Areas of extremely low adhesion between the wheel and rail can cause critical problems in traction and braking that can manifest in issues such as signals being passed at danger. There is currently a lack of real-time information regarding the state and location of low-adhesion areas across rail networks. The study presented here examines the scientific challenges of understanding the change in vehicle running dynamics with variations in adhesion using the latest thinking of adhesion at micro-slip. This understanding supports the generation of suitable low-order dynamic models for use with a model-based estimator that infers adhesion levels in the wheel/rail contact using signals from modest-cost sensors that could be fitted to in-service vehicles. This paper presents verification of this technique by using simulated inertial measurement produced from a high-fidelity multibody simulation in a series of blind tests.