Simulation-based optimization for time-of-day coordinated ramp metering of a large-scale urban expressway network
Date:
Ye, Y., & Chen, X.* (2018, January 7-11). Simulation-based optimization for time-of-day coordinated ramp metering of a large-scale urban expressway network [Poster Presentation]. The 97th Annual Meeting of Transportation Research Board, Washington DC, United States.
Abstract: This article adopts a computationally efficient simulation-based optimization (SBO) approach to solve the time-of-day ramp metering optimization problem for a large-scale urban expressway network. The decision variables are multidimensional, and the objective functions of traffic management authorities can be based on the perspectives of either the whole road network (including the expressways and local streets) or the urban expressway network, e.g., minimization of the urban network-wide average travel time, maximization of the network-wide total throughput, and maximization of the expressway network throughput. Different from the existing optimal ramp metering methods, which mainly optimize the ramp metering problem with explicit formulations, the SBO approach optimizes coordinated ramp metering for a large-scale urban expressway network with efficient searching and approximation by surrogate models. To demonstrate the SBO framework for solving the time-of-day coordinated ramp metering problem, this paper uses a mesoscopic simulator, i.e., DTALite, and adopts a calibrated dynamic traffic assignment (DTA) model to evaluate the performance of the transportation system. In a case study of a real-world large-scale road network in Hangzhou, China, a mixture surrogate model is employed to solve the optimization model, and the candidate approach (CAND) is utilized to infill the initially sampled points. The optimization results show there is 1.36% (0.186 min) reduction in the average trip travel time for the whole urban road network compared to the case with no ramp metering. The total travel time savings for all 309,671 travelers are around 961 hours during the extended 4-hour morning peak (6-10 AM). Given the value of time CNY 40 (US $5.9) per hour, the total reduction in the travel time spent is around CNY 38,440 (US $5676.5) for the study period. The improvements in the total throughput and expressway throughput are 0.54% (1,396 vehicles) and 1.24% (1,374 vehicles) during the extended morning peak, respectively. The proposed SBO framework for ramp metering can be incorporated along with other traffic control strategies to solve urban expressway optimization problems.