The optimal design of a physical distribution system on a multicommodity multi-echelon network

Strategic planning in physical distribution management requires decisions as to where intermediate distribution centers should be located, what sizes they should be, and how commodities should be routed from sources of production through the distribution centers and on to customers. The literature concerning the location of distribution facilities is extensive, covering a wide range of modeling options. This work has developed an optimization procedure for solving a multicommodity multi-echelon distribution/location problem with upper and lower bounds imposed on distribution center throughput.

An implicit enumeration algorithm is presented which uses a heuristic to generate an initial throughput-feasible solution. The algorithm employs a new hybrid price-directive/resource-directive scheme for efficiently solving the multicommodity capacitated transshipment problems which result from temporarily fixing the network configuration at each node of the enumeration tree. A number of branching and screen ing options are used in order to keep the tree search small. These have proven to be very effective in practice.

A software package, to implement the algorithm, is described and computational results are reported on a large-scale practical applica tion involving national distribution of ice cream products in Australia.