Optimizing Distribution Networks with Application to the US Postal Service

The research reported in this paper has two objectives—one conceptual and one applied. The conceptual objective is to represent the variable cost function of a complex multi-level distribution network in a way that abstracts from local geographic detail. A network is multi-leveled when objects proceeding from an origin to a destination must pass through intervening collection and distribution hubs. The cost function’s variables are aggregate measures of traffic, such as the mail loads on a postal network. The function’s parameters relate variable cost to the aggregate dimensions of the network, to the size of the geographic areas served, and to technical parameters characterizing network activities. An efficient network is identified by selecting the dimensions to minimize the total variable cost of operation for a representative load. The applied objective is to evaluate the US Postal Service’s plan to reorganize the domestic network it operates to collect, process, transport and deliver the mail. The main finding is that the Service’s plan will leave an inefficient network that will add several billion dollars to the annual cost of operation of the US postal network.