Facility Location Optimization for Last-mile Delivery

Technological breakthroughs and consumer preferences have led to e-commerce growth in the United States. Given the continued rise in e-commerce demand, the final leg of transportation of goods to the end recipient, or the “last-mile,” is of increasing importance to freight services companies. The last- mile is one of the most challenging and complex aspects of business for such companies, who must manage the cost of getting items to consumers’ doorsteps against an increasing expectation for level of service and delivery times. The sponsor company seeks to become more competitive by optimizing their delivery network through exploring alternative facility location options to reduce the cost to service last-mile deliveries. Seven metropolitan areas were chosen based on volume and significance to the business: San Francisco, Los Angeles, Chicago, Dallas, Atlanta, Newark and New York City. The research hypothesis was that creating a more strategically-located network of cross-docks, each more proximal to common delivery points, would generate cost efficiencies through a reduction in overall travelling distance. The analytical objective was to explore the tradeoff between the additional facilities’ operating costs and the implied savings on purchased transportation costs. We implored a two-step methodology to explore this tradeoff: a center of gravity analysis to identify potential facility locations based on the concentration of customer demand, and a mixed integer linear programming (MILP) model to identify the cost optimal solution given the balance between the transportation cost reduction and the new facilities’ lease costs. The model output resulted in a shortlist of facilities, locations, and costs for each region. The average distance to serve the customer fell by 14 miles (40%) compared to the baseline average distance of 35 miles, and only 8% of customers remained greater than 50 miles away from a cross-docking facility compared to 14% in the baseline. The proposed facilities were $15 per-square-foot-per-year less than baseline lease fees on a weighted average basis. The key takeaway from the facility location optimization model was that adding facilities reduced the overall network cost by 23%, as the reduction in transportation cost to serve the historical demand outweighed the incremental facility lease fees.