Watch the Sessions:
Talks added weekly from OpenAI, Rescale, AWS, Microsoft, Intel, Google and more…
Supersonic Commercial Flight is Possible with Big Compute
Blake Scholl, CEO of Boom Supersonic
Good morning. Really excited to be here talking about how advances in high performance computing are enabling advances in high performance aircraft. The story of high-speed aircraft is a very interesting story in technology and we’ll get to that in a couple minutes. But first I want to talk about why we’re doing this. At Boom, we believe that life happens in person. And that real face time, beats FaceTime. And so our vision is to make the world dramatically more accessible by removing the barriers to travel, which are principally time, money, and hassle. If we can chip away at those, we can build a world where more people can go more places more often, which is going to be a better planet for all of us to live on. So let’s look at that history. Last year was the 50th anniversary of Concorde, the only ever a supersonic airliner to have sustained passenger service.
But today, the most impressive airliner ever built is not found in our airports. It’s founded our museums. And I think this is crazy. And the tech world we’re used to technology is from cell phones to electric cars, to the internet itself, starting with niche markets at relatively high prices. And then we work out how to get the cost down and achieve economy of scale and bring more things to more people. But yet in aviation, we’ve actually gone backwards. And today we’re flying around with the basic capabilities that we had in the late 1950s. If you look back at how this happened, so Concorde did not come from a startup or even from an established company. It was a joint venture between two governments, the French and the British. And in a way, it’s remarkable that it ever flew because most joint ventures between France and Britain, of course have been wars historically.
So Concorde was established by treaty in 1962. And as a cold war era project, the goal was not to go into a new age of supersonic aviation, but really to show that Western technology was better than Soviet technology. And so incredible things were accomplished technically, but very little thought was given to the sustainability or the practicality of the aircraft. Built with 1960s technology. It was a gas guzzler, and that meant that ticket prices were high and today’s money, a ticket on Concorde would cost you about $20,000 round trip from New York to London. And except for a very tiny number of people. That’s more like a bucket list item, not something that really changes how you travel the planet. So 50 years after Concorde, lots of things have changed in aerospace though. While we have it sped up airplanes, we’ve changed virtually everything about how we design and build aircraft.
From an aerodynamic perspective, we’ve gone from developing in wind tunnels where every iteration takes months and costs millions of dollars. To being able to do things in computer simulation where you can test more iterations and arrive at a more refined, a more efficient design. We have new materials. We’ve gone from principally aluminum, to carbon fiber composites, which means you can build a strong lightweight structure that can better withstand the stresses and temperatures of high-speed flight. Propulsion has completely changed. As you may know, if you’re an airplane geek, Concorde was the only passenger airliner ever to fly around with afterburners. And if you’re an airplane nerd like I am, afterburners are cool. They’re rip roaring loud, there’s a flame coming out the back of the engine. You can’t miss them when they fly over. But if you’re an airline or a passenger, you maybe don’t love them because they’re loud, they bother people on the ground
The flame is kind of scary and most important, they’re incredibly fuel inefficient. And so today we have totally new kinds of engines called turbo fans that are quieter and more efficient. And since we’re here at Big Compute, we’re also going to talk about how computing is part of this. We’ve gone from being able to design on drafting paper to be able to do all kinds of design and CAD and then simulation, allowing us to test our designs faster to iterate faster and arrive at a more efficient aircraft. So at Boom, we’re putting all that together, to build a new, more efficient, most affordable ever supersonic airliner. And I feel very lucky to have fallen in love with this problem at a time when you can actually do something about it. Not only has the technology advanced since Concorde, but the market has also grown.
Since Concorde retired in 2001 there’s actually been a doubling in international air travel. And what that means is there are 65 million passengers per year flying on routes in business class where we can offer them nearly a doubling in speed. That turns into a $200 billion opportunity for Boom, which means despite the capital intensity of this effort, there’s a great return for investors there and we’re actually able to build a business. So at the intersection of proven technology, proven market, supersonic travel is truly inevitable. It’s only a matter of time. So this is overturned. This is our first airliner and the first commercial airliner, first commercial supersonic airliner since Concorde. So this will carry a 55 to 75 passengers in business class. And it will cut travel times in half flying at Mach 2.2 or about 10% faster than Concorde. So what does that mean in practice?
With fare similar to today’s business class more like $5,000 instead of $20,000 and with a smaller seat count than Concorde, there are over 500 routes with enough traffic to support daily supersonic service. So this can change the way tens of millions of people get around the planet. I’ll take a couple of examples. So New York to London today, you fly that as a red eye. It’s about seven hours. You leave in the evening, get there the next morning. But at Mach 2.2 that red eye flight actually changes into a daytime flight. Just three hours and 15 minutes gate to gate. So you could get up in the morning, catch a morning flight from New York. Even with the time difference working against you. You arrive at Heathrow in the early afternoon. And can get to central London in time to make a meeting, a business dinner, catch a return flight, be back in New York in time to tuck your kids into bed.
That’s like a day faster than the shortest round trip you can do across the Atlantic today. Or across the Pacific from here, San Francisco to Tokyo. Today, that’s about an 11, 12 hour flight. And if you’ve got a Monday morning meeting in Tokyo, you actually have to leave midday Saturday in order to get there in time. You arrive end of day, Sunday, Tokyo time. You go to your hotel room, you try to sleep, your alarm goes off, you try not to sleep in your meeting the next morning. And you can catch a flight back to the US and the whole thing takes a minimum of three calendar days and you’d better not make any important decisions later that week cause you’re really badly jet lagged. With supersonic, that flight shrinks to just over six hours. And it’s not just that savings in the air that really matters is that the savings compound because the flight schedules change.
So instead of leaving Saturday, you can leave Sunday, you leave Sunday morning, you get there, you feel like it’s Sunday afternoon in Tokyo, it’s actually Monday morning. You can do a whole day of meetings, catch an overnight flight back and you’re back in San Francisco 24 hours after you left, before any jet lag has set in. So this really changes what you can do in a day. Now there’s a no compromises passenger experience onboard overture as well. So we’re starting from a blank sheet of paper. Re-envisioning not just the airplane, but also what it will be like from the moment you walk onto the aircraft to the moment you step off. So we’re building a nice comfortable seat, plenty of room to spread out, relax, do work. And my personal favorite feature is a cup holder that’s nowhere close to where you put your laptop.
Now this isn’t just exciting for passengers because of the operating cost profile on the mainstream nature of this aircraft. Airlines are excited too. So we have preorders from both Virgin and Japan Airlines for 30 aircraft, which adds up to 6 billion in preorders. Now how do you actually get started on this as a brand new startup company? You don’t go off and build a 75 seat commercial airliner. That is one of the most complicated safety critical machines ever built by humanity. You actually start off with a subscale demonstrator aircraft that will look a little bit like a fighter jet, but fly like an airliner. And that is XB-1. So XB-1 starts with digital design. We’ve been working on this about five and a half years now. And we’re advancing into the build phase. So this is what XB-1 looks like, digitally it’s a primarily carbon fiber composite aircraft with bits of titanium here and there.
An advanced flight control system with a computerized stability augmentation, aerodynamics that have been proven out in simulation with confirmation in wind tunnels. And a propulsion flow path that is significantly more efficient than Concorde’s. So that’s XB-1, digitally. Now the aircraft today is about half assembled in our hangar. And I’ll show you some hot off the press is pictures of that. So in December we did our first major structural bond where we took the cockpit and the nose landing gear and integrated them into the fuselage of the aircraft. The team actually showed up at 2:00 AM that morning to start grit blasting the surfaces for bond prep, carried it into the hanger and then used precision tooling as well as metrology, like laser trackers and metrology arms to get every every piece exactly where it goes within the… Accurate to within the width of a couple of human hairs.
What we’re doing assembly wise is building out each subcomponent of the aircraft and then bonding it into the fuselage and then bringing the wings together. So what you’re looking at here is the main landing gear bay, which is about two weeks away from being integrated into the fuselage. The wings, this is what the wings looked like about two weeks ago. They’re actually getting built upside down. So the upper wing skins on the bottom of those tools. That we’ve put in the supporting structure, which is carbon fiber composite spars and some titanium spars as well. And then you close that out after you’ve gotten in all the critical systems that need to go in while you still have access to it. So this is a picture just from last week. So there’s wings are done now and in about four to six weeks they’ll actually be on the airplane.
Now how do we know this is all going to work? Well, it’s about testing both digitally and in physical hardware. One of the first things we built was a flight simulator. And this takes not just wind tunnel data, but a tremendous amount of output from simulation. We’ve done about 66 million core hours of computing, mainly through rescale since we started the design effort on XB-1. And if you asked yourself what that would look like and wind tunnel testing, it would be financially and timewise just absolutely impractical. We’ve been able to test hundreds of iterations of aircraft designs, which you just could not do with wind tunnels. We’ve gone to the wind tunnel just three times for XB-1 to get calibration data to confirm that we’re calibrated and CFD and then to get a final sign off of the exact design that we’re shipping.
We’ve also done engine testing, so XB-1 uses off the shelf general electric engines and we’ve been able to take them down to the US Air Force Academy just about 45 minutes South of our office and get them running to full power. We’re really proud that we’re able to do this on sustainable biofuels. So supersonic aviation is not just going to save you time, it’s also going to be good for the planet. This is an example of the kind of simulation that you’re able to do to understand what’s going on in the aircraft and a more detailed way than you could even get from a wind tunnel. So this is what XB-1 looks like as it breaks the sound barrier and accelerates to Mach 2.2 Those colors there show you the shockwave says they come off the aircraft. You’re also able to do structural simulations. So this is what the structure deformation looks like and an extreme high G pull up maneuver.
And we’re also able to simulate the propulsion system. One of the most difficult to design components of the aircraft is the intakes. So they have to take the oncoming Mach 2.2 air, slow it, condition it, compress it and feed it to the engines sub sonically. And getting that correct is very tricky. We were able to do it in simulation. So that’s the animation you just saw. And then we confirmed the design and a supersonic wind tunnel and the results matched within 1% which is phenomenal. Most design programs historically have had to go back to supersonic wind tunnels over and over again to get this working. It took the Concorde team about 12 years. We were able to do it in six months thanks to simulation. So let’s take a step back. That is XB-1. XB-1 is going to be fully assembled over the summer and we expect to be a taxi testing around the end of this year and in the flight middle of next year.
So this is coming together quite rapidly, but. We’re doing it really to pave the way towards Overture and Overture, we call Overture because it is an opening, not just of a journey for its passengers, but an opening of a new era of high-speed flight. When you look back in history, it’s easy to under appreciate how much aviation has done for the planet. Isn’t it interesting that we haven’t had a world war since the dawn of the jet age? When you reduce travel times, people go more places, more often. There was a six fold increase for example in travel in the first ten years of the jet age to places like Hawaii that were previously inaccessible. And we think that Overture will kick off a similar growth in air travel is similar increasing of accessibility of earth. We want to live in a world where our children have not just read about places like Cape Town and Tokyo and Mumbai in a textbook. They’ve I’ve actually been there. Imagine what it’s like when everyone has experienced the wonderful people, places and cultures our planet has to offer. And Overture is the first and the series of aircraft we’re going to continue to build them larger, more efficient, quieter and more environmentally friendly. And I think in our lifetimes we will see a world where every flight over about a thousand miles is supersonic. So thank you all very much. And I think we have a few minutes for questions.
Question from the audience:
Hi there. I remember the Concorde. I’m old enough. And I remember that it flew very, very high and the concern at the time was polluting the stratosphere I guess it was. Is that going to be an issue that you’re concerned about?
So we think, the question is what about the environmental impact? And we think a lot about this a lot and the major consideration is that the CO2 that you’re putting out and thanks to efficiency gains since Concorde we’re able to match the fuel burn per seat mile of subsonic business class. So if you look at it in a like to like basis, it’s about the same emissions that you’d have in the alternative today. And since the aircraft is getting designed from the ground up to run on sustainable alternative fuels, the total environmental footprint can actually be better. Now you’re asking specifically about the altitude. So similar to Concorde we’re going to fly at about 60,000 feet. So we will be up higher. The research on whether the emissions at different altitudes is significant is ambiguous today. We’re watching that carefully but we haven’t found a cause for concern. One question people often ask us about what about radiation? Because you’re higher up, you actually get a little bit more. But since the flights shorter, the total exposure is actually better than flying subsonic today.
Question from the audience:
Can you talk about the sonic boom, cause I know that was historically one of the challenges for supersonic flight.
Right? So what about sonic boom? So if you don’t know, so sonic boom is the sound that a supersonic aircraft makes when it flies over at any supersonic speed. So a common myth is that you only hear it right when you break the sound barrier, you actually hear it continuously above Mach one in flight. So it turns out that doing something about sonic boom is not part of minimum viable product. So we’re focusing first on transoceanic routes where you’re at high-speed over water. You’re high subsonic over land, so you’re not making a supersonic boom over populated areas. And that still gets you to 500 routes. And a market for on the order of a thousand to 2000 aircraft. So you don’t need to solve it in the first go. That said in my view this is something that will be relatively easily addressed, when you’re flying at 60,000 feet and you can do some things to the aircraft or sort of shape that sound.
This is something that can sound more like thunder and not something that’s actually going to be a problem. And I think when we have supersonic flight available at mainstream prices and you can for example, get from San Francisco to Tokyo faster than you can get from San Francisco to Washington DC. The current prohibition on supersonic flight over land is going to melt away. But we’re not counting on that and our business model.
Question from the audience:
Can you talk a little bit about challenges you face simulating at that kind of scale? And this is at the Big Compute conference, what you’ve found… Lessons learned about Big Compute.
Lessons learned about Big Compute. The data management has been a significant challenge. We have produced about a quarter petabyte of simulation data. So it’s a significant amount of storage and being able to manage that well, and be able to keep track of, what are all of the different results.
And know exactly what was run in which case, has turned out to be one of the most important things. And so leveraging rescale but then also having some custom developed internal tools. We are able to sort of version control the airplane and have the simulation results from a bazillion different iterations. Then tied back to those versions and built some tools that allow us to go compare and plot and analyze. So that was one of the big things. There have been also a few hiccups and fun things along the way where if you accidentally spin up the wrong job on a bunch of a massive number of cores, you can blow through capital relatively quickly. So getting some good QA around that has been important as well.
Question from the audience:
So I have one question. It seems regulatory challenges might be some of the biggest ones you’re facing. So when do you intend to go to market and how is the access to simulation shortening the time to get regulatory approval?
Regulatory approvals. So the regulatory strategy is we are only flying technologies that have been proven safe, reliable, and efficient on other aircraft. So we’re using similar materials to what it’s found on the 787. The digital fly-by-wire flight control system is standard on new commercial aircraft. The engines that we’re using are adapted from current production commercial engines. And so when you add all those things up, you have a new design aircraft and a new capability for passengers. But we’re not doing anything that hasn’t been done somewhere before. And so what that means is you’re fighting downhill from a regulatory approvals perspective. When you go to the regulators, they’ve got a long checklist that you have to go through to approve that you’re safe and that’s an expensive process.
It’s a time consuming process. But in the grand scheme of things, it’s a relatively predictable process. Stimulation contributes in some places to how you can get your approvals. So for every sort of safety line item that you have to demonstrate to the FAA, you negotiate what’s called a means of compliance. And sometimes that means you have to go off and do physical testing. But in some cases you can actually use analysis and therefore get a compute power to prove your compliance with the safety regulations. Now where we get the biggest leverage from computing power though, was really in the early design. It’s about being able to test many more iterations of the aircraft for less money and less time than otherwise would be required, which allows you to opt… Come up with a design that is just more fuel efficient and therefore more affordable and more environmentally friendly.
So where do you… Where are the limits between simulation and hardware testing? Well it really, it varies depending on what you’re talking about. So I can talk about aerodynamics. So with aerodynamics we did three low-speed wind tunnel tests. And the first one was really about getting calibration data. Because it’s important to have physical calibration data that you can then take back and make sure your CFD results are giving you what you want. And then the first iteration we were off by about 30%, which in the grand scheme of things is actually really good. Then we did a second iteration with a more advanced airplane design to confirm that we were well calibrated. We got excellent agreement and that second one and then that really gave us everything we needed to do the rest of the design iteration completely in simulation and we went back to the wind tunnel a third time just to confirm that real air was doing exactly what the simulated air was doing. And we’ve done wind tunnel tests for the whole aircraft.
Actually only subsonic. It turns out that subsonic simulations are actually more challenging than supersonic simulations. So we have done supersonic test of the inlet, the engine inlet specifically. Because there’s some unsteady flow phenomenon and that leads to, that can lead to uncertainty in simulation. But from a whole aircraft perspective, supersonic simulations actually work great. And you don’t have to go to the wind tunnel really any more for that. When you think about other aspects of the design. So structures for example, you build a finite element model and you can do simulation to look at how strong you are. You can do simulation to look at whether you’re going to have flutter and vibration issues. But it’s important to go off and do physical testing to allow you to take uncertainty margin out of the design. So what happens in the engineering process, of course, is you certainly want your structure to be strong enough and able to handle the vibration, but any margin you add for uncertainty turns out to be weight on the airplane and weight on airplanes is evil.
So what we do is we do a simulation, then we go do a build up testing and hardware. So for example, we’ve done about 1200 what we call coupon tests, where we’re looking at samples of carbon or samples of how a joint works. And that you build up from this coupon test to sub-component test. We built a test spar, we built an entire test horizontal tail and loaded it up until it failed. That’s a fun one. You do this out of carbon, that whole horizontal tail weighs 43 pounds, take 10,000 pounds of load at 300 degrees Fahrenheit. And when we did that test, the thing that actually failed first was the titanium attach bracket, not the carbon. But what you’re doing along way is you’re comparing your simulation to your hardware test results, and that allows you to then dial in the simulation, have more confidence in your results, and take out any uncertainty margin that just adds weight to the airplane. All right, well, I think we’re at time now. Thank you all for being a great audience.