Determination of Distribution Function Used in Monte Carlo Simulation on Safety Analysis of Hydrogen Vessels

The test data of static burst strength and load cycle strength of composite pressure vessels are often described by GAUSSian normal or WEIBULL distribution function to perform safety analyses. The goodness of assumed distribution function plays a significant role in the inferential statistics to predict the population properties by using limited test data. Often, GAUSSian and WEIBULL probability nets are empirical methods used to validate the distribution function; Anderson-Darling and Kolmogorov-Smirnov tests are the mostly favorable approaches for Goodness of Fit. However, the different approaches used to determine the parameters of distribution function lead mostly to different conclusions for safety assessments.
In this study, six different methods are investigated to show the variations on the rates for accepting the composite pressure vessels according to GTR No. 13 life test procedure. The six methods are: a) Norm- Log based method, b) Least squares regression, c) Weighted least squares regression, d) A linear approach based on good linear unbiased estimators, e) Maximum likelihood estimation and f) The method of moments estimation. In addition, various approaches of ranking function are considered. In the study, Monte Carlo simulations are conducted to generate basic populations based on the distribution functions which are determined using different methods. Then the samples are extracted randomly from a population and evaluated to obtain acceptance rate. Here, the “populations” and “samples” are corresponding to the burst strength or load cycle strength of the pressure vessels made from composite material and a plastic liner (type 4) for the storage of hydrogen. To the end, the results are discussed, and the best reliable methods are proposed.