Selective Maintenance for Systems under Imperfect Maintenance Policy

Abstract

Due to the advent of complex engineering systems in the last few decades, reliability and maintenance issues are gaining more attention than ever. Maintenance is important to keep a system running and ensure that it performs its functions satisfactorily. Often, there are limited resources to complete the maintenance of a system. Such a limitation may arise due to limited budget, time, or repairman availability, etc. Under these circumstances, it is required to optimally allocate the available resources in a way that selected components within a system and maintenance actions performed on the selected components assure satisfactory performance of the system after maintenance. This maintenance policy is called selective maintenance. At the time of maintenance, it is not always necessary that a system is replaced. It may undergo other maintenance options instead, such as minimal repair (when a failed system is as bad as old after maintenance) or imperfect maintenance (better than minimal repair but not as good as a new component). This Ph.D. research studies the selective maintenance modeling for systems when imperfect maintenance is also possible in addition to replacement and minimal repair. Models are developed in this thesis to reflect the effect of different maintenance actions on system reliability. In juxtaposition with these models, selective maintenance policy is developed for systems. A maintenance policy is influenced by several factors such as the age of the system, failure modes in the system, maintenance history, and the performance levels of the system. At first in this thesis, the effect of the system age and maintenance budget on the maintenance decision is modeled and selective maintenance is performed. A single mission selective maintenance model is developed for a system that can exist in any of the two possible states: working and failed (also called binary system), and all failures within the system are maintainable. Maintenance may or may not affect different failure modes in a system, and accordingly the system can be maintainable and non-maintainable with respect to these failure modes. Therefore, the presence of these two types of failures in a binary system is also studied and a single mission selective maintenance problem is solved. If maintenance is required more than once in a given planning horizon, then the single mission selective maintenance is no longer adequate. In these conditions, the time to perform maintenance is important in order to keep a system reliable throughout the planning horizon. Hence, maintenance scheduling is required for a system along with the selective maintenance decision during each of the maintenance breaks. Thus, the selective maintenance scheduling problem is solved for a binary system in this thesis. Conventionally, it is assumed that a system has binary states. But a system may also have more than two performance states. For such a multistate system, the binary selective maintenance model is not applicable. Therefore, a comprehensive model is developed in…