Statistical analysis of defects for fatigue strength prediction and quality control of materials

A wide range of studies and experimental evidence have shown that the lower bound of fatigue properties can be correctly predicted by considering the maximum occurring defect size. The estimate of this dimension can be done by analysing the defect sizes using the statistics of extremes. The scope of this paper is to discuss and investigate the two key points in a successful application of this technique: the first is the choice of statistical method for the analysis of data; the second is the knowledge of the minimum number of defects needed to obtain a good estimate of extreme defects. The results obtained in this study allow one to formulate a procedure: for estimating the extreme defects with a precision suitable for fatigue strength prediction.