How Network Rail CIO Aidan Hancock is using data science to speed up your commute

How do you make trains run on time? It turns out you use a mixture of data science, analytics and cloud, according to Hancock

"I was brought in to bring innovation. No, I don't like the word innovation, either. It has the connotation of endless pilots."

Aidan Hancock has been the CIO of Network Rail for just three months, but has already made it clear in this short time that IT buzzwords aren't his bag.

"I've heard from business leaders from multiple huge corporations, ‘We must do digital'. What do you mean by that? ‘I want machine learning and blockchain. Yeah - transformation'. Okay, but what do you mean? Everyone wants all that, but no one can really define it," he tells Computing.

Well Hancock is defining it and, for him, it's a relatively simple message, with a simple trajectory: At Network Rail, it's going to be all about the data.

After 12 years at BP, where he was tasked with selling a similar outlook (we get the "data is the new oil" jokes out of the way, and move swiftly on), Hancock is ready for the challenge of more actively digitalising Britain's rail network.

That's 20,000 miles of track, covering almost anywhere on the land mass of Great Britain where trains can deliver people and things, and of which 10TB of data is captured for every 400 mile stretch, every fortnight.

And with £969m budget for the space of time known as Control Period 6 - 2019-2024 - he's seizing the chance to innovate by the scruff of the neck.

So in his own words, how - exactly - is Aidan Hancock defining IT transformation?

"The way I'm going to define it is we have vast pools of data," he says.

"And I mean, vast. More than probably every other major organisation outside of, possibly, Google and Facebook and some of those other tech giants. But we're not using enough of it. We're not leveraging it. But we're going to change that."

Hancock says by "pure guesswork" he estimates Network Rail is leveraging about 10 per cent of the data it could be, but is already able to start scaling up to several projects - mostly AI and machine learning-based - to begin improving the function of services by removing manual operations.

Network Rail has been collecting data for decades. As an arm's length public body of the Department for Transport, it manages just the infrastructure of the country's railways, but doesn't own or run any trains.

Except, it does. Unbeknown to many, Network Rail has trains (essentially mobile field bases) trundling up and down the country measuring and checking that everything is working - and collecting "vast amounts of data," Hancock says.

He is confident this sort of mostly untapped legacy and current data can be used to more fully automate timetabling systems, or cut down on the dreaded points failures that scupper many a commute. There are even, he reveals, still a number of manual signal boxes in place.

"I'm not saying that's [even] all digital these days - this is a huge, complex, long running programme," says Hancock.

Living data

An IoT-focused programme called Intelligent Infrastructure has been live in "various places", since 2011. "Intelligent Infrastructure is exactly what you'd expect," says Hancock.

"You're moving from a manual, fixed, periodic - because you simply don't have any more data - system to an analytics-provided system support saying, ‘Actually, the status of those points is fine, or the status of that overhead line equipment is fine'.

"It doesn't need to be maintained once a week, you can maintain it once every quarter. And because the workers have so much to do, they can be doing other things."

How Network Rail CIO Aidan Hancock is using data science to speed up your commute

How do you make trains run on time? It turns out you use a mixture of data science, analytics and cloud, according to Hancock

By way of example, Hancock describes visiting Glasgow Central Station recently to observe maintenance work ("No matter how smart you are, you never really understand how the work is getting done until you get out there"), and discovered first-hand the three-hour window, between the last train at 1.15am and first train to Euston at 4.28am, in which critical maintenance work has to be done across up to 16 tracks.

"By moving to a predictive analytics and decision support model, we could focus what they're doing better. Rather than saying ‘You'll have to do the same thing today as you did this day last week' it's a more powerful and flexible way of doing things. And that means we should see a more reliable network, and more trains running on time."

At the moment, Intelligent Infrastructure is still only in operation in selected places, but Hancock describes the programme as being involved in a "vast scaling programme", governed largely by the "Control Period" five-year time chunks mentioned earlier. Around £200m of that is going directly on Intelligent Infrastructure.
But as he describes himself as a data-led "innovation CIO", Hancock "doesn't want to be involved in deploying sensors".

"I've been involved in that previously, and that's a logistics exercise - and needs its own expertise. My part is asking, ‘Do we have a suitably scaled, robust platform? And is that platform plugged into state-of-the-art machine learning services - say, Microsoft Azure?'."

Hancock is keen to point out that Microsoft's platform is only one of many suppliers and services that's being used, but is enthusiastic about Azure's potential, especially as workers get to grips with collaborating on Microsoft Teams, and the wider Microsoft ecosystem.

There's everything to play for when engaging with new cloud technologies, as Network Rail is currently migrating from legacy mainframe systems - some built around IBM's 1960s System/360 OS - to a hybrid cloud setup in what Hancock calls a "classic data centre rebuild programme - shrinking our data centre estate and physically moving, with a big cloud capability around that".

He talks of even moving into the forecasting behaviour of institutions like the Met Office, except powered by the cloud instead of supercomputers.

But some of the legacy is also newer than one would think.
"In terms of modern devices, we had this strange position in Network Rail where we were very early to adopt Apple mobile technology, so we've almost now got a legacy mobile app estate, because we've been doing it since 2012," reveals Hancock.

"I think we're the only people who are [at that point]. My predecessor, and now current boss [managing director of Route Services, Susan Cooklin, did a fantastic job with Apple, so we have a number of obvious mobility candidates - work planning, safety, management apps, so we've got 70+ serious apps - I don't mean the little ones where you're just presenting information - but serious, integrated apps."

In a "dichotomy", however, Network Rail "hasn't quite competed the move to Office 365", and Windows 10 is also yet to be rolled out across the organisation.

[Turn to next page]

How Network Rail CIO Aidan Hancock is using data science to speed up your commute

How do you make trains run on time? It turns out you use a mixture of data science, analytics and cloud, according to Hancock

Factory shop

Still, in terms of productivity tools generally, Hancock and his IT team have a more nuanced in-house approach.

He isn't fond of the "cognitive dissonance" of the enterprise employee expectation of applications that never turn out "as pleasant" to use as the consumer services on which they're often modelled.

"You can't predict who will use what, but usage tends to exist in pockets," he says.

"Now, the conventional way of doing [project] IT is, the IT business analyst goes and sits with Mrs Business Person and says, 'I hear you've got a problem', she says yes, the analyst says, 'Alright, let's define your requirements', they write some requirements down, go away, go out to the market, six months pass and something is produced, they go back to Mrs Business Person and she says 'Well that's not what I wanted'."

"Now we're very good at delivering projects in that conventional way at Network Rail, but there's other ways."

Hancock says he's reticent to use the word "bimodal", but that Network Rail has just started work on a "Digital Factory" approach of exploring and solving development problems.

"It's agile with a small ‘a' - because we're probably all sick of the capital "A" buzzword by now - but the Factory is a way of clearing the roadblocks from a set of agile teams, so that all the governance and procurement frameworks are all done and wrapped around the Factory."

If a problem needs solving, it's done by a multidisciplinary team and not "just IT", Hancock enthuses.

The result is development resources from IT, as well as knowledge of how to find the data, but everything else is controlled by project stakeholders. From data scientists to subject matter experts, to wider business people.

"All the people who need to solve a problem will be able to start solving the problem - or exploring it - on day one. If it's not exactly new, it's a new approach for us.

"And the reason I'm calling it a factory rather than an 'innovation garage' or any of these other names for it, is a factory delivers products, which are used."

Train to gain

In terms of collecting the wider talent to start filling in the data science parts of this agile new projects and home-grown solutions, Hancock has a two-stroke approach to finding and training data professionals.

"With data science, we're not going to transform overnight. On one end, you've got serious experts who live and breathe Python and R and so on, and they really love mountains of data, and I want them working here," he says.

"We've got mountains of data, we're part of UK plc, we're hugely attractive. What data scientists really want is data - big challenges. You can help train performance, and timetables, if you come and work here.

"But all the way at the other end, I'd like to get to a point where almost everybody, if you're sitting at a desk for your job, should be able to do some data science."

In terms of fully-fleshed data scientists, Hancock agrees there "aren't that many" in the south east of the UK.

"And I talk to a lot of CIOs - you look at the number who have been placed and the number we actually need related to that, and there's a big difference."

But Hancock believes it's easier to close than some believe.

"You could actually argue that anyone who's ever done anything fancy with a spreadsheet could be a data scientist. So we have graduates coming in who did something data sciencey, or at least they know how to code, and they're being presented with interesting challenges, or seeing the shape of something and saying to us, 'Actually, could we take a look at that?'"

Younger emerging talent is certainly helping Network Rail's data science grow "organically", but Hancock seems even keener on turning anyone and everyone else at the organisation into robust data scientists, too.

"At some point, we'll all do some data science," he enthuses.

"We're all getting a copy of PowerBI, so that's basically Excel on steroids. If we got our people working in PowerBI better, consolidating and making more sense of our data, the reports - instead of in a monthly periodic basis, cut and pasted and all that - would just be instantly done.

"It's a huge amount of time being saved, and more accurate. Almost everyone needs dashboards! I want them to be dynamic, and real-time, and accurate."

Hancock clearly has huge data-led ambitions as an "innovation CIO" at Network Rail, but after only three months in the post and good board relationships on an individual level he's certainly off to a flying start.

"I'm more than happy to put my hand up and say ‘I really do think digital can help run more trains on time and get better train performance. My challenge is enabling us to do that," he tells Computing.

Well, you heard it here first. Will Aidan Hancock's data-laden plans deliver bang on schedule? Only time will tell.