DYI Engine Balance

[mielgato]

I think the OP did a great job and it is really appreciated to see peoples going to such extent about balancing. (To be honest I suck at advanced maths so i'm not really in a position to criticize the job done here)

Still, i'm playing KSP for a long time now too (since 0.18 if I remember right), and right now in 1.0.5, i'm using all engines.

In a stock career save, even if I have unlocked twin boar overpowered engine, I might not use it for a simple munar satellite launch, but instead use a mix of LV-T30/LV-T45/LV-909 (or even smaller engines like the tiny rockomax or ant -like engines)

I definitly think it all depend on what type of launch you are going to have, if it's for a 50tons space station chunk, then yeah i'll probably be using twin boar or 3.75m engine. I even sometime had to use clusters of 9xLV-T30 engines on a 2.5m fuel tank as first stage, not as efficient in term of isp, but I was getting a better twr than with one mainsail (9xLV-T30 give slightly the same thrust, slighly less twr, and is cheaper than a KR-2L+ Rhino without the need for 3.75m parts)

So, lower tier engines are less powerfull in term of twr/isp compared to latest tier ones, I can only agree, but that don't mean they are always useless, they are cheaper and can still be used for a lot of missions involving light final stage.

Why would I launch a 150 000f munar orbiter made out of 3.75 and 2.5m parts when I can do the same for less than 50 000f with 1.25parts ?

Then of course, there are some "must have" engines depending on where you are going, even with all what i'm saying I can't go against this point either, and my way of thinking can only apply to early and mid career, if you are doing a fundless save (sandbox/science mode) or get to late career then some engines become useless and have no points in being used at all as you are not restricted by funds anymore, so you just pick the best suited for the task.

That said, I recognize the fact that engines need balancing and thanks [URL="http://forum.kerbalspaceprogram.com/members/39284-spikeyhat09"][B]spikeyhat09[/B][/URL] for such a great thread. Edited November 18, 2015 by mielgato

[Streetwind]

[quote name='mielgato']That said, I recognize the fact that engines need balancing and thanks [URL="http://forum.kerbalspaceprogram.com/members/39284-spikeyhat09"][B]spikeyhat09[/B][/URL] for such a great thread.[/QUOTE]

The thread is less about engines needing balance, and more about ways and reasons for tweaking things to one's own liking. spikeyhat09 specifically wanted to eliminate the inherent power progression in engine stats because he likes playing sandbox mode, hence his attempt at a unifying formula.

[mielgato]

Yah I realised this too late, the post was already made :blush: I tried editing to match the initial reasoning but I couldn't do so without completly changing the post's content, thus I left my initial post practically unchanged. I try to think twice before posting, but sometime I just don't.

Said like this I can only agree and have nothing more to say that could add to the conversation.

[spikeyhat09]

You are both correct. The engines need balancing for mielgato because he wants there to be, and they dont need balancing for someone who thinks they are balanced. Squad knows that they cant please everyone, which is why we have such fantastic modding support.

[quote name='theend3r']LV-T45 having more thrust than LV-T30 doesn't make any sense when the only difference is the added gimbal.[/QUOTE]

They are actually more different than that. The LV-T45 has more mass. If you reduce the LV-T45 mass to that of the LV-T30, you end up with a thrust of 220.313 KN. Conversely, an LV-T30 scaled up to the mass of the LV-T45 would have a thrust of 300 KN. Edited November 18, 2015 by spikeyhat09

[Reactordrone]

[quote name='spikeyhat09']


They are actually more different than that. The LV-T45 has more mass. If you reduce the LV-T45 mass to that of the LV-T30, you end up with a thrust of 220.313 KN. Conversely, an LV-T30 scaled up to the mass of the LV-T45 would have a thrust of 300 KN.[/QUOTE]

But the LV-T45 has more mass because of the gimbal system, not because the engine is bigger. The LV-T30 is a bigger engine on a lightweight, fixed, thrust plate and the LV-T45 is a slightly smaller engine in a heavier gimbal mounting.

[Norpo]

[quote name='spikeyhat09']If you're willing to dedicate the time and effort to create such a program, it would not go unappreciated.

I noticed this as well when trying out different solutions, and couldn't come up with one that keeps the Reliant above the Swivel without the rest of the engines becoming ridiculous. The main problem is that the Swivel weighs quite a bit more despite having inferior thrust in the base game. The result is that the Swivel actually has pretty terrible TWR in the base game, even in comparison to the Reliant. A Swivel with the same mass as the Reliant would have a max thrust of 220.313 KN according to the example model.[/QUOTE]

Alright, I'm working on a ModuleManager patch right now; it shouldn't take [I]too[/I] long. ~2 days maybe, if I get the time, but let it be known; It's being worked on!

I'll have to manually set a few engines to disobey the equation just to get around some limitations (The Twin-Boar, for instance would also take into account it's fuel-tank weight, making an absurdly powerful engine...), but that shouldn't be too hard. :)

[NecroBones]

[quote name='Norpo'] (The Twin-Boar, for instance would also take into account it's fuel-tank weight, making an absurdly powerful engine...), but that shouldn't be too hard. :)[/QUOTE]

It has the capacity of an orange tank, so I'd subtract out the mass of that. An empty orange tank is 4 tons, so the weight of just the twin boar engines should be 6.5.

[spikeyhat09]

[quote name='Norpo']Alright, I'm working on a ModuleManager patch right now; it shouldn't take [I]too[/I] long. ~2 days maybe, if I get the time, but let it be known; It's being worked on!

I'll have to manually set a few engines to disobey the equation just to get around some limitations (The Twin-Boar, for instance would also take into account it's fuel-tank weight, making an absurdly powerful engine...), but that shouldn't be too hard. :)[/QUOTE]

sweet!

[quote name='More Boosters']You do realize that your formula greatly favors vacuum engines right? For calculating thrust? Something that means the most to lifter engines? It's counterintuitive and unnecessarily homogenizing. It should be obvious that things with low TWR got better TWR and those with high TWR got worse TWR. Things with low TWR have low TWR for a reason and the opposite is also true.

Terrier and Poodle are just fine. The same can be said for the 3.75m engines as their TWRs are similar to 2.5m engines and only Rhino is considerably more efficient AND with better TWR than Skipper while having a similar role. The Mammoth is just fine really, it has slightly better TWR and Isp and that's it. The LFB still beats it which in turn comes with its own downsides.

I think a good indicator that this formula needs more work is that you're proposing a bigger buff to Swivel than you are to Reliant. The former is already a much superior engine in early game scenarios where you either don't have controllable canards or adding them would skyrocket the price of the booster to begin with.

I think you're putting too much stock in gimbal degrees. Not having gimbal at all is a major downside and it should count for something, but having any gimbal at all is generally good enough for control. You're also ignoring the cost aspect here, as relatively minor perks like high gimbal range and surface-attachability can be justified by the high price of Vector; as long as its not stuff like substantially increased TWR or Isp.

And well, the bit about NERVA... You really need to think this over man. Tweak the formula to make gimbal range matter less and having a gimbal at all matter more (log it up!), and try to make Swivel/Reliant better. They're really the only ones that need the help.[/QUOTE]

Valid points. I agree that there probably need to be multiple formulas for different groups of engines, depending on their purpose. The Swivel vs Reliant issue came as a surprise to me as well, and it comes down to the fact that the Swivel has 20% more mass than the reliant, so in order for the reliant to come out ahead, that difference needs to be made up in the difference of ISP and gimbal.

As for the gimbal discrepancy, you are probably right. I'm experimenting with logs and integrals of logs right now, but nothing I'm coming up with wouldn't be made easier on the mind and hardware with linear approximation. p=(-1/2)g°-5 seems reasonable enough, where p is the percent adjustment to the final thrust.

The Nerva and aerospike (as well as other unique engines not tested) do seem like they need a model of their own or individual examination, as I pointed out in the OP. Remember, that was just an example formula I came up with rather hastily for the OP. It's not meant to be the end all balance formula, and I encourage anyone who wants to make alterations to it or better yet derive a completely new and more comprehensive one.

I've also included a config file in the OP for anyone who wants to try the altered engines out (feedback section). It definitely feels more difficult to achieve what one could easily do with the default values, since the heavily relied upon engines are no longer such amazing workhorses.

[MaxL_1023]

Have you tried having gimbals simply subtract a value from the TWR? The physical gimbal mechanism would scale in size with the loads it would bear, which would roughly scale with the amount of thrust the engine is producing multiplied by the available gimbal range - a 5 degree thrust shift would basically put 5 times the load on the gimbal as a 1 degree shift, using the small angle sine approximation and assuming thrust tangential to the direction of motion would need to be transmitted through the gimbal to the rocket as opposed to the pre-existing structure which transmits prograde force.

Ex: Gimbal Size = 2-3% of engine weight (assuming gimbal tech scales with engine tech) * gimbal angle in degrees.

A 5 degree gimbal would add between 10-15% of the engine weight, with thrust and ISP being constant. A direct trade off, as all you are really doing is adding some heavy stuff to the engine, not changing the fuel flow, combustion chamber or other infrastructure that much.

[Norpo]

Alright, I made a ModuleManager config for setting (most) of the engines to their auto-balanced counterparts. I have no idea if it's compatible with mods (or technically, even stock, but it [I]should[/I] work okay...)

[B]Link to download [URL="https://www.dropbox.com/sh/7gudaxmdi0dl3w5/AABSKxA9HNOkW6dJiU6ul550a?dl=1"]here.[/URL]

[/B]A dependency is ModuleManager, of course, not bundled, but if you use mods you probably already have it in your GameData.
[SIZE=2]It also comes with a Python script that I used to make sure all the numbers were the same. Enjoy! Also, if you want to try out different constants and such you can edit the config, but be warned: it's big because I didn't optimize it at all, so you'll have to change the numbers in 4 places. Still, at least it's easier than manually going through every single part .cfg in the Squad folder. Good part about this is that it should stay current between versions, so it should still work once 1.1/1.2/1.pancake rolls around. :)[/SIZE] Edited November 19, 2015 by Norpo
like, I don't like it when people like, use like, like, it's like, a comma.

[spikeyhat09]

[quote name='Norpo']Alright, I made a ModuleManager config for setting (most) of the engines to their auto-balanced counterparts. I have no idea if it's compatible with mods (or technically, even stock, but it [I]should[/I] work okay...)

[B]Link to download [URL="https://www.dropbox.com/sh/7gudaxmdi0dl3w5/AABSKxA9HNOkW6dJiU6ul550a?dl=1"]here.[/URL]

[/B]A dependency is ModuleManager, of course, not bundled, but if you use mods you probably already have it in your GameData.
[SIZE=2]It also comes with a Python script that I used to make sure all the numbers were the same. Enjoy! Also, if you want to try out different constants and such you can edit the config, but be warned: it's big because I didn't optimize it at all, so you'll have to change the numbers in 4 places. Still, at least it's easier than manually going through every single part .cfg in the Squad folder. Good part about this is that it should stay current between versions, so it should still work once 1.1/1.2/1.pancake rolls around. :)[/SIZE][/QUOTE]

Fantastic! I've updated the OP to include your work.