Lifetime prediction and design of reliability tests for high-power devices in automotive applications

Abstract

Different procedures are defined and compared to extract the statistical distribution of the thermal cycles experienced by power devices installed in hybrid vehicles operated according to arbitrary mission profiles. This enables both to design efficient accelerated tests tailored on realistic data and to provide the input for lifetime prediction models. Initially, the system lifetime is predicted under the assumption of linear accumulation of the damage produced by low cycling fatigue. In the second part, a novel prediction model based on some fundamental equations is introduced which takes into consideration the creep experienced by compliant materials when they are submitted to thermal cycles.