Friends, Americans, Countrymen. Lend me your ears. I’ve come to provide constructive suggestions to SLS, not to bury it.
A couple days ago, a friend asked me what I knew about the DUUS “Dual-Use Upper Stage” for SLS, and if there was any super-sekrit info on NASA Spaceflight’s L2. I finally came up for air yesterday from some proposal writing, and I’m supposed to be working on my USML Category XV NPRM comments, so of course in a moment when I badly needed a distraction, I decided to take a quick look. While unless I missed something, L2 doesn’t seem to have much more info on DUUS yet than the public side. But what I saw got me thinking a bit.
What we seem to know about DUUS is that it’s an upper stage for SLS, that uses an 8.4m diameter LH2 tank, a smaller diameter LOX tank (5, 5.5, or 6.5m diameter), and 4x RL-10 LOX/LH2 rocket engines. At least from how I’ve seen it described/depicted, it kind of looks like a super-DCSS (Delta Cryogenic Second Stage–the stage that flies on the Delta-IV), with separate tanks and a truss structure tying them together (sort of like how Xoie was arranged). Personally, I’ve always been much more of a fan of the RL-10 than the J-2X, and putting some thought into a non-ridiculous upper stage for SLS is long past due.
As an aside, back when the Augustine Committee was going on, I decided to run some architecture analysis numbers. One of the key takeaways I had was that especially for a chemical propulsion architecture, what you really want is a big, high propellant mass fraction, high Isp upper stage, with the ability to top it up after the fact. How that upper stage gets to orbit, and how full it is on the launch pad or when it initially arrives in LEO are really secondary concerns. While an existing Centaur stage looks like it might be able to do some bare-bones deep-space exploration when combined with on-orbit refueling, a bigger stage that didn’t sacrifice the Centaur’s pmf (90%) or Isp (452s-462s depending on the model) would be very interesting. NASA has wanted to do a large upper stage for SLS, but the question has always been if they could do a large stage that was also efficient. Previous Cryogenic Propulsion Stage concepts we’ve seen batted around had pmf’s as low as 75%, and used J2-X rocket engines, so while they were bigger, they didn’t end up anywhere near as efficient of designs as the Centaur. When I saw that DUUS was going to be an RL-10 based stage, it piqued my curiosity a bit.
Getting back to the topic, and to my “random thought”, I saw that someone had asked the question on the public side, if any of the DUUS concepts used a common bulkhead, like the Centaur propellant tanks. Most of the reaction seemed to be that it was impractical to do a common bulkhead design when the two tank diameters were not the same. Now, I was pretty sure I had seen some Centaur variants (the ones for Shuttle and Titan-III/IV) that had larger diameter LH2 tanks than LOX tanks. Sure enough after digging I found some verification in this document on pages 2 and 3. So yes, you can definitely fly a common bulkhead upper stage where one of the two tanks is smaller than the other.
Here’s a picture of one of the Titan-IVB Centaur stages being prepared for flight:
The Titan-IVB Centaur had been derived from the Shuttle Centaur design (Centaur-G and Centaur-G’) which had been cancelled after Challenger. I had correctly remembered that the wider LH2 tank was in order to shorten the stack, since the shuttle payload bay was a lot wider than the 3m Centaur standard diameter. But I didn’t recall why they widened the LH2 tank and not the LOX tank. After all, as commenters on the NSF thread were pointing out, having both tanks the same diameter with a common bulkhead is a lot more structurally efficient. Also the larger aft bulkhead makes packaging engines and other hardware easier–Have you ever looked at the tail end of a flight rocket upper stage? Those things get busy fast!
The answer that was provided in the Titan-Centaur document I linked to earlier was what gave me my crazy idea (emphasis mine):
The Titan IV Centaur, shown in Figure F, consists of a 576 cubic foot aft liquid oxygen tank that shares a common intermediate bulkhead with the 1890 cubic foot forward liquid hydrogen fuel tank. To reduce overall length, the LH2 tank diameter was increased to 14 ft. compared to the 10 ft. diameter Atlas Centaur tank, taking advantage of the 16 ft. diameter Titan IV Payload Fairing (PLF). A conical transition section joins the LH2 tank with the 10 ft. diameter LO2 tank, which attaches to the Titan Forward Skirt Extension (FSE). This transition ring maintains the successful 10 ft diameter common bulkhead, which reduced development cost and risk.
If you look at where the development cost and risk are, the common bulkhead, and the arrangement of engines, RCS systems, pressurization systems, etc. on the aft bulkhead are far more expensive and risky than the rest of the tank walls themselves. This is why you notice that over time the most common Centaur modification is a barrel-stretch. Adding a little length to the barrel takes a little structural analysis, a bit more pressurant, and probably updating the GN&C based on new vibration modes and such. But going to a new diameter, or going to a new engine configuration is a lot more expensive and time-consuming, because there are a lot more items you need to requalify. Going to a two-engine Centaur design for Commercial Crew launches is only really affordable because Centaur used to fly with two engines, and going back to that design (while rolling in the upgrades they made for the Single-Engine Centaur) is easier to delta-qualify than a clean-sheet design.
So with all of that prologue, my crazy idea was what if you designed the DUUS as a common-bulkhead stage, where the LOX tank, common bulkhead, and aft bulkhead/engine bay area were the same as the 5m diameter LOX tank, common bulkhead, and aft bulkhead/engine-bay of the planned ULA ACES upper stage?
Why would you want to do that? Here’s some potential benefits:
- Even with the less structurally efficient taper-section, the design will likely be more efficient than a DCSS type separate bulkhead arrangement. Structural efficiency matters a ton for upper stages.
- A common-bulkhead design will also be shorter, enabling either more volume for payloads in the PLF, or less bending moments on the vehicle.
- Compared to an equal-diameters common bulkhead design, there will be significantly smaller whetted area between the LH2 and LOX tanks–this will reduce the heat leak from the LH2 to the LOX.
- It’s a lot easier with a common bulkhead design to use LH2 boiloff to reduce and/or suppress entirely the LOX boiloff. Hydrogen has a much higher heat of vaporization and heat capacity than LOX, and even boiloff hydrogen is much colder than the freezing point of LOX, meaning that you lose a lot less mass overall if you’re using the hydrogen as a heat spongue. And this is easier to do with a common bulkhead design.
- Using the “bottom half” of an ACES stage means you’ll also be getting the Integrated Vehicle Fluids system, which is a high-efficiency way of providing long-duration power, propulsive propellant settling, and repressurization.
- Not only is IVF more efficient, since it’s using LOX and LH2 from the main tanks for those three functions, you don’t need bigger pressurant bottles or RCS bottles for the two designs. You can either run the IVF system at a higher duty cycle, or maybe add extra IVF modules to deal with the vastly different tank volumes between an ACES stage and a DUUS.
- Manufacturing Commonality! If done right (no guarantees on that), the DUUS could use either an identical or a modularly upgraded ACES LOX tank and aft bulkhead design, which could be produced on the same line as the rest of the ACES stages. This should lower the marginal cost of the system by leveraging larger efficiencies of scale. The LH2 tank would be based on the Michoud SLS core stage tooling/fabrication equipment, meaning that even if you didn’t need a DUUS for a while, the capability wouldn’t just disappear. You don’t have anywhere near as much of a standing army that is DUUS-specific that has to be constantly fed even if you don’t need the DUUS at that given time.
- Design Commonality Using this approach means that you can get two great upper stages for about the cost of 1.25 stages worth of development. If NASA can take a light hand on the requirements side (stop laughing!), the DoD might be willing to split the development cost, because having the ACES stage could be very beneficial to the DoD.
- By helping get the ACES stage qualified, if you need a high efficiency third-stage on the design, for deep-space missions, you’ve got one developed for free, and being used commercially. And it will be much better than the DCSS-based iCPS that they’ll be using for the first SLS flights.
On the ACES side, getting NASA to cover some of the development costs means that ULA is more likely to be able to close the case for fielding ACES, which could provide some real benefits to ULA, DoD, and NASA:
- Having an ACES stage could enable larger payloads on Delta-IVH and could boost the payload capacity of Atlas-Vs without strapons that you might be able to eliminate the Atlas-V strapons entirely, saving some money.
- Or if you keep the Atlas-V solid strapons, the payload on an Atlas V 552 (though I guess it’d be a 554 in this case) would be high enough that you might be able to eliminate the entire Delta-IV line.
- Even without refueling, Atlas V Phase 1, which is what this would be, is a very capable vehicle for LEO, GEO, and deep-space missions. Not everything is big enough to justify SLS.
There are probably plenty of other benefits I’m not thinking of.
What about the challenges? Here are a few I can think of:
- ULA has moved back to the current CRES material used in Centaur for their ACES concept. The Michoud 8.4m diameter tank line uses Aluminum. Either 2219 or 2195 Lithium-Aluminum. This would require some sort of dissimilar metal connection where the taper section ties into the LOX tank. As I understand it, ULA has done some work already on explosive welding aluminum to their stainless material, as part of some of the common bulkhead work they’ve done in the past for ACES. While this is probably solvable, throwing some analysis and early prototype testing at this joint is probably a good idea.
- It doesn’t use the J-2X, which we’ve now spent a lot of time developing. The J-2X just isn’t a good upper-stage engine. Its Isp isn’t that spectacular (448s vs 451-462s for RL-10), and it is just way too big for most upper stage applications–a J-2X engine weighs more than the entire dry weight of a Centaur upper stage. You just don’t need a high stage T/W ratio for most upper stage applications. It might be ok for an explicit second stage that was only used to get something into LEO, but for going beyond you really want an optimized upper-stage system that’s as sleek and efficient as possible. Leaving J-2X on the shelf for other applications is better than compromising a design just to use a politically-favored component. Aren’t we doing that enough with SLS?
- Will NASA actually be able to take a light-enough hand on requirements to enable their DUUS LOX-tank and aft bulkhead/engine-bay to stay common-enough with the ACES hardware to get any net benefit from commonality? I hope so.
- Wouldn’t having both tanks the same diameter save even more weight and height, and provide more aft-bulkhead real-estate, enabling more engines if necessary, and generally just being better? Maybe. But at the cost of NASA having to qualify a bigger common-bulkhead, losing all commonality with commercial launchers, and having to foot the whole bill for the DUUS standing army. Is that worth a slight performance gain?
- Isn’t it impossible to fit 6 engines onto a 5m diameter stage? It probably depends on which RL-10 type you choose. The earlier ACES stages showed 6-engine options, but the more recent ones seem to focus on 4 engines, possibly with bigger nozzles. This is a good question though.
- Isn’t a taper from an 8.4m diameter tank down to a 5m tank a bit extreme? Glad you asked. I decided to make a spreadsheet to investigate this (had my VPN been working, I’d even have a SolidWorks 3D model tied to the spreadsheet, but for now you’ll have to live with the graphic in the third Excel sheet. Here’s what such a stage would look like with 300klb of propellant and a 30 degree taper angle:
And here’s what one would looke like with 450klb of propellant and the same 30 degree taper angle:
Neither of those looks particularly terrifying.
Anyhow, it’s a bit of a crazy idea, but I think it has some merit. Anyone who reads this blog regularly knows that there’s little love lost between me and the SLS, but this might be a way of doing things that if done well could actually be a real net-benefit to the community.

Jonathan Goff

Latest posts by Jonathan Goff (see all)
- NASA’s Selection of the Blue Moon Lander for Artemis V - May 25, 2023
- Fill ‘er Up: New AIAA Aerospace America Article on Propellant Depots - September 2, 2022
- Independent Perspectives on Cislunar Depotization - August 26, 2022
> It doesn’t use the J-2X, which we’ve now spent a lot of time developing.
This is a bit aside from your main point, but NASA said last year that it only intends to test the J-2X through 2015, then put it on the shelf at least through 2025. So the fierce urgency of pork might not be urgent enough to derail DUUS. Maybe.
Jon,
Check out “Taming Liquid Hydrogen” that is a part of the NASA historical series (it is free on NTRS). It tracks the development of the Centaur stage from when General Dynamics and the Air Force kicked it off through Shuttle Centaur and Titan IV. It also covers a lot of the political ideology differences between the Marshall viewpoint (which hasn’t really changed all that much over the years) and the Centaur team from an interesting (if not pro-Centaur) perspective.
Also, Bill Notardonato and I published some work similar to what you are suggesting for topping off upper stages at AIAA Space last year. We basically agree with your premise and published some architectures to back it up (granted I got confused on the delta V to Jupiter and in the rush to publish it crossed the first few sig figs with the C3). But we showed the huge improvement in performance to Mars (and back).
Wes,
I’ve saved a copy of Taming Liquid Hydrogen, but haven’t had time to read it all the way through yet. It looks like a fun read. I’ve heard about the Marshall take on the Centaur and Atlas design. I loved the hammer story, whether it was apocryphal or not.
Could you send me a copy of the paper you did last year? jongoff at gmail dot com
Thanks,
~Jon
Thanks for the informative post. To save further on development cost, NASA could use the Ariane 5 core stage. It’s of common-bulkhead design. I recently learned it also uses the pressure-stabilized, “balloon tank”, method a la the Centaur to save on tank mass. This would be of uniform 5.4 meter diameter for both tanks.
It would be about at the size of your 300,000 lb. version of the DUUS. The ESA also believes its Vulcain II engine can be made air-startable since this was planned for the Liberty rocket. The Vulcain uses a rather short nozzle since it is meant for ground launch, giving it a 432 s Isp. But simply giving it a nozzle extension would give it the ca. 462 s ISP of the RL-10.
You could also give this the RL-10 engines, instead of the Vulcain. The Vulcain weighs about 1,800 kg. Four RL-10’s would weigh 1,200 kg. So this would save 600 kg off the stage dry mass.
You mentioned the advantage of having different diameters for the hydrogen and oxygen tanks to maintain commonality with tooling of existing stages, and that is the reason for not having both tanks the same diameter. But someone noted on the Nasaspaceflight thread on this topic that NASA could just use the same tooling for both that is used for the 8.4 meter SLS core stage tank.
For any of these possibilities it would be very good if NASA could use the composite tanks Boeing is investigating. You mentioned in a prior post ULA estimated the ACES could get a 20 to 1 mass ratio by switching to aluminum-lithium for the tanks. According to Boeing an additional 40% can be saved off the tank mass by using composites, resulting in an even larger mass ratio than 20 to 1.
Scaling up your stage mass, such as to the DUUS size, is also known to be able to improve your mass ratio. Imagine then all these mass ratio improving factors being taking into account. How high could the mass ratio get, perhaps to the 25 to 1, or even 30 to 1 range???
Imagine what you could do with a hydrolox stage with an ISP as high as ca. 462 s with a mass ratio as high as 30 to 1. 😉
Bob Clark
Jonathan
I read your article with great interest. I worked for over 20 years building Centaur tanks and Titan Centaurs as well. I have constantly been amazed by the elegance of simplicity that the steel balloon architecture developed by Charlie Bossart and Kraft Ehricke.
Some issues that you might want to include for consideration include:
1) Secondary structures – Centaur presently attaches engine and hardware directly to the tank because it can. Spun form Aluminum domes have to be heat treated after forming and then need to be machined to get a minimum and consistent thickness. Once heat treating has occurred welding would destroy the heat affected zone tensile properties; adding secondary structures results in reduced propellant mass fraction.
2) Propellant feedlines – It is important to nail down the method for transferring propellant from the upper tank. Pass through feedlines and double-walled vacuum bulkheads can be very difficult because you have to tie in and carry that vacuum jacketing into the propellant feedline adding complexity. It is a consideration on which propellant is in the upper tank and which way the common bulkhead is oriented.
3) Structural Load path – What is the attach point of the US to the LV? Centaur and TC both were at the lower tangent plane of the LOX tank. Both had to carry ascent loads and bending moments through the entire structure. A DCSS style configuration would allow for reduced skin gages in the suspended portion of the tank. Centaur at 10 ft. has skin gages driven by L/D and bending moment issues along the LH2 tank sidewall.
4) Thermal Analysis – One of the primary reasons that the ACES tank reverted to 301CRES is the thermal conductivity between the LOX and LH2 tank in what is called the thermal barrier in the Centaur tank. Switching to Aluminum requires about three time the thickness with a material that has increased thermal conductivity.
Don’t discount the resistance welding process. Although FSW is new and sexy, it can’t weld steel (yet). Resistance welding in the tank sidewalls results in joint having 95% of the parent material strength. BTW, that material has a minimum FTU at 200ksi.
Cayman,
Sorry for not responding right away. I’ll send an email to you as well letting you know I responded.
1. I was assuming you’d still have the LOX tank be the CRES that they’re baselining for use on ACES. You’d have a metal transition from CRES to aluminum at or near the joint between the two tanks, but I think that’s actually feasible.
2. This idea pretty much assumes LH2 is on top, with the bulkhead pointing upwards. It would be hard to do it otherwise. The upper tank is the one that is able to be bigger. If you did LOX on top, it would probably mean that you couldn’t make the Super-ACES and ACES tanks have interchangeable bottom tanks/engine structures (which is where most of the savings comes from). What that means for feedlines, I’m not sure. For SuperACES, it would be easy to run the feedlines outside the LOX tank, since that whole area would be within the payload shroud, and currently has nothing else that needs to be there. For ACES itself though, my guess is that they’d run it internally. Admittedly, ACES might want to have the bulkhead sticking down into the lower tank (I think that’s what I’ve seen in a lot of pictures), so this would be a compromise from their ideal design, because, once again, I’m pretty sure that you’d want the dome curved into the upper tank for this concept.
3. For this DUUS as a Super-ACES concept, you would probably run the loadpath through the Upper tank, since it would be the same OD as the overall vehicle. Once again, think of the way you guys did the Titan or Shuttle Centaur designs with the much wider LH2 tank up top. The key difference between this and DCSS is that while we have two different tank diameters, I’m suggesting using a common bulkhead, and the CRES technology for the lower LOX tank, instead of using all isogrid aluminum with two separate tanks and an intertank structure.
4. Yeah, I’d leave the Y-joint where you join the aluminum tank to the CRES tank as CRES. I’m pretty sure you can take a big sheet of aluminum and a big sheet of CRES and explosively weld them together in a way that you could then machine a ring out, and weld the steel side to the CRES ring and the aluminum side to the rest of the aluminum tank’s bottom frustrum. Would obviously need a fair amount of analysis, but I’m pretty sure it’s doable. I know a rocket company that’s using a similar process to bond aluminum and copper together in a part of their rocket engine that really wants to be mostly aluminum for weight, but needs a portion copper for thermal reasons…
And yeah, I’m a big fan of RSW technology.
~Jon
Jon
1. The CRES to Al transition will be in the LH2 tank as I understand it. This naturally will require extensive analysis and qualification testing and may pose some challenges like thermal expansion coefficient consideration. Should be traded against an all CRES TC style tank. There are advocates in ULA for such a tank but they are not the ones that publish the papers via AIAA etc…
2. Bulkhead direction determines which tank has the higher pressure since you have to maintain the positive pressure differential across the common bulkhead system. RL-10 using LH2 expansion would be more efficient with higher LH2 tank pressures. On Centaur, the Lox tank pressure becomes a limitation. ACES proposes bulkhead down orientation to increase LH2 pressure versus LOX pressure. Bulkhead up requires a feedline outlet in the sidewall of the transiton (conical) skin (affecting the placement of the CRES/Aluminum transition placement) right next to the common bulkhead. Some analysts don’t like this crevice zone for flow characteristics but this configuration has proven itself in much higher flow rate designs like the Atlas IIAR/III Lox outlet.
3. I agree that the load path through the larger /upper tank sidewall would allow for optimization in the lower tank. Ditch the isogrid. If you have to perform tank pressure monitoring for the Lox tank/common bulkhead then use the Centaur approach. Isogrid adds cost for groundhandling only. All these Isogrid designs still require pressurization systems for flight loads. See Falcon 9.
4. Steel Balloon designs don’t have Y-joints. You just take three different plys of cheap sheetmetal and spotweld them together. Y-joints are an Aluminum GTAW/VPPA/FSW paradigm. This joint will be conical in nature and explosive welding needs the appropriate analysis as you surmised. See comments in Item 1 on LH2 temp concerns.
Cayman
Pingback: ULA IVF and their H2-O2 Thruster System program implications for deep space Cubesat missions | Yellow Dragon, a mission to Venus
Pingback: ULA IVF and their H2-O2 Thruster System program ; can Cryogenic Xenon be cooled with LH2? | Yellow Dragon, a mission to Venus
Pingback: Random Thoughts: ACES/EUS Public Private Partnership Idea | Selenian Boondocks
Thank you for at least being open to SLS as a hydrogen depot. It’s LH2 capability is the biggest reason I support it. EELVs assembly is too Rube Goldberg, and an SLS launched depot can be perhaps more robustly built, and have less problems with boil off, still retaining usable LH2 even with a delay.
Now if Elon would just lose that hydrolox phobia of his.