Sunday, 21 September 2003

Space Shuttle Economics

From Tech Central Station comes some unpleasant truths about the US Manned Space program, the Shuttle, and the Great White Hope that is the Orbital Space Plane (OSP) :
Depending on how you do the accounting, it costs around three billion dollars per year to operate the Shuttle, almost regardless of launch rate. That's simply the annual cost of maintaining the logistics and engineering "tail" that supports the space "tooth," as represented by actually putting people and hardware into orbit a few times a year. Because the only mission for the Shuttle now (particularly since we learned the danger of sending it to non-station orbits) is to support the station, and the station has so little purpose, the number of flights required is low.

This is a problem because of average versus marginal costs.

Average costs are the total annual program costs divided by the number of flights. For Shuttle, because of the high annual fixed cost, and the low flight rate, this results in a cost per flight of hundreds of millions of dollars -- over half a billion by even conservative estimates.

The marginal cost per flight is the part of the cost that has to be incurred in order to fly the next flight, given that the system is already operating. It's computed by adding up the total annual costs for N flights, and subtracting that from the total annual costs of N + 1 flights. It basically consists of things that are expended, or have to be replaced or maintained each flight, such as propellants, external tank, etc., and mission-specific crew training. This cost is more like a hundred fifty million per flight or less.


Unfortunately, OSP will not significantly reduce costs, and may even increase them, at least if one does normal (as opposed to governmental) cost accounting.

First, a mission that could have been performed with a single Shuttle launch will now require at least two, and perhaps three flights of a still-expensive expendable (probably on the order of a hundred million per flight). One to deliver the OSP, providing delivery/return of the crew, and one or two to deliver the payload that the Shuttle would normally carry in its payload bay.

Second, consider the cost of operating the OSP itself. Let's be generous and assume (improbably) that NASA will actually reduce the work force at the Cape and in Houston currently devoted to maintaining the Space Shuttle and training astronauts (it will be hard to do that, because of Congressional pressure to maintain the jobs). Let's be wildly optimistic and assume that they can cut it by two thirds. That's probably still several hundred million dollars per year.

That means that, given NASA current trivial plans (four annual flights to space station), the average cost of processing an OSP flight will be on the order of a hundred million per flight. So now our "Shuttle replacement" is up to three or four hundred million dollars per equivalent Shuttle mission.

But wait, there's more!

Estimates to develop and build the fleet of OSPs range from a few to a dozen billion dollars, and history teaches us that even NASA's high estimates often turn out to be low. That money will be spent mostly up front, before the system even flies. It will have to be amortized over the number of future flights.

As a simple sample calculation, suppose that it costs nine billion dollars, and we spend one and a half billion per year for the next six years. Let's say further that it will fly first at four per year for several years, then at eight per year out to the year 2030, for a total of 144 flights. One can build a spreadsheet to determine how much must be charged per flight for that amortization for various discount rates (i.e., the fact that future dollars are worth less than present dollars, reflecting the very real cost of money).

Even being fair, and discounting the development costs as well, a discount rate of five percent (meaning that a dollar this year is worth only ninety five cents next year and that in ten years its value has been reduced to sixty three cents) would require a per-flight charge of over a hundred and thirty million dollars. Seven percent yields a hundred seventy million, and a ten percent discount rate requires a whopping quarter of a billion dollars per future OSP flight just to amortize its development and construction costs.

Some might say that I've got an unrealistically low flight rate, but you'll be hard pressed to find anyone at NASA projecting a higher one, partly because they know that the marginal cost (the cost of the expendables needed to support it) is always going to be high, and not within any predicted budget projections. Only a truly fully reusable space transport has a chance of getting its per-flight costs down to the point at which elasticity of demand (the fact that when the price of something goes down, the demand for it goes up) can reasonably be assumed to kick in, thus permitting a high enough flight rate to make the amortization costs reasonable.

Add this to the operating costs already described above, and it's clear that this is at the very best a break-even proposition. Note that we don't have to worry about amortizing Shuttle development costs -- they're already, in accounting terms, "sunk," and unavoidable, whereas the development costs for OSP are entirely avoidable, given a little fiscal sense.

That, in a nutshell, is why the government has never funded a Shuttle replacement -- we simply don't plan enough activity in government manned spaceflight to justify it.
He's right regarding the problem. The US must either leave it to the Chinese (as the remaining Shuttles wear out or crash) or spend a lot more on a much more ambitious programme than the one they've got.
As regards the criticism of NASA in the article - my own experiences of NASA have been limited to co-operating with them on the GPS system on FedSat. Solely from that evidence, I have the highest respect for them.

No comments: