Optimal Stock Levels and Service Channel Allocations in a Multi-Item Repairable Asset Inventory System

Abstract A methodology is described for determining the optimum allocation of repairable item inventories and repair channels to support operating systems consisting of different components. The steady-state probability distribution for die number of failed components is based upon the M/M/s finite population queueing model. Dynamic programming is then used to allocate a fixed budget among the components in order to maximize system availability. At each stage of the dynamic program, optimum trade-offs between stock levels and repair channels are made. An illustrative example is provided.