Reliability analysis considering product performance degradation

This dissertation presents a statistical model and analysis procedure for product performance aging degradation data. This model takes into account the strictly increasing/decreasing nature of performance measurements at multiple observation times. Maximum likelihood estimation (MLE) is used to estimate the time varying parameters of the proposed statistical model. The analysis of both generated data and field data is presented. To demonstrate product reliability under aging, an analysis of surface mounted solder joints due to thermal fatigue is included in the dissertation. This analysis was done by first examining published life test data and then identifying the intermetallic compound (IMC) thickness randomness. Results indicate that the IMC layer thickness randomness may have significant influence on the Mean Time To Failure (MTTF) and the reliability at high thermal cycles. The analysis of products with competing hard and soft failure modes is presented in terms of distribution independence. Derivation and examples are included for the event when the product finally fails in a specific failure mode. Finally, an improved strength-stress interference (SSI) reliability model is derived for analyzing a more general engineering degradation problem. This model incorporates both stochastic strength aging degradation and the stochastic loading force directed at the product. Statistical inference for simple stochastic processes and numerical examples are analyzed and discussed to verify the model