Shuttle-Derived Disaster

Jeffrey Bell came out last week with another questionable op-ed on space, having to do with the adaptation of Shuttle hardware into heavy-lift vehicles to support the Bush space plan (which still doesn’t have a good name).

Fisking the whole thing would take too much time (which is still in short supply around here), so I’ll just focus on a few of the bigger howlers.

The underlying fallacy here is that most of the Shuttle’s problems are contained in the Orbiter, and that the rest of the stack can make a safe low-cost launcher. This is simply not the case. Recall that the two fatal accidents so far were caused by fundamental design defects in the SRBs and ET, not any deficiency in the Orbiter.

Except that it isn’t a fallacy — most of the trouble and expense with the Shuttle is in the Orbiter, naturally, it being by far the most complex element in the Shuttle stack. A few recent examples of Orbiter-related problems include flowliner cracks, wiring problems, misassembled speedbrake actuators, and nosecap hardware corrosion. Each of these deficiencies in the Orbiter had the potential to cause the loss of another vehicle if not found and corrected, much as the known and uncorrected problems with the SRB field joint design and ET foam shedding brought down Challenger and Columbia, respectively. While one might argue against these examples constituting design defects, one could just as easily claim the field joint failures were operations rather than design errors (launching in weather too cold for the seal system), and that the foam shedding was a problem not with the design itself but with its implementation (environmental considerations forcing a change in material), or even a design problem inherent to the Orbiter itself, with its fragile TPS materials.

The rest of the stack can, in principle, be made into a safe, low-cost launcher. Safe, because it is likely to be used only as a cargo launcher (no crew aboard), and because the Orbiter-specific failure modes (such as TPS damage) are eliminated with the removal of the Orbiter. Low-cost, because the maintenance, upgrade, and turnaround costs of the Orbiter are replaced with the cost of a shroud, thrust structure, strongback (for side-mount designs), engines and avionics, items which are (with the exception of the strongback) figured into the costs of any other expendible launcher and which could to some extent be derived from existing designs and tooling. In principle, because it’s hard to imagine the responsible parties not finding some way to “gild” it if it actually does get developed.

But even if the cost per launch equalled that of the actual cost per launch involving an Orbiter, the much larger payload makes it cheaper overall in price per pound — you would get 3-5 times the payload for the same launch cost. It’s hard to see how one can assert categorically that a safe, low-cost launcher, in comparison to the standard STS, cannot be built based on Shuttle elements.

Now all that is left of Shuttle is a modified External Tank. But this has its problems also. The absurd brown Styrofoam exterior has obviously got to go. And the production cost of each ET has ballooned up to over $60M — more than a baseline EELV booster complete with engines and avionics!

The cost per ET has risen as the result of reduced flight rates and the adoption of aluminum-lithium several years back. Aluminum-lithium is notoriously difficult (and therefore time- and labor-intensive, and thus expensive) to work with, but the higher cost was the price to be paid for the increased performance needed to build ISS in a high-inclination orbit. Changing back to AL2219 would bring the cost per tank back down, as would the implementation of advanced manufacturing methods such as friction stir welding…but we don’t need to wait for an SDV program, since those sorts of changes are already being implemented.

Again, it’s important to compare SDVs with other launch vehicles using valid measures. Even if an ET for SDV use did cost $60M, the comparison with EELV is invalid because it doesn’t account for the significant difference in payload capacity between, for example, Shuttle-C and a Delta IV Heavy (it’s also invalid because even the least expensive EELV for which astronautix gives cost figures, the Atlas V 401, costs nearly a third again the figure he gives for an ET, which directly contradicts his assertion).

As for the “absurd brown Styrofoam exterior”…I’m sorry if the ET’s TPS isn’t up to Mr. Bell’s aesthetic standards, but it does serve a purpose, and therefore is not absurd: the (visible) foam is on the tank primarily to maintain propellant conditioning and prevent the formation of ice, which (how ironic) could fall off before or during launch and damage the Orbiter’s TPS system, and also to provide some acreage protection against aeroheating (though the brunt of that is taken up by a silicon-silica-cork-phenolic ablative material buried under the foam in critical areas). Substituting a side-mounted cargo element for the Orbiter would eliminate much of the risk of damage from falling ice, allowing foam to be reduced in certain areas, and moving to a fully inline configuration would eliminate many of the protuberances which require special foam and ablator applications, but a foam layer would still be required (or at least desirable) for efficient and accurate propellant loading and level maintenance and increased propellant density. Of course, the foam could always be painted over, as it was on the S-II stage (and ET-1), if that would make it look less “absurd”.

(And to pick a nit, it isn’t expanded polystyrene, either — it’s a urethane material, mostly NCFI 24-124 spray-on polyisocyanurate.)

The reason here is the obsolete labor-intensive methods of manufacture, combined with a complex structure needed to absorb the forces generated by the side-mounted engines. Clearly a cheap SDV-4 needs new lightweight first-stage tankage designed to be as simple as possible and fabricated mostly by robots instead of people.

Robots. Riiight.

The result of this little exercise in back-of-the-envelope engineering shows the complete absurdity of a SDV.

No, it doesn’t. It only illustrates the result of an a priori hostility to the SDV concept.

I haven’t even touched on the selective use of certain elements (SSMEs have to be used, and if they’re used, they have to be used with a recovery module) and the questionable premises for some of his arguments (basing his dismissal of EELV/SDV mixed architecture on the 1960’s Gemini-to-Agena docking program), but I’ve already spent too much time on this. There’s plenty of howlers for everyone — feel free to tackle others in the comments.