Friday, September 09, 2005

Falcon 9 Upper Stage Recovery

The SpaceX annoucement is fluffy. Their big engine is late, and they're going to try to push their Merlin 1 based product a bit past its economic sweet spot.

Significantly, SpaceX has not specified the upper stage. The upper stage for a plain Falcon 9 to LEO lift will require at least two and perhaps three Merlin engines. The upper stage for a Falcon 9-S9 to LEO lift (required to deliver a Space Shuttle cargo to the ISS) will require at least five Merlins, which means it will actually be a Falcon 5 first stage, but with large expansion ratio nozzles and some ability to do in-space restarts (propellant settling, multi-use igniters, RCS without gimballing the main engine). If they are going to recover that stage, though, they'll need more than just the F5's parachutes, they'll need some really wonderful ablatives, somehow spread under all the delicate engine bits.

If you ignore the upper stage reuseability, the business scheme seems wonderful. NASA needs 20+ heavy LEO launches to lift the rest of the ISS. SpaceX builds a fleet of identical airframes and engines. Four airframes and 32 engines go up, three and 27 come back. They build about 25 airframes and 140 engines over several years, and each airframe gets an average of three flights, and each engine gets four or five. They might charge NASA $1.5B for the whole launch set and make a ton of money, and save you and me (i.e. the average taxpayer) about 75 bucks apiece.

But what's this about Full reuseability? Full reuseability is not a press mistake. The SpaceX press release says explicitly that

Falcon 5 and Falcon 9 will be the world's first launch vehicles where all stages are designed for reuse.

(Except Kistler's design, of course.)

This is crazy. If they take over lifting the ISS segments, they'll have a nice steady stream of launches. Over time, they're going to learn how to manufacture their engines better (more thrust, better reliability, lower manufacturing costs), and the later engines will be better than the earlier ones. The steep learning curve makes the early hardware depreciate fast, which makes spending more money to recover it less attractive. Throwing away the upper stage is a great way to clear out the obsolete inventory.

Heck, at some point (2007?) they're going to deliver the Merlin 2. My guess is that this will be a 600,000-lbf-thrust engine. Their next-gen EELV-class launcher will lift 25,000 kg to the ISS without strapons, with four Merlin 2's as the bottom stage and four or five Merlin 1's as the top stage. Why use the Merlin 1's in the upper stage? Because their business plan requires each engine to get used a bunch of times. The last thing they want to do is introduce a new rocket which obsoletes $60 million dollars worth of engine inventory. Instead of recovering a brand-new engine from the upper stage, it will be cheaper for them to use and expend their Merlin 1 inventory in the upper stages of their launches.

I bet the upper stage reusability verbiage is just there to appease some group at NASA. SpaceX may actually attempt a controlled upper stage reentry with ablatives when launching some slightly more lightweight ISS segment. They may get some interesting data from such experiments. But I can't see them pushing upper stage reuseability very hard. They need to concentrate on the easier and more lucrative problem of recovering their lower stages first.


  1. Iain,
    I discuss a little about this in my recent Selenian Boondocks post. Basically, I heard a little bit about the approach they are currently noodling, and I think that they'll likely make some attempts at seeing if it'll work, and trying to take data to improve the concept until it is eventually ready for prime-time. While I highly doubt they'll pull it off without a hitch the first time, I give them decent odds of getting at least some data, and if the idea isn't unworkable, eventually getting it figured out.

    Somewhere else, Elon mentioned that they were designing all the parts on the Falcon V first stage (and thus the Falcon IX too) for up to 100 flights, not just 3-4. Whether they can acheive this or not, and if it makes economic sense or not is anyone's guess at this point.

  2. Jon says that they're going to try a tail-first reentry. Seems reasonable, that's the heavy end of an empty rocket stage, and the nozzles are used to very high temperature operation.

    Bleeding a little gas into the back side of the shock would give them a way to get some heat out of that area. I don't think they'd need a lot of gas, maybe just run the gas generator for one of the turbopumps of one engine. This would cost fuel mass only, as everything else necessary is already there (and the fuel is snugged down against the fuel pickup because it's a tailfirst entry and we're decelerating).

    The extra gas might even make the front of the shock a bit wider and help get it past the bottom edge of the rocket. I wonder if you could push the idea farther and get a gas-phase parachute effect like the thrust reversers on a modern jet.

  3. Iain,
    There are dangers in running too much gas out of your nozzles for reentry cooling. Apparently, IIRC the research they did on the topic showed that it greatly decreases drag due to the shock changing (ie it simulates having a pointy instead of blunt reentry shape). While the vehicle is keeping a bit cooler at first, you end up going a lot faster in the lower atmosphere, which either defeats the purpose by causing worse heating and G-loads, or even could lead to lithobraking if you're not careful.

    But a little gas is probably a good thing.