Proportional hazard model (PHM) is a common reliability model in reliability engineering. This model considers the operating condition as covariates. The operating condition is assumed to affect the failure rate proportionally. In the state-of-art applications, this PHM is used as a data-driven model. Nevertheless, some literature argue the PHM has the advantage of physical explanation for its covariates. This paper chooses important failure mechanisms from engineering system to derive the failure model from the physical perspective. Later on, the model is extended to statistical model from the statistical point of view, with the aim to explain the PHM model. For these specific failure mechanisms, the paper investigates the accuracy of the PHM model compared with the physical model. A numerical example is presented to demonstrate the difference between the physical model and the classical PHM model.