I Know I'm Flogging a Dead Horse....

...but this post over on Burchismo was pretty much spot-on. A few excerpts:

It wouldn't have been that much more difficult to build a reusable lunar orbit-surface-orbit shuttle and leave it in lunar orbit after each mission. But NASA has an irrational fear of on-orbit refueling. For some reason that the space commentariat can't identify, they won't develop or test the technologies necessary to do it, and so they can always say it's untested and therefore too dangerous. This from the same outfit who designed and implemented the shuttle system so that it's first flight HAD to be manned. Huh? The same goes for the lunar transit step. A reusable Earth-Moon taxi is NOT beyond our technical competence. But NASA won't take the refueling step. So the program is designed in a way that makes it too politically easy to declare victory and quit at any point after a lunar landing or two or three with the new system has been achieved.

Regarding the Ares V HLV:

But it begs the question of the need for and economics of such a brute. The Russians couldn't make it pay with their equivalent, the long-dead Energia super-booster that only flew a few times and has now gone the way of the Saturn V. The demand for such heavy lift capacity is very, very small. In fact, it's only real because NASA chooses to make it so with the Orion mission design.

Ken was putting together a blog post that he never finished that had a rather informative quote from Scott Horowitz of NASA ESMD that illustrates rather clearly the biases and assumptions that helped them justify the decision they wanted all along:

The cost of breaking the exploration missions into numerous smaller pieces to accommodate a smaller launch vehicle is cost prohibitive. Each smaller element will have to become a complete spacecraft on orbit while performing an automated rendezvous and docking and be burdened with all the systems required to survive and operate in space including power systems, thermal control systems, propulsion systems, guidance navigation and control systems, docking systems, etc. Then there is the cost of the infrastructure required to support the surge rates needed for multiple launches of smaller launch vehicles that would be required during a lunar or Mars campaign. This combined with all of the associated operational costs make the use of smaller launch vehicles for exploration missions cost prohibitive. Add to that the impact on mission reliability as a result of performing so many launches and associated on-orbit assembly operations and one quickly realizes that the chances of accomplishing multiple moon or Mars missions using smaller launch vehicles is slim to none.

As we were driving down to Space Access, James Robertson was with us reading through some of the ESAS related documentation (or at least I think it was ESAS), especially the part leading up to the preference for the "1.5 Launch Architecture". I'm emailing him to see if he can find the relevant section again (because I sure couldn't find hardly any real substantiative discussion of that critical choice in the ESAS documents I skimmed tonight), but the gist of it was that the way they were treating mission reliability for multiple launch architectures was completely broken. They assumed that if you had a multi-launch architecture, that if a single launch failed, that the whole mission would be completely lost, as though all the hardware that you succesfully orbitted was somehow going to disappear instantaneously if you have a single launch accident.

[Updated Sept 12: James pointed me to the section, it's 6.4.4. It's worse and more garbled than even I remembered. Basically, they assume that all the assembly is going to be done autonomously (in spite of the fact that the US has little experience with Autonomous Rendezvous and Docking), and that any failed rendezvous event, or any delay in propellant delivery past the initial planned departure date, or any failed launch event means that the entire mission is lost. Now Rand points out that this may be valid for a Mars mission. But do you really think that by the time we do a manned mars mission that we won't have some sort of on-orbit infrastructure? That we won't have propellant depots that act like an UPS system for your computer, providing enough capacity to deal with temporary bumps in the supply chain? That we won't have the ability to "redock" components if the AR&D system doesn't work right the first time? Their analysis is valid if you accept all of their silly assumptions, but if you actually think it through, you come to completely different conclusions. ]

This is echoed in the last part of Scott's quote. If you assume a more sensible architecture, where instead of trying to make a whole bunch of mini-spacecraft and mini-tanks, and assembling it all together at some spacedock, that you instead just do a simple drylaunch approach like several of us have mentioned, that concern goes out the window. Most of the launches in a drylaunch architecture are non-mission critical, because they are propellant. Even with the worst realistic boiloff rates, a full month worth of delay wouldn't be an issue even for LH2 boiloff. Losing any propellant tanking launch is at worst going to delay your departure date. If you use a buffer/capacitor in the form of a propellant depot, you can even isolate your mission reliability from the launcher reliability further. And, if you use a reusable transfer stage, and reusable lander, and have a small fleet of three to four of each (launched over a year or two space), then even the accidental loss on launch of any given part of the hardware, the impact on the overall system is marginal.

So yeah, if you make unrealistically dumb assumptions about how a multi-launch architecture should be done, you can "prove" that a politically convenient Shuttle-Derived launch architecture is superior. Data never lies, but if you torture it long enough, you can get it to confess anything you want.