Dynamic flow control model and algorithm for metro network under FIFO condition

Implementing passenger flow control strategies is an effective approach to reducing commuter travel delays and ensuring crowd safety in a congested metro network. Due to the intricacy of the interweaving of passenger flows between various lines and stations, the development of a scientific passenger flow control strategy is challenging in the networked mode of operation. The first-in-first-out (FIFO) rule can ensure service fairness and optimal operation by accurate modeling passenger queuing dynamics, but it is rarely considered in existing studies. Inspired by the traditional dynamic traffic assignment models, we propose a novel passenger flow control model with the FIFO rule to find a more reasonable control strategy for a metro network. Unlike road traffic systems, the FIFO rule is formulated as a set of linear constraints to explicitly capture the passenger queuing properties at origin stations. The passenger flow control problem with the FIFO rule is then modeled as a mixed integer linear programming model , which can significantly reduce the model complexity. To reach a high-quality solution, we propose an efficient rolling horizon decomposition approach. In the algorithm, the planning horizon is rolled forward from the current time, and the effects of subsequent periods are considered at each iteration. Besides, a dynamic procedure for loading passengers is developed to evaluate the bounds between the proposed approach and other flow control strategies. The proposed model and algorithm are then applied to solve the problems in test and real metro networks. The numerical results demonstrate the validity of the model’s properties and the algorithm’s performance. • A passenger flow control problem with FIFO constraints is established. • A dynamic flow procedure is proposed to provide the upper bound of the model. • A rolling horizon decomposition algorithm is designed to solve the model. • Numerical experiments validate the effectiveness of the proposed approaches.