totm apr 2023 Engine plumes (Split from AMA questions)

[RocketRockington]

On 3/18/2023 at 7:15 PM, j12sfgd23 said:

 

What was the reason for this switch? Is this an aesthetic decision? Or have the kerbals just developed a different type of engine than we have here on earth? No judgement either way, I am just curious

Just FYI your screenshot looks like it was taken on the Mun.  As an aerospace engineer, you should know plume expansion changes based on ambient pressure - an underexpanded exhaust at ASL will not be underexpanded in a vacuum - in fact all engines in a vacuum will be overexpanded, as you will never have an infinitely-long nozzle.

 

——————-

MODERATOR NOTE:

This thread was split off from:

 

Edited March 22, 2023 by Gargamel
Added mod note

[j12sfgd23]

22 hours ago, RocketRockington said:

Just FYI your screenshot looks like it was taken on the Mun.  As an aerospace engineer, you should know plume expansion changes based on ambient pressure - an underexpanded exhaust at ASL will not be underexpanded in a vacuum - in fact all engines in a vacuum will be overexpanded, as you will never have an infinitely-long nozzle.

Hey there!

Yes, you are (almost) correct! Plume expansion is almost solely a function of nozzle geometry (fixed) and ambient pressure (dynamic), and so for any given engine the plume will expand and change its efficiency strictly as a function of pressure, assuming ideal flow and no reverse flow shenanigans. That being said, I think you may have your terminology wrong, and not quite understand the fundamentals, but dw its a super common mistake so I'll explain it here! The term "Underexpanded", "Overexpanded" or "perfectly expanded" is actually somewhat deceptive compared to the actual shape you see: When a fluid flows through a bell nozzle like a rocket engine, as the ExitArea(AE)/ThroatArea (AT) increases the velocity of the gas rises with it. This is why sea level engines tend to have smaller AE/AT than vacuum optimized engines: for a given throat area, a bigger exit area means a faster exhaust which means higher efficiency.

Unfortunately, as this velocity rises, the internal pressure of the gas actually begins to drop off rapidly. If the internal pressure of the flow matches the surrounding (ambient) pressure, say at sea level, then we call it "perfectly expanded" because it is at exactly the right pressure to be ejected in a perfectly straight line out of the back of the engine. However, if the engine bell is too large/small, things get wonky. If the engine bell is too large, the internal gas pressure drops below the surrounding ambient pressure (IE, the gas has "expanded" too much), which causes the surrounding atmosphere pushes inwards towards the flow and causes the classic "shock diamond" effect" seen in an overexpanded flow.

Conversely, if the engine bell is too small, the internal pressure of the gas will be considerably higher than the surrounding atmosphere, which will cause the exhaust to push "outwards" and create that classic expansion plume I attached earlier. This is exactly why plumes go from over-> perfect-> under when you see a rocket rise through the atmosphere; as the pressure drops, the relationship between the internal pressure (Pe) and the ambient pressure (Pa) goes from Pe<Pa to Pe=Pa to Pe>Pa. And yes, since the engine cannot ever be infinitely long, it is nigh impossible to ever get Pe = 0 in space with a conventional bell, so it will always have a bit of a parabolic plume. That being said, this scenario where Pe>Pa is called "under expansion" because it is not expanded *enough* to eject its fuel optimally.

As you can probably tell its a very weird concept, but TLDR an "Under-Expanded Nozzle" has a higher exit pressure than its surroundings, causing that outward expansion you saw in my post. And while yes, that effect will be dramatized further in space, that doesnt change the fact that it will still be under expanded in a vacuum. (Not to be rude, just trying to be helpful) So, I was correct in calling it under expanded at sea level in my previous post, as the engine they had shown was clearly underexpanded.

Just as a heads up, I take my profession very seriously, and while im sure you meant no offense, starting a comment with "As an aerospace engineer, you should know..." could comes across as quite demeaning. I am sure you didn't mean it in that way though, so please if you have any other questions about basically anything else rocketry related please let me know!! I love talking about this kind of thing, and would be happy to answer any questions you may have!

Edited March 21, 2023 by j12sfgd23

[RocketRockington]

1 hour ago, j12sfgd23 said:

Hey there!

Yes, you are (almost) correct!

*snip*

Ok yes I had the terms backward.  Thanks for the correction, my mistake. 

But then what was your point in your original post?  Showing a picture of the plume expanding outward on the Mun is expected behaviour.  Or was that not the Mun?  It didn't look like Kerbin or another planet with an atmosphere.

[j12sfgd23]

Just now, RocketRockington said:

Ok yes I had the terms backward.  Thanks for the correction, my mistake. 

But then what was your point in your original post?  Showing a picture of the plume expanding outward on the Mun is expected behaviour.  Or was that not the Mun?  It didn't look like Kerbin or another planet with an atmosphere.

the point of that second post was to show that no, the expected behavior for an underexpanded engine is *not* what we see in the game. While yes, it does expand the lower the surrounding pressure gets, the geometry/shape should not change entirely:

as you can see here, the plume shape IRL is parabolic, and while yes it does expand as it gets further from the engine, this "rate" of expansion should slows down to create that parabolic arch (shown in the bottom left). Instead, it seems to expand at an increasing rate, creating that wide spread hyperbolic shape, shown below on the right (Hopefully that makes sense)

Interestingly, the new shape is not far off when youre talking about landing a rocket, as the flow would probably behave in a similar way if the engine bell is close enough to the surface for ground effect to become an issue. Unfortunately that is not what we are seeing here, as this same hyperbolic effect can be seen high in orbit too: 

- Matt Lowne's most recent video, since I dont want to open the game rn

the "point" I guess of my original post was to ask what their reasoning behind this was. I personally find this new plume beautiful, but it gives me a bit of an uncanny valley vibe as its just not quite right

[Sea_Kerman]

Isn’t that parabolic shape due to the plume equalizing with the atmosphere and thus not expanding more? In a vacuum there’s nothing to slow the expansion so I’d expect a cone.

actually, since I assume the plume has some internal pressure, wouldn’t that cause the particles to accelerate away from each other and thus make that hyperbolic shape?

[Periple]

1 hour ago, j12sfgd23 said:

as you can see here, the plume shape IRL is parabolic, and while yes it does expand as it gets further from the engine, this "rate" of expansion should slows down to create that parabolic arch (shown in the bottom left). Instead, it seems to expand at an increasing rate, creating that wide spread hyperbolic shape, shown below on the right (Hopefully that makes sense)

That's very interesting! I thought it would be hyperbolic in vacuum. My reasoning was that as it exits the nozzle, the pressure in the plume will continue to accelerate it outward, and since there's no air pressure to counter that, it'll be expanding faster the further it is from the nozzle, until it becomes so diffuse it stops behaving like a gas (by the time I expect it would be invisible).

Without any pressure from vacuum to "push back" on it, what causes the rate of expansion to slow so the form is a paraboloid?

[asmi]

2 hours ago, Periple said:

Without any pressure from vacuum to "push back" on it, what causes the rate of expansion to slow so the form is a paraboloid?

It seems rather logical - right as exhaust leaves the nozzle, it's pressure is maximal and so initially it rapidly expands, but as pressure drops with distance from the nozzle exit, the expansion slows down. Hence paraboloid.

Edited March 21, 2023 by asmi

[Periple]

8 minutes ago, asmi said:

It seems rather logical - right as exhaust leaves the nozzle, it's pressure is maximal and so initially it rapidly expands, but as pressure drops with distance from the nozzle exit, the expansion slows down. Hence paraboloid.

Wouldn’t that make a hyperboloid? The initial acceleration would cause it to bend outwards, then as the pressure drops the rate of acceleration approaches zero which means the rate of expansion approaches constant, i.e. from a distance it would look like a cone.

I must be missing something obvious here, because it seems to me that to form a paraboloid the outward velocity itself would have to fall, i.e. something would have to slow down the gas along that axis. In the atmosphere it’s atmospheric pressure, but what is it in vacuum?

[darthgently]

I don't understand the hyperboloid either.  There is no energy being added so it shouldn't be accelerating radially.  A straight walled cone in vacuum makes the most sense to my gut, with some pseudo-hyperboloid when the nozzle is directed at, and near, a surface in vacuum

Edited March 21, 2023 by darthgently

[Bej Kerman]

Just now, darthgently said:

I don't understand the hyperboloid either.  There is no energy being added so it shouldn't be accelerating radically.  A straight walled cone in vacuum makes the most sense to my gut, with some pseudo-hyperboloid when the nozzle is directed at, and near, a surface in vacuum

I just figured the pressure just outside the nozzle causes a bit more expansion, seeing as each particle in the exhaust doesn't suddenly stop coexisting with the other particles after leaving the engine.

[RocketRockington]

1 hour ago, darthgently said:

I don't understand the hyperboloid either.  There is no energy being added so it shouldn't be accelerating radially.  A straight walled cone in vacuum makes the most sense to my gut, with some pseudo-hyperboloid when the nozzle is directed at, and near, a surface in vacuum

Yeah that's what I thought as well.  Here's a picture of a rocket firing in high altitude test chamber 

https://www.arnold.af.mil/News/Photos/igphoto/2000699792/

[darthgently]

5 hours ago, Bej Kerman said:

I just figured the pressure just outside the nozzle causes a bit more expansion, seeing as each particle in the exhaust doesn't suddenly stop coexisting with the other particles after leaving the engine.

That is what causes the straight walled cone is what I'm thinking, for it to be hyperboloid the pressure would have to be increasing relative to the decrease in pressure that must be happening with distance from nozzle.  An acceleration post reaction would have to be occuring.  Think of it this way; at what point in the process is each molecule going the fastest?  As it exits the chamber or meters behind the bell?  Once free of the bell it will not accelerate any faster radially than at that point where it exits the bell.  There is no source of additional acceleration and a hyperboloid shape suggests additional acceleration.

 

[RocketRockington]

25 minutes ago, darthgently said:

That is what causes the straight walled cone is what I'm thinking, for it to be hyperboloid the pressure would have to be increasing relative to the decrease in pressure that must be happening with distance from nozzle.  An acceleration post reaction would have to be occuring.  Think of it this way; at what point in the process is each molecule going the fastest?  As it exits the chamber or meters behind the bell?  Once free of the bell it will not accelerate any faster radially than at that point where it exits the bell.  There is no source of additional acceleration and a hyperboloid shape suggests additional acceleration.

 

I could see some amount of variance within a nozzle length of the exhaust exit, as there's going to be non-homogeneous pressures within the exhaust, it's not just molecules moving in a straight line and bouncing off of the nozzle walls, but also bouncing off of one another, the gas is atill dense enough near the nozzle.   So if say, the center of the exhaust stream is higher pressure than the edges, you could get some hyperbolic expansion.  But yeah, after some distance, it should be linear.

 

[Periple]

2 hours ago, darthgently said:

There is no source of additional acceleration and a hyperboloid shape suggests additional acceleration.

I thought there would be pressure causing acceleration as it exits the nozzle, which would form a hyperboloid! It’s only when it’s diffuse enough that it no longer behaves like a gas that there’s no pressure and it stops accelerating, i.e. expands at a constant rate, i.e. forms a cone.

However we have an expert assuring us that it’s not even a cone, it’s a paraboloid, and I still don’t understand how: to get there the outward expansion would need to slow or the axial expansion would need to speed up, and I can’t figure out what force would do either!

Please @j12sfgd23 enlighten us, this is really interesting and I have the feeling I’m about to learn something new!

[i dont know how to forum]

The devs have talked about this, see the dev diary here with accompanying diagram showing high-atmosphere vs vacuum plumes:

Spoiler

I believe this was in part inspired by a user response to a previous showcase of the old plumes here on the forums, which Nate showed an interest in learning more about:

I'm nowhere near an expert and I can't really speak to the accuracy of these plumes, but it seems to me the devs have done their research. I haven't been able to find much in the way of images of engine plumes in vacuum, especially since in real life they tend to be mostly invisible. Best I've been able to find is this video of a VASIMR test in a vacuum chamber, which looks slightly hyperbolic to me.

Spoiler

Edited March 22, 2023 by i dont know how to forum

[Periple]

I think the problem with testing plumes in vacuum chambers is that once you fire the rocket it won't be a vacuum very long

I wonder if anyone's been able to film a rocket burning in space from a point of view you'd actually see the plume? I was thinking about this and then I realized that it's actually really hard, you'd have to have a camera in space, close enough to the vehicle to see it, but not on the vehicle, nor so close it would endanger anything. 

I would imagine this could only be feasible on Moon missions, especially if there's a base – you could film vessels from the surface.

[j12sfgd23]

32 minutes ago, Periple said:

I thought there would be pressure causing acceleration as it exits the nozzle, which would form a hyperboloid! It’s only when it’s diffuse enough that it no longer behaves like a gas that there’s no pressure and it stops accelerating, i.e. expands at a constant rate, i.e. forms a cone.

However we have an expert assuring us that it’s not even a cone, it’s a paraboloid, and I still don’t understand how: to get there the outward expansion would need to slow or the axial expansion would need to speed up, and I can’t figure out what force would do either!

Please @j12sfgd23 enlighten us, this is really interesting and I have the feeling I’m about to learn something new!

Sorry for the confusion everyone, when I initially posted I used the term "parabola" to generally describe the shape, but as far as I know there really isnt a "true" classic shape that the fuel flows in, mainly because of the non-homogenous pressure in the exhaust plume:

 

As you can see here, it actually starts as mostly straight out, then quickly curves outwards before beginning to curve back inwards. 

I say "parabola" because its a fairly common shape that people know, but really that's not it at all, nor is there really any common geometric shape. Since the flow is supersonic, having a low/zero pressure boundary causes the flow to turn, causing all sorts of annoying-to-model oblique shocks. To make it worse, this changes drastically based on the fuel mix and exit characteristics, which makes answering all of these questions in one go tricky, but I will do my best.

Immediately after the fuel leaves the diverging portion of the nozzle, we have extremely high pressure gas, moving super fast. Since the surroundings are much lower pressure, the edges of this flow are bend outwards, causing an oblique shock and large pressure & velocity losses. These ripple out as we move further from the exit, and pretty quickly that high pressure region starts to drop off  to (what I am going to call) a medium pressure region shown in red/yellow/green. As more and more of this flow turns outwards, it reaches a point where there is almost no pressure in the center, which causes the flow to turn back in on itself, creating a new (smaller) high pressure region. This continues until the flow is no longer moving supersonic and is as a steady equilibrium, but I honestly don't know how long it takes to reach that point, and that would be on an engine by engine basis. It is important to note that while it is expanding and supersonic, the exhaust velocity is actually still increasing, since the "upstream" flow has a higher pressure than the "downstream" flow. This is where that magic acceleration comes from that causes the classic curve, basically converting high Pressure, Temperature, and Density into velocity. That being said, as it approaches equilibrium, the outer shape of the flow at & beyond will slow to almost nothing, and will be functionally linear (or conical if you'd prefer it in 3-d). While it will still be slightly changing in trajectory, it will be extremely small, practically unnoticeable, so its totally fair to call it linear/Conical as y-> infinity.

In regards to the internal pressure pushing the flow outwards, you are absolutely right, and technically the flow can even bend backwards during initial ejection:

(Note, this is just the initial expansion, in the 0-0.1D distance from the engine, this is not representative of the entire shape)

That being said, the more the flow bends, the higher the shock, and so the slower the flow actually will be travelling in these extreme backflow regions. in some situations, its so sharp that the flow literally cannot turn that far without having an absurd number of oblique shocks, and so it's practically stopped by the time it does so. To make it worse, those edge cases are sporadic, and fluctuate a ton with the pressure, so while yes they are technically there, they are so slow and so small that theyre basically non-existent. 

If I had to guess, the reason that high altitude test shows such a perfectly straight line is twofold: 1) I suspect it has been sized to *not* have a plume, just to fit within the test bed without burning it, and 2) it likely does expand like this a bit further down, and we just cant see it because the camera view is too short. That is a really cool picture though, I've never actually looked into high-altitude test facilities so I guess thats a new rabbit hole I am about to dive down lol

[Periple]

Ooooh, very cool! So much stuff going on in the plume. Now I get the inward-bending thing, the lower-pressure region forming at the center of the plume would do that! Thank you!

I actually dived into a little rabbit hole and skimmed through some papers I didn't understand but that had pretty pictures, notably this one, and it seems that this is really complicated and hard to model!

(Also some of the pictures looked bell-shaped, some looked like they fanned out, and others were almost conical. So from that it really looks like "it depends!") 

Edited March 22, 2023 by Periple

[j12sfgd23]

13 hours ago, i dont know how to forum said:

The devs have talked about this, see the dev diary here with accompanying diagram showing high-atmosphere vs vacuum plumes:

  Hide contents

I believe this was in part inspired by a user response to a previous showcase of the old plumes here on the forums, which Nate showed an interest in learning more about:

I'm nowhere near an expert and I can't really speak to the accuracy of these plumes, but it seems to me the devs have done their research. I haven't been able to find much in the way of images of engine plumes in vacuum, especially since in real life they tend to be mostly invisible. Best I've been able to find is this video of a VASIMR test in a vacuum chamber, which looks slightly hyperbolic to me.

  Reveal hidden contents

Yeah, they have really been doing a phenomenal amount of research and its super impressive, honestly all the kids who grow up on this are gonna be a different breed when they hit the industry haha.

I am definitely going to need to do more research into this, (hell, maybe ill do a sim of my own haha), but if I am wrong I will be happy to own up to it. That being said, streamlines can be misleading, especially when youre dealing with supersonics and visuals. Any time a supersonic flow turns, even if it is 0.0001 degrees, there are shocks, which lead to a massive pressure drop and deceleration of the flow, which leads me to believe that these streamlines are likely not representative of what it would actually look like, but I cant say for sure so I wont say anything concrete. All of the information I have given so far is just based on my current understanding of supersonic flow, but I am still not convinced it would not be parabolic.

Another potential hole in that thread you sent is the strange behavior from transitional -> Vacuum. Think about it, if the plume is able to expand at the exit due to internal pressure in transitional:

Why would it not also expand in vacuum at the exit, where it has even LESS resistance to its expansion? If what that person said was true, it would likely be closer to: 

Unfortunately I do not know much about ION engine, but AFAIK they are very low pressure, so it makes sense that it wouldn't have a substantial initial expansion. I have no idea though, that is a total shot in the dark cuz I know very little about those electric magic machines.

13 hours ago, Periple said:

Ooooh, very cool! So much stuff going on in the plume. Now I get the inward-bending thing, the lower-pressure region forming at the center of the plume would do that! Thank you!

I actually dived into a little rabbit hole and skimmed through some papers I didn't understand but that had pretty pictures, notably this one, and it seems that this is really complicated and hard to model!

(Also some of the pictures looked bell-shaped, some looked like they fanned out, and others were almost conical. So from that it really looks like "it depends!") 

ooooh more fun things to read, well guess I know what im doing instead of going to sleep tonight....

Glad I could help! Its a really complicated topic so i was honestly expecting to do a terrible job haha. But yeah, that's really the best answer, it depends on SOOO much, but generally its a bell that turns conical, with the size of the bell being dependent on the exit pressure of the flow

Quick question: I didnt see any that fanned out in that paper you sent, would you mind telling me what figure you are talking about? The only ones I see that actually seem to "fan out" are ones that simulate a collision, like a lander near the surface or an engine right after stage separation where the flow is hitting the lower stage

Edited March 22, 2023 by j12sfgd23

[j12sfgd23]

53 minutes ago, Periple said:

I think the problem with testing plumes in vacuum chambers is that once you fire the rocket it won't be a vacuum very long

I wonder if anyone's been able to film a rocket burning in space from a point of view you'd actually see the plume? I was thinking about this and then I realized that it's actually really hard, you'd have to have a camera in space, close enough to the vehicle to see it, but not on the vehicle, nor so close it would endanger anything. 

I would imagine this could only be feasible on Moon missions, especially if there's a base – you could film vessels from the surface.

Do I have good news for you:

Unfortunately, due to the fuel used in the ascent module and the massive bell of the engine, the fuel isnt really hot enough to see it with the naked eye, save for that split frame where it absolutely BLASTS the descent module. realistically, its very rare to see any plume in space, (heck, falcon 9 launches are barely hot enough to see once they are in vacuum) but thats no fun for a game like KSP, so I am totally happy with them making that change

That being said, there is still hope to see that one day soon! I believe there are a few fuels that do burn visibly in space, and I think hydrolox is one of them. if we ever do begin manufacturing fuel on the moon, there is a good chance we can actually see it

P.S. iirc the reason you see it so clearly on earth is actually not because the fuel is glowing, but its because its heating up the surrounding air enough that it glows!

Edited March 22, 2023 by j12sfgd23

[Periple]

38 minutes ago, j12sfgd23 said:

Quick question: I didnt see any that fanned out in that paper you sent, would you mind telling me what figure you are talking about? The only ones I see that actually seem to "fan out" are ones that simulate a collision, like a lander near the surface or an engine right after stage separation where the flow is hitting the lower stage

Figure 16(a) – the ion engine plume. Even though it only fans out a little.

[j12sfgd23]

4 minutes ago, Periple said:

Figure 16(a) – the ion engine plume. Even though it only fans out a little.

Ah I see what you are talking about, yes it does seem to be somewhat hyperbolic, but this sim is actually of a plume contacting a surface. Dont worry about missing it though, reading these papers is a hassle sometimes and i do the same thing all the time lol

[Periple]

Oh right, got it, thanks! 

If you do find out more about this please post what you discovered, this is interesting stuff (and way above my pay grade )

[Smegsy]

Thread = mind blown. Can't think of another game forum where you would see this sort of stuff. Really interesting.

[i dont know how to forum]

2 hours ago, j12sfgd23 said:

Yeah, they have really been doing a phenomenal amount of research and its super impressive, honestly all the kids who grow up on this are gonna be a different breed when they hit the industry haha.

I am definitely going to need to do more research into this, (hell, maybe ill do a sim of my own haha), but if I am wrong I will be happy to own up to it. That being said, streamlines can be misleading, especially when youre dealing with supersonics and visuals. Any time a supersonic flow turns, even if it is 0.0001 degrees, there are shocks, which lead to a massive pressure drop and deceleration of the flow, which leads me to believe that these streamlines are likely not representative of what it would actually look like, but I cant say for sure so I wont say anything concrete. All of the information I have given so far is just based on my current understanding of supersonic flow, but I am still not convinced it would not be parabolic.

Another potential hole in that thread you sent is the strange behavior from transitional -> Vacuum. Think about it, if the plume does expand at the exit due to internal pressure in transitional:

Why would it not also expand in vacuum at the exit, where it has even LESS resistance to its expansion? If what that person said was true, it would likely be closer to: 

Unfortunately I do not know much about ION engine, but AFAIK they are very low pressure, so it makes sense that it wouldn't have a substantial initial expansion. I have no idea though, that is a total shot in the dark cuz I know very little about those electric magic machines.

ooooh more fun things to read, well guess I know what im doing instead of going to sleep tonight....

Glad I could help! Its a really complicated topic so i was honestly expecting to do a terrible job haha. But yeah, that's really the best answer, it depends on SOOO much, but generally its a bell that turns conical, with the size of the bell being dependent on the exit pressure of the flow

Quick question: I didnt see any that fanned out in that paper you sent, would you mind telling me what figure you are talking about? The only ones I see that actually seem to "fan out" are ones that simulate a collision, like a lander near the surface or an engine right after stage separation where the flow is hitting the lower stage

This is almost entirely over my head, but what little I do understand is really interesting. Thank you for the elaboration! It seems clear this is a lot more complicated than I had thought, but of course it is rocket science. As I understand it the devs are always on the lookout for more in-depth information on things like this, so I hope they see your AMA question and/or this thread. While I'm sure it's near the very bottom of the priority list at the moment, I'd love to see them continue to develop the plumes to make them even more realistic.