Optimising the Roll-out Strategy of ERTMS

Abstract

Migration from legacy railway signalling to the European Rail Traffic Management System/European Train Control System (ERTMS/ETCS) must be phased while the railway remains operational. Existing deployment studies primarily address feasibility and cost, rather than using the performance of intermediate network states to guide implementation. This paper develops a multi-period optimisation approach that integrates network performance into strategic migration decisions. It is applied to the Dutch railway network, represented by 340 consolidated implementation units. Global efficiency and meshedness are converted into vertex-level gains that are either calculated once or updated after each implementation decision. The gains enter a lexicographic objective that prioritises network performance, followed by cost and completion time. A holistic mixed-integer linear programme is compared with sequential fixed-gain and dynamic-gain procedures under budget, implementation capacity, asset end-of-life, transition cost and regulatory target constraints. The holistic model achieves a 5.9\% (meshedness) to 9.4\% (global efficiency) higher cumulative network performance than the sequential fixed model. The sequential dynamic model produces more spatially connected intermediate networks and completes earlier and at lower cost than its fixed counterpart. Under global efficiency, its computation time is approximately one-tenth as long. Optimising global efficiency improves intermediate connectivity at a cost increase of approximately 14.6\% relative to cost-only optimisation. Sensitivity analysis identifies available implementation time, rather than budget, as the main constraint on target compliance. The results show that planning horizon and indicator updating materially affect migration sequences and should be considered explicitly in long-term railway signalling transitions.