Copula‐based Bayesian reliability assessment for series systems

Bayesian reliability assessment plays an integral role in determining the success of product development and deployment into the market, especially when limited information is available to estimate the reliability of products. Nevertheless, the Bayesian reliability approach has untapped research potential when it is used to evaluate systems consisting of dependent components, that is, a situation that reflects real‐world conditions. This study addresses the research gap by applying a copula to encode the dependency structure between the failure rates of the components, particularly for the simple series systems. Unfortunately, this study found that Bayesian aggregation error still occurs when information is used at different levels. By employing Monte Carlo simulation and classification tree learning, this study also investigates the key factors affecting such error over a wide range of possible scenarios and derives useful guidelines for reliability practitioners so that they can choose a more suitable analysis under certain circumstances.