Evaluation of flaw tolerance. Use of a sensitivity analysis on the input data in contrast to rigid safety factors

The acceptability of defects in structures, whether postulated or real, is most commonly judged on the basis of a fixed factor between the defect size and the critical defect size at which failure of the component might be expected to occur. In this paper that approach is shown to be somewhat arbitrary and not necessarily consistent with conventional design factors such as that on maximum stress. A sensitivity analysis, performed by systematically varying input parameters in ways justified by the status of the data, is encouraged by the CEGB defect assessment route. Such a procedure is carried out in this paper on a problem which could occur in a nuclear pressure vessel. It is demonstrated that confidence in the analysis can be increased and the true risks of the final judgement quantified. This type of analysis can be used to justify or discourage more detailed analytical procedures or experimentation, more accurate data collection or modification of plant operating regimes. (author)