Best listening experience is on Chrome, Firefox or Safari. Subscribe to Federal Drive’s daily audio interviews on Apple Podcasts or PodcastOne.
Both the armed services and contractors are trying to speed design and prototyping of new platforms and weapons using digital techniques. They’re taking traditional computer-aided design (CAD) and updating it with artificial intelligence and machine learning, giving development a sort of autonomy it’s never had, and building more sophisticated virtual environments in which to test new platforms....
Both the armed services and contractors are trying to speed design and prototyping of new platforms and weapons using digital techniques. They’re taking traditional computer-aided design (CAD) and updating it with artificial intelligence and machine learning, giving development a sort of autonomy it’s never had, and building more sophisticated virtual environments in which to test new platforms. For one view of how industry is doing this, Federal Drive with Tom Temin turned to the President of Raytheon Missiles & Defense Wes Kremer.
Insight by Leidos: As the Intelligence Community blends open source and public data with its own intel, balancing the use and development of AI tools to analyze data, and strategies for sharing data appropriately across classified teams, creates new challenges. Download this executive briefing to read more!
Tom Temin: Mr. Kremer, good to have you one.
Wesley Kremer: Tom, it’s great to be here with you this morning.
Tom Temin: And we should point out that as a one-time user and consumer of these types of weapons, this is something I understand is close to your heart, this whole idea?
Wesley Kremer: Yeah, absolutely. I served in the Air Force for 11 years, and was the weapon systems officer on the F-111. And the F-15E. So I feel like you know, I do have some experience in this. And now to be at the helm of one of the large defense partners for the Pentagon, it’s a real pleasure to be able to still serve in this capacity and be able to support our men and women in uniform.
Tom Temin: Now, CAD has been around for a long time. I remember in the 80s, Apollo workstations and looking at people doing wire diagrams. And artificial intelligence is not brand new in this context. So what is changing here? What is the new development, what’s the essential change that’s new here?
Wesley Kremer: Obviously, Tom, the the buzzword is digital design or digital thread. And the part that’s really different is the emphasis is now on the full product development lifecycle. So it’s not just doing mechanical CAD design, it’s not just the engineering piece, it’s connecting all of that together so that you design something in engineering, it flows seamlessly to your supply base, it flows into operations and being able to literally, like we talked about 3D print something, and carries all the way through the sustainment model of how you operate, how you maintain, and how you sustain. So that’s really what we’re trying to do with digital thread is connect that entire product development lifecycle.
Tom Temin: Because the sustainment part of the average thing that flies or rolls over the ground, or sails at sea, the sustainment is much greater cost in the lifecycle than the acquisition cost, isn’t it?
Wesley Kremer: Absolutely. In many cases, that is the largest part of the cost of a weapon system is that sustainment costs, it’s not the original procurement. And so that’s why it becomes really important to understand that. And we also talked about integrating that into the cost models and being able to represent the entire thing in a digital environment. And, as you said, many of the pieces of this have been around for a long time. And clearly, artificial intelligence, machine learning are not totally new concepts. But putting all of that together to really speed, the process by which we do that is what we’re really trying to do. And again, for somebody like Raytheon Technologies, we’ve been working on this for many years. But I would say, this year 2021, really, all of those pieces are taking off. And we’re seeing just an incredible acceleration, with the toolset and all the things and it really is speeding our designs.
Tom Temin: So it sounds like one of the essential challenges then is designing a testing virtual environment that you can have confidence would give you the same output, as if you were able to test a thing in a real environment.
Wesley Kremer: Yeah, Tom. Absolutely. And I think that it’s not actually that test environment that’s the hardest part of this. And let me talk just a little bit – we are very good at, for example, one of the programs we have today work every single night, we’re flying 6 million miles in essentially a virtual simulator. So you think of like a Microsoft Flight kind of thing, only doing this totally autonomously. You’re flying this weapon in a representative threat environment. And you’re seeing, “Hey, did the changes I made to the signature, the changes I made to, different parameters, or the sensor I put in – does that make it better than it was last night on the 6 million miles that it flew?” So that part, I think, is well understood. And we’ve had good advances in that. The challenge that I think industry is still facing is when you get into the lower level components. So I can take a mechanical structure, I can design that in CAD or in 3D, and I can go print that. But when I get down to things like a circuit card, even again, the tools are better for designing a circuit card. But when I have a digital stack of six or eight circuit cards, understanding all of those little electrical interactions, and second third order effects, coupling, things like that, I’m still to the point of where I have to build hardware and test that. And so that’s what we’re trying to get to is can I actually build the entire weapon system in a virtual environment and then go straight to print? And we’re not quite there yet.
Tom Temin: We’re speaking with Wes Kremer, he is president of Raytheon Missiles & Defense. And we should put some realism on this, you are trying these technologies on a new air launched cruise missile. And let me ask you this, when you say you test it by flying it 6 million miles every night, how in a virtual environment, can you simulate the wind pressure, the thermal dynamics, and back and forth, and all the other things that might face in flying millions of miles?
Wesley Kremer: That’s actually the easy part. We have very sophisticated models that not only replicate the environment that it flies in the physical environment, like you said, of air pressures, and clouds, or storms, or weather, but also the threat environment that it flies in. So those things are actually well understood. And part of it is you’re really flying a representation. And you understand the capabilities of the system, you understand the physical environment, and now you’re looking at different, you know, modifications to your platform or to your weapon system on that. So again, that’s very well understood. The challenge was what I was talking about earlier, is when you decompose that down into the individual components, and you start to understand, you know, what are the individual interactions on something – at a component level, is where we’re still advancing the technology to get to that point where you literally can go, a first-pass success. I can design something totally in the digital environment, and know that it will operate in the physical environment exactly as I had predicted.
Tom Temin: In other words, the pieces inside that are inside the skin, they’re not directly in the environment that it’s flying through, but they’re indirectly by being inside of it.
Wesley Kremer: Correct. Something like what we would call a mixed signal ASIC. So you know, an ASIC is an application specific integrated circuit – the little piece of silicon chip that does a specific kind of thing. Well, clearly, we understand how to do one of those digitally but in many cases, we have things that run both analog and digital signals on the same little chip. Understanding those interactions in a virtual environment is still – the models still aren’t perfect in that. And so we end up you know, we have a simulation, we have a model, we design the circuit, we go send it out to a fab [semiconductor fabrication plant], and we have some of those chips built. We come back and measure those, and they’re close, but they’re not always perfect. And when you take a stack up of say, a dozen of those, and you look at those little interactions, or those little, margins or tolerances, that you get, the variance between those, that’s the part where we’re still working to improve our models, and get to this ultimate thing that we want to do, which is to literally be able to pull a digital thread and go all the way from concept to fielding something.
Tom Temin: And therefore, one of the challenges then is the acquisition of the correct data that you can use to train the overall model?
Want to stay up to date with the latest federal news and information from all your devices? Download the revamped Federal News Network app
Wesley Kremer: Absolutely, that’s one of the pieces of being able to have all of those tools. And we call them, essentially virtual environments, or emulators, for example, one of the areas is field programmable gate arrays, FPGAs, which are essentially a reprogrammable chip that takes the place of an ASIC. And the technology in that in the last just four or five years, the emulators and the ability to emulate those in environment has come along tremendously. But they’re still not perfect, you still end up actually having to program one to see how it’s going to operate, especially if you’re operating a lot of them. And it’s in a very dense electronic environment.
Tom Temin: And can this concept work with, say a notch down from a completed system like a missile, to say, an engine? Because I know one of your sister companies, Pratt & Whitney, they make jet engines. And those are a long time in development and testing. And sometimes 15 years after they’ve been in service. They discover things “Whoops! We missed that or this particular metallurgy isn’t quite right,” and so on.
Wesley Kremer: Yeah, and even in hypersonic weapons, you know, in Raytheon Technologies, we’re developing a scramjet engine. Which, is essentially a jet engine that has no moving parts that operates that supersonic or hypersonic speeds. And that is totally 3D printed, the engine that we do for that. It used to be hundreds of parts and now it’s literally a handful of parts and you can print complex curvatures and things that are impossible to machine. And so there’s incredible advances that are happening here in the additive manufacturing area. When you tie that together with digital design.
Tom Temin: And adding this all up we see a DoD that is sort of littered with programs that are late and overdue. There are some flying platforms made by other companies that have been 10, 15, 20 years in development, and they’re still not really fully baked. Do you envision these types of digital designs, this end-to-end lifecycle, digital design, ultimately being able to get around that issue, so that you can have an idea and field it, maybe in five years instead of 25 years?
Wesley Kremer: Absolutely, Tom. I think that is really one of the great hopes that are out there for the digital thread or digital design. And one of the things is, it’s not just the speed on the battlefield, it’s the speed to the battlefield. And that’s one of the biggest drivers for digital design is to be able to field weapon systems faster to understand in a digital environment, exactly how they’ll perform so that you can limit the testing and not have to take as much time to test. And then as you point out, to then be able to iterate faster to make updates, modifications or improvements to that.
Tom Temin: And there’s some urgency to this in the national sense, isn’t there? Because when you’re dealing with software systems to drive physical systems, then the competitive advantage over say, China is not all that great.
Wesley Kremer: Absolutely, the cycles of Learning and Technology cycles have become so compressed that if you were a technology iteration ahead, that might translate to several years or, or even a decade, and now that translates to weeks, or maybe only months. And so clearly, we have to go faster. And this is one of the driving tools that we’re using. And I think that you know, this is a great partnership between government and industry, to be able to make this successful. We also have to change the acquisition process, right? We have to get away from the traditional waterfall concept, the preliminary design review, the critical design review, the initial fielding, the developmental test, the Operational Test. I mean, we have to all align on this model, or this digital simulation. And there has to be alignment, that that is the basis of what we’re going to do – not these other types of reviews. So I think that’s still part of the process that’s catching up is the contracting, and the oversight part of this of how government and industry works together to use digital technologies to field systems faster.
Tom Temin: That and also keeping requirements reasonable, and building a platform that you can get done quickly. And then maybe adding capabilities down the line.
Wesley Kremer: Yeah, and that’s where I think, Tom, there’s an actual advantage. You know, one of the things we always had in challenges in the past is striving to meet a requirement. And oftentimes there would be a lot of energy spent on trying to get that last 1% of capability or something. One of the advantages of digital simulations and the ability to apply artificial intelligence to it is, you know, I call it dollar-izing DB’s. You can actually understand, what does that last DB of performance actually cost you? And you can have that discussion early on before you’re actually building it to say, “Hey, is this a requirement that we absolutely have to have, or if we backed off by 2%, on this requirement, it might save us 10%, either in development time, or in production costs.” And being able to actually show that, I think, is also one of the improvements that we’re seeing with digital design.
Tom Temin: And finally, we have been using Air Force examples. And people cite the former official Will Roper, who pioneered a lot of this or pushed a lot of this through the Air Force. Are you finding, in your point of view that all the armed services have this idea?
Wesley Kremer: Absolutely. We’re seeing that all of the armed services are focused on this. Great example, like in the Army is their OMFV, the Optionally Manned Fighting Vehicle, which is the Bradley replacement. They have steady contracts to that, to several industry partners. And it’s totally digital environment. There’s absolutely, you know, no metal being bent. It’s all being modeled in a digital environment. And so we’re certainly seeing that, and that’s, you know, just one example, but across the services, this is definitely the way of the future. As you said, Dr. Roper, is truly a pioneer in this and I think it’s been a driving force. And we see huge advantages for this not only within our company, but again back to speed and being able to deliver capability to our warfighters and our allies around the world much faster.
Tom Temin: Wes Kremer is president of Raytheon Missiles & Defense. Thanks so much for joining me.
Wesley Kremer: Absolutely, a pleasure to be here.
Tom Temin: We’ll post this interview at FederalNewsNetwork.com/FederalDrive. Hear the Federal Drive on demand and on your device. Subscribe and Podcastone or wherever you get your shows.