Bi-objective optimization of traffic assignment with air quality consideration via CFD-based surrogate model

Abstract

Vehicular emissions form a major source to ambient air pollution, constituting the leading risk factor for public health globally. To mitigate this problem with sustainable urban mobility, the paper proposes an optimization framework to minimize the air pollutant concentrations at selected site(s), where the residents are susceptible to airborne hazards. Kowloon Peninsula, Hong Kong, is taken as the example. The on-road traffic volumes nearby are optimized to minimize the carbon monoxide (CO) concentrations as well as the accumulative travel time. The CO concentration is implicitly determined via the surrogate model (Gaussian process regression; GPR). Its dataset is established by the combination of traffic model, emission model, and computational fluid dynamics (CFD). The non-dominated sorting genetic algorithm II (NSGA-II) is adopted to solve the bi-objective optimization problem. The outcome helps advance traffic strategy with the newly proposed optimization objective, giving priorities to vulnerable recipients at selected sensitive locations. Enabling the capacity to directly evaluate the pollutant concentration, it is the first attempt to incorporate surrogate model for pollutant dispersion based on experimentally verified CFD results into traffic assignment. This framework will facilitate the transport authorities to assign the traffic volumes and address health equity issues in a more precise manner.