The paper proposes a robust Bayesian methodology developed for a real-world case to assess the failure rate of different kinds of low-pressure pipline use in metropolitan gas distribution networks. Assessment of the probability of failure allows effective and efficient maintenance and replacement policies to be set correctly. In this concern, the main problems involved are: identification of factors leading to failure; use of historical data on failures, generally limited and incomplete, and in any case of scarce value in view of the continuous change in running conditions.
展开▼