Markov Maintenance Models with Repair.
Abstract
In this study discrete time finite state Markov maintenance models are investigated. In each model, a machine is assumed to be operating over time with its condition deteriorating as time goes on. The state of the machine is observed at the beginning of a period. An operating machine can be sent to a repair shop at this time, whereas a failed machine must be repaired. When a machine is being repaired, the number of time periods that it is unavailable is usually assumed to have a geometric distribution. A repaired machine becomes available in its best state. An operating cost is charged while a machine is operating, and material and labor costs are charged when it is being repaired. The objective is to find a policy which minimizes the total expected alpha-discounted cost or the long-rung average cost. Special emphasis is being placed on finding sufficient conditions to assure that a control limit policy is optimal. The aforementioned model had only one machine in the system. Models with spare machines in the system are next studied. For these models a penalty cost is added when the system fails (only when all machines are inoperative). (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 05, 1976
- Accession Number
- ADA033427
Entities
People
- Yukio Hatoyama
Organizations
- Stanford University