Enhancing Tipping Point Detection and Prediction for Near-future Traffic Congestion

This project aims to advance the detection and prediction of traffic tipping points (TPs), at which traffic conditions transition between free-flow and congestion, and develop a practical predictive tool to support proactive traffic management and operational decision-making at National Highways.

The project will first enhance the existing rule-based TP detection model developed in an earlier project with National Highways through improved mathematical formulations and systematic parameter calibration. In parallel, supervised machine learning (ML) will be investigated as an exploratory and supplementary approach to assess whether it provides measurable improvements, particularly in detecting irregular and transient TPs.

The central deliverable is a predictive TP tool that combines the enhanced detection model with available traffic-flow forecasts or scenario-based inputs to identify potential future congestion and emerging hotspots. Supported by GIS-based visualisation and network exploration, the tool will enable users to investigate congestion risks under different traffic conditions and scenarios.

The project will then investigate irregular and transient TPs and their relationships with external factors, including roadworks, incidents, weather and traffic-management interventions. Alternative speed-density relationships will be explored to better represent observed traffic behaviour. These investigations will inform further refinement of the predictive tool.

Particular emphasis will be placed on practical development and robust evaluation. The tool will undergo iterative refinement, followed by dedicated testing using historical data, operationally relevant scenarios and diverse traffic conditions. Its predictive performance, robustness and practical usefulness will be evaluated in collaboration with relevant stakeholders.

The project will deliver a validated research prototype, GIS-based visualisation capabilities, documented methodological findings and recommendations for potential future operational application.

Read about our previous project here. 

Visit the National Highways website here.

Impact

The project is expected to enhance National Highways’ ability to anticipate emerging traffic congestion and identify potential hotspots before conditions deteriorate. By delivering a practical predictive tool with GIS-based visualisation and network exploration capabilities, it will support more proactive operational decision-making, scenario assessment and targeted traffic management. Improved understanding of irregular tipping points and their relationships with external factors, such as roadworks, incidents and weather, could further strengthen congestion prediction and network resilience. The project will also provide a validated prototype and evidence-based recommendations to inform future operational implementation, potentially contributing to reduced congestion, improved journey time reliability and more efficient use of road network capacity.