Geometric and operational optimization at reversible unconventional arterial intersection reducing traffic emissions

This study develops a novel unconventional arterial intersection with reversible lanes (UAI-RL) to reduce traffic emission, where six forms of UAI, conventional intersections (CI), contraflow left-turn lane (CLL), tandem intersection (TI), continuous flow intersection (CFI), parallel flow intersection (PFI) and upstream signalized crossover (USC) on each leg can be flexibly switched in a mixed assignment area between pre-signal and main intersections. Traffic emission of UAI-RL is minimized by a unified optimization for both geometric design and signal control. A machine-learning-based surrogate-assisted algorithm is established to solve the problem. The result demonstrates that optimal UAI-RL is useful for intersections where traffic demand frequently fluctuates. Decreasing vehicle stops is beneficial to reducing emission and CI on four legs is recommended in the case of low demand. Enhancing capacity to prevent intersection oversaturation is more helpful and TI, CFI, PFI and USC on four legs are recommended in the case of heavy demand.