Train operators typically prepare detailed timetables for daily operations. Nevertheless, the intricate network structure and extensive daily activities render railway networks highly vulnerable to disruptions. Large-scale disruptions necessitate rapid recovery actions involving coordinated adjustments to train timetables, rolling stock, and crew schedules. Even after revised timetables are computed, local operational conflicts−such as track and platform assignment constraints−may still render solutions infeasible in practice. This study addresses the real-time, integrated rescheduling of freight services and track assignments under multiple resource-specific disruptions. A mixed-integer linear programming model is proposed at the operational level, employing mesoscopic modeling, regulatory constraints, and business priorities. To ensure scalability, the model is embedded within a rolling-horizon framework with an adaptive spatial band, in which only a committed subset of decisions is executed and the remaining plan is re-optimized as new information becomes available. The framework enables priority-aware train sequencing and flexible use of infrastructure to enhance resilience. Using real-world data from the Netherlands railway network (423 trains, 60 stations), we demonstrate that the proposed approach generates operationally feasible and high-quality schedules within practical computational limits across a range of disruption scenarios. Results indicate that short disruptions are largely absorbed before trains reach their terminal stations, whereas longer disruptions exhibit persistent delay propagation downstream; moreover, flexible facility substitution significantly curtails network-wide spillover effects. A sensitivity analysis of the look-ahead horizon highlights computational trade-offs, providing guidance on the choice of an appropriate horizon length.
Haque et al. (2026) studied this question.