Capacity assignment in repair shops with high material uncertainty

We consider a group of identical systems, each consisting of multiple Line Replaceable Units (LRUs) that fail according to a Poisson process. A failed LRU is replaced by a ready-for-use one from a single stock point and, if not available, a backorder cost is incurred per unit of time. The failed LRU is returned to a repair shop, where it is inspected to identify which Shop Replaceable Units (SRUs) caused the failure, and is repaired by replacing the failed SRUs. After repair the LRU is ready-for-use again. Both the LRUs and SRUs are controlled by base stock policies. The repair shop is modeled as a two-stage service process consisting of an inspection and a repair phase. Inspection and repair are executed by one group of repairmen. The repair times depend on the time that elapses between inspection and the repair of a part. We model the total repair capacity as a single server and we compare policies that, based on the repair workload in the repair shop, give priority to either inspection or repair of parts. We suggest two approaches to set the SRU base stock levels, and simulate the system for multiple combinations of the repair workload threshold and predetermined vectors of SRU base stock levels. Based on the simulation results, the LRU base stock levels are optimized. We study a representative setting in which the repair shop faces a high material uncertainty, under different scenarios. We show that a scenario in which we maximize the SRU job completeness, combined with a repair priority policy (in which repair of jobs takes precedence over inspection of jobs) leads to the lowest total costs.