A strategy for optimal management of spares

This paper develops a model for optimal management of spares inventories in a network. This model is flexible and robust. It can be applied as local model or as a regional model supporting many local sites. The output includes a locally optimized list of spares that is optimized with respect to the frequency of failures, restoration time, and time to replenish the spares stock. If a regional spares stock is available, it is possible to take advantage of the idea of a "shared spare," which splits the sparing load between the local and regional offices. In addition to technical outputs, the model proposed here is a dynamic model, in that the parameters can be easily estimated and updated as experience is gained. Also, it allows assumptions regarding MTBF, MTTR, and lead time to be tested. A third benefit is that it allows policy to be modeled, and allows an organization to look ahead and forecast the impact of policy choices. Finally, the cost of not having a required spare in place can be included. While this model does not address every nuance of operating a spares inventory, it is reasonable to estimate that planning inventory using the structured approach advocated here can result in significant savings. Moreover, the model and approach advocated here are useful alternatives to setting spares inventory by specifying a percentage of the number of installed units, rounded up (e.g., if there are 57 units and 2% is the selected level, then 2 units would be the stocked number of spares). Setting inventory levels in this way will overstock reliable components and understock unreliable ones. It will also tend to understock critical units if no attention is paid to the probability of stock-out.