Saturday, December 20, 2008

Hypothetical Oxyhydrogen Propulsion Systems

Oxyhydrogen (LOX/LH2) is generally a very expensive system, but despite its overall impracticality, has an awful lot of experience in space because of the performance-oriented nature of space launch vehicles to date. Unfortunately, that doesn't make these systems practical for low-cost space access...until you are contemplating very large boosters and payloads, such as the Ascent Roadmap's Saturn-class Freezerburn. After Columbia, after almost six years of tinkering with booster concepts, has concluded the following about oxyhydrogen's proper use in low-cost boosters:

- A point will be reached where the reduction in mass of the upper stage for a given payload performance will outweigh the difficulty of increasing base area or total thrust pressure in the first stage. This is the point where oxyhydrogen becomes practical in upper stages. There is no analysis yet to back up this figure, but I currently estimate that it is somewhere between 100 and 500 tonnes LEO. Beyond LEO, oxyhydrogen may be welcome at a lower performance point. Let us assume for a moment that 200 tonnes LEO is the largest payload an all-oxykerosene launch vehicle can launch to LEO. By upgrading this launch vehicle (with the same lower stage system) to an oxyhydrogen system, a 300 tonne LEO performance can be achieved in a vehicle which is larger, but perhaps weighs less, since oxyhydrogen has a higher specific impulse. However, if one is looking for a performance beyond LEO (say, to geosynchronous) an oxyhydrogen transfer stage for higher energies may be desirable before oxyhydrogen is welcome in the booster itself. It is unlikely that this point in the Ascent Roadmap will be down where oxyhydrogen stages are currently used.
- Oxyhydrogen will never be practical in lower stages because it produces a lower total thrust pressure than other propellant combinations, including oxykerosene, which is currently the cheapest one.
- Oxyhydrogen will be best looked after by permanent launch site facilities. When selecting launch sites for oxykerosene-dominant booster, it will be important to make available the real estate that will eventually be required for oxyhydrogen. Since launch site real estate is dominated by the hazard area, and facilities can be inside it (preferably near its edge where the facilities can be practically armored against a booster failure), this should not be a problem.

The actual technical issues of using liquid hydrogen are rather unique. Some of its properties are nettlesome, but some are beneficial. Liquid hydrogen creates some problems while solving others, leading to what were once highly novel approaches to engine design. The Pratt&Whitney RL-10 is the first expander-cycle engine, one which is able to use a turbine that does not require combustion for its energy. This cycle is also being used in ESA's new Vinci engine for the ECA-B Ariane 5 upper stage. The weird properties of liquid hydrogen are listed below:

- It has a density of just 0.071kg/m3, one fourteenth that of water. This requires very light tanks with large volumes, but the low density means that these tanks usually do not require structures to handle propellant inertia or weight resulting from accelleration (such things as trusses and slosh baffles.) Their loaded vibration modes are similar to their empty vibration modes: in plain language, a tank loaded with liquid hydrogen can barely tell that it isn't empty. A disadvantage for turbopumps is that they need to generate fourteen times as much “head” as an equivalent water pump. A pump doesn't care as much about the discharge pressure of the fluid it is pumping, as much as it does its head, the product of acceleration and density. This is better known in earthly terms (since pumps have been around several times as long as rocket engines using them have been) as the column of fluid that a pump can raise against Earth's gravity. This column needs to be fourteen times as high to achieve the same pressure as for water, making hydrogen pump design a major challenge.
- It has a boiling point of just 25K, or -248degC...very, very cold. Liquid oxygen boils at a comparatively balmy 90K, and freezes well above hydrogen's boiling point. This makes materials selection and insulation a hazzle to make up for the tanks thinking that they are nearly empty.
- Hydrogen has the ability to form hydrides with many metals, very useful if you are contemplating a metal hydride battery, but very nettlesome if you're trying to build a titanium alloy tank. Most hydrides are far weaker and more brittle than the metals they are based on, and thus the tank is weakened and will fail. A nuclear reactor in Pickering, Ontario suffered enough weakening in one of its pressure tubes that it blew in August 1983 long before it was due for replacement. This was due to a small amount of hydrogen (actually, deuterium, the two-nucleon isotope of hydrogen) being dissociated from the cooling water by the reactor's radiation. This obscure accident is the closest a Canadian nuclear power reactor has ever gotten to a meltdown. Hydrogen embrittlement of pipe and pressure-vessel materials is a much bigger problem in sour gas wells and pipelines, where the metal can take hydrogen from hydrogen sulfide, which leads to both embrittlement and nettlesome sulfur deposits. Fortunately, all this experience means that while embrittlement is a nettlesome problem, it is one we are good at solving and mitigating in industries less expensive than that of space launch vehicles.
- Hydrogen has a remarkably high heat capacity. Both the specific and latent heat capacities hydrogen are far higher than those of most other substances, and may in fact be the highest of all substances. This makes hydrogen a very good coolant (although much to be desired as a refrigerant.) It is also these capacities that make possible the expander cycle. In the expander-cycle engine, the liquid hydrogen is first pumped to high pressures by the engine's turbopump (which is also pumping much denser liquid oxygen), and then fed into the nozzle jacket. Next to the main fire of the engine, the hydrogen is able to pick up a tremendous amount of heat while fulfilling its function as a coolant, keeping the engine's rocket motor from melting. This heat picked up from the nozzle is usable as enthalpy in the turbopump's turbine, and there is plenty to go around. After driving the turbine, there is still plenty of residual pressure, and the hydrogen is then injected into the engine, which can run at pressures as high as 5MPa before anything needs to be burned with the hydrogen to provide more power in the turbopump. This nozzle jacket enthalpy also makes staged-combustion engines like the SSME and RD-0120 much more efficient. Large engines have less nozzle area per unit flow of hydrogen, and so make it harder to stuff this usable enthalpy into the hydrogen, so there is a size limit to expander cycle engines with single motors. This limit could be exceeded by using multiple motors in the engine, or by convoluting the combustion chamber to have more area available to transfer heat into the hydrogen. Neither of these ideas have been tried as of yet (Ascent Roadmap is not contemplating oxyhydrogen engines large enough to bother.) The SSME and RD-0120, desiring 20MPa chamber pressures, add additional energy by burning a small proportion of the oxygen with the entire flow of hydrogen. The RS-68 is the laughing stock of oxyhydrogen engines for three reasons. First, the engine is too large to use the expander cycle effectively, and second, more importantly, the nozzle is ablatively cooled, so that transferring nozzle waste heat into usable enthalpy in the hydrogen is not even attempted. Finally, and most importantly, the RS-68 uses an open gas cycle, thus not using the entire flow of hydrogen to efficiently drive the turbine for the turbopump, but only a tiny flow of less efficient combustion gas, which is then lost to the engine by being vented out of a separate low pressure nozzle. This results in an abysmal specific impulse 9% lower than its competitors. This propagates into the almost totally ignored Delta IV launch vehicle (which piles on the bad decision of using oxyhydrogen in a first stage in the first place.)

So...what would the Ascent Roadmap ever do with oxyhydrogen? In order to consider it, we must look at the sources of expense in existing oxyhydrogen engines and seek to eliminate or reduce them. Ones that can be helped:

- Interpropellant seals: An interpropellant seal is a seal required to keep separate two fluids which can burn if they come into contact with each the rather undesirable location next to the shaft that connects pumps and turbines together. Inside a rocket turbopump, without getting into the specifics, you will almost always have a situation where the fluid being pumped will leak into the fluid driving the turbine because the fluid being pumped is at a higher pressure. The typical approach to interpropellant seals is to have a helium purge chamber and two seals: one to separate the turbine fluid from the helium, and one to separate the pumped fluid from the helium. As the two fluids leak into the helium chamber, the helium is circulated and vented to keep them from building up to concentrations where they can burn. The interpropellant seal almost never appears in industrial applications, but almost always appears in rocket engines. Life would be a lot easier if you could eliminate it. Let's see if we can.
- Diverse pumps: The properties of liquid oxygen and liquid hydrogen are so different that it is nearly impossible to achieve much commonality between the pumps and still have an engine of sufficient performance. Oxygen, with its much higher density, can be delivered by a pressure-fed system and still perform well. Especially since the hydrogen tank will be at least two and a half times as large as the oxygen tank in a typical oxyhydrogen stage, it may be possible to implement an engine that pumps only a single propellant.

So...can we design an upper stage that pressure-feeds its liquid oxygen, and uses a turbopump for its liquid hydrogen and thus eliminate these problems. Doesn't seem likely because the increased mass of the liquid oxygen tank at high pressures will make the stage too heavy. In this scenario, we would have to select a fairly low chamber pressure of perhaps 2MPa. The engine would use the expander cycle to drive its hydrogen turbopump, thus pure hydrogen will be in the turbine, with pure hydrogen in the pump at a slightly higher pressure. Since a small leakage of pumped hydrogen into the pure hydrogen turbine gas is acceptable, the interpropellant seal is eliminated.

What if we want to pump both propellants? The problem here of course is that in a typical staged or expander cycle engine, you will have a hydrogen turbine coupled to an oxygen pump, thus again requiring the interpropellant seal. Perhaps, though, since almost all of the pumping power required by an oxyhydrogen engine goes into the hydrogen turbopump, perhaps a less efficient cycle not requiring an interpropellant seal can be used for the oxygen pump. One possibility is to use a separate supply of hydrogen peroxide, as has been done in the Viking and RD-107/8 engines (famous for being economical in their time, on board the Ariane 1-4 and Soyuz boosters respectively.) Hydrogen peroxide breaks down into water vapor and gaseous oxygen vigorously enough to drive a turbine, but it has the disadvantage of requiring a third propellant to be carried on the stage (grant that it's relatively small, and the stage will probably already be using it for its reaction control thrusters, this is only a small problem.) As for the pumped liquid oxygen leaking into the turbine driven by hydrogen peroxide, the real problem is that the water vapor will freeze. This problem is bigger if it is the turbine gas that will leak into the pump. Rather than requiring a full blown interpropellant seal, the shaft can be packed with a dessicant (which removes the water vapor) and a heater (which keeps it from freezing.) Such a scheme may be overloaded, making the design of such a seal as much of a pain as an helium purged interpropellant seal, so it would help to find a way to drive the turbine with a cleaner gas mixture. The simplest solution is to burn a little bit of the hydrogen with either a generous dose of oxygen, or the entire oxygen flow. The latter is probably not practical, since oxygen does a relatively poor job of driving turbines compared to hydrogen, but don't count it out, since the RD-170 family drives their turbines with their oxygen flows burned with kerosene. The other possibility is to drive the turbine with a fraction of the oxygen flow burned with hydrogen in an open cycle (venting this oxygen rich turbine gas through another nozzle.) This turbine exhaust may also be useful as a pressurant in the oxygen tank, and of course, used for roll control during the engine's operation, saving on reaction control propellants. The turbine gas resulting from burning a little hydrogen with a lot of oxygen will still have a lot of water vapor, but not as much as hydrogen peroxide...enough to make a difference in the design of the oxygen pump's seal? We don't have to know, but we should save the pages in the recipe book, just in case.

No comments: