A multi-commodity network design for the Defense Logistics Agency

Abstract

The Defense Logistics Agency (DLA) currently operates 28 depots in the United States from which it supplies over 45,000 customers with over 3 million products procured from over 10,000 suppliers. DLA plans to reduce its infrastructure and proposes to analyze its distribution system using the Strategic Analysis of Integrated Logistics Systems (SAILS) model - a mixed integer linear programming model widely used by a number of civilian organizations to make facility location decisions. The size of DLA's distribution system precludes directly evaluating all possible depot, product, and customer combinations. This thesis derives a 29 product, 113 customer aggregation scheme which facilitates SAILS execution and appears to adequately model DLA. Extensive comparisons between this aggregation scheme and others (44- , 49-, and 67-product; and 199- and 113-customer aggregations) at 100, 90, 80, 50, and 30 percent of derived depot throughput capacity show solutions to different aggregations result in virtually identical closure recommendations and total annual cost. This thesis shows how DLA can save over 300 million dollars annually through depot closure and reorganization.