Launching into the future
By Cat Hofacker|October 2019
There are no small jobs in the satellite launching business, but United Launch Alliance CEO Tory Bruno might have one of the biggest. After building a reputation for ULA as a reliable provider with 135 successful launches and counting, Bruno is betting that reputation on development of the successor to the Atlas V rockets called the Vulcan Centaur. Set to debut in 2021, Vulcan’s first job will be to end the irony of the U.S. government launching defense and spy satellites on rockets powered in part by Russian-made RD-180 engines. In the long term, Bruno sees the Vulcan as a reusable, flexible launch vehicle for a bustling economy in cislunar space. I spoke with Bruno at the AIAA Propulsion and Energy Forum in Indianapolis.
IN HIS WORDS
Building the Vulcan
As we have new major systems for Vulcan, we are deliberately feathering them in on Atlas at least a year or more before Vulcan will fly. If I start at the top and work my way down, we developed a brand-new 5.4-meter composite payload fairing for Vulcan, which has a giant, cavernous payload volume. This payload fairing is built in half the time, it costs half the money and it performs better; it’s much lighter. That’s being fabricated right now by our partner RUAG [Space] inside our Decatur [Alabama] factory, and we will start flying that on Atlas next year. As I come down, the next thing would be the upper stage, Centaur. It is our existing Centaur III but three times the amount of energy, so much bigger propellant tanks. The one thing that will be new is a dual-engine version of that, which we will start flying on our new Starliners [Commercial Crew capsules] starting just in a few months, so that’ll have probably four or five flights under its belt before Vulcan flies for the first time. We’re reusing all the ordnance. We recently developed a new avionics suite and, since we were doing that, we made it common for Delta, Atlas and Vulcan. So that’s been flying for a year already, and because it’s the same hardware, 80% of software gets recycled. We’re using the same technology for the booster tanks, so rigid structure, 7,000 aluminum, different grid pattern, but basically the same technology. The only thing we won’t be able to fly first will be the [Blue Origin] BE-4 engine, and so we’ll test the heck out of it on the ground.
We are confident to go directly to real missions with real customers on our first flights. To be certified for Air Force national security missions, you must fly two times, first for somebody else, and so these are those flights. We just think these are great missions. Peregrine [made by] Astrobotic will be the first big step back to the moon. It’s a lunar lander; it’s going to carry a couple of dozen payloads to the moon. It just seemed fitting. We have great visions for what will happen on the moon, with cislunar space, so that was part of it. The second flight is with the [Sierra Nevada Corp.] Dream Chaser lifting body vehicle. And I’ll tell you, it’s the coolest bird. I have been a fan of Dream Chaser from the moment I saw it. We have supported them, we have helped them get to where they are today, and it was a great privilege to receive a block of six missions from them. These were just a really natural fit for us.
There’s three kind of basic ways to do reusability in a space launch vehicle. You’ve got glide back like the shuttle, you’ve got propulsive fly back like that other company’s doing, and then you’ve got component reuse. You know, there are things where you want to go first, be first to market, and then there are other circumstances where it’s OK to be second to this. In terms of reuse, I gave them to you in the order of how difficult they are technically and almost the order of how difficult they are to make a business case actually work. Component reuse is a much lower hurdle. You don’t have to sacrifice performance of the rocket because you can burn off all your propellant. We already know how to separate and recover bodies — we’ve done this in the past — so bringing back the [Vulcan] engine is a natural fit. It is two-thirds the cost and value of a first-stage booster. Take the engine away, what do you get? Empty aluminum tanks. Propulsive reuse is pretty tough, and we’re going to wait a little bit until we see someone show that as economically sustainable because it has not been shown to be that yet. Then as we learn more about our rocket and how it behaves in the environment it experiences, then we’ll move where that technology allows us to go.
One-stop shop for launches
Why one rocket instead of a whole family of rockets? It is the most efficient and cost-effective way to do it. It also greatly simplifies the infrastructure and allows us to focus all of our energy, our innovation, our investments on a single platform to get the most out of it. In advance of Vulcan, we really transformed our whole company. You know, the Atlas rocket launch service now is well over 35, 36% cost reduction from where it was when I joined ULA five years ago. That was largely about reorganizing the company and redesigning how we build the rockets, retiring the venerable Delta that we only maintained for assured access. That’s no longer our sole burden, so I don’t have to maintain a costly and sort of redundant product line. The second thing is the Vulcan itself. Vulcan is a much more powerful but also much more affordable launch vehicle, and it’s really a new class in that it is a single-core heavy. For the heavy missions, it collapses the cost by almost three-quarters in a more flexible vehicle. You will see people that do multiple payloads occasionally on a launch vehicle, dual launch, ride-sharing. We do that too, but one of the challenges in doing that is getting payloads going to the same place at the same time that are a good fit. Vulcan is especially flexible for that, so it will also make it much easier for us to fly two payloads of almost any kind of weight class, three payloads and so on, which further reduces the cost to each customer.
We are six times better at flying on time than anyone else in the industry. It’s natural for people to not care about the launch until the last few days before it goes, and that’s when everybody tunes in. What you don’t realize is people generally buy their ride to space about two or three years ahead of time. We set a launch date at that time and we are within, on average, a couple of weeks of that launch date two years later. So losing a few days to make sure it’s going to be successful is an obvious thing to do; that’s a no-brainer in terms of the decision you make. There are people who have been waiting three years past their flight date that was promised to go to flight. You’ll never see that with us. The other thing is with Vulcan, the cycle times are so much shorter. With Vulcan, we can fly 11 days on center off the same pad. Just 11 days apart. That’s sort of unprecedented. If you were talking about a scenario that you hear a lot discussed nowadays around operationally responsive space — for example, what if there’s an emergency mission, like we need to have a replacement satellite on orbit or we need to get out to ISS or eventually to Gateway — if you are willing to pre-position a relatively inexpensive payload adapter, we can go from this spacecraft being available to its destination orbit in eight days. If you spend a little bit more money and put a payload fairing there as well, five days. I’m feeling pretty good about being able to go from zero to five days in space against anybody in the marketplace.
You have to provide the means by which the capsule can continuously monitor and assess the health of the rocket. In the event that the rocket were to malfunction and the capsule would like to initiate its emergency abort system and fly away, it has to have real-time data coming off the propulsion systems, off of the other critical systems and the rockets. It has the few milliseconds required to accomplish that, so we provide monitoring, we provide cabling, we provide accommodation for the capsule’s computer system that does all of that. In addition to that, certain sorts of requirements are of a higher standard in terms of reliability or the environments that might be experienced. So you have to go back and analyze your systems, if you have not already, for those unique requirements to make sure they’re satisfied. In some cases, you have to redo a qualification test to demonstrate that the analysis is correct. We did not have to do this on Atlas and do not expect it on Vulcan, but if necessary you might have to make a modification if you found out it wasn’t. We’ll go through that whole thing on Vulcan should someone ask us to fly people on Vulcan.
Streamlining commercial launches
What we set out to do in this current update to regulations was to streamline the bureaucracy: single forms, simplified forms, not so many different people within the agencies approving things so things can be done more quickly, single licenses for multiple launch sites. That’s all streamlining. That’s what we wanted to do and largely was accomplished. There’s at least one more thing the FAA is still working on, which is coordinating with the Air Force ranges because as it turns out, there’s a lot of overlap. When you go to the Air Force, do your work and get their permission to fly, you’ve done a lot of what you have to do for the FAA. There’s an opportunity there for the two organizations to formally coordinate and recognize each other while not letting anything fall through the cracks. That’s why you can’t just do that instantly. The things that are in contention are things like a desire, for example, to lower the required reliability for a flight termination system, or perhaps to lower the test levels required to qualify a flight termination system. This is safety. The FAA’s primary mission is to protect public safety. The flight termination system is the thing that on a bad day when your rocket isn’t working and the thrust vector control is jammed and the front end is on fire and it is flying for the Kennedy Space Center Visitor Complex that their safety officer hits a button and that rocket stops flying. That must work. It has to be reliable. It has to be tested to rigorous conditions, and it has to be able to survive the abnormal environments that occur when your rocket is perhaps breaking apart or on fire. These are fundamental safety requirements, and they cannot be violated. The FAA rules allow for people to demonstrate that they comply with safety by alternate means; that’s already there, but to simply say, “We want a lower reliability requirement,” I don’t think we’re ever going to get on the same page with that.
Vulcan’s future customers
When they first started talking about proliferated LEO, the architectures were super exciting because they had 3,000 satellites and 4,000 satellites, and they were going to fly in like two years. Then it took them a little bit longer to get funding and to mature, and while they did that, the constellations got better and they all got way smaller because they designed them more effectively. We’re still talking about hundreds. They’re small spacecraft, and so a hundred satellites is not 100 launches; they will be launched in large groups. Vulcan is really well-suited for taking up the initial constellations. The downside of these is because they’re in LEO, which is an hour-and-a-half orbit time, you have no utility with one satellite. You have to get a critical mass of them on orbit before you can start doing a job, earning any money with them. Medium- and heavy-lift vehicles, we’ll do the initial population because of that reason. Once they’re up, there will be periodic servicing of them. Satellites will die. These constellations, by the way, are inherently robust against an individual bird going out because there’s overlap, and because quite frankly in LEO they drift a lot in their orbit with respect to their position in the orbit anyway, so it’s easy to re-space them. But if you get a cluster together, now you’ve got a dead spot, so you’re going to have to repopulate. There’s a mission there, almost a secondary mission, if you will, for maintenance that I think leaves room for small launchers, and there are a number of small launch companies trying to happen right now. There’ll be enough market for two, maybe three. Some of that will still be provided by the medium and heavy lifters, because we have so much capability that when we’re taking a primary payload up anyway and it happens to go past where we would drop off a replacement, we would just carry that as a secondary.
Artificial intelligence on orbit
These proliferated LEO constellations are a great application for AI because enormous quantities of data will be handled through space. AI will enable efficiency of processing some of that data on the orbit before having to transmit it, so you will absolutely see that there. In terms of AI in the design process, well, there is an application for big data. When we design rockets, we don’t like to leave out any learning that you could have leveraged, and so you are already seeing a certain amount of, I’ll call it, very simple AI involved in the data mining in order to help engineers do the best job when they’re designing their parts. More importantly at the moment, though, than that is actually more advanced tools in terms of our ability and coupled analyses that are fully dynamic of all the phenomenon that might happen at a single moment when the rocket is being evaluated. For example, being able to put the propellant fluid column in a single finite element model where it interacts with your structure, creating deformations and strains and stresses to help the design process. Today, that is still more important than big data or other forms of AI. You will see it creep in over time, more on the next generation of rockets, I think, than this one.
AI on the launch pad
What ends up delaying a bird on the pad is weather, it’s conditions, sometimes there is an issue that pops up in real time, even though it was tested before, that has to be solved. I would say not the only, but the most interesting application for AI in those circumstances is to be able to answer the question during the countdown before you run out of time. One of the reasons we are as on time as we are is because there’s not just the guys in the control center that you see on TV. When we launch a rocket, there’s another 150 engineers back in Centennial, Colorado, supporting those people in the control center. Every part that has a responsible engineer, that engineer and their teams are sitting on the console in Denver. They have at their fingertips, on their consoles, all of the data ever collected for their part so that they can decide if it’s in family, if it’s out of family and how to solve the problem. That’s an application that can benefit from AI and one that we’re actually working on, but don’t tell my competitors.