RELIABILITY ASSESSMENT IN THE PRESENCE OF RELIABILITY GROWTH

A methodology is presented for estimating current and future reliability of complex weapon systems that show reliability growth during their development and early operational phases. The study proposes four reliability growth models or patterns that can be fitted to actual data experience to determine the quantitative characteristics of the growth within relatively well- defined tolerances. This is achieved by defining appropriate parametric models and subsequently using maximum likelihood procedures to obtain estimates of the parameters. Comparison of the models shows that under the conditions set forth in this study, three are generally superior in their predictive and assessment characteristics to representative nonparametric methods and to an applicable Bayesian procedure.