This thesis studies the test methods and test plans of MIL-STD-781D and MIL-HDBK-781: Test methods are for evaluating data produced during reliability development and growth test, and environmental stress screening (ESS) programs. Growth data evaluation methods are the Duane Method and the Army Material Systems Analysis Agency Method, and ESS evaluation methods are the computed ESS time interval method and graphical methods. Test plans are for mean-time-between-failures assurance, and fixed-duration and sequential reliability demonstration and assessment tests, and all-equipment reliability tests.
For each of the four methods and four plans, its statistical and operational aspects are investigated and examples are given. The robustness of fixed-duration and sequential tests, when the basic assumption of exponentially-distributed time between failures is violated is studied by Monte Carlo simulation for the following two cases; i) the time between failures is a lognormal or gamma-distributed random variable, and ii) the number of failures during (0, t) is a non-homogeneous Poison process with a Weibull intensity function.
