Data Sandbox: Improving Network Performance

The reliability of train services is of paramount importance for customers and this is where, as an industry, we need to make the most significant improvements.

In October 2017, RSSB launched a £500,000 competition to identify novel data driven solutions to key network performance challenges and awarded funding to five projects.

The academic-led feasibility studies – due to publish their findings shortly - have successfully integrated existing datasets from a wide range of sources to provide new insights. Together they show that new approaches to data – such as machine learning, and graph theory – can play an important role in helping the industry drive performance improvements.

Three projects focused on understanding the impact of reactionary delays and developing techniques that could be used to identify mitigations. 

  • City, University of London and Risk Solutions with support from Great Western Railway - developed a method for investigating the causes, severity and likelihood of reactionary delay under different scenarios and understanding delay dependency between locations.
  • The University of East Anglia, with support from Greater Anglia, developed software to help train controllers determine the knock-on effect of primary delays
  • Liverpool John Moores University, with support from Merseyrail, developed an optimization model to determine the extent to which tactical interventions could limit the impact of reactionary delay.

Two projects explored how machine learning and data visualisation techniques can be applied to help reduce dwell time variability.  

  • The University of Southampton, with support from South Western Railway, developed high-resolution maps of dwell time hotspots that can be targeted for mitigation measuresThe
  • University of Middlesex, with support from Southeastern, identified correlations and recurring patterns to predict real to-the-second timing patterns of passenger services.

The findings from these projects were presented at RSSB's Enabling Better Performance event on 4 April 2019, where industry discussed the merits of the approaches and solutions as well as future development opportunities.

Topics covered

  • Reactionary delay
  • Dwell time
  • Machine Learning
  • Data Analytics
  • Data visualizations


Developing an intelligence ensemble system for predicting and preventing train delays (COF-INP-02)
Agent based modelling and visualisation of the causes and consequences of knock-on delays (COF-INP-06)
Predicting and mitigating small fluctuations in station dwell times (COF-INP-03)
spark bulb
Providing data analysis insights into real to-the-second timing patterns of passenger rail services using Machine Learning techniques (COF-INP-04)
Anticipating and mitigating reactionary delays – a case study on the Northern line of Merseyrail (COF-INP-05)
GMRT2472: Requirements for Data Recorders on Trains
ATOC Guidance Note: Use of Data Recorders
Haven’t found what you’re looking for?
Get in touch with our expert for more information
Andy Castledine
Cookies help us improve your website experience.
By using our website, you agree to our use of cookies.