We're really concentrating on a digital-first mindset. It doesn't matter who you are or where you operate within our company, we really would like you to foster a digital-first mindset.
AI may be the trend du jour at tech industry events, but there are end users out there that can stake a claim to have been on the learning curve for a long time. One such firm is Rolls-Royce. Talking at the recent Microsoft Envision conference, Thomas Lee-Warren, Digital Chief Technologist for Rolls-Royce Aircraft Engine, said:
AI and data engineering has been trending now for a while, but at Rolls-Royce, we've really been wrangling data and looking at algorithms for probably 30 years or so.
Founded in 1906, Rolls-Royce is the number two manufacturer of aircraft engines in the world with 900 planes in the air at any one time. Those planes are sending back millions of data points to the Rolls-Royce service centers , explained Lee-Warren:
What we're looking to do through all of this, through our data, through our AI, what's that really about for our customers? And it's really about increasing the availability. And we're really looking to reduce the engines on the ground, our EOGs, because that costs both the airlines and ourselves in terms of our end-to-end supply chain in an optimum way, that doesn't work for either of us. The key focus for our business is about providing optimized, predictable costs for our airlines.
Any discussion of cost leads to consideration of fuel as the major contributor to airline spend, accounting for around 40%, according to Lee-Warren. That means that any percentage point that can be saved could translate to a saving of around $250,000 per aircraft, per year. Lee-Warren painted a picture of how the firm’s data analytics can assist here, telling the audience to image they were on a flight:
We get an alert. Probably most of the people in here have been on aircraft where we have gotten an alert, and it's really nothing to be concerned about, nothing major has happened. But those sensors that we have on our engines have actually picked up an anomaly. And what they've done is with these small anomalies, we've probably bursted a bunch of information back to our service centers. Now, before we had the sensors and we had Big Data, we probably would have had a situation arising for a long period of time, which would have made our engines actually perform not on an optimum basis. And so what does that do? That actually increases the cost to our airline customers.
New world of data
But in a world of sensors and AI, such a scenario is dealt with differently, with service engineers able to delve down into the data around the alert that has been issued:
Our service engineer can start to look at putting a diagnostics together, which is then going to go off to our maintenance engineers… He's understood that this is about a low oil pressure problem, and then we've got an alert that's been raised. Very quickly we've got all the details through a templated e-mail, and now the service engineer is going to pass this across to a maintenance engineer with the click of a button, send.
What results is a visualization of all of engines which are on the planes on the ground, said Lee-Warren:
We're actually showing what time they're going to take off. We've got the departure and where they're heading to. Just coordinated with that is the type of engine that engine belongs to. We've also got all of the alerts, the ones that you're not worried about. We're still keeping track of those. Not only have we got the alerts in text form, but we're actually showing each of those alerts on which flight they're accompanying and where they are geographically located. We're starting to now bring in third-party information, which actually helps the engineers understand actually that plane with that alert is going into inclement weather, whether that should be a concern or not.
In terms of the data visualization, the service engineer is able to see where the plane took off from and where it’s heading to:
This starts to become important when we start to think about timing. What sort of intervention will we make? It's probably not going to be an intervention, but we could if we wanted to, send an engineer to go do some sort of an engine wash or maybe change a filter, which will increase the efficiency of that engine. Again, making sure that we don't have increased cost for our airline customers.
So then we can actually drill down and, again, what we're seeing is this richness of data. So our service engineers and maintenance engineers can actually look at that engine in detail. What's been happening over a number of days, weeks, and months?
All of this adds up to Rolls-Royce delivering a sense of safety to its customers, says Lee-Warren:
We're actually collecting millions of data points. What we've done is we've looked at what kind of intervention that we should carry out. And what we've done when we’ve detected an anomaly with the sensors that are on our engines, we've been bursting that data through our satellite connections and down to the IoT Hub.
We've then taken the data [and] we’re now starting to look at how we can put together our diagnostic networks and the algorithms that we've been building over the years and how we start to actually merge that with the machine learning algorithms that we have available to us from the Azure Suite. Because we quite like this plurality of different techniques that we can use to train our algorithms to really evolve us to the next stage in how we actually detect at the earliest moment any anomalies in the efficiency of our engines.
Really, what that's doing is bringing the digital world and the physical world as close together as humanly possible. In data innovation, we're really looking to see how we can use data to actually create the optimum value to our customers. So we're looking at the data we hold, we're looking at third-party data, how do we mash that together to really evolve our next level of after services?
And of course, there’s a focus on AI, he adds:
AI is really important to us and we're really excited about the opportunity it presents in the next evolution of our services in either designing or manufacturing. I think we're really looking to see how we can create new networks and partnerships, how we can sort of really drive this outside-in thinking. Because although we do things incredibly well, we can actually learn from others.
We're also really excited about how we can use some advances in AI. So whether it be genetic algorithms and we start to look at biological evolution and survival of the fittest, that's really getting us excited about how we can train our algorithms…there’s no better time to be coming to Rolls-Royce. We're doing amazing things. We're doing amazing things.