Accelerated Life Tests: Classical Methods for Design and Analysis

Accelerated life tests are used to obtain quickly information about the failure time distribution of very reliable technological products. Tests are conducted in conditions more severe than the normal operating ones with the aim of accelerating the mechanism of failure and thus shortening the life of components and materials of interest. An estimate of lifetime distribution at normal use conditions is extrapolated from data obtained at higher stresses by using appropriate statistical models, which incorporate special relationships accounting for the effect of stress on lifetime of products under study. After a brief introduction that describes motivations and objectives of accelerated testing, the following topics are addressed: • Methods for accelerating reliability tests. • Models for classical accelerated tests. • Procedures for point estimation, interval estimation, and model checking. Basic criteria for planning accelerate life tests. In this article, only classical fully parametric estimation procedures are discussed. Moreover, only models for constant stress ALT data are considered, that is, data obtained by running tests at constant stress with a single accelerating variable. All the models presented in this paper are for single failure mode: they cannot be used for products that may fail in more than one mode.