The vacuum optimized Raptor: No attempt at an expander-bleed cycle

[Benjies]

TL;DR Summary

Photos of SpaceX's vacuum-optimized Raptor being prepared for tests soon to come confirm an interesting point: ~380s Isp is good enough for them.

Hello all,

The merits of the full-flow cycle are without question and have been fielded with Starship tests to the point of my complete buy-in into Raptor's capability as a launch engine. Elon Musk loosely discusses further refinement of their vacuum-optimized engine that they have created, which he claims should bring the specific impulse of the engine to about 380 seconds in vacuum.

My question that I hope facilitates discussion and learning (certainly for myself as well) is what might be stopping SpaceX from honing R&D into an expander bleed cycle for their upper-stage engine. I pose this point because, while the specific impulse of the launch-engine Raptors is great, a 380s Isp engine falls well below the mark of expander-centric cycles used in other rockets.

In many respects, the idea of developing an entirely new engine for SpaceX's upper-stage, is a stretch that SpaceX altogether may not need to take at all (especially with methane, where all Expander cycles being ran today are on Hydrogen due to an extreme difference in specific heat). But the specific impulse of this expander bleed would not be a two-or-three second improvement on specific impulse compared to the vacuum Raptor, but potentially tens of seconds (in theory, see reference for postulations on Isp of this cycle already sitting at 360s). SpaceX already tackled the full-flow cycle, and I wouldn't put it past them to be able to make an expander-bleed work with Methane to a point of extreme specific impulse improvement (Yes, I am aware of the extreme loss of specific heat when comparing methane to hydrogen).

See the following paper written by European engineers for EUCASS in 2017 regarding general inquiry into the Expander Bleed Methane cycle:
https://www.eucass.eu/doi/EUCASS2017-332.pdfSO- I'm creating this post to hear what you might know, think, or have to say about the idea of this engine being developed:
Reasons why methane simply would not facilitate the expander cycle, even if run in a bleed configuration on the TPA
Reasons why the Raptor engine meets the mark and is the superior candidate from an engineering standpoint
Interesting facts and background regarding past research on expander cycle engines, methane combustion characteristics, etc.
Other?

And of course, feel free to ask me for follow-up on this post! I hope I haven't thrown too much at you all at once, and would love to educate if any part of this post makes you want to "raise your hand", so to speak.

Mods/Admins, as this is my first post, please feel free to inform me if I am stepping out of line by requesting such general feedback and discussion. I've simply been scratching my head and talking to myself about reasons why SpaceX may not be pursuing this, and looking to discuss with others. Thanks!

 

[jedishrfu]

At PF discussion usually doesn’t work this way. Someone starts a thread asking a question and others respond. In your first post, its more like you’ve setup a committee to discuss some topic complete with papers to read and are now waiting for folks to join the conversation.

Take a look at other PF threads here and see how they evolve. In general, students will post something that is causing them some difficulty in a STEM subject and our members respond with helpful hints.

 

[mfb]

Hydrolox (hydrogen+oxygen) expander cycle engines don't have a higher I_sp because of the cycle, they have a higher I_sp purely because they use hydrolox.

If you look at comparison tables like the one on Wikipedia, Raptor has the highest I_sp of any engine that's not using hydrolox.
There is a methalox (methane+oxygen) expander cycle engine in that list - at 360 s. Your reference discusses lower I_sp values (see e.g. Figure 3 and table 3), too.

The main reason SpaceX uses a full flow stages combustion cycle is reusability. You have more mass flow, so things stay cooler, which makes it easier for the engine to survive a flight undamaged.
Everyday Astronaut has a detailed discussion.

 

[Benjies]

Hydrolox (hydrogen+oxygen) expander cycle engines don't have a higher I_sp because of the cycle, they have a higher I_sp purely because they use hydrolox.

The main reason SpaceX uses a full flow stages combustion cycle is reusability. You have more mass flow, so things stay cooler, which makes it easier for the engine to survive a flight undamaged.
Everyday Astronaut has a detailed discussion.

Oh wow, I hadn't even realized a methalox expander cycle had even been developed for the Vega- closed cycle too. Thanks for this reference. I had assumed the theoretical Isp loss from the engine was due to them bleeding the TPA mass flow- this is a perspective changer! I have always viewed cycle first and propellant next when trying to gage a 'hypothetical' Isp for an engine.

The main reason SpaceX uses a full flow stages combustion cycle is reusability. You have more mass flow, so things stay cooler, which makes it easier for the engine to survive a flight undamaged.
Everyday Astronaut has a detailed discussion.

It's odd to find a staged combustion cycle with a higher Isp than an expander cycle, even if it's full flow. But I suppose I had shelved some of the merits Everyday Astronaut mentions in that article. The increased turbine mass flow is something I hadn't considered- wanting to displace much of the temperature and enthalpy change to the MCC and just minimize the actual combustion happening in the preburners- which they can now do because they're dealing with so much flow (fun fact, in my rocket propulsion class, my model for the Raptor engine found an O/F of ~50 for the ox-rich preburner!). I'd be happy to share the model if you'd like (MATLAB), but take it with a grain of salt, of course.

 

[mfb]

Sort the table I linked by I_sp and see how well it separates the fuel/oxidizer combinations. That determines the maximal I_sp simply from the available energy.
All hydrolox engines have a much higher I_sp than everything else. The second group is methalox, closely followed by RP-1/LOX and then the last group is N2O4/UDMH. A bit of overlap between the groups because engines are different and the table collects both sea level an vacuum performance.

 

[Benjies]

Sort the table I linked by I_sp and see how well it separates the fuel/oxidizer combinations. That determines the maximal I_sp simply from the available energy.

Yep, clear as day. Thanks again for that one.

 

[cjl]

Isn't Raptor far too large for an expander cycle? I'll admit it's been a while since I dug into the details, but I thought the Vinci at ~40klbf was about as big as you could make an expander, although an expander bleed cycle does let you push that up a bit since you have less turbine backpressure. I still can't find an expander bleed engine bigger than ~150klbf (the BE-3U), and with the bleed, you give up some of the efficiency advantage that you get from the closed expander cycle anyways. I'm not sure you could scale the bleed cycle up to the >500klbf range very well, plus (and again, all this is based on some off the cuff guesses, and it's been a while since I've really looked at rockets in detail) I'd imagine as you do scale it up, your turbine mass flow would have to increase (as a percentage of overall massflow), further harming efficiency. My guess is that an expander bleed is already barely better than a FFSC like the Raptor already uses, and it might even be worse at that scale.

(For those curious about why you can't scale an expander up very well, it's because an expander cycle flows fuel or oxidizer around the nozzle and combustion chamber, which both cools the engine and boils and superheats the fuel. It then uses this superheated gaseous fuel to drive a turbine, which runs the fuel and oxidizer pumps. The discharge of this turbine and the discharge of the oxidizer pump both dump into the combustion chamber, where it is all burned. This is incredibly efficient, since it basically uses the waste heat from cooling the engine to drive the pumps, but the amount of heat you can extract is based on the surface area of the nozzle and combustion chamber, which doesn't increase as fast as fuel requirements do as you scale an engine up.)

 

[Benjies]

Isn't Raptor far too large for an expander cycle?

Definitely way too large. No expander cycle with hit ~2MN thrust. I cite the expander cycle for its efficiency and likelihood of a high thrust/weight ratio and Isp, which is ultimately what we're looking for. Thanks again to mfb, however, for checking me on that assumption that the expander would have the superior Isp, as Methane seems to be the larger limitation to Isp rather than the cycle.

My guess is that an expander bleed is already barely better than a FFSC like the Raptor already uses, and it might even be worse at that scale.

Yep, right on! See previous comment.

(For those curious about why you can't scale an expander up very well, it's because an expander cycle flows fuel or oxidizer around the nozzle and combustion chamber, which both cools the engine and boils and superheats the fuel. It then uses this superheated gaseous fuel to drive a turbine, which runs the fuel and oxidizer pumps. The discharge of this turbine and the discharge of the oxidizer pump both dump into the combustion chamber, where it is all burned. This is incredibly efficient, since it basically uses the waste heat from cooling the engine to drive the pumps, but the amount of heat you can extract is based on the surface area of the nozzle and combustion chamber, which doesn't increase as fast as fuel requirements do as you scale an engine up.)

Yes, the Expander cycle is "thrust limited". I'd mentioned in my original post that the investigation into this cycle would have been for an upper-stage application where thrust reqs are lower. That being said, the upper-stage Raptor is still over 2MN, so their thrust reqs for Starship may, once again, simply make an expander cycle non-feasible. Unless, they chose to mount multiple engines on their upper stage... Which I believe they had already been testing with three raptors at sea level with prototypes "SN-X"... So that does bring things full circle, to some extent, as to why I brought up my original question: Since mission reqs are necessitating three raptors for the upper stage, why hadn't SpaceX investigated an expander-bleed cycle for returns on Isp?

Once again refer to mfb's informative response as to why the full-flow cycle is actually beating what they might be able to get with Methane in an expander cycle. Ultimately having to get upperstage ignition on fifteen expander engines rather than just three raptors which are staged combustion would be another laughably hard feat!

 

[mfb]

They already have six ~2 MN Raptors on the upper stage, three sea level engines and three vacuum-optimized ones. If you go to 700 kN (BE-3U thrust) you need 18 engines. That is not impossible, but it increases complexity. In addition it means development of another engine.

 

[Benjies]

No doubt. The Raptor does all it needs to right now on the upper stage, and efficiently. I suppose at this point my curiosity lays with the Isp of an expander-centric cycle with methane- and if it's even feasible to obtain higher than 380s Isp with an Expander cycle. I mean, hell; what sort of Isp's would we see with a hydrogen full-flow engine?

 

[mfb]

I still don't see why it should be higher. The one methalox expander cycle engine we have as comparison has 362 s, or ~15-20 s lower. Both numbers in vacuum with vacuum-optimized engines.

There is no full flow staged combustion hydrolox engine we could use for a comparison.

 

[Benjies]

After your initial response I'd taken back my assumption that the Expander would find higher Isp, but still don't consider it entirely unfeasible that an expander could match or exceed this 380s Isp. If I've missed some research or perspective that outlines why a full flow will simply have higher Isp, in theory, in all use cases with any propellant, I would love to read up on that.

I only asked what the Isp of a hydrogen full flow engine would be as curious question, I understand none exist. The curiosity stems from the Isp difference we're seeing with the methane expander vs. the methane full flow. If hydrogen were to act in a similar manner then a FF hydrogen engine might touch on 500s assuming the full flow cycle will exceed the closed expander.