Optimal preventive repair policy for a semi-Markov deterioration shock model.
| Main Author: | |
|---|---|
| Other Authors: | , |
| Format: | Thesis Book |
| Language: | English |
| Published: |
1987.
|
| Subjects: | |
| Online Access: | Link to ProQuest copy Link to OAKTrust copy |
| Abstract: | Most maintenance models for stochastically deteriorating systems consider to replace or not to replace as the only two actions that can be taken at any deterioration stage. In this research a preventive repair model that allows repair actions at each deterioration level, besides replacement or not replacement actions, is developed. This increase in the action space allows more flexibility to the decision maker. The preventive repair maintenance model developed is for a continuously-observed system subject to a sequence of randomly occurring shocks causing a random deterioration to the system. At every shock, the decision maker has several alternatives from which to choose: (i) the system may be replaced by a new identical system, (ii) the system may be repaired (thus decreasing its degree of deterioration), or (iii) the system may be left as is. The system deterioration without maintenance is modeled as a nondecreasing semi-Markov process with the state space being the deterioration levels of the system. The two main objectives of this research are: (i) to characterize the optimal state-dependent policy for the preventive repair maintenance model using the long-run expected cost per unit time, and (ii) to study an algorithm that could solve the preventive repair maintenance model more efficiently than the traditional Howard's policy iteration algorithm. Under some regularity conditions the optimal state-dependent policy is proven to be a pseudo-control limit policy, where a pseudo-control limit policy is a policy that prescribes no action up to certain deterioration level and calls for a repair or replacement above that level. An algorithm by De Leve, Federgruen and Tijms (1977a) is adapted to solve the preventive repair maintenance problem more efficiently than Howard's policy iteration. The two algorithms are compared for some randomly generated problems. The algorithm developed in this research proves to be much faster and requires less iterations than Howard's policy iteration. |
|---|---|
| Item Description: | Typescript (photocopy). Vita. |
| Physical Description: | ix, 80 leaves : illustrations ; 29 cm |
| Bibliography: | Includes bibliographical references (leaves 63-66). |