Britain’s railway network is made up of around 200,000 kilometres of track, most of it dating from Victorian times.
But the railways face a very modern-day problem: dealing with the impact of climate change and an increase in extreme weather.
Heavy rainfall, especially when it happens over a short period, poses particular risks for the railway’s cuttings and embankments. In a worst-case scenario, damage to earthworks can foul the track and cause a high-speed train to derail.
As the organisation responsible for the railway infrastructure, Network Rail has safety as its priority. Put simply, accidents on the railway are a risk to life.
But the company also has to keep the network running, avoiding delays and disruption. If there is a risk of damaged earthworks, the standard response is to put speed limits on the line: slower speeds shorten stopping distances, and minimise the chances of derailment.
And delays on the railway bring their own safety risks. To improve safety and reduce disruption, Network Rail, and the Rail Safety and Standards Board, are turning to data analytics. The project was presented as a case study at the recent Big Data and AI World conference in London, in early March 2024.
“Part of what we’re doing in Network Rail is reflecting on the fact that extreme weather in particular rainfall is an increasingly frequent and increasingly extreme event,” said Russell Shanley, weather risk task force lead at Network Rail.
“Some of the potential consequences of that are really severe. One of the challenges our engineering and operations colleagues have is how do we keep passengers safe and keep freight safe, but at the same time run a service… when we’re making decisions on whether or not to run trains, and at what speed we run them, we’ve got to consider a whole range of measures. That depends very strongly on structured engineering and operations judgement, risk-based decisions our people make on the day.”
Cuttings and embankments
There are more than 250,000 earthworks on Britain’s railways, comprising soil cuttings, which tracks run through, and embankments, which they run on top of. “These are the two types of assets that are most at risk of failure during heavy rainfall,” said Mike Briggs, director of data insights at the Rail Safety and Standards Board.
His team set out to create a model for rainfall and the likelihood of an earthworks failure. Each earthworks is monitored for 100 different parameters, and Briggs’ team had more than eight terabytes of rain data supplied by the Met Office.
For this, the country is divided into 1km square grids, with rainfall measured every five minutes. To make the task more complex still, not all earthworks have the same standards of data.
“It’s a really big data challenge,” said Briggs. “One of the other problems we had was in the period of analysis, there weren’t that many failures that occurred. We had to make sure that the models we developed to understand the risk of failure was during rainfall were statistically significant. We have to remember that we’re making the decisions on the speed of passenger trains based on the outcomes of these models.”
This, said Briggs, meant working closely with Network Rail’s own experts.
“The analysis we did was in two parts,” continued Briggs. “For those earthworks, the first one was a vulnerability model. For the amount of rainfall that’s falling on it, what’s the risk of that failing?
“And the second one was an escalation model. Something’s failed. What’s the risk of that actually becoming a derailment hazard? The approach that we took was a statistical approach. We looked at, for an earthwork of a particular size and shape, how many times was it subjected to rainfall of a particular intensity? And how many times did we see failures?”
This statistical model then formed the basis of a new tool, PRIMA – Proportionate Risk Response to Implementing Mitigating Speeds to Assets.
Delays and disruption
PRIMA helps Network Rail’s operations teams identify parts of the track that are at risk following heavy rain. But it also aims to give operators a more granular picture of risk and reduce the need for blanket speed restrictions.
This is important, because speed restrictions are not just inconvenient – they bring risks of their own.
“When you look at that primary risk, which is rainfall causing an earthwork to fail, the mitigation for that risk is relatively simple. We can slow or stop trains,” said Shanley. “But increasingly, we’re having to think about the knock on and the consequential risks we cause by stopping or slowing trains.”
Some of these risks are significant, such as a train passing through a red signal. But rail operators see other problems, especially if delays cause overcrowding at stations. These range from a heightened risk of someone falling on a track to increases in aggressive behaviour.
“There’s a lot of different things that can happen, with the primary risk of there being a derailment incident at that location, but also from the decisions that you make in terms of the speed and what else can happen across the network,” said Briggs. And delays have an economic impact too.
Driven by data
PRIMA works as a decision support tool that evaluates derailment risk, speed restriction risk and economic risk. It allows operations teams to make data-based decisions and use speed restrictions in a more calibrated way to suit the risk on a particular section of track. This ensures safety, but keeps the network running as well as possible.
In the future, Network Rail could add remote sensors and remote condition monitoring to its tools; it is already using satellite imagery and in the future, this could be checked by artificial intelligence (AI). “At the minute, we’re very dependent on engineers siting and looking at images,” said Briggs.
Network Rail is also looking at other ways analytics can improve safety and operations, as long as they can obtain the data. Rainfall and earthworks are just one hazard.
“They are a very narrow sliver of the weather-related risks that affect the railway,” said Shanley. He concedes he is often asked about “leaves on the line”.
“I’ll give you a new one: trampolines on the line,” he continued. “Every year, as a result of high winds, at some point somewhere on the network, we will have disruption because at least three trampolines have blown onto the railway. Where do you even begin with the data for that risk?”