Railway networks are complex systems designed to facilitate the movement of passengers and goods for multiple origin-destination pairs. Since these networks are jointly operated by numerous interconnected components (e.g., railways and stations), loca...
Railway networks are complex systems designed to facilitate the movement of passengers and goods for multiple origin-destination pairs. Since these networks are jointly operated by numerous interconnected components (e.g., railways and stations), local disruptions often cascade through the system, potentially causing delays for journeys that do not directly involve the affected assets. To effectively mitigate such system-level risks, this study proposes a decision-support framework that identifies critical disruption scenarios characterized by high likelihood and high consequence, and then develops an optimal dynamic train scheduling model specifically designed to operate under disruption scenarios. To accurately assess cascading impacts while quantifying the propagating uncertainties of local disruptions, this study proposes a modified Multi-Commodity Network Flow (MCNF) formulation that simulates demand dynamics and minimizes compensation costs associated with unmet demand. The framework identifies the most critical scenarios by applying a genetic algorithm to the space of reliability index β and the redundancy index π^SL. Specifically, β represents the likelihood of an initial disruption scenario and π^SL measures the expected system performance loss under the scenario. To further assess rail disruption dynamics and optimize detailed re-routing strategies under critical disruption conditions, the study proposes a dynamic train scheduling optimization model formulated as a dynamic MCNF model. The model provides re-routed schedules at the individual-journey level while minimizing system-wide losses caused by train cancellations, disruption induced truncations, passenger delays, and detour usage. The proposed methods are demonstrated by their applications to two railway systems: the Korean railway network and the Great Britain railway network. The numerical investigations reveal practical insights into the relationship between network topology and systemic risk. Furthermore, under disruption conditions, the dynamic train scheduling model yields system-wide decisions that are more consistent and efficient than heuristic operational practices, thereby improving service reliability and resilience in large railway networks.