Evaluating the Series or Parallel Structure Assumption for System Reliability

摘要

The structure of a system is important for specifying a mathematical form for calculating reliability from component level data. For some systems, full-system reliability data are expensive or difficult to obtain, and using component-level data to improve the estimation of system reliability can be highly beneficial. However, if the intended structure of the system during design does not match the actual implementation after production, using the component-level data to estimate system reliability can give misleading results. Hence, we propose a test for assessing the appropriateness of assuming both independence of component failure mechanisms and a series system structure for a system when data at both the system and components level are available. Estimates are given for component and system reliabilities under the assumption of a series system. The test can also be easily adapted to test the assumption of a parallel system. Examples for both series and parallel systems are illustrated.