Environmentally sustainable transportation network design

Abstract

Transportation network design problem aims to find optimal strategic decisions of either link expansions or additions in networks, while accounting for the reactions from network users. These problems have been receiving attention because of its vital practical significance and theoretical complexity of the decision model. In recent years, the fact that road transportation negatively affects the quality of environment and deteriorates its bearing capacity has drawn a wide range of concern among researchers. This thesis contributes to the literature of transportation network design with a more comprehensive environmental consideration. A Network Design Problem with Environmental considerations (NDPE) is proposed, so as to not only alleviate congestion but also mitigate impacts on the environment. The proposed NDPE models serve as tools for planners seeking for a solution or decision-making aid for minimizing total emission cost, total noise cost, and total system travel time cost at the same time. This thesis addresses the importance of instilling environmentally sustainable considerations into road network design by demonstrating the existence of an analogue of Braess’ paradox regarding traffic noise, named excessive noise paradox. It is identified that in traffic assignment with non-cooperative behavior, some network improvements or traffic control strategies can cause the total amount of excessive noise and its cost to increase. The simultaneous and non-simultaneous occurrence of excessive noise paradox, emission paradox, and Braess’ paradox are also examined using integrated estimation models. It is concluded that the negligence of consideration of environmental sustainability in road transportation network design may lead to counterproductive and inefficient outcomes.

Both the weighted sum approach and the Pareto optimization method are used to propose comprehensive bi-level NDPEs. The upper level problem determines the optimal design for road network improvement for minimizing travel time as well as environmental measures. The lower level problem captures the route choice behavior of transportation network users reacting to the decision made by the transportation planner from the upper level problem. Environment-associated costs caused by the traffic are examined to be substantial in quantity and influential enough to alter the optimal design scheme. Different conflicting relations are discovered among different objectives under different demand situations. To solve this proposed bi-level NDPE with the inherent non-convexity and nonlinearity, a novel metaheuristic named Chemical Reaction Optimization (CRO) with widespread employment and remarkable performance is borrowed by this research. An enhanced version and a multi-objective version of the original CRO, namely Non-dominated Sorting Chemical Reaction Optimization (NSCRO) are proposed to solve the formulated bi-level problem using the weighted sum approach and the Pareto optimization method, respectively. Benchmark road networks with different demand levels are used to evaluate the performance of the proposed algorithms. The proposed enhanced CRO is demonstrated to acquire a comparable or even better objective function value than Genetic algorithm (GA). The proposed NSCRO shows decent effectiveness and efficiency in approximating the Pareto front compared with Non-dominated Sorting Genetic Algorithm II (NSGA-II). Both enhanced CRO and NSCRO are shown to be suitable methods for solving the bi-level NDPE.