Research in action
Tools to combat cognitive underload | Driving policies being reviewed in light of research findings | Predicting delays using machine learning: from feasibility to first operational use
Repetitive tasks can cause drivers to lose concentration, but there are ways to stay alert.
Some aspects of train driving involve repetitive work for extended periods of time, requiring little mental effort. This cognitive underload may make it difficult for drivers to maintain concentration, which can lead to mistakes.
In 2019, RSSB worked with the rail industry to explore the issue of cognitive underload. As a result, industry now has much more information available on how and when cognitive underload can impact drivers. The resources include a toolbox of mitigations developed by drivers for drivers, with practical tips for staying alert. They include everything from opening a window to speaking aloud in a different accent, from drinking coffee to exercising at stations.
Southeastern recently introduced RSSB research material on cognitive underload into its programme of driver skills enhancement. The operator produced a bulletin featuring some of the key causes of underload on its service and some techniques to reduce it.
The new limits will then be usable across the network, without the need to go through the Network Rail service plan review process to agree an individual increase.
We encourage all operators to implement our findings on cognitive underload if they haven’t already done so. Find out more about the research and download the toolbox from rssb.co.uk/research-catalogue (search for T1133).
To discuss the research, or for support with implementation, contact Marcus Carmichael, Professional Lead, Operations and Performance:
Marcus.Carmichael@rssb.co.uk
We are using our new understanding of the whole-system effects of defensive driving to update driving instructions.
As a response to past high-profile incidents, defensive driving has been an integral part of operators’ driving policies for 25 years. This research was requested by industry to understand whether the current approach optimises the balance between signal passed at danger (SPAD) prevention, train performance and its associated safety risk, and network capacity utilisation. These areas are linked, and too much weighting in one area may have unintended consequences that increase whole-system safety risk rather than improve it.
A review of driving policies found that defensive driving instructions have remained largely unchanged since the concept was conceived in the late 1990s. This may mean that improved braking characteristics and equipment on new trains, such as sanders or wheel slide protection, are not being exploited.
The research also found that driving policies are often developed without sufficient input from a wider set of stakeholders. Both fleet and signalling design have valuable knowledge to feed into driving requirements, but policies are generally developed without any involvement from relevant engineers.
The research simulated the impact of specific driving policy instructions at two case study locations: the Derby area and the South Western Mainline between Woking and Wimbledon. The simulations were designed to evaluate how a range of driving styles most likely to result from different instructions—particularly different braking rates and timings—impact safety, performance, and network capacity. Findings showed that restricting the braking ‘steps’ that drivers can use increases the whole-system SPAD risk, supporting the case that defensive driving may have wider unintended consequences.
Following publication of the research, all passenger operators have started to review their professional driving policies in line with the good practice guidance to understand the opportunities to optimise the instructions they are offering their drivers. They will report back to the Ops Standards Forum before the end of the year.
Good practice guidance is available for passenger, freight, and on-track machine operations. Go to rssb.co.uk/research-catalogue and search for T1305.
To find out more, contact Marcus Carmichael, Professional Lead, Operations and Performance:
The National Rail Communication Centre is gaining first-hand experience of more accurate and timely delay predictions.
Predicting how delays will propagate across the rail network is a complex challenge with many variables. An abundance of data is captured from trains passing through timing points across the network, but they are difficult to interpret in a meaningful way. Presenting an aggregated, AI-powered view of the data can deliver a more holistic picture and generate new insights, which can help operators to make more informed decisions.
Building on a feasibility study that was funded through RSSB’s Data Sandbox+ competition, Frazer-Nash Consultancy has developed a new delay prediction tool. PREPAIR applies a machine learning model, trained on historical delays, to live Network Rail data, which generates predictions on how delays might develop in the coming hours. Results can be visualised across the whole network, so users can quickly identify delay hotspots that could impact their operations.
The National Rail Communication Centre has installed PREPAIR on a large screen in its control room, which shows current and predicted rail traffic and updates every 15 minutes. Staff are also using PREPAIR on their iPads, which means they can interact with the data and drill down to get a more detailed understanding of reactionary delay in congested areas.
Operational control centres around the country could make use of PREPAIR to better understand when significant train delays are likely to occur, identify options to mitigate the effects of those delays, and deliver more accurate information to help customers with onward travel plans.
PREPAIR has been developed in collaboration with Lampada Digital Solutions, using the NR+ network database, and Frazer-Nash Consultancy is working in partnership with Thales Ground Transportation Systems to further develop and refine the solution.
For more information about PREPAIR and how you can use it, contact Richard Wheldon: R.Wheldon@fnc.co.uk
Read more about the initial Data Sandbox+ research that led to the PREPAIR tool. Go to rssb.co.uk/research-catalogue and search for COF-DSP-08.
For more information about the Data Sandbox+ competition projects, contact Melissa Frewin, Senior Partnership and Research Grant Manager:
Melissa.Frewin@rssb.co.uk
