An accuracy analysis of Army Material System Analysis Activity discrete reliability growth model

Abstract

The accuracy of the discrete reliability growth model developed by Army Material System Analysis Activity (AMSAA) is analysed. The mean, standard deviation, and 95 precent confidence interval of the estimate of reliability resulting from simulating the AMSAA discrete reliability growth model are computed. The mean of the estimate of reliability from the AMSAA discrete reliability growth model is compared with the mean of the reliability estimate using the Exponential discrete reliability growth model developed at the Naval Postgraduate School and with the actual reliability which was used to generate test data for the replications in the simulations. The testing plan simulated in this study assumes that the mission tests (go-no-go) are performed until a predetermined number of failures occur at which time a modification is made. The main results are that the AMSAA discrete reliability growth model always performs well with concave growth patterns and has difficulty in tracking the actual reliability which has convex growth pattern or constant growth pattern when the number of failures specified equal to one. Keywords: Reliability growth, Estimate, Mean, Standard deviation, Thesis