Bus priority: lane-based pre-timed signal timing with buses in different directions at isolated intersections

The existing research on bus signal priority (BSP) with exclusive bus lane (EBL) often assumes that all buses travel straight through, neglecting the presence of turning buses. This oversight can lead to the issues such as BSP green time waste and traffic conflicts with buses turning, resulting in delays and safety concerns. This research investigates the applicability of BSP and EBL combined with lane-based pre-timed signal timing at isolated intersections to accommodate bus and car movements to various directions. By considering the constraints related to lane markings, signal timing, and traffic flow saturation, we develop the models aimed at maximizing passenger throughput and minimizing passenger delay, specifically through mixed-integer linear programming (MILP) and mixed-integer nonlinear programming (MINLP), respectively. The branch-and-bound method is adopted to solve the MILP, while the extended cutting plane method is employed to solve the MINLP with convexity validation against specified lane assignment. To enhance computational efficiency, three acceleration strategies of surrogate model, branch pruning, and parallel computation are implemented for the MINLP. Case studies reveal that EBL is not adopted even bus ratio is high up to 56.5%, different from the previous research on BSP and EBL that assumes all buses travel straight. In contrast, when left- or right- turning cars are prohibited at the target intersection, BSP and EBL are always adopted to enhance passenger throughput by 6% to 15%, or reduce passenger delay by 9% to 21%. This study contributes to the redesign of BSP and EBL for intersection efficiency maximization when buses are allowed to make turns.