Throughput-based importance measures of multistate production systems

Abstract

Many production systems are multistate, with a finite number of performance levels that are between perfect functioning and complete failure. Importance measures are often used in the maintenance planning of complicated systems, to observe the criticalities of components, reveal the system weakness, and thus to guide the allocation of limited maintenance resources. This paper compares several commonly used importance measures for multistate systems, and investigates their effectiveness and limitations with a simple example. These existing measures focus on the states of a system at some moment, while ignoring the dynamic behaviours in the long-term. For a production system, however, its throughput in a certain period, rather than the instantaneous performance, is the system property of interest. Therefore, two new long-term throughput-based importance measures: total throughput importance measure and maintenance effect importance measure are proposed in this paper, to answer the questions about the criticalities of different components and the long-term effects of successful maintenance activities on the throughput of a production system in a certain period. A case study on an offshore production system is conducted, to illustrate how the new importance measures work and what kind of implications can be provided to the maintenance crew.

Keywords:

• multistate production system
• importance measure
• maintenance
• performance analysis
• total throughput importance measure
• maintenance effect importance measure
• Markov model

Acknowledgement

The authors really appreciate for the insightful comments from the three reviewers.