A Multivariate Gamma Process for Dependent Degradation Modelling and Life Phenomena

Remaining service life of infrastructure assets is governed by functionality and structural integrity, both deteriorate with time. In this paper, we propose a multivariate gamma process model to model the stochastically dependent deterioration phenomena that collectively define the asset life. The temporal uncertainty is characterized by nonstationary gamma processes with independent increments while dependence among degradation processes is characterized with a correlation matrix in the copula space. Parameter estimation is done by the maximum likelihood method. For the lifetime prediction, a component experiencing multiple degradation phenomena is said to fail based on a number of scenarios. So the remaining lifetime distribution will be based on the current state of the component as well as failure thresholds of all phenomena. The proposed methodology is illustrated with a case study of a highway pavement experiencing multiple degradation such as rutting, cracking, and surface smoothness