When travelling by train, you want to know that you’re going to arrive at your destination on time. In fact, along with security and cleanliness, reliability consistently ranks amongst the top concerns for rail customers.

Reliability is important to us, too. That’s why we release monthly reports to showcase just how punctual we are. And it’s why we have a close working relationship with Network Rail and the Department for Transport, with an aim of 90% punctuality between Monday to Saturday. 

Whilst some disruption and delays are unavoidable, we are looking at new ways to reduce or mitigate these by trialling new-to-the-rail-industry technologies and innovations, including machine learning. By taking a test-and-learn approach, we see which solutions have the biggest impact before scaling up or taking learnings forward.

From ticket to track – here are some examples of where we’re experimenting.

Before ticket purchase

Before you’ve even thought about planning your journey, we’re focussing on where we can put things in place to maintain an on-time arrival.

For example, the geospatial analytics platform that we’re working on to trial. It will feed information on live journeys back to our performance team, who will be able to analyse the data to understand where things aren’t running as smoothly as they could be along the line.

Reliability starts with a clear picture of the service. With live, geographically mapped train location data, we’ll be able to drill into the details of even the smallest delays. If a section of track consistently takes two minutes longer than it should, the data will help us to build up a picture - and we’ll be able to work with our teams on mitigations. We will learn from the past – and adapt to deploy a better service in the future, stopping delays before they even happen.

In the station

In-station delays can be caused by anything from slow-opening doors - making it harder for customers to board the train - to issues such as passengers with bikes being in the wrong place to board, causing a delayed departure time. The more we understand about the causes of in-station delays, the better we understand the measures and mitigations we need to put in place to reduce station dwell time and get trains departing at their booked time.

We’re undertaking a couple of things to address this. Firstly, our Machine Learning Team is developing a tool that looks at mass amounts of data to predict where delays are most likely to happen and why.

A solution to address one cause of delays in the station is our new digital coach signposts – being trialled at platforms 4 and 8 in Doncaster Station. They’re improving the boarding and alighting process of passengers by giving them the information they need for the next train into the platform.

By providing this information in a simple, easy to understand way, it helps passengers find the right place to wait on the platform to quickly find their reserved seat, the Café Bar, where to store their bike or where to find an unreserved seat. And this isn’t just for LNER’s trains – they show information for any train operator’s train that stops at those platforms.

And don’t forget our app! It delivers notifications with the latest platform or disruption information - keeping everyone informed of any updates to the service, so nobody’s left running for the door at the last second.

On the journey

Once you’re on the move, sometimes unforeseen delays are unavoidable, but we’re exploring new, innovative technologies to mitigate some of the most unpredictable incidents. Like, for example, deer on the line.

These cameras detect where they are approaching the track and use variable ultrasonic frequencies to safely deter them away. That means we’re saving deer from being struck by trains, reducing the associated delays for our customers and the cost and big knock-on effect to our timetable of trains being out of service for repairs.

To support our busy Service Delivery Team, our Machine Learning Team collaborated on a tool that enables informed decision making, once an incident happens that causes disruption. It processes around two million data points every hour, giving the on-duty Operating Support Controller a complete overview of where delays are happening, potential pathing clashes that can cause delays, possible decisions they could make, and any associated consequences.

It learns from historic data - from the impact of broken doors to downed overhead lines - to make smarter decisions today. So even when incidents can’t be avoided, their effects are lessened.