TWI Industrial Member Report Summary 618/1997
H G Pisarski
Conventional deterministic flaw assessment procedures which incorporate partial safety factors, as recommended in BSI PD 6493: 1991, have been conducted. The target reliabilities implied by these procedures were compared with reliabilities derived from probabilistic fracture mechanics analyses. Significant differences are noted and a cautious approach in applying the partial safety factors is recommended.
Background
To ensure a conservative outcome, deterministic fracture mechanics assessments often use worst case assumptions for the input variables in order to decide the acceptability of a known flaw in a welded component. With the BSI PD 6493 flaw assessment procedure, the flaw is considered acceptable provided that the assessment point falls on or below the failure locus of the relevant failure analysis diagram (FAD). The risk of failure is not quantified. Where a higher degree of confidence in the result is required, Appendix A of BSI PD 6493 recommends use of partial safety factors which are applied to flaw size, stress and fracture toughness. Their use is intended to ensure that the component achieves a certain target reliability. Choice of a target reliability depends on the consequences of failure.
The partial safety factors were derived from reliability analyses using procedures different from those in the current document. Consequently, it is appropriate to examine the implications of applying the partial safety factors in the current procedure on the probability of failure.
Objectives
- To examine the implications of using currently recommended partial safety factors on the probability of failure when conducting flaw assessments to the Level 2 FAD in BSI PD 6493.
- To assess the significance of using the minimum of three fracture toughness test results on the calculated probability of failure.