Bill Ruh on Digital Transformation, IoT and the Future

Session Description
Bill Ruh, CEO of GE Digitial, talks to Techonomy's David Kirkpatrick about the future of data analytics and connected machines. Ruh makes the argument that the market winners of the future will be companies that can make their assets as productive as possible.
 
Kirkpatrick: The theme of our conference, and in fact our entire year at Techonomy is what we’re calling man, machines, and the network. It’s a way of sort of taking a slightly bigger view of what’s often called the Internet of Things and a lot of related phenomena that are central to a tech-based society that we’re all moving into. And in fact, Bill Ruh, who is the CEO of GE Digital, and is both running that as a business unit for customers, also in charge of digital transformation of GE’s entire operations, so it’s a very interesting dual role. Three years ago, roughly, Josh and Simone and I sat down with Bill at CES, and it was the first time anybody really kind of shook us by the shoulders, saying this IOT thing is really big, and it really, really got our attention because he’s quite passionate and eloquent about what’s happening. And it really sort of changed the course of our company. So it’s especially gratifying to have him here today to open this conference and to talk about why the Internet of Things, and what GE calls the industrial Internet, is so central to its evolving self-conception.
I’d just like to start by asking you, Bill, to talk about, at the highest level, what is happening, in your view, in terms of digital business and social transformation?
Ruh: I think that one piece of data that I heard a while back that really I think crystalized what’s going on is that over a 20-year period, from about 1990 to 2010, industrial productivity was at 4% year over year. And when you’re growing at 4% year over year, either your top line or your bottom line, life can be very good, and so you continue doing what you’re doing. But about 2011, industrial productivity went down to about 1%. And so, you know, it appears that everything that had been done with lean and process improvement, etcetera, had sort of run its course, and we don’t see that kind of productivity that companies need to grow and to be successful.
Kirkpatrick: In fact, just today there’s a report that US productivity in 2015 is probably going to have declined.
Ruh: And it’s not surprising. The interesting thing from my perspective is if you look at the last decade, for all the talk we have about the technology, it’s really been about the consumer. And that’s actually why we say the industrial Internet, because there’s not an Internet. It’s the consumer Internet. Whether the industrial world takes advantage of the technology to drive productivity, it hasn’t yet. But I think that we are at a point where, fortunately, the price points driven by cellphones and networking have come to a point where we can now do these interesting things.
And the interesting things are this. Look, I think that if we—what we believe is that if you can connect machines and pull data off them—which is sort of where IOT stops in many ways. We never use the word IOT with an industrial customer. It just doesn’t—the conversations don’t go well. We really talk about this idea of driving productivity through data and analytics. And it’s a very simple thing. But the reality is that’s where we think the biggest opportunities are going to be, and, you know, we’ve done it for ourselves first to show people. But this idea that I’m going to connect my machines, gather data and drive really interesting analytics to drive outcomes is probably what’s going to drive productivity going forward. And we think the most interesting applications are not going to be necessarily the consumer Internet in the next ten years, but it will be on the industrial side.
Kirkpatrick: What’s an example of an outcome that you think is happening that is otherwise unattainable that you’re excited about?
Ruh: I’ll give you an example, one that we’ve done is that, you know, we have been—our aviation services business is going through a digital transformation. And they are rethinking how you maintain a jet aircraft engine, and it’s really moving, I think, into a world where we treat every engine individually. I want you to imagine this in the auto world, because this is where the auto world has to get to. If we look at it, there’s no good reason to change your oil every 6,000 miles other than it says to do that. The fact is, how you drive, where you drive, when you drive, some people ought to be changing it earlier and some people could go much longer. There’s nothing magic about those dates. It’s the operation that matters.
So we do the same thing now with jet aircraft engines, where every engine is individually being managed with a unique maintenance program. Now, what’s happened is that we’re giving our customers the ability to have more time on wing on their aircraft because they’re—
Kirkpatrick: You mean flying?
Ruh: Flying. They call it time on wing.
So that’s what they care about. That’s the outcome they care about, how much time on wing do I get for that plane. We’re giving them more time in the air of their jet aircrafts.
But I’ll tell you what’s interesting is we’re now customizing how we maintain it. So if you look, we—if you start looking at the data, you find that one kind of engine that deteriorates more than others operates in what’s known as hot and harsh environments. And we have more flights going through Dubai and Abu Dhabi and other places, and these are hot and harsh environments and you find that the dust is much finer grain than the dust that’s used actually to test these engines. And what we’re finding is that they clog the airfoil, the engine gets more damage. So what we do with those is we water wash them. And when you water wash them, or then it turns out you can bend the curve on maintenance and make it look like engines that operate say in the Northern hemisphere. We found—
Kirkpatrick: But you discovered that only by measuring them more exactly using sensors?
Ruh: Yes. And in fact, we found that we can now figure out what our hot and harsh environments—which may seem simple. It’s not as simple as it sounds to look across the world and begin to find the places where you know when you operate you’re going to get more cumulative damage. And in very high altitudes, we found this was occurring. And so now the airlines are able to adjust the way they fly, and they have a better product, they—and I think where we think we’ll go with this is that we’ll go to zero unscheduled downtime. So 41% of all delays are mechanical. We actually believe as we use data, we’ll get to the point where we can predict when something’s going to break before it breaks.
Kirkpatrick: Wow. And the same thing is being done, for example, with locomotives, which I think you have something like 95% market share in the United States on. That sort of thing is also being applied in a somewhat simpler environment, which is on the ground.
Ruh: Yeah, and then you start even to think about logistics of moving locomotives around, where we’re going beyond just the locomotive itself to rethinking how you plan the network traffic. So for example, we know you can save—we have a sensor-based application that can save you up to something like 10% fuel burn. And given fuel is a big cost in a locomotive for a rail operator, the idea that you can adjust the speed to take advantage of the terrain, the fuel efficiency, and give them back a significant savings on fuel burn, that’s a huge outcome. That’s productivity.
Kirkpatrick: Okay, so both of those are examples inside GE with GE’s own products. I know you’re working very aggressively to apply this learning to your customers as well. Talk about how you’re doing that.
Ruh: Yeah. I think that—I’ll tell you one thing we’re finding as we work with these companies, that there’s a pattern emerging that we’re talking to a lot of companies about. And I would say the pattern we discovered, looking at the consumer Internet, is this: the companies who can figure out how to make an asset more productive are going to be the big winners. Now, that seems maybe obvious, but if you just look at Uber, Uber doesn’t own any assets. Why do we use them? It’s because they make it more productive to get a taxi. We see that with Airbnb. Apple does that with other people’s assets on their cellphone applications. Companies who can figure out take an asset and make it more productive are really going to win, we think, in the market. And the interesting thing is you don’t have to own the asset. And that’s a little scary, as a company that owns a lot of assets and sells a lot of assets. We want to be the best at asset productivity. But that’s—I think if you boil it down, my belief system is that’s what every application will be about. So what we’ve done is we’ve talked to companies like Pitney Bowes. You know, Pitney Bowes has been big in big mailing machines, mail sorters. Physical mail’s not going away, despite the digital economy. You know, there still is a lot of mail that goes out. They now think about how do they make their asset more efficient and provide more throughput. But they’re now building out new kinds of applications to make this idea of mail and mail logistics more productive and building out their own software business around it.
So I’d say the companies like Pitney Bowes—we work with companies in bath and kitchen fixtures like LIXIL, Toshiba in elevators and things like that. Any company who wants to get in the industrial world, whoever can figure out how to make an asset more productive, those are the things that the customers who own those assets, they’re going to buy that. They’re going to use that. And quite honestly, when we look at what’s happening with the consumer, they may become more valuable than the folks who make the asset. So our sales pitch to people who make the assets is you better be the best at making your asset more productive.
Kirkpatrick: Okay, we’re going to bring up our three other panelists in just a minute. But before we do, I want you to just explain this concept of the digital twin, which is becoming a big part of how you’re achieving what you just described. Because I think it’s a very novel concept.
Ruh: Yeah. So for us—I mean a lot of talk about AI and deep machine learning. We’ve run a lot of tests, and when you’re in an industrial firm, the one thing that—most industrial firms have been doing analytics for a very long time. They haven’t been doing AI-style algorithms. They’ve been doing a lot of statistics and advanced math kind of algorithms. And the other thing that you do is you do physics-based modeling. When you design a product now, you know, especially the more expensive products, I mean they all get developed online using design tools. We run super computers. You model it out, you run a million tests to try to get the design right before you move into that physical world. And by and large, those models were used in the design phase. Well, what we’re doing now is we’re using them in the operation phase, which no one has ever thought about in this way before. And then we’re coupling it with machine learning.
So the idea here is we use the machine learning and a lot of statistics—so not everything is AI. And we bring in lots of data, and what that will tell us is what’s happened in the past. The statistics is good when you have large volumes of events and you want to correlate them. The machine learning is good if you have a jet aircraft engine and there’s only let’s say 29 events you’re looking for out of millions. The statistics doesn’t give you it as precisely where the machine learning does. So we get really good understanding of what’s happened in the past, and then we couple that with taking and using the physics-based model to figure out what’s going to happen in the future. And the physics-based models, you pump that learning into it, and now suddenly you can run what if analysis for a million cases and say, okay, what is the optimal way to configure this machine. And so we believe the future is you’ll have a digital twin of AI, statistics, and physics-based modeling coming together, and that these twins will actually come together in systems, so a power plant will be made up of systems.
And let me tell you, one of the first ones we did this on was in our wind turbines. So we have now got the models, and the models work at the turbine level and at the windfarm level, and using these and using new kinds of sensors, every wind turbine is dynamically adjusted with zero additional stress put on it. Let me say anybody can make a windfarm generate more energy. It’s just the blades will break quicker, so how do you do it without maintenance problems. If you can do that, we found when we started with the algorithms, we were generating 5% more electricity with no physical change, and now we’re up to 20% more electricity we believe we can generate on a windfarm just through this digital twin technology and using it that way.
Kirkpatrick: Okay. Could you guys please come join us? But as they’re walking up, so basically the idea is you’re going to be able to give that digital twin capacity to more and more other companies in the economy, you believe, in order to generate efficiencies. Everyone will be able to increasingly model their ecosystem, their infrastructure before they even use it.
Ruh: And make their assets more productive.