Data-Driven Decision Making for Resilient Freight Transportation Under Cascading Disruptions

Intermodal freight transportation is essential to the US economy, enabling the efficient movement of goods through interconnected road, rail and waterway networks. However, extreme weather, infrastructure failures, accidents and other disruptions can propagate rapidly across these networks, causing delays, increasing costs and reducing supply-chain reliability. Existing planning and scheduling approaches are often unable to adapt quickly to large-scale disruptions that evolve over space and time. As transportation networks become increasingly interconnected, improving the ability to anticipate, manage, and recover from disruptions becomes a critical challenge for freight operators. This award supports research that aims to create new data-driven models and decision-support tools to improve resilience of intermodal freight transportation under cascading disruptions. The research will integrate simulation, machine learning, optimization and reinforcement learning to create a hierarchical framework that supports both long-term planning and real-time operational response to disruptions. The resulting tools will help freight operators better prepare for, respond to, and recover from disruptions while maintaining efficient freight movement across interconnected transportation networks. The project will engage undergraduate and graduate students in research activities to educate and train the next generation of engineers. Research findings will be broadly disseminated via publications and open-access tools. The project will engage a board of industry advisors to help ensure practical relevance and facilitate technology transfer to transportation practitioners. This research aims to improve the resilience of large-scale transportation networks affected by cascading spatiotemporal disruptions. The project will integrate deep learning, deep reinforcement learning, large-scale simulation, and decomposition-based optimization to develop a hierarchical transportation management framework that links planning and response decisions. The research will advance network modeling and optimization under spatiotemporal uncertainty, develop new approaches for hierarchical resilience decision-making, and identify dynamic policies that combine data-driven learning with mathematical optimization. The framework will be evaluated in three applications in the U.S. intermodal freight transportation system: dynamic freight rerouting; freight rerouting and service rescheduling; and coordinated planning, rerouting and rescheduling. By integrating data-driven learning models with mathematical optimization, this project will contribute to a new generation of efficient and scalable computational methods for safeguarding interconnected networks against cascading disruptions. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.