Repair Inventory Management System

Researchers: Frannie Fang, Ruthairut Wootisarn, and Yugesh Asokan (with John Maleyeff)

This project seeks to complete the development of a repair inventory management system inspired by the MBTA, an inventory of parts needed to repair trains, buses, and other equipment in Boston. At MBTA, repairs required equipment purchases, resulting in a considerable inventory investment constituting tens of million dollars over many decades. Demand for tens of thousands different parts is intermittent, and procurement lead times can be long.

In this study, we will develop a non-terminating simulation that analyzes inventory policies based on service levels for each part in a repair kit. The simulation was written in Python using NumPy, Pandas, Matplotlib, and SciPy for data analysis. It assumes that repair demand is Poisson and that part lead times and other parameters are consistent with circumstances existing at the MBTA. Users enter the annual repair demand rate, part lead times, part costs, fixed ordering costs, and annual holding cost percentage. Users have two options when running the code: (1) specifying a constant service level for every part in the repair kit or (2) specifying a customized service level for each part in the repair kit. As a result, the simulation generates six outcomes and calculates a confidence interval for each outcome. The simulation was used to compare two options and found that assigning a lower service level to expensive parts lowers relevant costs.

Current efforts concentrate on quantifying repair delay cost due to missing parts in the repair, analyzing the impact of fixed ordering cost, creating the simulation optimization model, and estimating the cost saving if the repair inventory management system is optimized.