Competing failure modes in accelerated life testing

There is a vast amount of literature on accelerated life testing, however, most of this literature ignores the possibility of competing modes of failure. The literature that attempts to address this problem often uses a maximum likelihood estimation method which may require large samples. Even in the case of a single failure mode, a small sample data is expected to be generated by an ALT procedure applied to expensive components. In this article we present a Bayesian framework for the analysis of ALT data with possible multiple failure modes. We illustrate the applicability of our model on some competing risk data sets available in the literature.