Improving Rail Safety Using Machine Vision to Manage Scour

Publisher

CGI

Executive Summary

CGI and Network Rail developed a real-time scour management system using Machine Vision, AI, and geophysical techniques to improve bridge and rail infrastructure safety.
Scour — the erosion of riverbed material around bridge foundations — is the leading cause of bridge failure worldwide.
This solution integrates real-time monitoring, sensor data, and digital twins to detect risks early, prevent service disruption, and support Network Rail’s £3.8 billion efficiency goal under Control Period 7 (CP7).

Solution Overview

Network Rail operates over 30,000 bridges, tunnels, and viaducts. Increasing extreme weather has amplified scour risk, making manual inspections insufficient.
CGI’s Machine Vision solution provides a data-driven, automated method to continuously monitor bridge foundations and predict potential failures before they occur.

Key Features

• Continuous real-time monitoring above and below water levels.
• Machine Vision AI detects movement, sediment loss, and visual changes.
• Sensor integration combines hydrological and geophysical inputs for deeper insights.
• Digital twin visualisation delivers 3D situational awareness of bridge assets.
• Unified management platform handles detection, escalation, and resolution workflows.

Deployment and Results

Piloted on three representative bridges, the system demonstrated:

• Early detection of high-risk scour conditions.
• Improved safety for engineers and passengers.
• Reduced disruption through predictive maintenance.
• Scalable architecture for nationwide deployment across the UK network.

The pilot validated the effectiveness of combining AI, multi-sensor fusion, and digital twins for proactive infrastructure management.

Technical Architecture

| Component | Function | |------------|-----------| | Machine Vision Cameras | Capture and analyse live imagery for erosion and movement. | | Flow & Geophysical Sensors | Monitor hydrology and subsurface change. | | AI Data Fusion | Combines visual and sensor data to generate risk alerts. | | Digital Twin Environment | Provides an interactive model for engineers. | | Decision Support Dashboard | Displays alerts and supports prioritisation of maintenance actions. |

Microsoft Azure Integration

The solution can be deployed as a cloud-enabled service using:

• Azure IoT Hub / Event Hubs for data ingestion.
• Azure Machine Learning for AI model training and inference.
• Azure Digital Twins and Power BI for visualisation.
• Secure REST APIs for integration with asset management systems.

This architecture enables scalable, secure, and interoperable deployment across large infrastructure networks.

Customer Benefits

• Early risk detection and improved safety outcomes.
• Lower maintenance and inspection costs.
• Fewer service disruptions through predictive management.
• Enhanced resilience and sustainability in flood-prone areas.
• Proven impact supporting Network Rail’s CP7 efficiency targets.

Recognition

The collaborative research underpinning this project received the Best Paper Award (2025) at the Railway Engineering Conference for pioneering the use of Machine Vision and geophysical methods in scour detection and management.

Resources

• CGI Case Study
• Award-Winning Research Paper (PDF)
• Demonstration Video