Engine balance TWR/Isp

[thereaverofdarkness]

It's been bugging me very much that the engines don't seem to have TWR balanced with Isp very well. Some of them are quite overpowered and can allow you to do things that normal rockets can't do, while others are just underpowered and as a result are rarely used. I'm not here to say what should be the right ratio, but I'd like to see some sort of consensus, so that the engines can be brought more or less in line with each other.

Here's my concept sketch for a ratio between the two ratios:

I'll start with the LV-909 and the LV-T45 as my base, as they are close to the middle in value. Since they are close in Isp, I'll use the Rockomax 24-77 as a low Isp engine also for my values. I'm simplifying TWR to 10m/s/s gravity for easier calculation purposes. We have:

24-77: 22.22 TWR and 250-300 Isp

LV-T45: 13.33 TWR and 320-370 Isp

LV-909: 10.00 TWR and 300-390 Isp

Looks like our launch engines are high TWR with moderate Isp that doesn't carry out well into vacuum. The landing engines have the best Isp in vacuum but lose the most from atmosphere. These engines have pretty decent TWR but nowhere near the best. A simple calculation that carries these values in a somewhat similar fashion to the three engines here is as follows:

divide TWR by 10 and take the 4th root of the value (LV-909 is unchanged). New values: 24-77: 1.221 // LV-T45: 1.075. Now divide the min and max Isp of the LV-909 by these values:

24-77: 245-319 Isp

LV-T45: 279-363 Isp

Problem with this system is that it doesn't take into account the increased Isp range of the LV-909 as a dedicated landing engine. Since it has the same median Isp as the LV-T45, lets use the LV-T45 Isp as the base instead:

24-77: 262-303 Isp

LV-T45: 300-344 Isp

Now that looks about right!

If we use these calculations, we'll be raising the in-atmosphere Isp of the Rockomax 24-77 by a bit. We'll also be lowering the Isp of the LV-T45 and the LV-T30 significantly. I think that brings these engines more in line with each other and the others. Given how easy it is to make a SSTO with T30 engines, I think they could use a bit of an Isp decrease. But lets try these numbers on other engines!

First up, the big ones:

Mainsail: 25.00 TWR and 280-330 Isp 1.2574 New Isp: 255-295

Skipper: 16.25 TWR and 300-350 Isp 1.1291 New Isp: 285-330

Poodle: 8.80 TWR 270-390 Isp .9685 New Isp: 310-400

The mainsail gets a huge reduction in Isp. The overpowered engine we all love is put in its place! I don't know whether to laugh or cry. The Skipper actually has its Isp reduced. I didn't expect it but I'm not too surprised. I don't use Skippers much when I can use Mainsails, but in the absence of Mainsails, one discovers that Skippers aren't a bad engine. The worst thing about them is their thrust-to-mount space ratio. I get more thrust with 4 LV-T45s. The Poodle's slightly sub-par thrust pays out with its new slightly higher than LV-909 Isp values.

Don't like these Isp values? I know I don't find them completely satisfying. Lets adjust some other things and see where that takes us...

I'll leave the thrust of the Mainsail unchanged, but lets increase its mass to 7.5 tons. Now lets bring the Skipper up to 900 thrust and we'll leave it around 15 TWR, so 6 tons mass. And finally, lets just increase the Poodle's thrust to 250 so it's exactly like a big LV-909. I bet everyone will love it now. Actually, it might be a bit low on thrust to mount space ratio, so lets try making a bigger version of the same thing, too:

Mainsail: 20 TWR 1.1892 Isp: 270-310

- Thrust: 1500, Mass 7.5 tons

Skipper: 15 TWR 1.1067 Isp: 290-335

- Thrust: 900, Mass 6.0 tons

Poodle: 10 TWR 1.0000 Isp: 300-390

- Thrust: 250, Mass 2.5 tons

Big Poodle: 10 TWR 1.0000 Isp: 300-390

- Thrust: 300, Mass 3.0 tons

One possible problem with my calculation is that it will reduce the Isp of most engines by at least a little bit. Perhaps I should use the LV-T45 as the medium instead of the LV-909. Oh well. I'm just estimating here. What are your guys' thoughts on this? Got any suggestions for better calculations? Different ideas on how a consensus should be reached? Think I'm wrong and that the engines are all balanced? All input is encouraged, I want to know what you guys think! And I hope that this discussion can be used to help improve the engine balance in furute KSP updates!

[UmbralRaptor]

Your analysis is bad, and you should feel bad.

Okay, that's a bit ranty, but the approach chosen is deeply flawed. It does at least fix the poodle, but uses questionable assumptions and ignores how smaller engines interact with larger ones. The net result is making the (currently more or less balanced) Mainsail and (currently marginally underpowered) Skipper all but useless. Unless the smaller engines take similar nerfs, there will be little reason to use the size 2 parts. I suspect that it also makes the lifter parts no better at getting into LKO than the lander ones(!)

I'll start with the LV-909 and the LV-T45 as my base, as they are close to the middle in value.

In general (Or at least when Supernovy and I have been discussing balance), the LV-T30 gets thrown around as the baseline. It's the oldest liquid fuel engine, and the first one on the tech tree. This suggests that it's what the others should be balanced around.

I'm simplifying TWR to 10m/s/s gravity for easier calculation purposes.

Why not go with TMR and to a first approximation ignore gravity? Though for a more in depth analysis, you might want to block out some rocket designs.

I'll use the Rockomax 24-77 as a low Isp engine also for my values.

Careful -- the 24-77 may be worse in both TWR and Isp than the Mainsail, but it remains useful due to the radial mounting and much lower mass. Before the 48-7S was added, the 24-77 was used quite a bit for smaller craft, or augmenting LV-Ns in some SSTO designs.

divide TWR by 10 and take the 4th root of the value (LV-909 is unchanged). New values: 24-77: 1.221 // LV-T45: 1.075. Now divide the min and max Isp of the LV-909 by these values:

What is the reasoning behind this? It doesn't seem to correspond with the rocket equation.

Given how easy it is to make a SSTO with T30 engines, I think they could use a bit of an Isp decrease.

Absolute difficulty is a separate issue from relative engine performance. Whether the default setting for the game should be easier/harder is an interesting question, though.

The overpowered engine we all love is put in its place!

Uh. In 0.17 that would have been the Aerospike. In 0.22 it's the 48-7S. But the Mainsail...? o_O

Let's look at the stats of the lifting engines you proposed.

[table=width: 600, class: grid]

[tr]

[td]Name[/td]

[td]Thrust[/td]

[td]Mass[/td]

[td]Isp[/td]

[/tr]

[tr]

[td]Mainsail 1[/td]

[td]1500 kN[/td]

[td]6 t[/td]

[td]255-295 s[/td]

[/tr]

[tr]

[td]Skipper 1[/td]

[td]650 kN[/td]

[td]4 t[/td]

[td]285-330 s[/td]

[/tr]

[tr]

[td]Mainsail 2[/td]

[td]1500 kN[/td]

[td]7.5 t[/td]

[td]270-310 s[/td]

[/tr]

[tr]

[td]Skipper 2[/td]

[td]900 kN[/td]

[td]6 t[/td]

[td]290-335 s[/td]

[/tr]

[/table]

And using them as SSTOs:

(download)

I've found that approximating Isp as being at 0.2 atm gives very close to correct numbers. The lower section is for craft that I've actually flown to LKO (aside from the 0.21 48-7S, which I couldn't find a flight record of). If you insist, I suppose generating charts of various 2/3/4 stage designs and landers would be doable...

As for balance changes I would actually want to see:

* Buff the Poodle. I've toyed with a 260 thrust, 410 s version, but am unsure if that's overdoing it. Something to bring its TWR in line with (if not marginally better than) the LV-909, and do something about the LV-T30 being a better lander engine...

* Buff the Skipper. Say, to a 700 kN, 3.5 tonne engine. The current lower Isp and and TWR than an LV-T30 situation is embarrassing. It has TVC, yes, but for craft in the 1-Skipper size it doesn't matter. And larger ones can eat the TWR hit of an LV-T45 and still come out ahead. Especially with the Isp difference...

* Nerf the 48-7S. At least back to its 0.21 stats, and maybe even poke away further at the Isp and/or thrust.

* Make the Mk 55 useful. eg: bring the TWR up to Mainsail levels, or Isp up to LV-T45. Reducing (or eliminating) TVC on the Skipper or especially Mainsail would help(!)

Edited December 9, 2013 by UmbralRaptor

[thereaverofdarkness]

The numbers you show look good to me. If an aerospike has under 10% SSTO payload fraction, then I don't see why bell engines should be much over 5%. Aerospikes are more expensive, a higher tech, and less efficient in space. They should be significantly better for an SSTO on Kerbin I think. But also I could see upping the values on my calculation by increasing the Isp of everything by x1.075 to bring them all in line with the LV-T45 (except lander engines, they stay at the old value because they have min-max stretched outward by 20). The reason I chose the LV-T45 and not the LV-T30 is because the LV-T30 has no thrust vectoring, and thus has a greater TWR than the LV-T45. I think it's too good for its lack of thrust vectoring, an ability that's not really useful. As long as the LV-T30 maintains 215 thrust and the same Isp as the LV-T45, I think it should have 1.5 tons mass, or if it is to stay lower in mass, its Isp should be decreased a bit.

I never intended to suggest that there are not more unbalanced engines. The Rockomax 48-7S is quite overpowered as its TWR is ridiculous right off, but its Isp is actually pretty high, not extremely low like it should be. I think it should have its mass increased to 0.15 tons as opposed to getting a huge nerf to Isp. Making powerful engines heavier isn't as bad as you might think. It might decrease its SSTO payload fraction by a lot, but it doesn't substantially affect the rocket size it can lift. Screwing with its Isp allows it to propel tiny payload fractions really far, but makes it useless as a heavy lifter. It turns it into a stunt engine. The mainsail will be fine at 7.5 tons. It could do 9 tons but with its high thrust vs mount size, I think it should have significantly less Isp than the Skipper. Also, definitely should take out thrust vectoring. Can give it a bit more Isp back for that.

I think the Skipper should be at least 900 thrust. With a mass of 6 tons it'd be fine, if it actualy had 900 thrust. It'll lift bigger things farther than it did as a 650 thrust engine at 4 tons. It'll be a good mid stage engine, as opposed to (now) a crappy late-stage engine for people who don't want to wait on a Poodle for the higher Isp. Poodle should definitely be at least 250 thrust but I could see up to 400. It is 2x the width of a LV-909 and actually well more than 2x the height, making its volume considerably more than 8x a LV-909. 6-8x thrust leads to it being useful as a large-mount size lander engine, able to lift your 4 ton capsule with a 720/880 fuel tank easily. You won't have to make a long burn on descent to the Mün, but can make a quick burn right before you hit the ground, just like with a little lander that uses an LV-909.

Now the worst balanced engine is the LV-1 and LV-1R. Their TWR is terrible and their Isp is worse. Leaving their mass and thrust the same, I'd bring their Isp to 560 - came up with that a while back because it puts them as a proportional median between the LV-909 and the LV-N in terms of TWR vs Isp. But that's a bit high Isp for a cheap liquid fuel bell engine, so we could just decrease the mass by 50%-60%. Even after doing that, I would still increase the Isp.

I use the Rockomax 24-77s all the time. I love being able to mount a small yet powerful engine. I wish the Mk-55 engine had more thrust. I try to use Mk-55s in conjunction with Skippers to make a "light mainsail" when a mainsail is overkill, but I end up with something that has way less thrust and significantly higher mass, and its Isp is about the same at launch and only gets worse. But with a 900 thrust Skipper and 4x 180 thrust Mk-55s with reasonably-balanced mass and Isp, we'd get 1620 thrust for probably similar mass and Isp to a Mainsail. Of course I'd probably be willing to use the Skipper by itself if it had 900 thrust.

Say we leave the mass and Isp of the Mk-55 alone and change only its thrust to 180. Now it has a high TWR to compensate for its lousy Isp. It also has a low Isp range, making it better as an early launch engine and also useful for EVE launches.

Mainsail: 1500 thrust, 7.5 tons

Skipper +2 Mk-55s: 1260 thrust, 7.8 tons

The Skipper setup would have slightly better Isp. I'd still take the mainsail most times, but nice to have the midway option. It wouldn't be a bad choice on Kerbin, and it might even be a better choice on EVE.

Edited December 9, 2013 by thereaverofdarkness

[Moonfrog]

I think you guys use far too much maths and over complicate matters. Engines should just be able to do what they were designed to do. Should follow a general principle. Solid boosters have a high TWR rating, but low ISP. lander engines are good in vacuum, poor in atmosphere, launch engines poor in vacuum but powerful, decent enough in atmosphere. If u look at the engines u can see that's roughly what they go by. the larger a engine the more powerful it should be.

engines like the 909 and poodle, if u follow what most small and tiny engines go by, they lose ISP. SO the poodle ISP to compensate if its currently bad, should have its ISP increased, everything else remain the same. If u think about it, if your going to use a poodle engine, it should be for a big ship travelling in a vacuum, Thrust is sacrificed for ISP, This means in larger ships poodle will be worth it. This same principle can be seen in the nuclear engine, it has very high isp but very low thrust, compared to its mass.

Why not stick to the way things are now, instead of all this complicated maths stuff. boost isp of poodle problem solved. wasn't that hard.

[thereaverofdarkness]

The Poodle's Isp is fine, and in fact probably should not go over 400. Liquid fuel engines are supposed to have limited Isp, while it is other technologies like the LV-N which have the high Isp. The reason the Poodle is underpowered is because you can get about the same thrust with less mass and height by using a large quad-coupler and putting on 4 LV-909s. The extra mass of the quad-coupler is still less than the Poodle, and the Poodle is still bigger. It should be more powerful, else it just gets in the way.

I see what you're saying, and to a big extent the ay to design an engine is by trial and error. I base my mass and thrust adjustments of engines around serving their purpose, or making them to serve a purpose at all, in some cases. The TWR/Isp balance is mostly just to prevent cheaty min-maxing with certain imbalanced engines, resulting in rockets that do things they shouldn't be able to do. Good example: 48-7S + fuel tank + EAS-1 OKTO2 makes a rocket with far too much delta-v for its high rate of thrust. Increasing the mass of the 48-7S engine substantially reduces its delta-v, as the high delta-v is reliant on the tiny dry mass fraction, being barely higher than the fuel tank itself with a 360/440 tank, and still getting enough thrust when full to lift off the ground on Kerbin.

Here's pics of what a 48-7S SSTO can do:

It has 6003 delta-v and a TWR on Kerbin of 1.90. But when I add 0.15 tons mass (to simulate the effects of increasing the engine mass by 0.05 tons), the TWR becomes 1.84 and the delta-v becomes 5545. Even that is pretty high, and is why I think this engine's Isp should be lowered even after increasing it's mass. It's a heavy lifter, not a SSTO engine. It should not be beating the Aerospike at its own job, a job which the 48-7S wasn't even built for.

Edited December 10, 2013 by thereaverofdarkness

[UmbralRaptor]

Thereaverofdarkness: I suppose this should be more in depth, but...

Cost does not seem like a good metric at this time. If (when?) there is more of an economy, then it will certainly need to be considered as a balance factor, but I'm hesitant to throw it around in the interim.

As they currently exist now (especially if unchanged), aerospikes would be better in all phases of flight, especially in space. The modified engines you propose don't allow for the "worse in space" comment to hold at all, given their narrower Isp ranges. For multistage craft, payload fractions seem to be ~1.5-2x the single stage values, but this depends on details, and would require several test flights to verify with the modified engines. (I suppose that the aerospike would be easier to balance if Isp affected thrust rather than fuel consumption...)

For augmenting a Skipper while building rockets, I'd suggest attaching LV-T30s via any of several structural/aerodynamic parts. Or even strappping a few boosters on. The current thrust is hilariously convenient (1 orange tank is about the perfect amount of fuel), so I'd rather a TWR tweak, rather than scaling up.

I'd much rather see the 48-7S have reduced thrust and perhaps Isp than increased mass. Between the 24-77, LV-909, and LV-N, it's already in a rather crowded thrust range. Meanwhile, there's a rather large gap from the 24-77 (or old 48-7S) and LV-1/1R...

The LV-1 is if not fine, near fine. I suspect you are building craft too heavy for it. It has stats that suggest a higher performance (higher thrust and Isp, lower mass) RCS port. If you're building a craft light enough (and admittedly with modest enough ÃŽâ€V requirements) to seriously consider using RCS, then you're in the right area. Upping the Isp by 1.9x to 560 s sounds like something to make it directly take on the LV-N and PB-ION(!), and breaks with your earlier comments about realistic engines.

In general, I'd want to see a focus on fixing the balance issues with the Mk 55 and 48-7S (in that order). The issues with the other engines are comparatively small. Your suggestion of upping the Mk 55 thrust by 50% would probably make them useful.

Moonfrog: Yes, but no.

While there are a lot of ways to make all parts useful ('balanced'), going with "yeah, this looks right" can cause issues, especially with the variety of engines in current versions of KSP. (Consider that most upper stages/landers that use the Poodle would see a TWR and ÃŽâ€V increase from switching to the LV-T30)

[John FX]

What we need implemented is exhaust speed. Then some engines will simply stop pushing as hard when you get faster which will make some engines better for liftoff as they have a slow exhaust speed but a good TWR and some for traveling at 5km/s where a fast exhaust speed would offset a relatively low ISP and TWR.

Then thrust would be calculated as a function of your speed, the atmospheric density, throttle, and mass. So there could be a case where you have your throttle on full and you are not getting any faster because you are traveling at your exhaust speed...

This would let you optimise the mainsail for liftoff in an atmosphere and other engines for vacuum or high speed travel independently without having to mess about with a global attribute like ISP or TWR to get the result you desire..

[TheExplosionist]

Realism is far more important than "balance". This isn't a multiplayer game.

[UmbralRaptor]

a fast exhaust speed would offset a relatively low ISP

Isp is exhaust velocity.

Realism is far more important than "balance".

Both -- it would be sort of silly to have all these engines if only 1-3 of them are ever used.

[John FX]

Isp is exhaust velocity.

Specific impulses are often measured in meters per second and are often termed effective exhaust velocity. However, if propellant weight is used instead, an impulse divided by a force (weight) turns out to be a unit of time, and so specific impulses are measured in seconds.

NASA uses seconds to measure their ISP for this reason and KSP reduces fuel flow rate when ISP changes instead of increasing exhaust speed and so the ISP in KSP is measured in seconds instead of m/s.

I`ll just repeat myself and hope that my point gets addressed, being that what we need implemented is exhaust speed.

[foamyesque]

Specific impulses are often measured in meters per second and are often termed effective exhaust velocity. However, if propellant weight is used instead, an impulse divided by a force (weight) turns out to be a unit of time, and so specific impulses are measured in seconds.

Uh. Aside from the bolded mutually contradictory statements, a velocity divided by a force is not, in fact, a unit of time, but seconds per kilogram. What this might physically represent I leave as an exercise for the reader.

NASA uses seconds to measure their ISP for this reason and KSP reduces fuel flow rate when ISP changes instead of increasing exhaust speed and so the ISP in KSP is measured in seconds instead of m/s.

How KSP implements Isp variance has less than nothing to do with why Isp is measured in seconds, and for that matter neither does propellant weight. It's done the way it is because it makes it irrelevant if you're using metric or Imperial units elsewhere. Isp is nothing more or less than the effective exhaust velocity divided by Earth's surface gravity. As such the two are interchangeable as long as you apply the correct conversion factor.

What we need implemented is exhaust speed. Then some engines will simply stop pushing as hard when you get faster

Rockets don't work that way, champ. It's one of their most noticeable principles; a rocket's speed has exactly no impact on their thrust. That's the entire reason the Oberth effect works in real life.

So there could be a case where you have your throttle on full and you are not getting any faster because you are traveling at your exhaust speed...

No there could not. That's a fundamental misunderstanding of rocketry.

This would let you optimise the mainsail for liftoff in an atmosphere and other engines for vacuum or high speed travel independently without having to mess about with a global attribute like ISP or TWR to get the result you desire..

Isp is not a global attribute and variations like the one you want are the entire reason we have an Isp curve.

Edited December 15, 2013 by foamyesque

[UmbralRaptor]

Specific impulses are often measured in meters per second and are often termed effective exhaust velocity. However, if propellant weight is used instead, an impulse divided by a force (weight) turns out to be a unit of time, and so specific impulses are measured in seconds.

Yes, it's alternate way of measuring. Hence the unit conversion in the rocket equation when you use seconds, but not when you use ft/s or m/s. Note that rocket engines do not know how fast they are traveling, just the relative speed of exhaust leaving the nozzle.

NASA uses seconds to measure their ISP for this reason and KSP reduces fuel flow rate when ISP changes instead of increasing exhaust speed and so the ISP in KSP is measured in seconds instead of m/s.

Yes, NASA does, but they're clearly using it more like m/s.

NEXT has a maximum specific impulse (Isp) of 4,190 sec, a maximum thrust of over 236 mN, and a peak efficiency in excess of 70%. The project requirement for xenon throughput is 300 kg (1.23 x 10^7 N-sec total impulse)...

To provide 1.26e7 N*s, 300 kg would have to be moving at 4.1e4 m/s. That's 4190 s to rounding error.

KSP engines historically didn't deal directly with Isp at all, just thrust and fuel consumption. At one point, the Isp system was just a hack to generate fuel consumption, though that's less clear now. C7 has in the past mentioned that the current system (Isp changing fuel consumption, rather than thrust) may be changed, but if/when is unclear.

I`ll just repeat myself and hope that my point gets addressed, being that what we need implemented is exhaust speed.

Only for jet engines (where how fast you're moving does directly affect engine performance), and it already partially exists in the part.cfgs.