Reusable Launch Vehicle Engineering: As I See It

I am not, repeat not, a rocket engineer. The purpose of this page is not to train you to be a rocket engineer; it is to make my SA '05 web site self-contained by showing how much I do (and don't) know about the subject.

The Rocket Equation

You can't understand sailing ships without thinking about weather and you can't understand RLVs without thinking about the rocket equation. One form of it is

            delta-V = Ve * ln ( M0 / M1 )

where: This simple equation glosses over the fact that Ve will usually change during a flight, and leaves out the fact that a launch vehicle is held back by air resistance and pulled back by gravity during some of its flight. If these factors are taken into account, the delta-V you need to get into orbit is usually about 9,000 metres per second. The basic problem facing anyone who wants to launch into orbit is that this number is much larger than the Ve available with common propellant combinations. Using LOX and kerosene gives about 3200 m/s; LOX and liquid hydrogen give about 4400 m/s. Both of these numbers vary slightly from engine to engine. Either way, we soon see that M0 has to be anywhere from eight to twenty times M1. (This ratio, M0 to M1, is known as the "mass ratio" of a vehicle.) If M1 is to include all the tanks, engines, plumbing, electronics, heat shields, etc. etc. etc, and some payload, then either the vehicle will be very fragile or the payload will be very small.

The traditional answer is to drop some hardware off in mid-flight so as to take them out of M1. So you have a two-stage vehicle, where the piece that drops off is a vehicle in its own right, or perhaps a one-and-a-half stage vehicle where you just drop some pieces. The simple equation above no longer applies, and useful payloads can be launched by a vehicle smaller than a mountain, but no airline would make money with a vehicle that takes itself apart and has to be put back together after every flight. Nobody has shown, or even (as far as I know) suggested, a way of doing this that approaches the simplicity and cheapness of landing the vehicle in one piece and re-filling the tanks. However, lately I have not seen much optimism that a true SSTO RLV can be built with acceptable safety margins.

This is not quite the same thing as saying that an SSTO ELV cannot be built. Vehicles dating back decades had designs that allowed one stage to reach orbit with a tiny payload, though they didn't in fact do it. But they had nothing with which to land: no heat shields suitable for re-entry at orbital speeds, nor any wings, parachutes, or rocket fuel with which to arrange a gentle touchdown. Adding these would most likely have increased M1 to the point where performance fell below what was needed to SSTO, even with no payload at all.

One obvious answer is to try and raise Ve. However, with one exception the proposals I have seen for doing this all involve propellants that are very expensive, very toxic, very corrosive, nuclear, or several of the above. I am pro- rather than anti-nuclear, but the mere thought of cleaning up after a crash of a nuclear-engined launcher is enough to fill me with lack of enthusiasm for that kind of thing. As for the chemical alternatives, they might be tolerable for experimental vehicles, but for an industry which I want to see flying many times per week, and which will inevitably have some accidents, they sound like bad ideas.

Another answer is to try and reduce the delta-V needed. There have been some promising approaches suggested here, most of which involve doing what Burt Rutan did: igniting the rocket engine after climbing above much of the atmosphere, thus reducing atmospheric drag and allowing the use of a more efficient engine. It turns out that an engine designed to operate in thin air or no air can often achieve slightly higher Ve than one designed to operate at sea level. You'll see several air-launched vehicles proposed here; there have also been proposals for a rocketplane to be towed behind a 747, and for one to take off with mostly empty tanks and then fill itself up from an aerial tanker.

Catapults and guns of various kinds have been proposed too. But I no longer see much serious work being done on them. The electrical kind (railguns and coilguns) involve engineering challenges that nobody seems anxious to take on, and large initial investments to get anything working. Guns using hot hydrogen or chemical explosives have a rather limited following, though I saw one proposed at SA '05.

Hungry aerospace engineers periodically propose to get around the rocket equation altogether by not using rockets. That is to say, they propose vehicles that breathe air for a subtantial proportion of their acceleration. But they do this only when they are spending the government's money and are under no real pressure to produce results. Jet engines as we know them are good to about 1,000 m/s and they are much heavier than rocket engines for a given amount of thrust, so a true SSTO that breathed air would incur a rather high M1. Proposed engines that can work at higher speeds (scramjets) run into another problem: if the air is dense enough to provide a useful amount of oxygen (don't forget that atmospheric oxygen comes polluted with roughly four times its own mass in nitrogen, which provides no energy to speak of and reduces Ve), then it is dense enough to impose a harmful amount of air resistance, increasing the required delta-V. I have not heard of anybody even thinking that they have succeeded in squaring this circle.

No account of RLV design would be complete without mentioning two more disagreements: between hydrogen and kerosene, and between horizontal and vertical lift-off.

As you will have noticed above, hydrogen gives a much higher Ve and therefore requires a lower mass ratio. But it is far from clear that hydrogen is the fuel of choice. Apart from csoting rather more than kerosene, it is far bulkier and has to be stored at much lower temperatures than just about any other propellant. The bulk means a bigger tank; and since the mass of a tank that has to hold pressure is proportional to the volume times the pressure, this alone makes the hydrogen tank heavier than a kerosene tank, so that although M1 can be higher, you have to allocate more of your M1 to tanks and less to payload. The pumps and pipes, and the rocket engine itself, also become larger and heavier for the same amount of thrust. Finally, the tank has to be heavily insulated. People who have run all the numbers say that, in the end, there is little to choose between the two. Hydrogen also causes design problems in that metals which come in contact with it tend to get brittle; kerosene causes other problems in that when it gets wamr it can leave nasty deposits in your pipes and engine. The final result, as I observe, is that hydrogen is preferred mostly by designers who know their design will be built with government money.

The old Saturn V used kerosene in its first stage and hydrogen in its second stage (I think). There was some talk of designing an SSTO RLV with a "tripropellant" engine that could burn LOX/kerosene or LOX/hydrogen, and providing a kerosene tank for initial ascent and a hydrogen tank for subsequent acceleration. Since most of the propellant is burned early on, this could have allowed for a much smaller hydrogen tank while retaining some of the advantage in Ve (since Ve can be averaged, roughly speaking, over velocity rather than over mass). But I have neve heard of such an engine being made to work.

The DC-X, and some more recent vehicles lifted off and landed vertically, using their engines to touch down gently; the Shuttle, Rutan's SpaceShipOne, and some other recent vehicles have wings and land on wheels. The advocates of "spaceplanes" used to ask whether you could be sure your engine would restart when you needed it, with an implied claim that wings were inherently safe, but that argument lost some of its force when the Columbia disintegrated. Wings add to M1 and increase air resistance somewhat; powered touchdown also increases M1 because extra propellant has to be carried. The jury is still out on this as far as I know. It is widely asserted that you should, at the least, take off the same way up as you land (i.e. don't do what the Shuttle does). I hear two reasons for this: one operational, namely that tipping the vehicle onto its tail before each flight adds work, and one developmental, that if your first test flight involves a transition from vertical to horizontal, you'll be taking a much bigger risk on that flight than the DC-X crew (or the Wright brothers) took on theirs.