UBC Theses and Dissertations
A dynamic programming - Markov chain algorithm for determining optimal component replacement policies Young, G. Glen
An algorithm is developed to determine the optimal component replacement rules to follow in managing a particular class of equipment. The work follows basically the models developed previously by S.E. Dreyfus and R.A. Howard. However, a different Markov state description has been used to extend the application of these models to systems of more than one component subject to stochastic failure and for which the failure of any component renders the entire system inoperative. The model, in effect, selects optimal replacement alternatives as individual components fail, under the uncertainty of further failures occurring in the same transition interval. The model was programmed for an I.B.M. 360/67 computer and the results for a hypothetical problem were checked through renewal theory.
Item Citations and Data