Our project demonstrates the value of using an agent-based model, combined with powerful interactive visualisation techniques, to reveal insights about the common factors that cause and contribute to reactionary delay. The objective was to find the most effective interventions and contingency plans to control and recover from delay, and improve network performance. Our partner GWR wanted to understand how random and multiple everyday incidents on the railways affect delays. They wanted to use this knowledge to help determine the most effective interventions – or contingency plans – to reduce delays, reduce the number of cancelled services, and to improve their service for rail customers.We worked closely with GWR to:a) Design, implement and demonstrate an agent-based model characterised by existing railway data, to simulate and explore the range of possible reactionary delays that can cascade from known primary delays, across key routes. The model simulated the many possible outcomes that can occur for a given day, due to the many random events that can affect delay. It produced detailed output data describing the many ways in which train services can be delayed.b) Identify, design, implement and evaluate techniques for quantification, visual depiction and visual exploration of delay in a visualisation tool. The tool interrogated the ABM delay output data so that it can be easily interpreted by our rail partners, who can explore and identify the common factors that are more likely to be contributing to reactionary delays.c) Model and quantify the effects of possible interventions on reactionary delays. Key interventions or contingencies can be tested by the model, and the impact of these interventions explored using the visualisation tool.