Reliability modeling and prediction of systems with mixture of units

Abstract

Traditional reliability analysis and predicting are performed by utilizing observed failure time or degradation data of test units or systems. However, in many cases, the reliability testing is performed during the entire life of the units by testing different samples with different characteristics at different times. This system is referred to as "system with mixture of units". A typical test is performed by either testing all units in the system or by taking a sample. In both cases, the history of units under test is known including the age if the unit, time of previous repair (if any). This paper investigates an effective approach for estimating the system reliability by taking the characteristics of this mixture of units into account. We obtain expressions for reliability metrics for different scenarios.