Search

In this article, a bi-level transportation network design problem with environmental considerations is investigated. To explicitly reflect various requirements of environmental sustainability from planners, total emissions costs and total excessive noise cost are minimized along with total system travel time while performing optimal capacity expansion. To leave space for additional information on decision making for planners and provide a more generalized description of solution optimality, the Pareto optimization approach is adopted. The study proposes a multiobjective variant of a new metaheuristic named Chemical Reaction Optimization as the optimization tool to solve the formulated network design problem. Pareto front approximations have been successfully acquired for each scenario tested using the Sioux Falls network. The results show that the relations between pairwise objectives can change significantly (e.g., from conflicting to aligned) with regard to different demand situations. The newly proposed multiobjective metaheuristic succeeds to produce better approximations of the Pareto front than NSGA-II with comparable runtime for the considered road network. [ABSTRACT FROM AUTHOR]

Nowadays, the decision makers in the transportation industry are being urged to incorporate environmental costs into road network design decision making because road traffic affects the environment and health. The design of a road network should not only be cost-effective but also environmentally sustainable. This article proposes a new network design problem (NDP) that takes both vehicle emissions and noise into account. This proposed environmentally sustainable NDP is formulated as a discrete bilevel program. The lower-level problem is formulated as user-equilibrium assignment. The upper-level problem determines the optimal road capacity expansion to minimize the total costs of emissions, noise, and travel time with the considerations of budgetary and capacity improvement constraints. The proposed problem is solved by an enhanced version of a new meta-heuristic named Chemical Reaction Optimization (CRO), and its parameters are tuned by our proposed tuning procedure. Two benchmark road networks with different demand levels are used to evaluate the performance of the enhanced CRO and illustrate the properties of the problem. The results show that there were tradeoffs between emissions, noise, and travel time costs, and that the enhanced CRO outperformed Genetic Algorithm (GA) on more than half of the testing scenarios and had a comparable performance on certain test scenarios compared with GA. [ABSTRACT FROM AUTHOR]