Multiattribute Multicommodity Flows in Transportation Networks

Previously developed minimum cost multicommodity network flow models do not simultaneously consider the weights, volumes, and inventory holding costs of the commodities. ignoring one or more of these attributes may prevent detection of potential savings however, simultaneously accounting for all three attributes leads to a problem considerably more difficult to solve. This thesis examines a multiattribute multicommodity flow formulation of a transportation network with transshipment terminals, which seeks to minimize total vehicle and inventory related costs. First, a decomposition strategy transforms the model formulation, P, into an equivalent formulation, P. In P, the vehicle flow variables may be found as function of the commodity flow variables furthermore, the vehicle capacity constraints need not be explicitly considered. The decomposition, however, creates a situation whereby a commoditys incremental cost function on a given arc may contain concave andor convex portions. This feature implies the presence of numerous local optima, over which an exhaustive search for a global optimum is computationally infeasible. Computational results from a series of test problems measure solution quality and algorithm efficiency.