Going to Space - The Right Way

Alex Tabarrok has a response to a couple of his critics about his statistical analysis of rocket failures at Tech Central Station. Since I just now have the time to write a complete rebuttal of Professor Tabarrok's original article, I will include further rebuttal of his additional post.

The first problem is in comparing orbital Expendable Launch Vehicles (ELV) with sub-orbital Reuseable Launch Vehicles (RLV). The 40 years of data appear to be based on the Atlas, Titan, Redstone, Saturn, Delta, and Space Shuttle lines of vehicles. All of these vehicles are one use items. This prevents extensive testing as would be expected of durable goods such as airplanes and automobiles. Even the supposedly re-useable Space Shuttle has proven to be at best a re-furbishable vehicle requiring as much expense to prepare for the next flight as some expendable rockets cost to build new. Rand Simberg is absolutely correct to suggest that everything changed with SpaceShip One. The engineering and management problems of producing and using durable goods are entirely different than the problems of disposable goods. Not only that but the whole philosophy behind ELVs is that of artillery. Some of the ELVs were designed and primarily used as ICBMs. They were never intended to be safe, but rather have decent odds of getting a very large warhead to the Soviet Union. 5% failure is perfectly acceptable as a tradeoff for a few more pounds of warhead to the target. Even though the Space Shuttle was supposed to be a departure from the ICBM mindset, the political reality during its design and construction encouraged it to remain mired in the ICBM philosophy.

On the other hand the various RLVs being built today are designed with the philosophy of airplanes. Burt Rutan is an accomplished airplane designer, not a rocket scientist. Even the vehicles most like the ELVs, such as the ones being built by Armadillo Aerospace and Masten Space Systems, are based on aircraft principles of small crews, quick turnaround times, and long lifetimes. Spaceship One and XCOR's Xerus even look and operate like airplanes. These are vehicles that are supposed to be landed, re-fueled, and sent right back up again.

This difference in design philosophy results in major changes to the engineering trade-offs that are considered acceptable. When it is cheaper to do a second flight to loft the rest of the payload than to reduce the expected life of the vehicle, the vehicle reliability goes way up. The engineering tradeoffs are that it is better to use larger safety factors and lose some payload capability.

The difference in reliability testing between re-usable and expendable should be rather obvious. The cost of a re-useable flight is the cost of propellants, crew labor, and the risk of damaging the vehicle. The cost of an expendable flight is the cost of the vehicle, plus propellants, and crew labor. The reliability of a vehicle design that can and is tested frequently will greatly exceed that of a vehicle design that cannot be tested repeatedly. The test plan for MSS's XA-1 will give it more flights than any one of the Space Shuttles before XA-1 even reaches revenue service. Shortly after being put into revenue service XA-1 is expected to have more flights than the entire fleet of Space Shuttles. MSS's test plan is not unusual, other RLV's that are being built have similar test plans.

Management is an other issue. Comparing the safety record of a NASA managed Space Shuttle operations and the operations of SpaceShip One is questionable at best. SpaceShip One is operated by a small ground crew and a couple pilots, each with clear ownership and responsibility for their portion of the project. The shuttle has an org chart from Hell with no one person responsible for anything, even with all the paperwork and processes that are supposed to provide auditable records of responsibility. When the Space Shuttle has an incident it takes several months for an investigative team to figure out went wrong and who all had responsibility for the cascade of errors that led to the incident. For the unexpected behaviors of SpaceShip One during it's test flights, it took a day or two to figure out what the errors were and who needed to to do what to make sure that would not be a problem again. The enormous effects on quality by organizational culture are well documented. NASA as a government agency spending other people's money for other people's benefit cannot develop the culture of personal responsibility and ownership required to provide the level of quality and thus reliability necessary for commercial operations. Both Shuttle accidents have been attributed to an organizational culture that is not capable of doing the right thing at the right time.

Finally, there is the problem that historical evidence from other things, such as airplanes, that says the improvements in reliability won't come in a nice gradual curve like Prof. Tabarrok's graph, but rather with dramatic steps along the way. I fully expect reliability to greatly increase with the increase of small entrepreneurial firms designing and flying RLVs.

Share this

[...] tic Filed under:

[...] tic Filed under: General — mmealling @ 3:43 pm David Masten, our CEO, has responded to Alex Tabarrok’s critique of the safety record of [...]

Tabarrok actually says it

Tabarrok actually says it himself when he's responding to Rand:

These are statements of faith not of reason. Simberg has no data to back these claims because none exist.

He is correct. Very little data exists concerning vehicles that Rand is talking about and thus there is no statistically valid way of applying data associated with a completely different vehicle design philosophy to a new one. What Tabarrock is doing is essentially attempting to make statistical statements about cars by commingling the data about over the road trucking, passenger cars, and Formula 1 race cars. We went over the errors inherent in that approach in my Statistics 101 class.

Michael, you're overlooking

Michael, you're overlooking an extensive record for manned, suborbital reusable rockets- the X-15. In 199 flights, several vehicles were damaged, but only one pilot killed, for a fatality rate of 0.5% The data exists, and it proves Tabarrok wrong.

Game over.

Andrews Space did a very

Andrews Space did a very good risk analysis of the Space Shuttle system in their study for restructuring the NASA space program back in 2001. It is apparent that the Shuttle system is about the worst possible combination and arrangement of systems, placing risks on each other and summing up to total crew loss in most launch failures, so you can't even blame the shuttle losses on an ICBM mentality. The much more sane new proposal for a shuttle derived launcher using a single SRB with a liquid upper stage and an Apollo-style capsule CEV brings risk of crew loss up to 1 in 1300+, and that is the expected risk of a mature ICBM mentality.

The problems of the Shuttle, and most liquid fuelled heavy booster systems, is that the system is structured on two priorities which shouldn't even be in the top 20 on any launch ventures priority tree.

Firstly, NASA designers focus on maximizing Isp and negating environmental impact to the exclusion of all else. This has led to an improper obsession by NASA on using LH2/LOX fuel or other mixtures which produce as harmless an exhaust trail as possible, with as much thrust as possible.

What this leaves out is that the size of the vehicle is dependent upon the density of its fuel. Liquid hydrogen is the least dense fuel around, and an obsession on its use results in vehicles with massive fuel tanks and terrible aerodynamic cross sections. NASA does know better, it purposely built the first stage of the Saturn booster to burn kerosene so the fuel tank would be smaller and thus the first stage structure would be half the size (and mass) that it would have been with a hydrogen fuelled first stage.

They stick to LH2 for environmental reasons, but also for Isp reasons, but the Isp advantage of LH2 gets negated by poor aerodynamic coefficients in the bottom 20 miles of the trajectory that its low density forces in having large fuel tanks. Diborane, for example, offers significantly better Isp than hydrogen, but is several times denser, AND is liquified in the same range as LOX rather than the ultracold requirements of hydrogen cryogenics. Diborane's only drawback is an tendency to produce borax precipitates in the exhaust column, which is non-toxic, but hardly unnoticable like water.

The other priority that doesn't belong is a somewhat unintended systemic tendency towards complexity. In order to get governmental budgetary buy-in by enough special interests to pass congressional muster, any government space program needs plenty of hands in the cookie jar, and thus with lots of people making work for themselves to justify a government-spending-derived paycheck, systems tend to gain excessive complexity. The space shuttle is a prime example of of this: a not-so-heavy lift launcher, capable of carrying some scientists in space to do some make-work government grant science, with enough systems, subsystems, and auxilliary systems to fatten the payrolls of contractors in dozens of Senators districts, to make the DoD somewhat happy with the ability to launch large spysats (but never letting them put military space stations or weapons systems up), to make the academic lobby happy by adding value to Americas university system in giving students the ability to put experiments into space, paid for with government research grants, and to fatten the workrolls of government employee unions...

Walter Mondale once complained about "spending billions in space, when there are far better purposes to spend it here on earth". While I wasn't aware of any contractors operating in orbit to spend money with, it is clear that the government space program has a much higher priority on spending taxpayer dollars on anything that puts pretty pictures on newspaper front pages and paychecks in the pockets of American technical workers, than on actually successfully expanding our presence and infrastructure in space. Why should they? An uppity population of workers on a lunar base might get ticked off with paying exhorbitant income taxes and decide to do what Americans do: have a revolution. Keeping them all under the thumb here on Earth is so much easier and cheaper.