In the first decade of the twenty-first century, the doubling of the global retailing from seven to fourteen trillion dollars has been accompanied by a soaring competition in the marketplace. Further, ever-rising customer expectations for the availability of in-store products and on-time delivery of online purchases have intensified retail competition. Nimble distribution tactics are essential to manage retail delivery and replenishment operations in such a competitive environment. This dissertation investigates two new distribution tactics for retail demand fulfillment of fast moving full-pallet products, and commit-to-delivery online purchased products. The first distribution tactic is that of the shipment of full-pallet products via a subset of retail stores instead of established distribution centers. The tactic is studied by developing a solution methodology which employs both clustering and optimization. The methodology is applied in a computational study which required generating instances for multiple U.S. census regions, and successive optimization of multiple mixed integer multi-commodity network models for each problem instance. A Java application is developed which uses US Census Zip Code population demand based data to generate retail distribution regions. The application then clusters a distribution region into several service areas. Finally it optimizes each service area separately using the IBM ILOG CPLEX libraries, and visualizes the obtained solution as well as the clustering stages of each problem instance. The computational study reveals that through-store-transshipment achieves up to 11.6% cost reduction in distribution of full-pallet products. The solution methodology can solve the model instances faster than the exact method by an order of magnitude. The second distribution tactic is the use of an alternate distribution channel for online order fulfillment. The problem of identifying online order fulfillment channels is modelled as a two-stage location-routing problem for which a heuristic solution algorithm is developed. The algorithm rests on the selection of candidate fulfillment centers first, and successive optimization of stores reassignment at subsets of fulfillment centers next. It is implemented in Java using the IBM ILOG CPLEX solver libraries with automated solution visualization using Tikz libraries. The performance of the algorithm is tested in a computational study that includes three prototypical chains of retailers with presence in New England and Mid-Atlantic, South Atlantic, and Pacific census regions. Computational tests verify the viability of the developed solution algorithm both in terms of obtained solution quality and computation time-efficiency.
| Identifer | oai:union.ndltd.org:UMASS/oai:scholarworks.umass.edu:dissertations-6969 |
| Date | 01 January 2013 |
| Creators | Ebtehaj, Milad |
| Publisher | ScholarWorks@UMass Amherst |
| Source Sets | University of Massachusetts, Amherst |
| Language | English |
| Detected Language | English |
| Type | text |
| Source | Doctoral Dissertations Available from Proquest |
Page generated in 0.0011 seconds