Abstract
Increasing traffic density and the deployment of digital signalling technologies are transforming railway capacity from a static infrastructure constraint into a real-time controllable system-performance variable. This critical review provides a system-level analysis of dynamic headway control and its impact on effective railway capacity within closed-loop operational environments. Modelling paradigms, including analytical methods, microscopic simulation, optimisation-based traffic management, and artificial intelligence approaches, are comparatively assessed with respect to control capability, scalability, real-time applicability, and sensitivity to disturbances. The review shows that conventional planning-oriented capacity assessment systematically overestimates achievable throughput because it neglects traffic stability, disturbance propagation, and feedback-based regulation. The evolution of control architectures from fixed-block signalling to moving-block systems, predictive traffic management, and virtual coupling is interpreted as a transition from infrastructure-limited to control-limited capacity. A unified conceptual framework is proposed that links control architecture, headway dynamics, traffic stability, and effective capacity within a cyber-physical control loop. This perspective enables the identification of key research gaps, including the lack of integrated real-time models and the limited incorporation of artificial intelligence into safety-critical control. Future intelligent railway operation is expected to be based on digital twins, reinforcement learning, and distributed coordination, enabling continuous capacity regulation under stochastic conditions.