The SLS Empire Strikes Back

So there was a leak of an internal NASA document a few weeks ago which showed that space transportation architectures that employed the use of orbital storage of propellants, rather than lifting them all at once on a heavy-lift vehicle (such as the Senate Launch System), would save the taxpayer tens of billions of dollars and accelerate the schedule for manned trips beyond earth orbit by half a decade or more. It was information that California Congressman (and former Chairman of the Space Subcommittee) Dana Rohrabacher had been demanding from the agency for weeks, to no avail until it was leaked. The political effect would be that the SLS is unneeded, which would be a devastating blow to those senators and representatives who had continued to support it as an earmark for jobs in their districts.

Well, supporters of the “Big Monster Rocket” have struck back. Previous NASA administrator Dr. Michael Griffin, and former NASA associate administrator under him (and current head of the Space Policy Institute at GWU) Dr. Scott Pace, have jointly written an editorial defending it over at Space News. Despite the credentials and experience of the authors, however, the arguments presented are flawed. I’ve written a letter to the editor, but it was restricted to 500 words, and there are far too many flawed arguments to address them all in that space, so I thought I’d dissect it here, with a thorough fisking. [In the interest of disclosure, Dr. Pace is a former colleague of mine at Rockwell International Corporation, and a friend of almost three decades.]

Considerable recent attention has been devoted to the possible use of orbiting fuel depots for human exploration beyond Earth orbit. In this concept, large propellant tanks are placed in a suitable low Earth orbit (LEO), to be filled by multiple launches of medium-payload-class vehicles, i.e., a few tens rather than a hundred or more metric tons of payload capacity. These depots are then used to refuel upper stages, which arrive empty in LEO after launch from Earth, after which they are launched outward to the Moon or beyond.

Actually, they wouldn’t be just in LEO — most serious proposals have them at the earth-moon L-1 point as well, where they could better utilize lunar resources as a propellant source, and have monthly opportunities for trips to the rest of the solar system.

Advocates for this approach believe that the money saved by not building a heavy-lift launch vehicle such as the Space Launch System (SLS) will more than compensate for the cost and operational inefficiencies entailed in bringing the required total mass of propellant to orbit in smaller individual packages.

That is actually only one benefit. Others are that in-space vehicles (such as lunar landers) launched unfueled can have less structural mass, because they don’t have to sustain the high loads imposed by full propellant tanks in the high acceleration of launch. This improves overall propellant efficiency of the in-space transportation architecture. And depots will be required eventually, anyway, because no matter how large a launch vehicle is, one can always come up with a mission that needs a bigger one. So we might as well learn now how to do deep-space missions with multiple launches. But their biggest benefit is that they will drive down the cost of access by providing a healthy propellant-delivery market for a competitive robust domestic launch industry. They also provide opportunities for cooperation with other nations, without putting any company, or nation, on the critical path.

Whether fuel depots make sense in the near term depends upon what question we are trying to answer. If the question is, “What kind of space architecture will generate a high traffic model for private space firms without having to pay for missions that actually go beyond LEO?” then fuel depots are an attractive concept. But if the question is instead, “How can we efficiently create the strategic space transportation capabilities to enable humans to explore beyond LEO?” then they are not.

Actually, that’s a false choice, and in fact it’s a straw man, because I don’t know anyone who is asking the first question. Though in fact depots do nicely answer the second one. Here’s the question we should answer: “How can we maximize the amount of human space activity for a given federal budget?” The Space Launch System not only doesn’t answer this question, it is in profound opposition to it.

The fuel depot concept may be — we think will be — valuable when propellant can be harvested from in-space resources, such as water trapped in lunar craters or oxygen extracted from the regolith. Unfortunately, we are not yet in a position to exploit such resources, and so for now fuel depots are an answer to a question that is at best premature. The SLS and the Multi-Purpose Crew Vehicle (MPCV) are needed today. Fuel depots will be needed tomorrow, when a robust space operations infrastructure has been established and operations beyond LEO are common.

That is simply not true. I have already described the benefits of depots above, even without the use of extraterrestrial resources. And if we really need SLS today, then we’re in trouble, because NASA tells us we aren’t going to get it until the end of the decade. But fortunately, we don’t need it today (or any day). In fact, as I noted over at Popular Mechanics today, Dr. Griffin recently testified before Congress that one can go to the moon without SLS, at least if one is Chinese:

Q: I know the Chinese Long March 5 rocket is in development. I wondered if you could compare that to anything we have in the American inventory. When it’s built will it really be larger than anything we have? And why do you think that the Chinese are building such a large rocket?

Griffin: Well, the Long March 5 is comparable in scale to today’s Delta IV Heavy or to the Ares I crew vehicle—which we were going to build and which was cancelled. So it’s on the order of, and of course until it flies regularly we won’t actually know, but it’s on the order of 25 tons of payload to LEO. So it’s not in the class of, say, the Saturn V or the new SLS [Space Launch System].

But it’s a very significant capability and in fact by launching and rendezvousing four of those in LEO it would be possible for the Chinese to construct a manned lunar mission with no more than that rocket and no more than Apollo technology. And I have in the past written up on how that mission would work from an engineering perspective. So with the Long March 5 the Chinese inherently possess the capability to return to the moon should they wish to do so.

Q: And you are saying that we do not have anything comparable to that other than what had been talked about?

Griffin: We do not. Well, we have nice view graphs (laughter in the background).

[My comment] Actually, contrary to Griffin’s implication, the Delta IV Heavy has flown, so it’s more than “view graphs.” And the Long March 5 isn’t scheduled to fly until 2014. But even in that timeline, China could be thinking about a moon visit relatively soon. In the U.S., by comparison, the Space Launch System NASA is now mandated to build couldn’t return Americans to the moon until at least the late 2020s (and would add tens of billions to the cost), according to a recently leaked NASA internal document.

So, Mike, if those inscrutable Chinese can get to the moon without a Big Monster Rocket, why can’t we? Are they that far in advance of us?

But the next section is where the op-ed really goes off the rails:

The challenge for fuel depots is simply that the marginal specific cost of payload to orbit is generally lower for larger launch vehicles. There may be exceptions, but the trend is clear. Moreover, this same trend is observed for other forms of transportation — road vehicles, trains, ships and airplanes. Without exception, larger vehicles are used whenever possible for long-haul transportation. In evaluating depot concepts, one must then ask: Why will space transportation be an exception? Is it really an exception? Or are we missing one or more crucial points in the analysis?

When depots or nodes are used in transportation architectures, they must be supplied by the least expensive means available rather than the contrary. There is an old joke about clothiers in the New York garment district who professed, it is said, to sell each garment at a loss, but would “make it up on volume.” Fuel depots seem to exemplify that joke. It is very difficult to see how putting propellant in orbit in small quantities at higher marginal cost can be cheaper in the aggregate than putting it up in larger quantities at lower marginal cost, even without factoring in the cost of the depot itself and its own operational requirements.

The economic attractiveness of propellant depots depends strongly upon the price claims of commercial launch companies for fuel delivered to orbit. At this point, such claims should be considered highly suspect. Even a signed contract offers little assurance, because if the supplier requires additional funds to continue service and there is no government capability available as a backstop, the money will be tendered. Thus, price claims made by companies that are not yet conducting routine operations at that price should be regarded with skepticism.

Where to start? Well, first of all, it makes no sense to talk about marginal costs in a decision like this. As I wrote in my letter to the editor of Space News (which I hope will be published next week):

Marginal cost (the cost of the next flight, given that a system is already operating) makes sense in deciding which existing vehicle to fly on, but it’s useless by itself in deciding whether or not to develop a new vehicle. For that decision, one must take into account the total life cycle cost, flight rate, and average cost per flight, including amortization of development costs and annual fixed costs.

Per Congressional mandate, SLS is going to cost a minimum of $18B just to be developed to its first, smaller version. Total cost estimates into the decade of the twenty twenties to get to full capability range from twice to three times that amount. NASA says it will fly once every year or two. If they get an annual flight out of it for thirteen years, that would imply a cost of at least three billion per flight. That would imply a cost per tonne (for 130 tonnes) of $23M, or about $10K/lb. Even if they somehow fly four times a year, and I generously grant a marginal cost of zero (unlikely with a large expendable vehicle), it’s still $2500/lb when all costs are included.

In contrast, the SpaceX Falcon Heavy is priced at $120M for 53 tons, or a little over a thousand dollars per pound — that is, ten percent of the cost of the SLS, with no development costs funded from the taxpayer. If you don’t believe that this vehicle will ever fly (we’ll know in a year or two, because that’s when first flight is scheduled), then use the now-existing twice-flown Falcon 9, whose quoted price is $60M for 23,000 pounds or $2600/lb – about the same as the most optimistic case for SLS, and available today, not a decade from now.

Note that it’s even worse, because I didn’t discount the future dollars — they’re all current-year, whereas in reality the up-front cost of the development loom even larger (not counting opportunity costs if we were spending it on actual space-exploration technology development and hardware).

Now I suppose it’s possible to think that SpaceX’s (and United Launch Alliance’s) prices will magically increase by the large percentage required to make the SLS’s numbers look attractive, but they offer no reason to do so, other than Fear, Uncertainty and Doubt (FUD), and it seems quite unlikely to me, given that they are in competition with each other and driven to keep prices as low as possible.

But the other problem is that they are making a theoretical argument (big rockets have lower marginal costs of payload than smaller rockets) with little empirical data to substantiate it, and as shown above, there is actually real-world data that says it’s wrong. But it goes beyond that — it’s not even a theoretically valid argument. Here’s why. They are saying that there are economies of scale with vehicle size, and generally there are, though for launch vehicles, there are limits to how well they scale, in terms of structural efficiency (high hydrostatic pressure in giant propellant tanks increases needed structural weight), ground support equipment, processing facilities, etc. For example, one of the problems that the SLS has is that while it’s the same class of rocket as the Saturn V, the Saturn was all liquid, and fueled at the pad, whereas the SLS has Shuttle-like solid boosters (because otherwise ATK wouldn’t get to keep their pork flowing) that are mated in the Vehicle Assembly Building, and then the whole vehicle is rolled to the pad on the same crawler and crawlway (road for the crawler) used by Saturn and Shuttle. But it wasn’t designed to handle that load, and studies have indicated that both will probably have to be upgraded (just one of many reasons that SLS will cost so many billions to develop).

But the other problem with their argument that bigger is better is that it comes with a caveat — all other things being equal. And as we’ve seen in the real world, they’re not. For one thing, we don’t have to pay development costs for existing vehicles. But as Jeff Greason noted in the Augustine hearings (and fellow panel member Sally Ride agreed), even if Santa had delivered Constellation fully developed for Christmas, NASA wouldn’t have the budget to operate it, because it was so manpower intensive (which was the point — it is about jobs, not spaceflight), and SLS won’t be any better in that regard.

Another way that they’re not equal is flight rate. Here’s the real problem. At this stage of the industry, there simply isn’t enough demand to justify a vehicle of that payload class, particularly if it’s expendable. At this stage of the industry, the only relevant scale to increase to achieve economy is not vehicle size, but flight rate. As I noted a few years ago at The New Atlantis, we saw this in the Space Transportation Architecture Study in the eighties:

These studies considered a wide variety of vehicle types—reusable, expendable, single- and multiple-stage, various propellant combinations, air-breathing, rocket, horizontal takeoff and landing, vertical takeoff and landing, and more—the entire range of conceivable ways of getting crew and cargo into Earth orbit and (when necessary) back using semi-conventional aerospace vehicles. These studies also considered a range of potential “mission models,” with different types, mass, and volumes of payloads, over the next few decades. The models ranged from the minimal (with no commercial activity and little or no growth in NASA or military space budgets) to the expansive (with major new civil space initiatives, including crewed lunar and Mars missions, and large-scale commercial activity).

As we looked at all the combinations of architectures and models, we discovered something interesting. While some vehicle design concepts were clearly better than others, they were all extremely expensive per-flight for the low-activity scenarios, and they were all much less expensive for the high-activity scenarios. Using the space shuttle as a reference, we developed a notional architecture that had sufficient facilities and vehicles for a hundred shuttle flights per year. (That sounds ridiculous today, since there have never been more than nine shuttle flights in a single year, but in fact the shuttle was originally intended to fly once a week.) Surprisingly, the per-flight costs that we estimated were much lower than the actual shuttle costs at the time. The same was true of other launch concepts we studied. The cost per-flight or cost per-pound varied dramatically—in some cases by a factor of ten—depending on the level of activity for a given vehicle in each mission model.

This means that even the theoretically best vehicle concept, if flown rarely, will be unaffordable to fly. A mediocre design, flown often, will beat it in cost per flight. How frequently we used the hypothetical launch system was much more important than what kind of propellant it used, or how many stages it had, or whether it took off or landed horizontally or vertically, or any other design choice. This, to me, was the key insight from all of those studies, and it’s one that remains true to this day. For example, the costs associated with the space shuttle largely go to pay the army of personnel and associated infrastructure needed to keep the shuttle fleet operational at all, even when the shuttles don’t fly. This doesn’t mean, of course, that we should ignore vehicle design, but it does mean that we need to pay much more attention to the dynamics of the market.

And it remains true today. SLS proponents are proposing a vehicle that will fly very rarely (NASA says once every year or two). What will the standing army of personnel be doing between flights to maintain their proficiency? How can such a vehicle possibly be as reliable as a system that flies dozens of times a year? How can its costs possibly compete with one that has a high utilization rate of manufacturing and processing personnel and facilities? And in fact, this explains the empirical reality, described above, that today’s existing rockets are much cheaper per pound of payload delivered than SLS can ever hope to be.

And here’s one more question, that SLS proponents never answer. If a vehicle of this class is essential for spacefaring, then why don’t we need two of them? Why is there no concern about launch-system resiliency, and redundancy, and architectural robustness?

After all, the Shuttle spent about a quarter of its life cycle unable to fly due to problems such as losing a couple, or mysterious hydrogen leaks, or whatever. The delays caused by Challenger and Columbia were at least two and a half years each. Do they really believe that somehow, this time we’ll get it right, this time the vehicle will have no problems that keep us from flying it? Really? So if we can’t explore space without it, and we are willing to risk not being able to explore space for months or years at a time due to problems with a non-resilient launch system, then the message I take from that is that we don’t think that exploring space is as important as building and (occasionally, but not very often) flying big rockets.

They go on to sow more FUD.

Issues of technical feasibility and practicality also exist. When cryogenic fuel is stored on-orbit, in whatever vehicle, the ability to maintain it in its cryogenic state is crucial. With today’s capability, we might achieve liquid hydrogen boil-off losses of about 0.35 percent per day, or about 10 percent of the fuel each month. At a boil-off rate of 0.1 percent per day — a capability not yet demonstrated — 10 percent of the fuel will be lost in three-and-a-half months. Completely closed-cycle systems, or those that are nearly so, are possible with active refrigeration. This technology absolutely must be pursued, as it is necessary for missions beyond the Moon. But we should be skeptical of unproven claims about extremely low boil-off rates, such as the 0.5 percent loss rate per month assumed in one recent study, until and unless the technology is demonstrated.

First of all, we don’t have to use hydrogen. Yes, it reduces the propellant amounts needed, but we could be doing exploration with storables to get started, and transition to cryos later after we mature the technologies. Also, I don’t know where their boiloff numbers are coming from. I don’t think that the people at ULA would agree with them, and while they indeed haven’t demonstrated the low rates, they have modeled them, and I’m unaware of any reason to think them unachievable.

The most reasonable claim made in support of fuel depots is that if they are employed to the exclusion of a heavy lifter, one saves the cost of building the heavy lifter. This is certainly true — but then we do not have a heavy lifter! Heavy-lift launch is a strategic capability for a spacefaring society, and its absence severely constrains any plans. The 130-metric-ton SLS capability should be regarded as the floor of space-lift capability for exploration, not the ceiling.

Note that this is an completely unsupported assertion. In what way does it constrain our plans? In terms of sending people to other locations in the solar system, or building off-planet facilities, what can we do with it that we cannot do without it? What is magic about 130 tonnes? Where is the analysis to support this? I’ve never seen it.

The kind of space program that we need requires transportation of much that is not fuel. While the international space station offers an existence proof that one can build a 400-metric-ton object in space using pieces weighing less than 15 tons each, the time, money and programmatic risk required for assembly offers the clearest possible demonstration that it was not the best approach.

What specific pieces of exploration hardware weigh more than a Falcon Heavy can toss? Where is the analysis? A lander with no propellant weighs a dozen tons at most.

And with modern high-flight-rate vehicles why would it involve more time to assemble a mission? If there is a launch failure, wouldn’t we rather lose a single element that could be cheaply replaced than everything at once? And if it’s not replaceable, with more coming off an assembly line, then how fragile is our human spaceflight program?

This is all just more FUD.

Thus, those who argue that we could save money by using fuel depots and not building a heavy lifter seem willing to ignore a key theme: We need a heavy lifter for reasons going far beyond the transportation of fuel. It may be that in future space architectures the flexibility offered by fuel depots will compensate for their inefficiency. But they are not an appropriate feature of the developmental systems and architectures we need to build now.

This is just a repeat of the previous paragraph. What pieces? How much do they weigh? How “inefficient” are depots, and by what measure? How can that “inefficiency” possibly cost us the tens of billions of up-front undiscounted dollars that they propose to spend on SLS? Do they really believe that developing and demonstrating cryogenic storage technology is going to cost tens of billions? If so, why? And why do we need to build SLS “now,” but we don’t need to build landers, departure stages, and things that we actually need to explore and develop space “now”?

Do you know when we need to build heavy-lift vehicles? When there is enough traffic to justify flying them more than once every year or two. And when we have figured out how to do them without throwing the vehicle away every time, so we can really have the “marginal cost” benefits that they falsely claim for SLS.

Fuel depots as an element of a near-term space architecture are an example of magical thinking at its best, a wasteful distraction supported by the kinds of poorly vetted assumptions that can cause a concept to appear deceptively attractive. We in the space community are especially prone to such behavior. If we actually want to accomplish anything, it must cease. We need to do the right stuff, right now. When we have settlements on the Moon and Mars, the use of fuel depots will make sense. But for today, the last thing we should do is to put one of the hardest problems — long-term cryogenic fuel storage — in series with our next steps beyond LEO.

You know what’s really magical thinking? Clark Lindsey described it yesterday:

They are the ones living in a marvelously magical land in which:
/– NASA’s budget is $30B and growing rather than $18B and dropping;
/– NASA’s overhead and fixed costs are not counted in the cost of development of NASA’s vehicles;
/– development, overhead and fixed costs are not counted in the operation of NASA’s vehicles (thus making for magically low marginal cost estimates);
/– the public gives a damn about seeing a Saturn V wanna-be take three or four astronauts to the Moon every few years;
/– and flying a totally expendable vehicle costing several billion dollars constitutes “spacefaring”

If these are the best arguments that can be made for SLS, it’s doomed. But unfortunately, probably not before Congress insists on wasting a few billion more on it.