An Example of Analyzing Accelerated Life Test Results with a Kinetic Model.

A central problem in reliability modelling is "How does one design and analyze accelerated life tests so that one reasonable belief in the extrapolations from the analysis?". This paper describes a first cut at the problem. In this paper we demonstrate how acceleration transforms, which are statistical models of data derived from kinetic models of degradation processes, can be used to model data from accelerated life tests. An example is given using data from an experiment investigating the reliability of printed wiring boards. We then consider the model in more depth and show that constant stress accelerated life tests do not provide sufficient information for modelling the kinetics of degradation, but (at least for this example) step stress accelerated life tests do.