Inventory management for perishable goods using simulation methods

Abstract

Amazon.com is the world's largest online retailer, and continues to grow its business by expanding into new markets and new product lines that have not traditionally been sold online. These product categories create new challenges to inventory and operations management. One example of this new type of products sold online includes the category of perishable goods. Perishable goods provide a unique inventory challenge due to the fact that products may expire at unknown times while in stock, making them unavailable for the customer to purchase. This thesis discusses a method for managing perishable goods inventory by characterizing the key variables into empirical probability distributions and developing a computational model for determining the key inventory attribute: the reorder point. This model captures both the demand and loss due to shrinkage based on the age of the product in inventory. The resulting model results in a 25% improvement in simulated inventory levels with more accurate results than current methods. This improvement is shown to come from accounting for the known variability in lead time, as well as survival rate of the product.