Unbiased Variance Estimates for System Reliability Estimate Using Block Decompositions

An unbiased estimator is proposed to calculate the variance of a system reliability estimate based on the estimated variance of component reliability estimates. The method does not require any parametric assumptions for component reliability or time-to-failure, and it allows Type-I and Type-II censored data. The approach can be applied to many situations as long as the system can be appropriately decomposed into series or parallel configurations. The new model is compared with existing methods using different reliability data and system structures. The empirical results show that the new model is generally superior in terms of computational efficiency, and estimation accuracy.