abaqus and a novel method using user defined variables to control the material properties was used to implement the material boundaries without the need to include the material boundaries in the mesh. This avoided potential difficulties in achieving a high-quality mesh at locations where the crack front intersects with a change in material.

The investigations confirmed the conservatism in the R6 approach. However, it was found that, with the more representative CRS profile, it was not possible to support fully removing the CRS as a frontline argument in assessments, although the levels of non-conservatism are small. For assessments with low margin, a sensitivity study with CRS removed may be useful, especially if there are other conservatisms in the assessment as the studies have shown that, for SIFs approaching typical parent toughness, the KJ estimates following the R6 procedure are within 5–10% of the values obtained by FE even when CRS is not included.

The conclusions are currently only supported for bulk clad regions — ie cylindrical, spherical or ellipsoidal regions with no underlying structural weld, but could be extended to other geometries in the future.

This content is only available via PDF.
You do not currently have access to this content.