Safety-oriented multi-depot integrated vehicle and crew scheduling problem

Safety in public transportation operations remains a critical concern. The state-of-the-practice countermeasures primarily focus on real-time warning strategies, but there is still room for improving bus operation safety through reallocating fleet resources. This study introduces the Safety-oriented Multi-depot Integrated Vehicle and Crew Scheduling Problem (S-MD-IVCSP), a new generalization to the IVCSP where the safety risks of trip tasks are explicitly considered. To this end, we use the survival analysis model to assess safety risks during drivers’ work processes using warning data. Particularly, a safety risk transition model is developed based on the changing patterns of safety risk in single-line bus operations, which can be applied to driver crossline scenarios under regional scheduling. A time-space network model is devised to represent the S-MD-IVCSP. The drivers’ risk probability is incorporated into the objective. An enhanced adaptive large neighborhood search algorithm, featuring multiple operators, is developed to reduce the fleet size, optimize crew working hours, and mitigate drivers’ duty risks. The model is applied to 3 real-world bus lines in Guangzhou, China. Results show that our approach can reduce safety risks by at least 30% compared to the traditional cost-oriented approach, without significantly increasing operational costs. There is strong compatibility between regional scheduling strategies and proactive safety measures. By leveraging the warning data, regional scheduling can reduce the crew’s safety risks effectively, improving safety performance by over 19% compared to single-line scheduling mode. We also investigate the impact of crew compositions and limited driver resources on overall safety performance.