Modelling systemic risks to inform a repowering decision

This article describes a method for estimating the reliability of a system under development that is an evolution of previous designs. We present an approach to making effective use of heritage data from similar operational systems to estimate reliability of a design that is yet to realise any data. The approach also has a mechanism to adjust initial estimates in the light of sparse data that becomes available in early stages of test. While the estimation approach, known as empirical Bayes is generic, we focus on one shot devices as this was the type of system which provided the practical motivation for this work and for which we illustrate an application.