Reliability assessment method for high-reliability relays in zero-failure accelerated testing

This study proposes an innovative modified one-sided lower confidence limit approach that incorporates virtual failure integration to address uncertainties in confidence evaluation arising from zero failure observations during accelerated testing in high-reliability relay assessments. A suitable composite failure rate is derived using the virtual failure dataset in conjunction with the Bayesian parameter estimation method, applied in the context of three vacuum circuit breakers for which no failure events were recorded. Utilizing this method, the mean time between failures (MTBF) is reduced by 71.5 percent compared to results obtained using the conventional one-sided confidence limit approach. This reduction is further supported by an increase in engineering conservatism relative to standard methodologies. The proposed approach offers a statistically robust framework for evaluating confidence levels in components tested under small-sample, zero-failure conditions.