Using data to create predictive safety models

We’re very proud of our safety record at NATS and, right across the industry, aviation has never been safer. However as the volume of air traffic continues to increase in the UK we have had to find new ways to ensure those standards are maintained and improved.

Alongside our existing safety management systems, the NATS Safety Strategy and Safety Plan talk about something called “Safety Intelligence”. This involves sourcing, analysing and making value-added safety data available to everyone in a safety improvement role across NATS.

It’s an irony that as we get safer – which is obviously what everyone wants – we have less and less data to learn from. This is where the Safety Intelligence approach is helping as we can define and track lower-level safety events and leading indicators, using that information proactively to try and predict and then mitigate situations where risk could manifest itself.

It’s not quite Minority Report, but data can help us predict and then mitigate areas of possible risk.

For example, although aircraft rarely breach vertical and horizontal separations simultaneously, the occasions when aircraft maintain horizontal separation but “bust” their cleared flight level by 300-400ft can be monitored. Going a step further, if level busts are more prevalent when aircraft are instructed to change levels, the most common regions for climb/descent instructions can be identified and taken into account when redesigning routes or procedures, to ultimately lower the probability of a loss of vertical separation in future.

The Analytics Safety team where I work plays a key role in defining how our safety and systems data is shared via our Business Intelligence data warehouse and dashboards – the one-stop shop for operational data across NATS. Once initially defined, data can be streamed into the warehouse each day and used with a variety of complementary information to help people make quicker and better informed decisions.

The shift in making this safety data available across NATS at the click of a button has vastly increased the potential for people to quickly identify areas of the operation which could be further improved and then assess the impact of recent safety improvement actions.

This was a concept that I recently presented at the UK Safety-Critical Systems Club to an audience made up from the aviation, maritime, rail and nuclear industries, who were looking to learn how NATS creates and disseminates this safety intelligence.

Better use of data is important, but we also recognise that fundamentally it is people who create and ensure safety. NATS has a very open safety culture and one industry-leading initiative that the Analytics and Human Factors teams here have recently developed is “Day to Day Safety”. This focuses on identifying and sharing the methods which individuals use to maintain an extremely high level of safety in their day to day roles.

We use trained observers to record the positive techniques being employed and the anonymised results over a period of time are then shared across the entire team for others to learn from and adopt in order to strengthen the team’s safety resilience. This is a technique that we regularly use for our air traffic control teams and that we also offer as a service outside of NATS, with a number of airlines and airports currently benefiting from this approach.

Thankfully, safety has become something that most people who fly can take for granted, but it takes a huge amount of work and on-going diligence to ensure it stays that way.