Dynamic Rolling-Stock Assignment for Minimising Originating Train Delays: A Case Study of a Congested Indian Railway Terminal

Originating train delays at congested terminal stations are a major contributor to delay propagation across railway networks. This study investigates terminal-level delay characteristics at New Delhi Terminal (NDLS), one of the busiest stations in Indian Railways, using a comprehensive operational dataset covering the full year 2023. The analysis reveals a strong interdependence between terminating and originating delays and shows that a substantial portion of scheduled buffer time remains underutilised under conventional static rolling-stock assignment practices. To address this issue, the terminal rolling-stock assignment problem is formulated as a constrained bipartite matching problem, and a dynamic rolling-stock assignment framework is proposed. A hybrid optimisation algorithm is developed to minimise originating delays while satisfying operational constraints including platform availability, maintenance requirements, and departure-window restrictions. The proposed approach is evaluated through detailed case-study analyses for three representative high-traffic months—January, July, and December—and compared with existing static and first-in-first-out (FIFO) assignment strategies. Results demonstrate that the dynamic assignment framework increases originating-delay absorption from 60–79% to 88–99%, corresponding to improvements of 19–40 percentage points across different seasonal congestion scenarios. The findings indicate that dynamic rolling-stock assignment can substantially improve departure punctuality at congested terminal stations without requiring timetable modifications or additional infrastructure capacity.