Joint optimization for crowd evacuation and vehicle scheduling at multimodal transportation hubs

Transportation hubs are critical nodes that accommodate substantial passenger flows, which will lead to significant congestion during peak hours, predictable events (e.g., holiday, extreme weather) or emergencies (e.g., operational disruption). It is crucial to design a collaborative evacuation strategy that fully utilizes the multimodal transportation capacities at hubs. Considering the impact of various transportation operations on crowd evacuation, this paper proposes a mixed-integer linear programming model that integrates pedestrian flow assignment and multimodal vehicle scheduling to efficiently evacuate the crowd. In the model, a demand-switching strategy among modes is incorporated, and various operational characteristics of transportation modes including departure times and fleet sizes are optimized for vehicle scheduling. Throughout the evacuation process, pedestrian dynamics are formulated by the cell transmission model (CTM). To solve the large-scale problems, a tailored Variable Neighborhood Search (VNS) algorithm based on decomposition is developed, where the subproblem is reconstructed on a time-expanded network to accelerate the solution process. The effectiveness of the proposed method model and algorithm are validated through a series of numerical experiments. The results show that the tailored VNS algorithm can effectively solve large-scale problems within a reasonable timeframe. The case study also demonstrates that the demand-switching strategy could optimize the use of available transportation resources, reducing the clearance time for taxis by 17.2%. Furthermore, the findings highlight the importance of adapting evacuation strategies to different emergency scenarios. This approach can be potentially applied to enhance emergency crowd management responses at transportation hubs.