ULA has often used the AIAA SPACE conferences as a venue for discussing technical ideas they were working on. In fact, I’ve written several blog posts over the years summarizing or commenting on previous versions of their papers. This year’s papers represent the first batch of AIAA SPACE conference papers since Tory Bruno took over as CEO in 2014, and to me show how strongly he’s backing his team’s efforts to accelerate implementation of some of these ideas that they’ve been pursuing for years.
You can find all of these papers on their publication page, here: http://www.ulalaunch.com/Education_PublishedPapers.aspx. At my request, ULA was kind enough to label all of the SPACE 2015 papers so you can pick them out of the crowd. I haven’t read all of the new papers, but here are three I wanted to provide summaries for:
- ACES Stage Concept: Higher Performance, New Capabilities, at a Lower Recurring Cost
- Distributed Launch – Enabling Beyond LEO Missions
- Launch Vehicle Recovery and Reuse
There was also a paper on the Emergency Detection System for commercial crew flights, and a presentation by George Sowers talking about potential cis-lunar architectures enabled by their Vulcan/ACES vehicles, but I won’t review those here. I should also note that while I’m a big ULA fan, I’m also a SpaceX fan, so if there were any SpaceX papers from SPACE 2015 that people would like me to review, please let me know (via email if you have it, twitter, or in the comments).
ACES Stage Concept
ULA has been interested in doing a larger upper stage to replace Centaur since shortly after this blog was created 10yrs ago. While progress has been slow and mostly theoretical for a long time, the changes at ULA have made ACES a much higher priority. While Vulcan without ACES would allow them to retire the Atlas V and Delta-IVM families, without ACES they can’t retire the Delta-IVH, which is something they really need to do to get their launch costs competitive with SpaceX.
For those of you haven’t seen any previous articles of mine about ACES, think of it as an enlarged Centaur, with a wider diameter (5.4m–same as the Payload Fairings on Vulcan), more thrust, and the Integrated Vehicle Fluids system replacing the existing RCS, battery power, and pressurization systems (and some of the avionics). According to the paper, they’re still trading 1, 2, and 4 engine versions with at least three potential LOX/LH2 upper stage engines: the Blue Origin BE-3, the AerojetRocketdyne RL-10, and XCOR’s piston-pump fed RL-10 competitor. By using a lot of lessons learned from Centaur and DCSS, the ACES stage should be one of the highest performance LOX/LH2 stages to fly, be able to operate far longer than any other high-performance upper stage in history, have very low LOX/LH2 boiloff, and be surprisingly cost competitive.
I’d suggest reading the paper for more details, but some of the highlights that stuck out to me, as someone who has been following previous iterations of this concept, include:
- Their goal is to have the ACES stage actually be comparable cost to the existing Centaur stages, in spite of having 3x the propellant load, and 4x the thrust.
- Part of this is by automating more of the tank welding steps, simplifying the structure to minimize the number of attachment points to the forward and aft bulkheads, and going with a concave-up common bulkhead with a centralized LH2 sump, among other things. While bigger, the structure will be a lot simpler,and consequently easier to manufacture, than Centaur.
- Going with integrated vehicle fluids (IVF) system instead of the existing Hydrazine RCS, high-pressure Helium pressurization system, and large one-use batteries, both saves a lot of mass and cost, and particularly saves a lot in integration and testing. ULA is working with Roush to develop the IVF modules as an integrated and separately tested module where most of the testing happens before integration with the ACES stage.1 And those cost savings are on top of the huge performance and capability increases from IVF.
- Part of this is by using aft-mounted avionics and encapsulated payloads to avoid needing to assemble the stage in a cleanroom. The avionics have also been modernized, and some of the avionics capabilities are being offloaded to the IVF controllers. The avionics for ACES should be both cheaper and far more capable than what is currently flying on Centaur.
- Probably the part I was most skeptical about was how they were going to get the engine costs down–if they go with RL-10 class engines, ACES would have 2-4x as many engines as a Centaur. There are definitely efficiencies of scale, since at their planned flight rate they’d be using 6x as many RL-10 class engines per year as they currently are. But some of the pricing may also be from the fact that Aerojet Rocketdyne knows they have to compete with both XCOR and Blue Origin, so everyone is trying to provide the best realistic deal. The engine cost is probably the area I’m least convinced on, but hopefully there will be more in the future about how they intend to keep the engine costs down for ACES.
- As mentioned above, they’re going with avionics mounted on the aft bulkhead, as a way of eliminating cleanroom requirements for the stage production.
- The wider tank diameter means that even with 3x the propellant mass, the stage is actually almost the exact same length as Centaur, possibly making ground interface modifications less drastic than it would be if the tanks were wildly different lengths.
- IVF will add all sorts of new capabilities, including durations > 1 week, making refueling (either from depots or the “Distributed Launch” concept I’ll discuss later) far easier since you only have two fluids, and as I’ll discuss later, significantly enhanced maneuverability capabilities–up to and including rendezvous. In a way, IVF turns ACES into a sort of service module for medium-duration (up to weeks) spaceflight.
- For long duration, low-boiloff missions, they’re looking at two options for MLI technologies that can function exposed to aerodynamic forces on the OML. They didn’t mention the vendor by name, but I have written previously about one such company working on that type of MLI technology…
- Apparently they also have a trade for doing a smaller 2-engine ACES variant (assuming the main ACES stage is a 4x RL-10 stage), to address the lower end of the market. There wouldn’t be much savings in anything other than the engines, but that might matter for some lower-end missions.
- They mentioned my old company (Masten Space Systems) and their XEUS horizontal lander concept that could turn an ACES stage into a lander for large lunar payloads. They did mention that once IVF is working, that IVF might be able to help provide Oxygen and Hydrogen to the landing thrusters, allowing for a much higher performance version of XEUS using O2/H2 thrusters instead of storables. With the power capability from IVF, they could possibly run electropumps for high performance landing engines 2.
All told, it’s cool to see this idea finally take shape. While Centaur class upper stages can enable some manned BLEO mission concepts (when refueled on-orbit), the ACES upper stages have enough higher performance that they make such missions much easier, and they’re genuinely better for the application too. I really hope they can find a way to accelerate the development of ACES compared to their previously announced plans, because ACES opens up so many cool new mission possibilities. And if they can really keep ACES cost competitive with their existing Centaur stages, that’ll be even more amazing (though going through the details provided, it sounds like they have a realistic shot at pulling that off).
Which brings me to the second paper. This one was written by Bernard Kutter, who I’ve previously done a propellant depot paper with (at SPACE 2009 while I was still at Masten). His paper discusses an updated concept for in-space refueling using expendable drop-tanks, which they call “distributed launch.” I’ll first summarize the concept and then discuss the pros and cons compared to using a depot.
The primary application of Distributed Launch that was described uses an ACES-derived dual-fluid LOX/LH2 tanker that gets launched to orbit, followed by a separate Vulcan/ACES launch with the payload. The two stages would then rendezvous, transfer propellant from the tanker to the ACES stage with the payload attached, and then the ACES stage would then do the earth-departure burn to send the payload to GEO, lunar vicinity, or beyond. There are plenty of variations on the theme (using multiple tankers, having the tanker launch vehicle be something other than a Vulcan/ACES, etc), that’s the general concept. The picture below illustrates the concept with a Cygnus-like payload on ACES3.
It’s interesting to note that even though a Vulcan/ACES based tanker can only partially refill a Vulcan upper stage (30.5mT of usable propellant vs. the ACES capacity of ~70mT4), it still enables sending almost the full maximum payload launchable to LEO on a Vulcan/ACES vehicle all the way to escape velocity. If I’m doing my math right5, that’s double the escape-velocity payload of a max-performance, expendable Falcon Heavy for probably only a bit more than 2-3x the price… On the other hand, a partially-reusable Falcon Heavy will drop the price by a decent amount, but at the cost of some non-trivial payload performance. But on the gripping hand, a partially-reusable Vulcan vehicle can provide at least some fraction of the reusability savings of a partially-reusable Falcon Heavy, but at a lower performance hit. Anyhow, that observation isn’t from the paper explicitly, but was an interesting aside.
Ok, now for a few more details describing how the system works that I found interesting:
- Assuming both the tanker and payload launch vehicles are Vulcan/ACES vehicles, they need to be able to handle as much as 1 month between the tanker launch and the payload vehicle launch. Which means they need to hit an aggressive boiloff rate (no LOX boiloff, and less than 0.7%/day LH2 boiloff, for a combined 0.1%/day boiloff rate) with the LH2 tank a little oversized to compensate for boiloff.
- They make a pretty believable case that this is achievable based on previous Titan/Centaur data and the following modifications:
- 20 layers of MLI instead of 3 for Titan Centaur, to cut down on radiative heat transfer from Earth and the Sun.
- The tanker propellant tanks are based on an ACES stage tanks, but with no MLI penetrations on the LH2 tank6, and only a ring of low-thermoconductivity struts connecting it to the launch vehicle, cutting way down into heat leaks from the rest of the vehicle into the tanker7.
- The common bulkhead insulation is designed so that the heat leak from the LOX to the LH2 tank balances out most of the heat leak from the vehicle and outside world into the LOX tank. The boiloff GH2 is run through vapor-cooling systems on the struts connecting the LOX tank to the vehicle, intercepting any remaining heat, so the LOX tank doesn’t heat up so long as there is LH2 on board.
- The stage stays settled using a transverse (end over end) rotation scheme. By leaving the delivery upper stage attached, with the nice heavy engines at the bottom, the CG for the stack once the upper stage propellants are mostly empty is somewhere near the center of the tanker LOX tank. This means the tumbling will keep LOX on most of the walls of the LOX tank, but the LH2 will be up at the “top” of the tanker tank, with a GH2 barrier between it and the LOX, which should cut down on heat transfer from the LOX to the LH2 somewhat8. The tanker would de-spin and transition to a 1 milligee axial settling acceleration once the payload Vulcan was launched and nearing rendezvous.
- They suggest placing the drop tank into an orbit with a repeating ground-track, with a low altitude, so that when the payload Vulcan launches, it has fast direct rendezvous windows once every day or two (depending on the orbit you pick). This minimizes the time the payload has to wait in LEO before departure.
- Distributed Launch leans heavily on the new maneuvering capabilities provided by IVF to enable the two stages to rendezvous and then formation fly in settling mode during propellant transfer operations. I’m actually pretty confident that the rendezvous and closing operations are doable with IVF, but I’m more skeptical about the ability to formation fly while a single set of fluid hoses connect the two vehicles. The fluid hoses will be under at least some pressure, which means you’ll be transmitting forces and torques between the two vehicles, and the coupled dynamics of such a formation flying situation with those disturbing forces scares me a bit. I’m not saying the problem isn’t solvable–I haven’t run numbers on the scenario in question, so it might be totally doable GN&C wise. If it isn’t, I know exactly how I would solve the problem, but that’s a topic for a venue other than a review of their paper. Suffice it to say I think that this is a solvable problem.
So, how does this compare versus using a dual-fluid depot like the ones we’ve blogged about previously here on Selenian Boondocks?
Benefits of Distributed Launch over Depots:
- Easier to deal with low flight rates due to lower fixed-infrastructure costs, and lower boiloff between missions.
- Easier to place the tanker into the optimal plane to enable low-penalty BLEO launches to destinations with short launch windows and tricky departure declinations (NEOs, Comets, and some planetary missions).
- The short duration minimizes requirements on the depot itself–most of the spacecraft controls can be handled via IVF if the tanker delivery vehicle has IVF, minimal need for MMOD protection, no need for the tanks to be both filled and detanked, less need for liquid mass gauging since they’re filled on the ground and you can measure boiloff.
- Distributed Launch can also be repeated at non-LEO locations. For instance, doing this twice could enable placing a tanker in EML-1 or 2, and then sending a payload there to rendezvous with it and refuel it.
Drawbacks compared to traditional depots:
- If you have high demand for distributed launch, launching a new tanker each time starts becoming tedious and expensive.
- The tanker doesn’t really save that much over a depot, and what savings it does provide rapidly go away if you do end up having enough flight rate to justify a depot.
Depots make it easier to handle propellant deliveries in a launcher-agnostic fashion, including using smaller vehicles to perform the deliveries.Once you go to more than one tanker transferring propellant in distributed launch–remember that a Vulcan/ACES V564A can only delivery ~30.5mT to orbit, but the ACES stage can use 70mT of propellant–you start adding more docking events to the payload vehicle. It might be preferable to to have the depot take the heightened risk of multiple tanker deliveries than to have the payload delivery upper stage take that risk. A depot can probably afford more robust rendezvous and interface hardware than a single-use drop-tank setup.
All that said, distributed launch is a fascinating idea, and it helps put almost all the technology on the shelf for future depot missions while allowing you to start when there isn’t enough demand for a full-blown depot. Also, it’s interesting to note that getting ACES to the point where it can rendezvous with another space object means you could use it almost as a space tug for delivering bigger payloads to space facilities9, enabling delivery of larger cargo, station modules, and raw materials to orbital manufacturing sites in addition to just propellant tanking. This concept of using upper stages to deliver payloads directly to another vehicle or facility without the need for tugs or prox-ops vehicles is a concept near-and dear to my heart at Altius, and a direction we want to encourage over time.
I’m pretty excited to see where Bernard and his team take this between now and next year. I really think that this approach of using orbital refueling to enhance launchers’ BLEO capabilities is a intriguing one. With luck, maybe I can finagle my way into being involved in next year’s paper.
Launch Vehicle Recovery and Reuse
This last paper is an update on a concept ULA first presented in 2008, with more discussion of alternative approaches and why they think this approach is better. In review for those who haven’t read this before, ULA’s “SMART” (Sensible Modular Autonomous Return Technology) reuse concept involves recovering just the first stage engines instead of the whole first stage like SpaceX is trying to do with Falcon 9. The first stage engines would be connected to the stage via separable structural and fluid connections. Once first stage burnout was complete, the engine pod would separate from the stage, inflate a Hypersonic Inflatable Aerodynamic Decelerator, and then once it was going subsonic, it would release a guided ram-air parachute. A recovery helicopter would then recover the engine-pod in mid-air, like the old Corona spysat film capsules were recovered during the early space age. This would allow the engine to experience a recovery environment that is very benign relative to flight, without expending a lot of propellant or other mass for the recovery.
The concept here is that the engines are half the cost of the first stage, but less than a quarter of the mass. And by doing mid-air recovery, you can keep the environments benign enough that reuse should be straightforward, and requires the minimum payload hit. You use the HIAD to decelerate instead of supersonic retropropulsion, and you have the recovery helicopter down range so you don’t need any boostback.
It’s an interesting idea, but it’ll also be interesting to see what SpaceX manages with Falcon 9. Coming from a background of VTVL powered landers at Masten, I’m definitely biased towards the SpaceX approach. It does require more of a performance hit, and it’s less clear if propulsive landing is going to do bad things to the engines, but RTLS propulsive landing removes constraints with downrange recovery–which we’ve seen can be a big deal from previous SpaceX recovery attempts on their Autonomous Drone Ships. And while the engines are most of the cost, the rest of the hardware is non-trivial. To me the real questions are going to be: how high of a flight rate will there really be demand for–the higher the flight rate, the more gas-and-go reuse makes sense, and how much refurbishment time will SpaceX’s approach take. If the refurbishment time is low, I’m not sure how SMART will compete with that long-term.
I’m not trying to rip on SMART–most of the developers of the technology are people I’d consider friends. I’m just expressing my biases coming from a VTVL powered landing background.
Ultimately though, it’s good having different groups trying out different approaches. We’re still in the infancy of reusable orbital vehicle development, and the more ideas tried, the more likely we’ll find the right answer–and it may be possible that there are more than one right answer. Lastly, if it turns out SpaceX makes rapid progress with Falcon 9 reuse (which I wouldn’t bet against), ULA has demonstrated its ability to adapt and come up with clever outside-the-box solutions.
While ULA has presented some really interesting ideas over the years, this year’s presentations are all the more exciting because there’s a real chance we’ll get to see ULA actually try these technologies. Their situation is such where they have to innovate, but fortunately they’ve got an extremely talented and created team. I hope these reviews were interesting to readers, and I hope they encourage everyone to read the whole articles. They’re well worth the time.
- My startup, Altius Space Machines, is involved in the design and development of some IVF subsystems, so I’m very biased about how awesome the technology is.
- Though that last one is my idea not from the paper, I haven’t run the numbers to see if makes sense.
- And yes, that totally looks like a Sticky Boom connecting the two stages, at least to me…
- And yes, for all the metric unit pedants, mT isn’t the correct annotation for metric tonnes, but it should be pretty clear we’re not talking about magnetic flux densities from the context.
- Based on Table 1’s numbers from the paper, not on any numbers I’ve run myself.
- [Clarification 9/6/2015]: I originally had stated “no penetrations at all”, but what I really meant was no MLI penetrations on the external surfaces of the tank, so I changed the phrasing
- The nearest heat-generating system may be on the aft LOX bulkhead of the launching upper stage… That’s a lot of thin gauge stainless steel to conduct through.
- Though I do wonder what happens when you despin the stage and transition to lateral settling prior to propellant transfer–that would seem to lead to a decent amount of boiloff when the LH2 hits the now warmer cylinder section down near where the common bulkhead attaches, where the GH2 was previously
- At least facilities not afraid of having a big LOX/LH2 stage sidle up next to them