Railway digital twins (DTs) are multi-physics dynamic models of rail vehicles that feature feedback from wayside or vehicle on-board sensors, enabling fine-tuning and calibration of the dynamic models to accurately simulate the physical system behaviour. The current DT implementation concept is typically asynchronous, given the intense computational power requirements of multibody dynamic simulations, thus limiting potential applications that could improve safety and efficiency of railway systems. This paper presents a real-time digital twin implementation concept for railway systems, including a faster-than-real-time multibody dynamics simulation of a railway bogie model and a 1:4 scale physical bogie rig provided with an axle box acceleration sensor. The real-time DT was able to recreate the dynamics of the bogie throughout the experiment. The developed real-time DT enables systematising the platform requirements of physics-based vehicle system dynamics models for DT architectures and opening the discussion on their possible practical application areas.