That a piston pump can be built light enough surprises me. I always figured something like a gear pump or lobe pump would be lighter than a piston.
It is a bit odd that we’re still using the same pump idea that Von Braun hit on by walking down to his local fire chief and asking them what fire trucks use. A turbopump is such an odd, indirect way to compress fuel. You burn propellant way off optimum (to keep the exhaust temperature down), run the exhaust through a set of two or three turbines, use the turbines to turn a shaft, use the shaft to spin a centrifugal pump which spins the fuel to give it a high tangential velocity, which you then stop in the volute to turn the velocity into pressure. Then in some pumps you try to repressurize the burner exhaust so you can dump it into the engine’s combustion chamber. It seems like an overly complicated, roundabout device.
I’m still crunching numbers on my direct-action rocket pump idea (blast the fuel with rocket exhaust). If it worked it would be simpler and lighter (and from the early numbers looks to be more efficient than a turbopump), so I must be missing some obvious violation of a principle of thermodynamics, heat transfer, or fluid flow, but I’ll be darned if I can’t figure out where the screw up is. Today I timed-stepped fuel droplet positions and velocities in an exhaust stream, hoping I’d found a show-stopper in the lack of radial inward droplet motion, but that looked fine, too. Perhaps I’ll have to build a water droplet/compressed air model out of PVC pipe just to prove to myself that it can’t work.
Something like 20 years ago I ran across an article describing an experiment with a fluidic injector pump suited to feed rocket engines. Seems the key was separating out and exhausting a portion of the gas stream after it accelerates the liquid being pumped. The volume of the stream entering the high pressure section must be less than the volume leaving the high pressure section as driver gas.
From some earlier calculation I did on an oxidizer pump, a 40:1 LOX to exhaust mass ratio using LH2/LOX in the burner (4,000 m/sec exhaust velocity) should result in a final velocity of about 96 m/sec for fuel and exhaust, a rise in LOX temperature by about 30 Kelvin, hopefully with a complete condensation of the exhaust steam into ice which would mix with the LOX and thus be injected into the main combustion chamber. There wouldn’t be any gas left, given enough time, so you’d get a staged combustion cycle for free. My calculations included the various specific heats, heat of vaporization, etc. That idea would only apply to pumping cryogenics, however.
My early thoughts on doing a water/compressed air experiment suggested some other outcomes. In the initial stages the 40:1 mass ratio of LOX to exhaust would also be a 1:27 volume ratio, so the droplets would be like one cell of a Rubic’s cube surrounded by gas. As the fuel velocity increases and the gas velocity plummets, the volume ratio would increase so the driver gas wouldn’t occupy as much space.
The seperation and pressurization problem boils down to trying to extract head pressure from raindrops by flying a plane through a thunderstorm, or from the spray of a garden or fire hose. One simple way to do that, which unfortunately reintroduces roller bearings, is the equivalent of putting a paint can on a shaft and spraying the water at the inside, just above the rim, slightly down and to the side, so that the bucket spins up. This is how you’d use a garden hose to spin a can in your yard, and once it spins up the water is slung to the outside with little loss of velocity. Momentum and gravity will carry it to the bottom of the spinning bucket, where a conventional centrifugal pump volute can convert it back into pressure. If the bucket’s bottom center has holes, any gas could escape via that route while the liquid stays slung to the outside rim. That idea seems bulletproof enough for a proof-of-concept experiment.
But such a mechanism might not be required. If the exhaust gas condenses to ice because you’re pumping a cryogenic, then the gas goes away and you’re just collecting a fluid stream. I have my doubts about that in practice because vapor pressure rules out a fluid in a vaccuum, but it might shift the design quite a bit. Also, once the gas goes subsonic you could neck the pipe down without introducing a normal shock, perhaps enough to where the LOX or fuel is almost a continuous stream, and perhaps boosting the velocity by a bit prior to complete exhaust condensation.
And another question that occured to me last week is whether you really have to convert the velocity back into pressure, since the point of the pressure is to reconvert it back into velocity for injection into the combustion chamber. If the fluid already has enough velocity for direct injection, all the pressure differential between the combustion chamber and input stream could do is to try and stop the inflow of fluid, and in so doing raising it’s pressure, which was the whole point of trying to raise the fluid pressure in the first place. Is letting the combustion chamber pressure stop the fluid to convert velocity to head pressure any different from using a mechanism to stop the fluid, as long as the fluid stream completely blocks the injectors so there’s no backflow of low density gas?
If that idea doesn’t have a major flaw in it, then the design would be small thrusters with low pressure exhaust blasting LOX and LH2 down pipes that neck down and bend (to seperate gas and liquid) and feed directly into the injectors. The pressure to run the thrusters could come from thrusters that are 40 times smaller, etc, till you hit two small pressure fed tanks for the first pair. The only moving parts would be valves.
One nagging concern is if the LOX or LH2 infeed would completely vaporize as a startup transient, and that would be quite exciting. Running an excess of fuel to exhaust at start up might prevent that.
Anywayk, given a 40:1 fuel/exhaust mass ratio and a 4,000 m/sec exhaust, 2.5% of the propellant is consumed for something like a 2,000 pressure (I haven’t checked back on that particular spreadsheet in a while), which is on par with the Rocketdyne F-1 which burned about 3% of the propellant in a preburner for a lower pressure.
It seems like a much simpler and more efficient method, which tells me I must be missing something obvious, and I can’t shake the feeling.
The driver gas condensing, like in a steam injector pump, works great if the propellant can absorb the heat. With standard practice cyrogen storage near boiling point, rather than deep subcooled, that doesn’t work so well.
Yes, deeper cooling would be required. I didn’t bother figuring in how the drop in gas pressure during condensation might cause subsequent low-vapor pressure boiling effects in the propellant. The idea already has plenty of unknowns piled up, such as what the droplets would actually do when hit by a supersonic hot gas stream and how they’ll behave when seperating gas and liquid phases. Given my incomplete knowledge of how it might behave, experiments with a small thruster and a pipe would probably be more fruitful than further ponderings unless someone here with a deeper background can flag whatever it is that I must be overlooking.
Back to the basic design idea, the initial propellant injector I envisioned around the pipe looks like a volute (tapering pipe wrapped around a cylinder), with a pattern of injection holes drilled in the cylinder before the volute is welded on. My last number crunching had injection pressure 20 to 40 psi above the exhaust nozzle exit pressure (ambient, but greater efficiency would be gained by near vacuum expansion). Since a massive amount of cryogenic propellant is being injected at the top of the pipe, it shouldn’t require any other cooling. Since it contains very little pressure it should be light enough to where length issues wouldn’t doom the idea if it takes a bit more length to accelerate the droplets up to speed.
A simple curved section at the end (to help seperate liquid and gas) with louvers acting like the blades in a volute might prove sufficient for pressure recovery without adding the requirement of the spinning bucket with precision high-speed bearings. Experimenting on that aspect of the problem could probably be done with a garden hose.
Given that it eliminates the most expensive and difficult part of a liquid rocket engine, can be built out of welded pipe, and might have greater efficiency than a turbopump, I can’t shake the thought that somebody should do some cheap experiments with the idea. If nothing else to uncover the reason why it can’t possibly work so I can quit doodling on it. ^_^
From the video shown the motor operation changes significantly from the beginning of the run to the end of the run, I’d say the plume is 60% as long at the end as the beginning….. I though a pumped system would have constant output?
The plume does indeed change length during the run, although the chamber pressure changed only slightly. The reason for the plume change is that the O/F changed a bit due to the fuel pump slowing down, which in turn was caused by the helium drive gas cooling as the He tank was emptied. This will not happen when we close the loop and reuse the helium.
Could you imagine where we’d be today if conditions had been right to create a space startup culture in the ’70s in parallel with silicon valley? The work these guys (and SpaceX, and Blue Origin, and Armadillo, etc.) is tremendous, it’s an exciting time to be paying attention to spaceflight.
That a piston pump can be built light enough surprises me. I always figured something like a gear pump or lobe pump would be lighter than a piston.
It is a bit odd that we’re still using the same pump idea that Von Braun hit on by walking down to his local fire chief and asking them what fire trucks use. A turbopump is such an odd, indirect way to compress fuel. You burn propellant way off optimum (to keep the exhaust temperature down), run the exhaust through a set of two or three turbines, use the turbines to turn a shaft, use the shaft to spin a centrifugal pump which spins the fuel to give it a high tangential velocity, which you then stop in the volute to turn the velocity into pressure. Then in some pumps you try to repressurize the burner exhaust so you can dump it into the engine’s combustion chamber. It seems like an overly complicated, roundabout device.
I’m still crunching numbers on my direct-action rocket pump idea (blast the fuel with rocket exhaust). If it worked it would be simpler and lighter (and from the early numbers looks to be more efficient than a turbopump), so I must be missing some obvious violation of a principle of thermodynamics, heat transfer, or fluid flow, but I’ll be darned if I can’t figure out where the screw up is. Today I timed-stepped fuel droplet positions and velocities in an exhaust stream, hoping I’d found a show-stopper in the lack of radial inward droplet motion, but that looked fine, too. Perhaps I’ll have to build a water droplet/compressed air model out of PVC pipe just to prove to myself that it can’t work.
Something like 20 years ago I ran across an article describing an experiment with a fluidic injector pump suited to feed rocket engines. Seems the key was separating out and exhausting a portion of the gas stream after it accelerates the liquid being pumped. The volume of the stream entering the high pressure section must be less than the volume leaving the high pressure section as driver gas.
From some earlier calculation I did on an oxidizer pump, a 40:1 LOX to exhaust mass ratio using LH2/LOX in the burner (4,000 m/sec exhaust velocity) should result in a final velocity of about 96 m/sec for fuel and exhaust, a rise in LOX temperature by about 30 Kelvin, hopefully with a complete condensation of the exhaust steam into ice which would mix with the LOX and thus be injected into the main combustion chamber. There wouldn’t be any gas left, given enough time, so you’d get a staged combustion cycle for free. My calculations included the various specific heats, heat of vaporization, etc. That idea would only apply to pumping cryogenics, however.
My early thoughts on doing a water/compressed air experiment suggested some other outcomes. In the initial stages the 40:1 mass ratio of LOX to exhaust would also be a 1:27 volume ratio, so the droplets would be like one cell of a Rubic’s cube surrounded by gas. As the fuel velocity increases and the gas velocity plummets, the volume ratio would increase so the driver gas wouldn’t occupy as much space.
The seperation and pressurization problem boils down to trying to extract head pressure from raindrops by flying a plane through a thunderstorm, or from the spray of a garden or fire hose. One simple way to do that, which unfortunately reintroduces roller bearings, is the equivalent of putting a paint can on a shaft and spraying the water at the inside, just above the rim, slightly down and to the side, so that the bucket spins up. This is how you’d use a garden hose to spin a can in your yard, and once it spins up the water is slung to the outside with little loss of velocity. Momentum and gravity will carry it to the bottom of the spinning bucket, where a conventional centrifugal pump volute can convert it back into pressure. If the bucket’s bottom center has holes, any gas could escape via that route while the liquid stays slung to the outside rim. That idea seems bulletproof enough for a proof-of-concept experiment.
But such a mechanism might not be required. If the exhaust gas condenses to ice because you’re pumping a cryogenic, then the gas goes away and you’re just collecting a fluid stream. I have my doubts about that in practice because vapor pressure rules out a fluid in a vaccuum, but it might shift the design quite a bit. Also, once the gas goes subsonic you could neck the pipe down without introducing a normal shock, perhaps enough to where the LOX or fuel is almost a continuous stream, and perhaps boosting the velocity by a bit prior to complete exhaust condensation.
And another question that occured to me last week is whether you really have to convert the velocity back into pressure, since the point of the pressure is to reconvert it back into velocity for injection into the combustion chamber. If the fluid already has enough velocity for direct injection, all the pressure differential between the combustion chamber and input stream could do is to try and stop the inflow of fluid, and in so doing raising it’s pressure, which was the whole point of trying to raise the fluid pressure in the first place. Is letting the combustion chamber pressure stop the fluid to convert velocity to head pressure any different from using a mechanism to stop the fluid, as long as the fluid stream completely blocks the injectors so there’s no backflow of low density gas?
If that idea doesn’t have a major flaw in it, then the design would be small thrusters with low pressure exhaust blasting LOX and LH2 down pipes that neck down and bend (to seperate gas and liquid) and feed directly into the injectors. The pressure to run the thrusters could come from thrusters that are 40 times smaller, etc, till you hit two small pressure fed tanks for the first pair. The only moving parts would be valves.
One nagging concern is if the LOX or LH2 infeed would completely vaporize as a startup transient, and that would be quite exciting. Running an excess of fuel to exhaust at start up might prevent that.
Anywayk, given a 40:1 fuel/exhaust mass ratio and a 4,000 m/sec exhaust, 2.5% of the propellant is consumed for something like a 2,000 pressure (I haven’t checked back on that particular spreadsheet in a while), which is on par with the Rocketdyne F-1 which burned about 3% of the propellant in a preburner for a lower pressure.
It seems like a much simpler and more efficient method, which tells me I must be missing something obvious, and I can’t shake the feeling.
The driver gas condensing, like in a steam injector pump, works great if the propellant can absorb the heat. With standard practice cyrogen storage near boiling point, rather than deep subcooled, that doesn’t work so well.
Yes, deeper cooling would be required. I didn’t bother figuring in how the drop in gas pressure during condensation might cause subsequent low-vapor pressure boiling effects in the propellant. The idea already has plenty of unknowns piled up, such as what the droplets would actually do when hit by a supersonic hot gas stream and how they’ll behave when seperating gas and liquid phases. Given my incomplete knowledge of how it might behave, experiments with a small thruster and a pipe would probably be more fruitful than further ponderings unless someone here with a deeper background can flag whatever it is that I must be overlooking.
Back to the basic design idea, the initial propellant injector I envisioned around the pipe looks like a volute (tapering pipe wrapped around a cylinder), with a pattern of injection holes drilled in the cylinder before the volute is welded on. My last number crunching had injection pressure 20 to 40 psi above the exhaust nozzle exit pressure (ambient, but greater efficiency would be gained by near vacuum expansion). Since a massive amount of cryogenic propellant is being injected at the top of the pipe, it shouldn’t require any other cooling. Since it contains very little pressure it should be light enough to where length issues wouldn’t doom the idea if it takes a bit more length to accelerate the droplets up to speed.
A simple curved section at the end (to help seperate liquid and gas) with louvers acting like the blades in a volute might prove sufficient for pressure recovery without adding the requirement of the spinning bucket with precision high-speed bearings. Experimenting on that aspect of the problem could probably be done with a garden hose.
Given that it eliminates the most expensive and difficult part of a liquid rocket engine, can be built out of welded pipe, and might have greater efficiency than a turbopump, I can’t shake the thought that somebody should do some cheap experiments with the idea. If nothing else to uncover the reason why it can’t possibly work so I can quit doodling on it. ^_^
From the video shown the motor operation changes significantly from the beginning of the run to the end of the run, I’d say the plume is 60% as long at the end as the beginning….. I though a pumped system would have constant output?
The plume does indeed change length during the run, although the chamber pressure changed only slightly. The reason for the plume change is that the O/F changed a bit due to the fuel pump slowing down, which in turn was caused by the helium drive gas cooling as the He tank was emptied. This will not happen when we close the loop and reuse the helium.
Could you imagine where we’d be today if conditions had been right to create a space startup culture in the ’70s in parallel with silicon valley? The work these guys (and SpaceX, and Blue Origin, and Armadillo, etc.) is tremendous, it’s an exciting time to be paying attention to spaceflight.