Effect of Ranking Selection on the Weibull Modulus Estimation
Determination of mechanical properties of materials requires repeatable and consistent results. Therefore, the Weibull statistical distribution is widely used to verify the confidence levels of the test results where probability of failure is predicted. The reliability of results is characterised by Weibull distribution where Weibull parameters are determined and compared with the benchmarks. Recent studies showed that the size of the population and the method chosen to estimate the Weibull modulus play an important role. Therefore, in this study, the effect of ranking selection over the Weibull parameters (alpha-characteristic life, beta-shape parameter, R2 and survival probability) was investigated. The data from authors’ recent research were re-evaluated. In addition, randomly generated data with population from 5 to 50 sample sizes were studied. The results showed that the mean rank regression to estimate the failure probability had the highest R2 and the lowest shape parameter. It was also found that the results were independent of sample size.
Keywords: Estimators, mechanical test, prediction limit, Weibull distribution, Weibull modulus