Managing service parts logistics: Inventory sharing and rationing in decentralized and centralized distribution networks
Abstract
In this dissertation, we consider an inventory sharing and rationing problem in continuous-review, infinite-horizon, distribution networks. Unlike previous literature on inventory sharing, we model inventory sharing as a multiple demand classes problem in which each dealer faces two classes of demand: his own customers and inventory sharing requests from other dealers. We first focus on a decentralized system in which each individual dealer is independent of the manufacturer and thus makes his own inventory decisions in order to minimize his own cost. We analyze inventory sharing in two different settings: a make-to-stock decentralized system with exponential replenishment lead time; and a decentralized system with constant replenishment lead time. For these two models, we use different approaches to obtain dealers' performance measures and cost functions, and then use a game theoretic approach to characterize the equilibrium behavior of the individual dealers. Comprehensive computational studies are conducted for both models, highlighting the impact of the manufacturer's incentives and the cost of sharing on dealers' sharing behavior in a decentralized system. We believe that these two models are the first attempt to simultaneously consider inventory sharing, inventory rationing, and decentralized decision making. To complete our analysis, we also consider an inventory sharing and rationing problem in a centralized network. We focus our analysis on a make-to-stock inventory sharing system, which enables us to consider capacity issues in inventory transshipment, which has not been studied in previous work. Compared to previous literature, we consider transshipment in a much broader sense. Transshipment can be used for transferring inventory for emergency demand filling and balancing inventory allocation between different locations, as well as achieving production capacity flexibility. For the centralized inventory sharing system, we prove the structure of the optimal policy, analyze structural properties of the optimal policy parameters, and provide conditions that simplify the optimal policy. We also conduct a numerical sensitivity analysis of the optimal solution to different cost parameters. We conclude with a discussion of a heuristic developed from our understanding of the true optimal policy.
Degree
Ph.D.
Advisors
Deshpande, Purdue University.
Subject Area
Industrial engineering|Management|Operations research
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.