RELIABILITY ANALYSIS FOR NONREPAIRABLE SYSTEMS SUBJECT TO DORMANCY (WARRANTY, WEIBULL, COVARIATES)

Dormancy is a factor impacting the reliability of military systems, private aircraft, emergency, and seasonal equipment. This research investigated reliability analysis and modeling for nonrepairable systems designed for a life cycle of two consecutive phases, dormancy and active use.;Extensive quantitative studies were performed on a warranty data set for air-conditioning compressors. The compressors were installed early in the model year, and experienced a dormant period prior to seasonal active use. Analyses included nonparametric estimation of life distributions by life-table and product-limit methods. Estimates of the Weibull parameters and covariate regression coefficients were determined by the method of maximum likelihood, using a Newton Raphson algorithm.;These studies affirmed the relevance of dormant reliability, Weibull and proportional hazards concepts. The research demonstrated that the reliability models are applicable to field warranty analysis on actual commercial equipment.;Parametric studies and sensitivity analyses were performed on the models, using parameter estimates for space-based systems, air-conditioning compressors and accelerated testing models. A computer program was developed for use on a microcomputer as an analytical tool. This software is potentially useful by practitioners for reliability predictions and design trade-off studies.;Point-estimate reliability models were developed for multiple-phase life cycles, based on the Weibull life distribution. Proportional hazards were employed to formulate models relating the Weibull characteristic life, or scale parameter to explanatory covariates. This approach facilitated the use of heterogeneous field failure data to estimate the model parameters. It also led to models for accelerated dormant aging (preconditioning) and life testing.;The modeling and analytical concepts resulting from this research were integrated with military standards for prediction and testing, and with existing practices of reliability and design engineers.