A UNIFIED FREIGHT TRANSPORTATION PLANNING MODEL

The objective of this research is to propose a unified freight transportation model which simultaneously solves all freight planning tasks. The fundamental idea which links the separate planning tasks together is the development of a new flow demand model. Based on a profit maximizing motive of every producer in every market, the equilibrium solution to this model solves for total supplies at flow origins, total demands at flow destinations, user-equilibrium flow distribution between origins and destinations, as well as user-equilibrium route and mode selections.; Existing network improvement models choose network improvement alternatives without considering the impact of network changes upon flow supplies, flow demands, and flow distribution. This shortcoming is overcome in our unified model because our network improvement model is integrated with our new flow demand model. A solution algorithm which decomposes the flow variables and the network improvement variables is presented. A flow assignment subproblem and a network improvement subproblem are solved iteratively. The equilibrium flow distribution, flow assignment, and the network improvements are obtained upon convergence of this algorithm.; A new assignment algorithm is developed to solve the flow assignment subproblem. It is an improving feasible direction method which includes flow increases on shortest paths and flow decreases on longest flow carrying paths between O-D pairs. Computational results show that our new assignment algorithm converges the fastest when starting from a good starting solution.; A new network improvement algorithm is presented to solve the network improvement selection subproblem. It involves the application of generalized upper bounding techniques in the solution of a separable programming problem. Our network improvement subproblem can handle contingency constraints and also can handle situations where one improvement activity affects more than one transportation link. The computational time is comparable with existing models, if not more efficient.; The unified planning model is demonstrated on a problem from the Multi-State Transportation Corridor Research Program.