Statistical analysis of interlaboratory dynamic fracture toughness data

This paper describes the first phase of a large test programme designed to determine the mechanical behaviour of nuclear pressure vessel steels in which the main emphasis was on the measurement of dynamic fracture toughness. In addition to more conventional techniques the instrumented precracked Charpy impact test technique for the determination of fracture toughness from small specimens was used extensively. The programme described here deals with the interlaboratory testing of a single material, A533B, Class 1, steel from HSST plate 02. The programme set out to establish appropriate statistical techniques for treating toughness data and to determine whether the methods proposed for fracture toughness measurement were equivalent. The results of the programme show that there are real differences between the testing methods. There was no evidence of bias among the different test laboratories, but some had significantly lower experimental error than others. These observations suggest that some of the scatter used to determine fracture toughness curves derives from testing and not materials. Improved procedures should result in less stringent limits. A set of acceptance criteria for the fracture toughness tests was examined. Data points which did not meet these criteria were found to have no significant influence on the elastic or elastic-plastic results. (author)