Railway landslip prevention; Railway landslip prevention;

Network Rail predicting rail earthwork issues, Various

Predicting rail earthwork problems with data analytics

Digital tools excel at revealing insights from within masses of disparate data. At Arup we are bringing this power to bear in physical contexts that previously depended heavily on judgement and experience alone. In a research project for Network Rail in the UK, we have used data analysis techniques to provide valuable insights about earthwork failures, a major threat to safe and reliable rail operations. 

As a national body Network Rail is responsible for over 190,000 earthworks covering 9700 miles across the country, and many of these date back to the network’s Victorian origins. Also, regional variations in weather mean that rule-of-thumb calculations about the likely threat posed by extreme rainfall needs to take account of a huge array of local factors. Every year the organisation typically has to deal with 120 to 150 incidents, leading to potential accidents, lost services and revenue. 

Our approach married geotechnical analysis with cloud-based data analytics. The system analysed regional rainfall data and correlated this information against local geologies and earthwork failures. The scale of this level of analysis is only possible using cloud-based calculation – the weather data alone accounts for terabytes of information. The tool can predict which sections of track face risks given current weather conditions. 

Project Summary

120 to 150 incidents a year

190,000earthworks Network Rail responsible for

Train travelling through a deep cutting in the UK Train travelling through a deep cutting in the UK

Data analytics to support decision makers

In matters of safety, decisions must be taken by people, not technologies alone. Our ethos on this project was to support decision makers with data in order to expand their individual awareness and understanding of potential issues, not replace their decision-making role. Data analytics at this level have the potential to radically transform how managers approach their estates or portfolios of assets, improving safety, extending lifespans, targeting maintenance more efficiently and helping them to prioritise their efforts. 

There are many other forms of physical infrastructure that need greater maintenance as they age, and this approach would be relevant for any major legacy asset management need, as well as new assets required to operate in an increasingly uncertain future.