Sequential inspection sampling to avoid failure critical items being in an at risk condition

The problem of how to extend the time interval of fixed time period maintenance for items whose failure can be catastrophic is considered in this paper. The paper proposes a coherent methodology particularly applicable to switchgear used within electricity distribution networks. The methodology involves taking a sample of the items and, based on their observed conditions, deciding whether to (1) maintain; (2) defer maintenance; or (3) take another sample. Consideration of the precise problem to be solved leads to a Bayesian formulation. The predictive distribution is then used to determine the expected outcome of taking further observations. Results using simulated and real data are reported. This sequential sampling approach seems particularly appropriate for distribution networks where inspection costs can be relatively high.