On the Expected Number of Failures and Maintenance Cost Prediction of Repairable Systems From Life Cycle Cost Modeling Perspective

Abstract

The objective of this paper is to provide some useful insights on how cost driving events are related to the characteristics of failure distributions and the product lifetime (design life) in case of repairable systems. Repairable systems are those that can be restored to their fully operational capabilities by any method, other than the replacement of the entire system. In case of repairable systems, the components can be repaired or adjusted rather than replaced, whenever a breakdown occurs and thus such systems experience multiple failures over their life span. For majority of repairable systems, the life time maintenance and repair costs dominate the life cycle cost. To predict the maintenance and repair cost, failure data, maintenance data and repair time data is needed which is not readily available at the system design stage. When a repairable system is put into service, how many times it will fail over its life span depends on its reliability. Similarly, how fast the system is restored to its working condition when it fails (maintainability), also affect the costs incurred. Thus, the expected number of failures, time lost in restoring the system after each failure and cost per failure are important from life time maintenance cost prediction viewpoint. The expected number of failures depends upon the time to failure distribution of the system components and the after repair state of the system. In this paper, a modeling methodology is suggested for prediction of life time maintenance and repair cost of repairable systems based on expected number of failures. The repairable system lifetime is modeled using a two parameter Weibull distribution. The expected number of failures are estimated for renewal process (as-good-as-new after repair state) and minimal repair process (as-bad-as-old after repair state). The expected maintenance and repair costs are also evaluated for six different failure distributions. The technique has been illustrated through a specific application, namely an industrial pump and the results are presented.