A time-discrete extended gamma process for time-dependent degradation phenomena

The non-stationary Gamma process is a widely used mathematical model to describe degradation phenomena whose growth rate at time t depends only on the current age of the item and not on the accumulated damage up to t. Nevertheless, the Gamma process is not a proper choice when there is empirical evidence that the variance-to-mean ratio of the process varies with time, because the Gamma process implies a constant variance-to-mean ratio. This paper proposes a generalization of the nonstationary Gamma process, which can be viewed as a time discretization of the extended Gamma process and allows one to describe time-dependent degradation phenomena whose variance varies with time t, not necessarily in proportion to the mean. A way to approximate the exact distribution of the degradation growth over a given time interval is given and a test for assessing whether the assumption of the Gamma process can be rejected or not is discussed. Finally, the proposed model is applied to a real dataset consisting of the sliding wear data of four metal alloy specimens.