A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Transportation Engineering, School of Engineering and Technology

Thesis (M.Eng.) - Asian Institute of Technology, 2008

Traffic assignment model plays an important role in network design problem by producing a user equilibrium flow pattern for finding the optimal network improvement. The classical traffic assignment technique assigns a fixed OD table onto the network, so that a set of average link flows are generated. However, the demand for travel encounters many uncertainties in the real world. Therefore, the OD demand cannot be fixed and using it may provide some misleading information. To account for such uncertainties, a stochastic OD table with the assumed Normal distribution in each OD demand is introduced. The modified traffic assignment with a stochastic OD table was examined in several experiments. It has been found that the probabilistic link flows tend to follow the Normal distribution. Furthermore, the different decision results were observed between using vie ratio and probability of congestion. Eventually, the proposed idea was further extended to the network design problem and led to the formulation of new network design problem concept which aims to minimize the system probability of congestion (SPC). Furthermore, some more experiments were conducted for making a comparison between proposed and conventional NDP. Finally, in association with the SPC, the new concept of Braess' s paradox is introduced.