Hacker News new | past | comments | ask | show | jobs | submit login
The Tech Behind SpaceX’s New Engine (hackaday.com)
443 points by _JamesA_ 36 days ago | hide | past | web | favorite | 123 comments

“Benefits of the full-flow staged combustion cycle include turbines that run cooler and at lower pressure, due to increased mass flow, leading to a longer engine life and higher reliability.”

TL; DR Full flow lowers turbine temperatures at the expense of parts complexity. Given turbopumps are the devil’s ass part of rocketry, this has been a sought-after technology. The pay-off isn’t so much efficiency as much as longevity. (This also explains why full flow hasn’t been a priority for anyone until SpaceX.)


> TL; DR Full flow lowers turbine temperatures at the expense of parts complexity.

Ehhh. I'd argue it also lowers part complexity, because it eliminates interpropellant seals.

> The pay-off isn’t so much efficiency

It improves achievable chamber pressure (assuming similar maximum pressures and temperatures at the turbines), which improves both thrust and Isp.

> (This also explains why full flow hasn’t been a priority for anyone until SpaceX.)

The reason it hasn't been a priority is more that no new development has been a priority since the 70's.

> ...it also lowers part complexity, because it eliminates interpropellant seals.

But it adds the whole second gas generator and turbine. No, it doesn't lower part complexity as a whole :) .

> It improves achievable chamber pressure

Yes, I'd even argue that's the penultimate goal of full-flow scheme.

1) For a rocket engine you usually (almost always) want as high Isp (which is approximately the speed of gasses flowing from the rocket engine) as possible.

2) To get that Isp, you need as high pressure in the chamber as possible.

3) To get that high pressure, you want to supply your pumps with as much power (in Watts) as possible from the given overall fuel flow (in kilograms per second) - while not melting turbines blades (so fuel:oxidizer ratio is limited).

4) To get most power to the turbines blades at fixed (maximum) gas temperature and fixed overall fuel flow, you want to use all of that flow (classical staged combustion uses only one component, which is essentially not using the whole fuel flow) and adjust ratios fuel:oxidizer at both gas generators so that total power would be maximum. You want maximum flow both because that increases turbine efficiency and because that actually provides more power to the turbine.

> But it adds the whole second gas generator and turbine. No, it doesn't lower part complexity as a whole :) .

Because the fuel side and oxidizer side typically have a very different flow rates, many engines have separate pumps and turbines for them on separate shafts even if both sides use same fluids in the preburner. For example, see the SSME.

With FFSC, each of the turbines have no sealing requirements. This means, among other things, that bearings become easier and safer, as you can afford larger margins in your design. In many ways, FFSC allows the individual parts of the engine to be simpler.

The main cost of doing it over gas generators (other than oxidation-resistant superalloys...) is that all parts of the engine sort of circularly depend on each other -- when the Merlin engine was in very early development, they drove into the desert and ran the gas generator and fuel pump on it's own with basically no support infrastructure. Testing Raptor in a similar way is just not possible. It would have never gotten built without the NASA Stennis facility that can provide all the intermediate fluids at the pressures required.

I fully agree with the rest.

> Because the fuel side and oxidizer side typically have a very different flow rates, many engines have separate pumps and turbines for them on separate shafts even if both sides use same fluids in the preburner. For example, see the SSME.

By "many" you perhaps mean "many American". In Russian engines it's mostly a single shaft.

> The main cost of doing it over gas generators (other than oxidation-resistant superalloys...) is that all parts of the engine sort of circularly depend on each other...

Yes, rocket engine is usually a complex dynamic system, with deep feedback loops. You can still get gas generator and turbines with pumps tested separately, but you need to measure dynamic properties - like resonance frequencies - in order to have better chances of the good work when the system is integrated.

> By "many" you perhaps mean "many American".

And by "many American," do you perhaps mean "pretty much just the Space Shuttle Main Engine?" Since that one, has America launched any new staged combustion engine off the ground?

I believe GP meant that not for closed cycle engines, but for all turbo-pump driven engines, which in USA mostly means open loop (gas generator) engines. In USA the tradition was to add complexity in form of gears and extra axes to get pumps running with more optimal speed, while in Russia (USSR) it was to reduce efficiency and gain in simplicity, robustness and mass.

So even U.S. gas-generator engines like the RS-68 have this extra gearing?

Was curious about your mention of Stennis, so I looked it up:


Sounds like SpaceX has, with NASA and the State of Mississippi modified the test stand to support methane.

Interpropellant seals require an entirely separate inert fluid to be continually injected into the seal. That's a lot of added complexity.

The implication here seems to be that the seal is the part that is either most likely to fail or hardest to manufacture or both. But I would have thought that manufacturing a high rpm turbine blade exposed to extreme temperatures and concentrated oxidizers isn't exactly child's play either.

So for the layman, are you saying that they've got a simpler turbine, use more of them, but not twice as many, because the design means you need fewer engines to lift the same payload?

They do avoid a failure mode associated with seals. I'd think modern seals shouldn't leak that much - they don't - but it's still better to have it eliminated in design.

Full-flow scheme has 2 gas generators - with different fuel:oxidizer ratios, and 2 turbines, each driving one pump of the same propellant which constitutes the main flow through the turbine. Staged combustion has 1 gas generator and (usually) 1 turbine with 2 centripetal pumps, so one of the pumps has the flow of the propellant opposite to the main propellant of the turbine. Here is the danger, if seals don't hold.

I'm confused; wouldn't the oxidizer-rich half run hotter? Isn't that challenge the Russians overcame but the US punted on back in the days of the space shuttle?

Your post and the referenced wikipedia article claims the turbines run cooler. Wouldn't that only be true for the fuel-rich side?

> wouldn't the oxidizer-rich half run hotter?

You try running it as hot as possible, but without burning metal. You have about twice as much liquid oxygen by volume than kerosene in a typical engine, but you can heat kerosene to higher temperatures without burning. So... I don't think you'll get higher temperatures with oxygen gas generator?

The hackaday article explicitly states the oxygen-rich preburner exhaust was extremely hot and required metallurgical advances before turbines could survive it.

How is the Raptor's oxygen-rich side any different from what the Russian's were dealing with?

The article left me with the impression that the Raptor is using similarly resilient materials for the oxygen-rich side, with its language describing the US and Russian designs solving the two critical halves of the problem.

There just seems to be a contradiction in the statements surrounding the turbines running cooler.

> The hackaday article explicitly states the oxygen-rich preburner exhaust was extremely hot

I didn't find this exact quote, but the idea is the following: oxygen-rich flow can get hot enough to ignite metal. What Russians did was to find the materials withstanding ignition best and find temperatures which are still safely avoiding ignition. This combination turned out good enough to use oxygen-rich gas generators.

Raptor may be using similar approach. "Two sides of the problem" language is, in my opinion, about the tradition to have fuel-rich turbines in America and oxygen-rich in Russia, and for full flow engine you need both.

The turbine can run cooler, because you don't need to have gases from gas generator that hot in order to get enough power to the pumps. You're getting that power from the fact that you use full flow of both fuel and oxidizer, and since you have that additional source of power, you can relax requirement of the gas generator exhaust temperature.

> but the idea is the following: oxygen-rich flow can get hot enough to ignite metal.

A relevant example of igniting metal being the first Raptor test firing, which was slightly "engine rich". Which we could see by the green given off by burning copper (and later confirmed by Elon, also somehow fixed).

Video: https://twitter.com/elonmusk/status/1092270756715737088


>According to SpaceX founder Elon Musk, this design was explicitly intended to achieve full reusability of all rocket stages and, as a result, "a two order of magnitude reduction in the cost of spaceflight" (wikipedia also)

And they hope to make the oxygen and methane on Mars for the return journey, using CO2 and ice found on there plus energy from solar cells.

It will be impressive if they do any of those.

To model Raptor's hypersonic turbulent combustion SpaceX used an internally developed simulator, which uses wavelet compression to vary resolution across many orders of magnitude in both time and physical dimensions:


Here is a fantastic talk from the NVIDIA conference:


https://www.netflix.com/title/80119093 My rocket-science knowledge is abysmal, but I thoroughly enjoyed this article and the Netflix documentary that I have linked to was incredible. Anyone even remotely interested in rockets should check it out :).

Most of my rocket-science knowledge is from Kerbal Space Program, which I also recommend to anyone with a passing interest in space :)

Unavailable for me in the UK, what's the name of the documentary?


Can't find it anywhere on the internet, not even Netflix here in my country. :(

I cannot access it as well, but it should be "Cosmodrome"

Excellent, thank you. There's a parallel between Sergei Korolev and Elon Musk, I believe.

To amplify just a bit - I watched the linked video. The parallel that struck me was the description of Korolev as an amazingly multitalented person. I regard Musk as a polymath as well. It interested me that these two chose work that requires understanding of many disciplines (fluid dynamics, combustion chemistry, materials science, control systems, etc.)and were so effective as individual contributors and as leaders.

If you live in SoCal (or even if you're just visiting), definitely take a quick trip by the SpaceX facilities in Hawthorne [1] where they have a Falcon rocket sitting outside. From a distance it looks like and industrial chimney, but as you pull up, you can see it's an actual rocket. Standing next to it gives a great sense of scale the next time you're watching a SpaceX video.

1. https://www.google.com/maps/place/SpaceX,+Rocket+Rd,+Hawthor...


Fake? Isn't it the first stage that ever landed!?

It is -- it's B1019. I suspect that's why they are a janitor.


If you provided supporting reasoning or evidence of any kind I believe you'd have better reception. Your unqualified assertion does not resemble the statement of a scientist, and you should not be surprised that relying on an appeal-to-authority of a janitor drew ridicule.

What proof/rational do you have that it's fake?

You're kinda dodging the important point that your original assertion that it's a "fake rocket" is entirely false.


It wouldn't be shocking if they've removed some ITAR-covered parts so no one can sneak up in the middle of the night and steal rocket tech, but it appears to be the actual booster.

What do you mean by 'fake rocket'? A prototype built for display, not for launching?

> While the Space Shuttle has long since retired, a variation of the engine itself will go on to power the Space Launch System. It will be the most powerful rocket NASA has ever built and is slated to begin missions in 2020.

2020? I honestly doubt we'll see the SLS launch before 2024 at it's current rate.

I would bet money that SLS+Orion launches before Falcon Super Heavy+Starship. It looks like SLS is already doing integration testing for the various cores and starting to assemble the main components. I'd guess 2021 at the latest, as long as there's no multi-month government shutdowns in the meantime.

Well, sanity check.

In 2017 the SLS launch was about 2 years out, but likely to slip: https://www.nasaspaceflight.com/2017/11/sls-managers-troops-...

In 2019 the SLS launch is a bit under 2 years out, but likely to slip: https://arstechnica.com/science/2019/02/nasa-still-working-t...

This strongly reminds me of the fact that late software projects are promised to be on time until about 6 weeks before launch, and then launch keeps getting delayed. And this is true no matter how late it eventually turns out to be. The reason why 6 weeks is the magic figure is that for a software project, that's the point where you can no longer paper over the inevitability of failure with wishful thinking. Rockets have slower schedule, but it is strongly looking like 2 years is a similar magic figure in that industry, for similar reasons.

In that light, the money quote from the second article is this:

However, the agency and its prime contractor for the core stage, Boeing, are on a tight timeline that has little margin for technical problems that might occur during the structural tests of the tank or the green run tests. Historically, during this integration and test process with other large rocket programs, major problems have often occurred.

I am generally a believer that what happened historically shouldn't be ignored. There is therefore no way that the SLS will launch in 2020. Or 2021. In fact nobody really knows when it will launch. Furthermore the upper stage, aka Orion, is apparently in even worse shape. Right now they are going back to the drawing board to try to find a design that gets costs down.

Admittedly SpaceX itself is promising the BFR in about 2 years, and also had a history of overruns. I don't think that they will launch on time. But they have a better history of getting launch vehicles up.

I will therefore happily take your bet for $100, but I would like to formalize it a bit. I win if Falcon Super Heavy+Starship or whatever it gets renamed to gets successfully launched to orbit first OR if SLS+Orion gets canceled first. Vice versa you win if SLS+Orion gets successfully launched first OR if Falcon Super Heavy+Starship or whatever gets canceled first. Note that "SpaceX goes out of business" counts as canceled, even if someone else (eg Bezos) buys the remnants and then launches something based on the work.

If my version is acceptable, you can contact me by email per my profile.

That's not quite correct. SLS is still slated to launch June of next year. It may get delayed to 2021 if problems are found. Also, Orion already flown in 2014, and will undergo an abort test in April.

You are right. After poking around for a bit, I found that SLS+Orion can launch without the upper stage. The upper stage allows more to be carried on the launch.

That said, I'm still on the side of betting with history. In most organizations, the deadline is the first date that nobody can (yet) disprove. When a deadline depends on problems not happening on this project that historically have been common, I think it is safe to bet that history will repeat itself.

This goes doubly for the SLS. Which is more ambitious than past launch systems, and is being built so long after the last new launch system was designed by the companies involved that there is little institutional knowledge left about how to do it. (Furthermore building with competing companies contracting for pieces that need to integrate just sounds like a recipe for expensive overruns to me.)

As opposed to the BFR. Which is being designed by a company with more recent experience of how to build new launch systems than the rest of the planet put together.

Most of the structural and integration tests are already complete though. We're far into the "beta testing" phase as it were. Also, the BFR isn't anywhere near the state that the SLS is in right now.

Do you think this is a Lindy effect type of thing, or something different?

I wouldn't have drawn that comparison. But I wouldn't rule it out as a reasonable comparison without some data to point to.

SLS Block I maybe, but definitely not SLS Block 2.

The Starship hopper will definitely be working long before either of those two take flight.

They're nearly finished actually. The core stage is in final assembly. We'll probably see a launch in 2020 or 2021.

The article is a bit incomplete. The space shuttle main engine already had two main turbopumps. IIRC this is because the optimal pump speed is different for pumping hydrogen and oxygen. They have very different densities. You want to avoid gearing as much as possible.

And the main problem problem in the staged combustion cycle - in feeding the gas generator exhaust to the main chamber is not that it's fuel rich or oxidizer rich - it's that it's usually much lower pressure than in the main combustion chamber, because it had to go through the turbine, which causes the pressure to lower.

That's why engines like NK-33 have a separate boost pump for the gas generator. RD-180 has more pump stages for the fuel entering the gas generator (all oxidizer passes through the gas generator).

Full flow staged combustion is another way to solve this - put all the propellants through the gas generators and turbines.

I realize it's hard to write popular technical articles about medium complexity subjects and sometimes you have to take some shortcuts.


The F1 engines of the Saturn V 1st. stage used a rich-mixture gas generator to drive the turbopump, but its exhaust was fed into the engine's nozzle about halfway down, through an annular manifold [1]. Up to that point, the combustion chamber and nozzle were cooled by circulating fuel through them, but beyond that point, the cooler turbopump exhaust layer protected the nozzle extension.

In pictures of launches [2], you can make out the brown smoky annulus of the turbopump exhaust, for a distance about equal to the length of the nozzle extension, until it either mixes with the hot exhaust, or with ambient air and then burns, at which point the smoke particles become incandescent.

[1] https://history.msfc.nasa.gov/saturn_apollo/documents/F-1_En...

[2] https://images.nasa.gov/details-ksc-69pc-442.html

This is an excellent article. Just curious, how much more efficient is the full-flow engine?

And, what exactly is the deal with the seals the article is talking about?

Anyone have more info?

Scott Manley has a fantastic video explaining all of this: https://www.youtube.com/watch?v=4QXZ2RzN_Oo

In terms of efficiency, it's worth mentioning that even tiny gains have big impacts because of the rocket equation. Having a 1% boost the entire time the rocket is running can have an enormous impact in the overall velocity, or reduce the overall amount of fuel needed for the same targets.

A staged combustion engine is intrinsically a lot more efficient than an engine where you dump some of your fuel to power the pumps, like in a tapoff or gas-generator engine.

Compared to a oxygen-rich full flow you can reach a higher pressure while having better safety margins which will allow you to re-use the engine. The higher your chamber pressure the closer the rocket's thrust at sea level is to its maximum thrust in vacuum. Also higher pressure tends to correlate with higher thrust to weight ratios.

Also, having both inputs already be gasses mean they mix better. This will have a tiny benefit in them burning more completely. But more importantly they'll burn better across a wider range of thrusts, which could be important for throttling down the engine for landing.

> Also, having both inputs already be gasses mean they mix better. This will have a tiny benefit in them burning more completely. But more importantly they'll burn better across a wider range of thrusts, which could be important for throttling down the engine for landing.

Also, and importantly for the design of the engine, it's much much easier* to computationally model the mixing of two gases, to ensure complete mixing and no combustion instability (especially when the reaction components are limited to ch4, o2, and partial- and complete-reaction products like c02, c0, h2o, oh-, ...).

* by much easier, i mean less impossible

Have you seen SpaceX's talk at some NVidia conference about their homegrown Computational Fluid Dynamics simulator? It was really interesting.


I hadnt seen that exact article, but i recall reading (or listening to a lecture) that mentions how one advantage of methane is the possibility to use CFD in design, while that was still outside the realm of possibility with RP-1 due to complexity of reactants (differing phases and many compounds)

Liquid-liquid mixes extremely well. The hard part is droplet atomization, not mixing. I’m not knowledgeable about gas-gas injectors but I find it hard to believe that you’d be able to easily make them mix well.

Wikipedia has some stats comparing their current engines (probably speculative for the raptor)

Raptor: Thrust 1900 KN / 330 specific impulse

Merlin 1D: 480KN / 275 s.i.

i can't find the weight of the Raptor. Merlin is 630Kg

To put this in perspective, the F-1 engine of the Saturn V (one of the biggest rocket engines ever made) only put out about 3.5 times more thrust (6.7 MN or 1.5 Mlbf) than the Raptor (1.9 MN or 0.43 Mlbf).

Also some people might not realize how much liquid is being pumped by rocket engines (the Saturn V F-1 could drain a 30,000 gallon swimming pool in 10 seconds):

Each second, a single F-1 burned 5,683 pounds (2,578 kg) of oxidizer and fuel: 3,945 lb (1,789 kg) of liquid oxygen and 1,738 lb (788 kg) of RP-1, generating 1,500,000 lbf (6.7 MN; 680 tf) of thrust. This equated to a flow rate of 671.4 US gal (2,542 l) per second; 413.5 US gal (1,565 l) of LOX and 257.9 US gal (976 l) of RP-1. During their two and a half minutes of operation, the five F-1s propelled the Saturn V vehicle to a height of 42 miles (222,000 ft; 68 km) and a speed of 6,164 mph (9,920 km/h). The combined flow rate of the five F-1s in the Saturn V was 3,357 US gal (12,710 l) per second,[4] or 28,415 lb (12,890 kg). Each F-1 engine had more thrust than three Space Shuttle Main Engines combined.[5]


So the methane-oxygen Raptor at 380 SI is almost as powerful as the hydrogen-oxygen Space Shuttle SLS at 453 SI, if I have the math right:


The tradeoff is reduced price and improved safety with the choice of methane over hydrogen, but with a reduced specific impulse.

Another tradeoff is size. You can pack about 4 raptors in the same area you can fit a single SSME. (Or, 8 in the space you can fit a single F-1). This is relevant because to lift the shuttle on 3 SSMEs, or the SLS on 4, they need to pack in massive amount of thrust in SRBs.

Instead of doing that with all it's cost and safety implications, SpaceX intends to put 31 raptors on the bottom of their rocket. You could not fit the same thrust in other major US engines without making the rocket comically large.

And, with 31 engines instead of 4, you get a massive increase in redundancy. If you lose one SLS SSME, you can push the others extra hard (they won't be reused anyway) and still make it to orbit (or land in the Atlantic). You'd need to lose 7 Raptors to be in the same bad position.

I wonder how nobody talks about ill fate of N-1 with 30 engines on the first stage with the idea that the number of engines was the reason for breakdowns.

N-1 steered by differential thrust, meaning every engine points directly backwards and losing one on the rim means you have to turn off another on the opposite side to compensate, and now you've lost all the margin there is in the design and quite a lot of control authority too. All SpaceX rockets steer by gimbaling engines, meaning you lose one on the rim and you just adjust the thrust vectors to compensate.

Apparently the N-1 never actually fired all 30 engines simultaneously before launch, so there was no way to detect plumbing issues.

Additionally, their control scheme was designed to shut down the engine opposite to any failed engine, meaning that they had very little engine-out margin in the inevitability of an engine failure.

All in all, we can (and have, with the Falcon Heavy) do much better.

> meaning that they had very little engine-out margin

N-1 could complete the mission with up to 4 engines turned off from the start. And I guess it it's not from the start, you could turn off even more engines and still have a successful flight.

Even though Saturn-V is heavier, N-1 has more lift-off thrust. Partially to have this kind of redundancy.

The relevant tech - computer control, metallurgy, etc. - has come a ways since the mid 1960s.

SpaceX's lots-of-engines approach has been well proven on the Falcon 9 thus far.

That's fine, but it means you can't blame the sheer number of engines on N-1 to its poor performance. There is no law which forbids launching, e.g. Soyuz with 32 working from the start chambers, or Energiya with 20 chambers, or Falcon Heavy with 27 chambers successfully. It's other reasons - maybe computer control (KORD was a complex system to create), maybe metallurgy (even though NK-33, manufactured in ~1973, manage to fly after 40+ years in storage), maybe something else.

This is just to compare optimism for heavily multi-engined BFR with suggestions that N-1 was doomed just because of the number of engines.

As I recall, nobody could stop talking about it in the weeks before the Falcon Heavy demo.

Methane also means they can refuel in-situ on Mars, provided the infrastructure is there.

So, for the folks like me who don't know anything about rockets, it seems like "specific impulse" is the measure of how much "impulse" is generated per unit of fuel.

So 330/275 = 20% better.

Most of that is actually because of the difference in fuel -- Raptor runs on Methane, which produces more H2O and less CO2 than RP-1, and H2O is a smaller molecule and therefore more efficient.

Yep. The actual physically relevant number is exhaust velocity, which is the mean velocity of the particles in the rocket exhaust.

Isp = Specific Impulse = exhaust velocity/(9.81ms^-2).

9.81ms^-2 there is not any actual acceleration, just an agreed conversion factor. The unit of Isp has no physical significance, the convention just arose because early rocketry was a collaboration of people who all used different units, and they wanted to be able to compare numbers. Since the unit they all shared was a second, and they all agreed (roughly) on the real value of g, they just divided exhaust velocity by their g to get a number that could be compared. The convention stuck.

Note that when using different fuel or oxidizer, specific impulse isn’t the only thing that matters. Density matters too due to its effect on non-payload mass ratio.

> 9.81ms^-2 there is not any actual acceleration, just an agreed conversion factor.

I'd look it up but my textbooks[1] are in boxes, but the units everyone uses are wrong. Actually inappropriately simplified.

[1] I have one and exactly one textbook that has the full units for gc.

Isp often defined as "thrust per unit of mass flow". That is, how many Newtons of thrust engine gets from each kilogram per second of propellant spent. Since N = kg * m / s^2 , and mass flow is in kg / s , Isp becomes (kg * m * s) / (s^2 * kg) = m/s , i.e. Isp has units of speed. In vacuum Isp is equal to the speed of gases flying from the engine.

>So 330/275 = 20% better.

due to the exponential nature of the rocket equation, 100 tons of fuel will result in 2 tons to LEO in the former case and just 1 ton in the later.

A full-flow staged combustion engine has more efficiency due to both propellants being fully vaporized before they mix, rather than a liquid-liquid mixture (as in a gas generator engine) or a liquid-gas mixture (as in a normal staged combustion engine).

The seals are between the turbine side (hot gas) and compressor side (liquid fuel) of the turbopump. In a full-flow staged combustion engine, you don't have to worry about accidentally getting a little of the hot gas into the liquid propellant, because that would be mixing fuel with fuel-rich gas or oxygen with oxygen-rich gas. But in a normal staged combustion engine, you have to carefully seal off the oxygen from the fuel-rich gas (in a fuel-rich staged combustion engine) or the oxygen-rich gas from the fuel (in an oxygen-rich staged combustion engine.

No, this is all just wrong. The efficiency is because you dont throw a load of enthalpy overboard as in a gas generator. The phase is _exremely_ second order by comparison.

You’re all over this thread being wrong and confused.

Source: am a rocket engine designer.

I'm not clear what the efficiency gains are here at all over a SSME style staged combustion cycle. In both designs all propellants go through the chamber. Comparing to a gas generator cycle there are gains, sure.

It seems like the major win here is running oxygen rich on the oxygen side and fuel rich on the fuel side to reduce the chance of a leak destroying the whole vehicle.

Am I missing something?

> I'm not clear what the efficiency gains are here at all over a SSME style staged combustion cycle.

You have higher pressure in the chamber with full-flow approach. Since the mass flow is the same, and Isp is higher, thanks to higher pressure, you're winning.

And you have higher pressure because you supply more power to pumps.

Reducing risks to destroy the vehicle due to leaks is, I agree, very secondary benefit. May be even tertiary - how about possibility to optimize turbine frequencies for both propellant flows without losses on gears?

But... you don't.

The turbines in any kind of staged or full flow system run at a pressure higher than the chamber is at... I don't remember the numbers offhand but I want to say the SSME's preburners were at 8000ish psi, with a drop to 4000ish through the turbines, and finally to 3000 in the main combustion chamber. The SSME was only partially preburned, and so the flow through the turbine has to expand over a higher pressure ratio, but I still don't see a fundamental difference.

Let me explain. The principal advantage of full flow system over regular staged combustion is higher Isp of the engine.

Higher Isp of the engine is achieved because the pressure in the chamber of full flow engine is higher, than chamber pressure of staged combustion engine.

The pressure in the chamber is higher because pumps produce higher pressure.

Pumps can produce higher pressure, because with full flow pumps have more power driving them than with staged combustion. Power here is roughly equal to volume flow (m^3 / s) multiplied by pressure drop (N / m^2). Full flow scheme gives you more volume flow, since all the propellant is used to turn turbines.

So even though all the propellant in both cases is burned in the chamber, Isp is higher with full flow scheme. Rocket engine maximizes momentum of the exhaust, not the energy, which is conserved.

I was under the impression that the question was regarding FFSC versus non-FF staged combustion.

You say the main benefit is just increased pressure?

does the mass of the exhaust fuels count as well? whats the percentage of thrust being lost in a gas generator?

It definitely counts. The effect it has on exhaust velocity is maybe 5-10%, which is a big deal when you consider that has an exponential effect on your mass fraction, i.e the percentage of total lift-off mass that is fuel vs machinery and payload.

Scott Manley has done a few videos (youtube) recently around rocket architecture; including several videos about the Raptor engine. Well-worth checking out. (I don't have a link at the moment, sorry.)

Scott Manley's channel : https://www.youtube.com/watch?v=Sdwy9fzQzl4

This guy has some simple short videos on rocketry. this one is on combustion cycles


I think this is about safety. You need two pumps, one for fuel, one for oxygen. The compressor side of the pump will be full of whatever it is you're pumping. But if you use a single type of preburner, then the turbines of both pumps are full of either hot fuel gas, or hot oxygen gas. That means you have one pump where you have hot fuel or oxygen right next to cold oxygen or fuel. If the seals leak, the pump blows up.

In short: you accept a higher failure rate (twice as many pumps) to get a lower explosive failure rate (much less explosive pumps). A very good tradeoff when building a rocket with a high number of engines and some redundancy which does not help against explosions. It gets even better when the pumps turn out to be more reliable individually due to relative simplicity.

Rocket lab's Rutherford engine uses a closed cycle with battery powered fuel pumps. They were actually to fly an engine like this in their "electron" rocket, as early as 2017. https://en.m.wikipedia.org/wiki/Rutherford_(rocket_engine)

This is only viable for small engines, like expander cycle is only viable for small engines. The specific energy of electric batteries cannot compare to that of kerolox or methalox.

Closed cycle is not a new idea: https://en.wikipedia.org/wiki/Staged_combustion_cycle#Histor...

I still have a soft spot for expander cycles - they seem so elegant compared to staged combustion cycles, and apparently i am not the only person who thinks that:


Interestingly the expander bleed cycle mentioned at the end is going to be used in the Blue Origin BE-3U, a change from the combustion tap-off of the older BE-3.

Excellent read, thank you.

"For example, the turbine of the V-2 rocket was spun with steam ..."

Steam technology and rocket technology have a shared history.

That kind of bottleneck shape that a is quintessential shape of a rocket engine, is actually a Steam-Engine-era technology called a de Laval nozzle.

I didn't know they used a tiny steam engine inside a V-2.

I think it's cool that the design of such an old technology, the steam engine, lives on inside the design of such a new technology, the rocket.

>I didn't know they used a tiny steam engine inside a V-2.

it wasnt steam engine per.se., it was catalytic peroxide decomposition.

The Soyuz first stage uses the same system even today.

Love this article and if you read it with Curious Droids' voice it's even better: https://www.youtube.com/channel/UC726J5A0LLFRxQ0SZqr2mYQ

Either approach, whether it recaptures the oxidizer or fuel rich preburner exhaust, is clearly an improvement over dumping everything overboard. But neither is an ideal solution as there’s still potentially combustible products being wasted.

Are there? Here's a diagram of the RD-180, which uses an oxygen-rich preburner:


Where is anything escaping other than through the combustion chamber?

There's no wasted propellant with normal staged combustion.

But full flow staged combustion lets you fully vaporize both propellants before they mix, leading to more optimal burning.

More importantly, supply more power to the pumps thus allowing higher pressure in the chamber, and so higher Isp.

Scroll one paragraph down from that quote and the RD-180's approach is mentioned.

> In the 1950’s, Soviet scientists came up with something of a compromise. Instead of using a fuel-rich mixture in the preburner which produced an exhaust that couldn’t be safely recirculated into the engine, they experimented with running the preburner oxygen-rich.

> Eventually the Soviets mastered the metallurgy required to build the turbine and developed several engines that operated on the oxygen-rich preburner concept. The exhaust was piped into the engine’s combustion chamber and recovered at least some of the propellants which would otherwise have been dumped overboard. The modern day Russian RD-180 engine, currently in use by the American Atlas V, is a continuation of this technology.

The diagram doesn't show that the fuel used in the preburner is combusted less efficiently than the fuel used in the main combustion chamber. It comes out through the nozzle, but isn't fully combusted (it comes out as liquid, while the main fuels are gasses) and so is wasted energy.

From what I understand, the fuel and oxidizer that has been through the preburners to spin the turbines is basically fungible with fuel and oxidizer that is used in the combustion chamber.

I'm not sure where you're getting that.

If it's incompletely burned in the preburner, it gets completely burned in the combustion chamber.

And even if that were true that combustion happening in the preburner was wasted, then full-flow staged combustion would be equally bad because it still has preburners.

As only a casual fan, this was revealing on what makes a full-flow staged combustion engine so promising. Thanks for sharing

Looking at the list of cycles on Wikipedia, I'm surprised that nobody seems to have used preburners to pressurize the tanks. Use a fuel-rich preburner to pressurize the fuel tank, and an oxygen-rich preburner to pressurize the oxygen tank.

Mixing should be limited even if nothing special is done, due to the temperature and phase of matter and short timeframe. One could of course pay the weight penalty of a piston (need not have a perfect seal) or collapsing bag.

Doing a heat exchanger (to boil and thus pressurize) is another option, but then you're back to needing a place for the exhaust. It would let you do a sort of full-flow engine without turbopumps however, which is great. All those issues with cavitation and lubrication and stress cracking just go away.

> I'm surprised that nobody seems to have used preburners to pressurize the tanks

This combines the worst of both worlds. You get the weight and complexity of turbopumps. And you get the weight, explosion risk and leakiness of pressure vessels.

Skip the turbopump. That is the whole point of pressurizing.

Pressurizing has weight advantages. You can use a balloon tank. The tank no longer has to have the rigidity to support itself.

Rockets with typical cycles have used balloon tanks. Even modern ones like the Falcon 9 are partially that way, with just enough rigidity to be erected empty on the pad. The Falcon 9 uses helium to pressurize; that could be changed to preburner exhaust even if you did keep the moderate pressure and the turbopumps.

I think you're underestimating the pressures involved here by several orders of magnitude.

To drive the flow of fuel and oxidizer into the combustion chamber, you must overcome the pressure inside. To drive this flow, the turbopump output pressure will be of comparable magnitude to the combustion chamber. This pressure is in the realm of a few 10s of megapascals, or a few thousand PSI. This is a hard design requirement.

If you're proposing to "skip the turbopump", then necessarily you'd have to pressurize the entire fuel and oxidizer tanks to these high pressures. This is, to put it very mildly, utterly infeasible.

This is one of the reasons SpaceX uses supercooled fuel and oxydizer (and one of the reasons for going methalox in the first place) - the tanks self-pressurize, so they can skip the Helium system.

Pumps are still good because you really don't want the tanks to pressurize at the same pressure as the combustion chamber. If you do, you'll see a rapid disassembly, but your claim on it being unscheduled will be questioned.

This is completely, completely wrong. Really honestly confused nonesense. Supercooling lowers the ullage pressure, if anything increasing the need to pressurise (self or otherwise), because the self pressurisiation (the vapour pressure) is lower. So you have to do additional work to feed the pumps. The only reason to supercool is to increase the density. Your comment is quite incorrect.

I really appreciate your expertise and that you're taking the time to correct the sometimes overeager amateurs in this thread. But can I suggest you tone down the unnecessary invectives? It's true that there ought to be some social penalty paid when people give confident wrong answers -- it's a pet peeve of mine -- but I think it's sufficient to just say something like "Your comment is incorrect. Here's why, in detail. Please don't project so much false confidence next time." That still stings for the other person to hear, as it should, but it leaves them much more likely to want to learn more. And it's much less likely to escalate into a fight that distracts from the main subject.

All fair, valid and noted. It’s the eve of Opportunity’s demise, and i worked on mars EDL, and the infinite tide of javascripters making statesmanly-yet-quite-mistaken proclamations on physics can feel quite disrespectful to the actual engineers who work on this stuff now, worked on the stuff on both sides in the cold war, and everyone back to goddard, tchiolkovsky and moore. But you are right. More whisky and no more HN.

> nobody seems to have used preburners to pressurize the tanks

IIRC, Block D taps some preburner exhaust for that...

Nice article, but this part is a bit misleading:

> American engineers went in the opposite direction. They believed that a fuel-rich mixture in the preburner was possible and could be done with existing metal alloys, so long as hydrogen was used as the fuel instead of kerosene. This ultimately lead to the development of the Space Shuttle Main Engine, which to date remains the most efficient liquid fuel rocket engine ever flown.

SSME performance was due to H2 vs kerosene, it was not a full flow engine.

Edit; also no mention of Blue Origins BE-4 which is also a full flow engine.

The article does not claim that the SSME is full flow; as described in the quote you pulled, it's fuel-rich.

BE-4 is not full flow, but oxidizer-rich (https://en.wikipedia.org/wiki/Staged_combustion_cycle#Oxidiz...).

Not true. Be-4 is a staged combustion metholox engine.

Be careful with terminology. BE-4 and Raptor are both staged combustion methalox engines. Raptor is full-flow staged combustion, while BE-4 is oxidizer-rich staged combustion.

The reason this is important is not the really the few % savings in fuel weight.

Any small improvement in exhaust velocity of the engine makes a huge difference(sort of exponential) in the amount of payload it can take to orbit. In this case the 2 pre-burners also make relighting the engine in a vacuum a lot more reliable.

For more on the math: https://en.wikipedia.org/wiki/Tsiolkovsky_rocket_equation

Right, another reason to use full-flow scheme is that both components get into chamber in the gaseous form - the heated gaseous form. And fuel usually doesn't have a problem igniting in the hot oxygen flow, especially if that fuel is in gas form itself, within milliseconds.

The RD-270 got me scared. That much hydrazine flowing through anything can't be good.

its not impossible, considering that both the US and the russians built them (but never flown). These articles usually go overboard with the hero-worship and fail to mention that those are incremental improvements on the immense rocketry feats of the 60s.

I don't like the usage of 'impossible'. Impossible by what standard?

Laws of physics forbidding it is understandable. We may likely never see the development of FTL because our understanding as we know it would see it impossible to do due to its strange implication about cause and effect.

someone didn't read the article, or notice the quotes around the word "Impossible"

There wasn't quotes around 'impossible' in the article.

Literally in the headline.

And they used other words next to the unquoted impossible in the article itself like "deemed by scientists to be next to impossible". Context matters.

The US scientists didn't think there was a metallurgy that could survive the oxygen rich environment, but the Russians did the work to find it.

The part I'm confused about is how it talks about all of these 60s rockets that close the cycle but aren't as efficient because they mess up the mixture in the combustion chamber, but shouldn't the injectors be tuned for the excess fuel/oxidizer from the get go? It doesn't seem like you should be exhausting unburnt fuel/oxidizer if you have it tuned correctly.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact