Thomas A. Heppenheimer, The space shuttle decision: NASA’s search for a reusable space vehicle, 1999. (Full text free online, since the book was government funded.)
I used to think the story of the Space Shuttle was that it promised to lower the cost of space launches, but sadly turned out to be more expensive than planned. And that is true enough: a cost-effectiveness report to congress in 1972 planned for $7.45 billion development costs and $9.3 million marginal cost per flight. They estimated that the project would remain cost-effective as long as costs did not grew by more than 20%. The actual program cost was $36.2 billion (no breakdown in recurring/nonrecurring available) and the marginal cost per flight $83 million (in 1972 dollars).
Heppenheimer’s book shows that the story is wrong. The cost was far higher than originally planned, for sure, but the original plan also did not add up!
The original idea for the space shuttle came from inside NASA. In early 1969 the Apollo program was nearing its climax, a new president was inaugurated, and the NASA administrator Thomas Paine asked his department heads to “outline bold objectives … worthy successors to those of Apollo”.
The plan they eventually recommended in the summer of 1969 was bold indeed. It called for a funding of $5-6 billion/year to develop: a re-usable space shuttle; a space station (starting at a crew of 12 and expanding to 100); a space-tug for near-earth applications; a lunar orbiting station; a lunar surface base; a nuclear space shuttle to ferry things between LEO-GEO-moon; and finally a manned flight to Mars in the mid-1980s. The first step of the plan was the space shuttle, because launching all the rest using expendable rockets would be unthinkably expensive.
$6 billion/year sounds suspiciously cheap for so much stuff, but even taken at face value it would mean doubling NASA’s 1969 budget of $3 billion, and (adjusting for inflation) exceed even the peak budget of 1966, back when NASA accounted for 4.4% of the total U.S. government spending.
In 1969 Americans were more concerned about Weathermen and Vietnam than about Whitey on the moon, and Nixon had absolutely no intention of paying for NASA’s proposal. On the other hand, it would be embarrasing to follow the triumph of the moon landing with the cancellation of all manned spaceflight.
The solution, arrived at by Nixon’s ad-hoc Space Task Group (STG) chaired by Spiro Agnew, was to fund just the space shuttle. The re-usuable shuttle was supposed to reduce launch costs by orders of magnitude and usher in a qualitatively new era in space flight—surely this would be a worthy new phase after Apollo.
There’s a snag, though. A re-usuable vehicle makes each individual launch cheaper, but the engineering work to design it is a big investment. So in order for the development to pay for itself there needs to be a lot of launches.
In Paine’s original proposal that went without saying, but now the STG planned to cancel almost all manned activity except for the shuttle development itself; even the small initial space station would be put on pause. So in fact, very few launches were planned. E.g., of NASA’s $3.3 billion budget request in 1970 only $125 million (3.7%) was intended to buy launch rockets, which is not enough to justify a many-billion shuttle program.
A space shuttle puts you in a peculiar kind of bind, compared to other types of projects. If your space station or space telescope is expensive, well too bad, you can’t put a price on science. But the whole point of the shuttle is to lower the cost of spaceflight, so if the price is too high the entire project becomes meaningless.
What’s worse, one of Nixon’s key priorities in 1969 was spending cuts, so the Office of Management and Budget (OMB) had a lot of clout, and they intended to check NASA’s numbers.
In May 1970 Mathematica Inc., a wonky Princeton-based consulting company founded by Oskar Morgenstern (of game theory fame), was contracted to analyze NASA’s plans using cutting edge economic theory.
Depending on the shuttle design the numbers come out slightly differently, but in general these analyses required at least 50 launches/year for the shuttle to be cheaper than the expendable Titan-III rockets that the Air Force had already developed.
For comparison, the highest rate NASA has ever achieved was 35 launches in one year, in 1966 at the peak of the Apollo program.
The years 1970-71, then, featured an expanding set of fudges. In order to get more launches, NASA postulated that the shuttle would handle all launches for the Department of Defense as well as civilian ones.
This had extensive consequences for the orbiter design: in order to be able to launch from Vandenberg air force base in California and glide back home from the Pacific Ocean, it was given large and heavy delta wings. (In the end, it never launched from Vandenberg even once.) And to launch military spy satellites the payload bay needed to be 60 feet long, while for NASA’s purposes 40 feet would have been enough. (The largest payload ever launched was the Chandra X-ray telescope, at 45 feet.) These constraints made the shuttle heavier and more expensive, but the DoD negotiation position was excellent; even though they were not paying a penny, they could in effect veto the project by declaring it unfit for military launches.
Unfortunately, even assuming the military launches, the number was not enough. At the NASA headquarters in Washington DC, a group (headed by engineering director Robert Lindley) was established to carry out economic studies in liaison with the OMB. In practice, one NASA manager recalls, this involved large amounts of motivated reasoning as the group scrambled for ideas that could justify the program
Lindley had this group of people from all over Headquarters, and he was drawing from people their estimates of “How much could be saved?” He was an extremely charming and extremely shrewd man who was getting out of this group of people a set of numbers for what the economics of the shuttle might be downstream. I mean, people would object about “This is not knowable, or if it is knowable, we don’t have the information yet; we would have to do a study.” But he really charmed them and said, “Hey, you know, let’s just get an estimate.”
It was obvious to me what he was doing was focusing, steering this group of Headquarters people into a totally subjective, qualitative kind of justification of the shuttle, without any real basis at all. And he went through about four or five iterations of this thing, finding out where the critical problems were, and finding solutions to these little problems. [The] problems, from their point of view, in justifying the shuttle.
For example, one NASA assistant administrator suggested they should push the space station connection. He postulated that if America built a space station, then an additional 26 launches/year would be needed to supply it, which would make the shuttle worthwhile. However, selling an expensive shuttle program by tying it to an expensive space station program seemed unpromising.
In retrospect, the projections of around 50 launches/year were extremely high. In reality the space shuttle averaged 4.5 launches/year over the lifetime of the program, and 4.2 launches/year in 1981-1990 (it was supposed to have payed for itself by 1990). The total launches from the U.S. was 17.3/year in 1981-1990, and 21.7/year in 1981-2010. Even at the time, the predictions probably involved some amount of bad faith. Robert F. Thompson, the space shuttle program manager, commented in 2003:
Mathematica discovered that the more you flew, the cheaper it got per flight. Fabulous! So they added as many flights as they could. They got up to 40 or 50 flights a year. Hell, anyone reasonable knew you weren’t going to fly 50 times a year. The most capability we EVER put in the program is when we built the facilities for the [External] Tank at Michoud – we left growth capability to where you could get up to 24 flights a year by producing tanks, if you really wanted to get that high. We never thought you’d ever get above 10 or 12 flights a year.
Apart from conjecturing a large number of launches, NASA also tried to argue with the OMB about discount rates.
These were a central part of the OMB/Mathematica analysis, because the spending happens over an extended period of time. E.g. in the two scenarios compared in the graph below, there is a big peak in shuttle spending around 1977, while the cost of buying expendable rockets is spread over the 1980s.
The money saved by not developing a shuttle could instead be invested in something else and pay interest, so to take this opportunity cost into account, Mathematica discounted all cash flows at a 10%/year rate. The higher the rate, the harder it would be to deem the shuttle cost effective.
In business, discount rates are a measure of the risk of a company defaulting on their payments.
A risky company that wants to borrow money have to offer investors a high rate of interest, or else the investors can lend to some safer company instead. (In June 1970, medium grade bonds yielded 9.2%, prime bonds 8.5%, and “risk free” treasuries 7.8%, although I guess these include inflation expectations.)
This concept was cutting-edge science at the time, as the CAPM was first published in the 1960s. In a government context, though, the meaning of discount rates is more dubious: the government is not supposed to invest on the stock market, or indeed make a profit at all.
The OMB instead used discount rates based on “national priority”, where programs considered important by the government—interstate highways, construction of schools and colleges—were evaluated using a low discount rate, while more frivolous programs like the space shuttle had to justify themselves at 10%. NASA argued that this rate ascribed too much risk to the shuttle program.
The OMB was implacable, but ultimately this didn’t really matter, because by most calculations the shuttle was unprofitable even at 5%, less than what treasury bonds were going for. So just paying down government debt would have been a better “investment” than developing the shuttle.
In the end, Lindley hit on the winning idea, which was named “payload effects”. If the shuttle did not sufficiently reduce the cost of launching stuff, then maybe it could reduce the cost of the stuff.
For example, because the Shuttle would have a ridiculously large cargo bay, there would no longer be any need to make satellites compact, so one could use really big circuit boards which would make them cheaper to design. (Would it? Also, the VLSI revolution happened just about when the Shuttle first flew, shrinking electronics to microscopic proportions.)
Or maybe, because the Shuttle was manned, it would be constantly sending up astronauts, so there would be no need to design satellites to be reliable–one could just debug them in orbit. Or when a satellite failed one could send up a shuttle to return it earth for refurbishment, and then re-launch it instead of building a new one. (These cost saving strategies also did not become popular.)
One must admire the chutzpah: usually we think the shuttle was expensive because it was too big and needed a crew, but now those two things were sold as the key to its cost effectiveness.
NASA commissioned a report from Aerospace Corp., subcontracted to Lockheed, which came to a suitably optimistic conclusion. Gazing into their crystal balls, they drew up a list of all space vehicles (both civilian and military) which would be launched by the U.S. in 1978-90, and estimated how much it would cost to build them using either legacy methods or by the new techniques that the shuttle’s large payload bay would enable.
They found the total cost of payloads would be $58.295 billion with expendable rockets, and only $44.170 billion with the shuttle. Then the cost-effectiveness report to congress could find the shuttle program cost effective even at a 10% discount rate—with the catch being that it assumed 47 launches/year and considered “total space program cost”, defined as “the sum of (1) launch system life-cycle costs and (2) payload system life-cycle costs”.
Aerospace Corp. and Lockheed were paid by NASA and wrote what their customer wanted to hear, but to outside observers the reasoning was obviously dubious, and the OMB was by no means convinced. In 1971 the director of the OMB Evaluation Division, economist William Niskanen, completely dismissed the Aerospace Corp.’s payload projections: “My impression is that the mission models that NASA is projecting for the 1980s are unrealistic.
They start at a number that strains credibility and go up from there.”, as well as the payload effects: “A large part of the presumed savings come from relaxed design, repair, and refurbishment of satellites.
I was struck, however, with the fact that payload design is so far down the road – in miniaturization, sophistication, and reliability – that you wouldn’t get manufacturers or users to go for much relaxation.” (Indeed, this proved to be the case.) In May 1971 the OMB completed their own in-house cost effectiveness analysis of the shuttle, which unlike the Aerospace/Lockheed/Mathematica report found the shuttle would not be cost-effective compared to the Titan-III.
Also in mid-1971, the shuttle gained a second enemy in the White House. In July 1971 Edward E. David Jr., the science advisor to the President and director of the White House Office of Science and Technology, created a new panel to review the shuttle program. The panel was chaired by Alexander Flax (president of a Pentagon think tank), and in practice it cooperated closely with the OMB and shared the OMB’s skepticism about the Aerospace/Lockheed/Mathematica study.
(One panel member is quoted “No one believed all the fancy economics and no one believed the mission model. I think they were on hemp when they were talking about sixty flights per year.”) On October 19 1971 Flax sent an interim report to David which analyzed NASA’s latest shuttle concept (here referred to as “Mark I/Mark II”), and completely dismissed the economics.
Never mind a 10% discount rate, considering likely cost overruns it would not pay for itself even at 0% undiscounted dollars. Notably, he correctly predicted that refurbishing the thermal tiles between flights might be expensive.
Considering all of the technological and operational unknowns involved in the shuttle development and the fact that no vehicles of similar function have ever been designed before or have ever operated over the range of flight regimes required for the shuttle, prudent extrapolation of prior experience would indicate that estimated development costs may be 30 to 50 percent on the low side. Thus, the estimates of $6.5 billion in RDT&E 34 for the Mk I/Mk II shuttle program may range between $8.5 to $10 billion, reflecting increased program costs of $2.5 to $3.5 billion. Similar uncertainties must be considered to apply to other non-recurring costs such as production and facilities (amounting to about $4 billion). Thus a possible cost uncertainty of about $5 billion for total program costs might be envisioned giving a high estimate of total non-recurring cost of about $15 billion.
At a launch rate of about 40 per year (DoD, NASA and other) over the 13 years used in the NASA cost model and an average payload cost of $30 million (not unrepresentative of the mix of current unmanned payloads), the total payload costs would be $15.5 billion. Thus, even if the total payload cost were saved (including those launched to Mars, Venus, etc.) over a 13-year period by recovery and reuse at zero refurbishment cost, it would, in the case of the high-end cost estimate, barely offset the cost of the shuttle program without discounting. A more realistic (although probably generous) estimate of the savings possible through payload recovery might be 50 percent of payload costs which could account for only $7.5 billion.
The other area of savings which is offered by the shuttle is its launch cost. Average launch cost with current expendable boosters is $12 million (projected into the 1978-90 era in the NASA cost model). Thus, with current expendable boosters, the annual launch cost will be $500 million. The cost of Mk II shuttle operation per flight is usually cited at $5.5 million; thus the cost for 40 flights per year will be $220 million. The saving of $280 million annually for 13 years amounts to $3.6 billion. However, a doubling of the operational cost would reduce the saving to $60 million annually or $780 million….
The operating cost estimates of $5.5 million per flight for the shuttle, within narrow limits, must be considered to be a very rough estimate at this time, particularly for the early years of shuttle operation. The actual value will depend upon the time between overhaul of equipment not yet designed, refurbishability of thermal protection system materials not yet out of the laboratory, and on the feasibility of operating in the shuttle in an “airline” mode radically different from all past experience in space operations.
In August 1971 the OMB hinted that they would recommend a NASA budget cut for 1973, which would effectively kill the shuttle program, and people within NASA had to seriously consider their alternatives. One interesting possibility was championed by George Low, the deputy NASA administrator. He proposed a program centered around a small cheap space station and a smaller gliding space plane launched using Titan III-L expendable rockets:
The glider itself would look somewhat like the shuttle, but would be smaller. It would carry a payload of 12 feet by 40 feet and a payload weight of about 30,000 pounds. It would have sufficient propulsion for on-orbit maneuvering but would not have the engines or propellant tanks required to propel itself into orbit. It would make use of current technology in avionics and other on-board systems. The glider would be placed into orbit with a two-stage vehicle of the Titan IIIL class. The glider and its payload would be reusable but both booster stages would be expended. The requirement for a 15 ft. by 60 ft., 65,000-lb. payload, could be met with the same expendable launch vehicle.
The Martin Marietta company liked this idea very much, since they were the ones producing Titan III-L. The Flax committee was willing to endorse it, provided the orbiter was scaled down even further, to a 10,000 lb payload.
And in retrospect, this does seem like an excellent idea. Putting the engines on the booster rocket instead of on the orbiter does mean that they are not reusable, but on the other hand, that lets you use the booster rocket alone to put cargo into orbit instead of having to waste payload capacity and maintenance budget on the orbiter.
When the Soviet Union developed their shuttle-clone Buran they did it this way, and given how expensive it turned out to be to refurbish the shuttle after each flight, NASA probably wished they did too. And after the American space shuttle was retired, and an older and more experienced NASA procured new vehicles for re-supply missions to the International Space Station, one of the three new space craft they funded was the Dream Chaser—an engineless space plane with an 11,000 lb payload, just as the Flax committee wanted. It would seem Low had exactly the right idea.
Low debated the glider with NASA administrator James Fletcher for several months, but in the September 1971 budget request, Fletcher still decided to risk pushing for the full shuttle. Inside NASA most people were unimpressed by the glider, e.g. Leroy Day, deputy director of the Shuttle program:
You had to put this thing on top of an enormous booster which you had to throw away each time. And so you had an operating cost that was getting to be kind of ridiculous. The vehicle size and everything-it didn’t have much utility. It was kind of a demonstration. It would certainly have been a research vehicle that you could have studied re-entry with. When you got all through with that then you would have said, “Gee, that would be nice if it was big enough to really do something.”
A second new option came to the forefront in September and October of 1971, when the director at Mathematica in charge of shuttle economics studies, Klaus Heiss, pushed NASA to consider a “Thrust Assisted Orbiter Shuttle” (TAOS) concept. This is the configuration we recognize as the shuttle today, where the orbiter still has engines, but is boosted by expendable rockets instead of a reusable booster.
Previously NASA had suppressed studies of this alternative as they were pushing for a fully reusable shuttle, but now they faced the possibility of either the glider or nothing at all.
By now the staffers deep inside the OMB had done the math, and in terms of dollars, the less ambitious options were better. Thus their reports recommended not developing a shuttle at all (and do all manned launches using a “Big Gemini” capsule launched by the Titan III); or failing that a small glider; or failing that a TAOS shuttle, but smaller than the 60ft/65,000lb payload that NASA was asking for. One memo, in arguing for a shorter and more narrow payload bay, eloquently sums up the entire shuttle program:
Since the Shuttle is not an economic system under optimistic assumptions, the importance of whether or not all of the payload benefits are realized becomes less significant. The important factors are really such considerations as national prestige, continuation of a manned space flight program, and advancement of technology. Any of the reusable vehicles discussed in this paper (even a 10’ by 20’ glider) provide this type of intangible benefit. A 10’ x 20’ glider would provide virtually the same intangible benefits as a 15’ x 60’ orbiter for less than half the investment cost.
At this point, the story changes from one of numbers to one of people.
Although the Flax panel was strongly opposed to the shuttle, this had little effect because the science advisor to the president did not have much clout. In the word of one insider, Edward E. David Jr. was “noticeably quiet, measuring his words, and repeatedly saying that he only represented science and that other factors are also involved. … Not really plugged into the President.”
The OMB was another matter. The OMB staffers were negative to the shuttle, and Donald Rice, an assistant OMB director, was the main exponent of the staff opinion. But as these recommendations filtered up to the OMB director George Shultz and the deputy director Caspar Weinberger, other considerations came into play. One might expect that the White House would choose between the NASA and the OMB proposals, but Nixon did not like to act as a referee and generally had NASA take budget disagreements to the OMB leadership, so the decision in effect rested with Shultz and Weinberger.
The stakes were high—after the super sonic transport plane (canceled in March 1971) and an abortive attempt in fall 1971 to fund a New Technology Opportunities Program, the space shuttle was the last remaining big aerospace project—and according to the same insider, Weinberger was “a real space buff. The only one in OMB really positive toward the NASA program. Causes Rice to over-balance in the opposite direction. Everybody lower in OMB is negative.”
The first key meeting took place in early August 1971. The OMB were about to propose a budget which would kill the shuttle, so James Fletcher, the NASA administrator, immediately met with Caspar Weinberger and John Young (head of OMB’s Economics, Science, and Technology Programs Division) to argue NASA’s case. Fletcher told Weinberger that without any new starts like the shuttle, the entire manned space flight program would be “out of business”.
Weinberger was convinced, and wrote a memo to the president which would keep the shuttle in play for the rest of the year:
eliminating or sharply reducing the balance of the Manned Space Program […] would be a mistake. The real reason for sharp reductions in the NASA budget is that NASA is entirely in the 28% of the budget that is controllable. In short we cut it because it is cuttable, not because it is doing a bad job or an unnecessary one. […] The Space Shuttle and NERVA particularly offer the opportunity, among other things, to secure substantial scientific fall-out for the civilian economy at the same time that large numbers of valuable (and hard-to-employ-elsewhere) scientists and engineers are kept at work
The final showdown took place in a series of meetings from November 1971 onwards, between Peter Flanigan (assistant to the president); James Fletcher and George Low (administrator/deputy administrator of NASA); and George Shultz and Caspar Weinberger (director/deputy directory of OMB). They had to decide between the pitches from Fletcher and Low on the one hand and Rice on the other.
There followed a tense game of poker. A NASA document sets out five possible shuttle sizes, and in private NASA carefully explored how small a shuttle they could accept.
Fletcher met with the secretary of defense to explore how rigid the payload bay requirements were, and found that both the launch weight and length requirements were “somewhat arbitrary” and could be reduced without jeopardizing the intended military missions. This was not something they would let on in communications with the White House/OMB, however, where they categorically stated that going to a 40’ bay would “lose most DOD payloads, all manned payloads (including resupply logistics vehicles), most planetary payloads, and many science and applications payloads.”
Rice’s staffers constantly probed the aerospace subcontractors and the NASA suboffices, asking for new studies and trying to prove that NASA had overlooked some cost-saving possibilities; the NASA directors tried to lock down information flow as much as possible, and to find embarrasing errors in the OMB staffer studies in order to undermine Rice’s confidence in his subordinates. At the same time, Fletcher tried to undermine Rice by “going over his head all the time”, frequently meeting with Weinberger and Flanigan.
Gradually, NASA won the OMB directors over. Fletcher and Low became confident that they would get at least the 2A option, which they already considered quite acceptable. The crucial final meeting took place in George Shultz’s office on January 3, 1972, where it seems Rice made a misstep. In the recollection of three NASA officials who had spoken with Fletcher and Low:
Rice got a little bit confused. There was a feeling with Low and Fletcher that Rice got too carried away, moving toward misstatements, trying to exaggerate some things. George Shultz picked up the phone and called Morgenstern during the meeting and asked him about it. Now, NASA had prepared for this kind of thing-to be sure Morgenstern was fully knowledgeable. And Shultz got off the phone and made the decision-we’ll go this way, and to prepare the papers for the approval.
I get this directly from Dr. Fletcher. He was telling his wife the story of how he went over there with the information that I supplied him from our shuttle program studies that showed that there wasn’t all that much to be gained by going 45-foot length and, furthermore, it invalidated a lot of missions that the Air Force claimed they needed for the sizing of the satellites. So Shultz said, ‘Well, what are you fooling around with that 45-foot configuration for? It doesn’t cost that much more. Why don’t you get the one you want-take the 60-foot one.’ And Fletcher came back with that message. That’s how it was settled.
George Shultz-he had personally spent almost as many hours going through all the planning studies, and especially this famous economic analysis that Klaus Heiss at Mathematica had done. And he personally called up Oskar Morgenstern and other people there, and he was satisfied, finally, that it was a reasonable proposition. So when it was clear that all the boys made their case, Shultz said, ‘If we’re going to do it, let’s do it right; let’s do the big shuttle and forget about the Bureau of the Budget shuttle.’ So that’s how we ended up with what it was.
The decision being made, all that remained was to tell Nixon, and then announce it to the world. Fletcher flew out to Nixon’s home in California to show him a shuttle model, and to discuss Nixon’s announcement statement.
Nixon stated that NASA should stress civilian applications but should not hesitate to note the military uses as well. He showed interest in the possibility of routine operations and quick reaction times, for he saw that these could allow the Shuttle to help in disasters such as earthquakes or floods. He also liked the idea of using the Shuttle to dispose of nuclear waste by launching it into space. Fletcher mentioned that it might become possible to collect solar power in orbit and beam it to earth in the form of electricity. Nixon replied that such developments tend to happen much more quickly than people expect, and that they should not hesitate to talk about them.
He liked the fact that ordinary people would be able to fly in the Shuttle, who would not be highly-trained astronauts. He asked if the Shuttle was a good investment, and agreed that it was indeed, for it promised a tenfold reduction in the cost of space flight. He added that even if it was not a good investment, the nation would have to do it anyway, because space flight was here to stay.
Dear reader, have you ever seen a space shuttle up close? If you’re ever in D.C., Cape Canaveral, Los Angles, or New York City, it’s worth making a daytrip. When you read that the shuttle is 37 meters long and weighs 78 tonnes those are just abstract numbers, but when you stand under the wing, it feels huge. The blocky contours and the ceramic tile covering makes it look even heavier, as if someone had once seen an airliner and then tried to build a brick house in the same shape. Now consider that we strapped this thing to a set of rockets weighing 1,900 tonnes, and accelerated it to 8 km/s, repeatedly. If the shuttle program was motivated by “important factors such as national prestige”—well, it’s hard to think of a better symbol of the American Way.