The results of an experimental test program to investigate methods of predicting the influence of specimen size on the cleavage fracture toughness of A533B-1 steel are presented. Three-point bend specimens from 10 to 230-mm thick were tested; to minimize scatter in the results, these were all extracted from a single plate of A533B, such that their crack tips lay on the midplane of the plate. A total of 217 tests were performed so that enough data could be obtained to allow statistical analysis.
At any one temperature there was considerable scatter in the results, which did not diminish with increasing specimen size. However, at a particular temperature, the mean or lower-bound fracture toughness at cleavage decreased with increasing specimen size. The fracture toughness was analyzed as a function of size and temperature using exponential curve fitting procedures, multiple linear regression analysis, Weibull statistics and the βIc correction, and the significance of each of these methods is reviewed. The size effect can be quantified in terms of a shift in temperature for a given toughness level. The shift, however, is dependent on toughness, being smaller at the higher levels of toughness, and there is a tendency toward saturation as the fully ductile state is approached.
The implications of this work in respect to obtaining data for use in structural integrity analysis is discussed.