Risk-informed maintenance for non-coherent systems

Abstract

Risk Importance Measures (RIMS) obtained from both qualitative and quantitative aspects of Fault Tree (FT) analysis can be used to identify weak links in a system. Information from RIMS can be used to direct resources towards the components that deserve the most attention. When RIMS are used to make maintenance-related decisions, it is referred to as risk-informed maintenance. Risk importance analysis for coherent FT has received much attention over the years. However, non-coherent FT does occur in real systems due to either the nature of the system or poor design. Non-coherent FT introduces difficulties in terms of both qualitative and quantitative assessment, and the importance analysis of noncoherent FT is rather limited. In this paper, eight most commonly used RIMS are investigated and extended to noncoherent forms. They are the Birnbaum's Measure (BM), Criticality Importance Factor (CIF), Improvement Potential (IP), Fussell-Vesely Measure (FV), Risk Achievement (RA), Conditional Probability (CP), Risk Achievement Worth (RAW) and Risk Reduction Worth (RRW). The feasibility of the extension are proved and presented throughout the analysis and applications. Furthermore, they are classified with respect to risk significance and safety significance. The CIF, IP, FV and RRW are identified as risk significant measures, while BM, RA, CP and RAW are identified as safety significant measures. Since maintenance can normally be categorized as corrective maintenance and preventive maintenance, it is concluded that risk significant measures contribute most information to corrective maintenance and safety significant measures contribute most information to preventive maintenance. An Automatic Power Control System (APCS) for an experimental nuclear reactor is used as a case study to demonstrate the theoretical development.