Repairable Systems Reliability

Abstract

Abstract Nonrepairable systems fail just once, so models for nonrepairable systems must account for random lifetimes. It is often reasonable to assume that different units have random lifetimes that are independent and follow the same distribution, leading to the usual i.i.d. (independent and identically distributed) assumption. By contrast, models for repairable systems must account for successive failures in time. For a given system, these times between failure are often not independent and not identically distributed. Various assumptions about the failure process lead to different models for repairable systems. In this article, we present the nonhomogeneous Poisson process (NHPP), the renewal process, the piecewise exponential model, and the modulated power law process (a compromise between renewal process and NHPP). Inference for a single system as well as for multiple copies of a system is discussed. We also discuss briefly the use of covariates or concomitant variables.