Heat is driving me bananas

[CaptainArbitrary]

So I'm working on some mod engines, and in the process I build a sort of reference stack with an engine on the bottom, then a few small fuel tanks above it. I do some flight tests and all is well.

I start working on my next engine, so I just pop the engine off that reference stack and replace it with a different engine … which I've set up using the exact same config parameters. You know, just to have someplace to start, right? It puts out exactly the same thrust, it has exactly the same heatProduction. It varies only in the specific impulse, because the new engine has a smaller nozzle than the other one so it gets less specific impulse at sea level for the same exhaust pressure.

I do my first flight test, and … kaboom. It's not the engine that explodes, though. No, it's the propellant tank immediately above it. It exceeded its temperature tolerance. (I play with Deadly Reentry, so propellant tanks weaken when they get hot.)

I figured out, after a hell of a lot of testing, that the difference between these two engines is their center of mass. Not mass itself; both of these engines have exactly the same mass, 'cause I set 'em up that way. No, the only difference is where their center of mass is, 'cause they're using different 3D models.

Heat conduction in KSP depends on where your center of mass is.

We all know how heat dissipation works; that's heat flowing out of your ship and into the environment (or vice versa) lowering (or raising) the temperature of each part independently. It's just Newton's law of cooling with a fixed cooling constant: ÃŽâ€T = (Tâ‚‘ – Tâ‚€) k where k is fixed, by the game, at 0.12 per second. That's an annoyingly inflexible approximation, but it works, so no biggie, we can work with it.

But conduction is a different animal. You'd think that heat conduction between adjacent parts would depend on (a) the difference between their temperatures, ( some thermal properties of the parts, which might be approximated as a universal constant the way the cooling constant is, and © the surface area of contact between the parts. Since the game doesn't really do surface areas of contact, we can imagine abstracting that away such that, for sake of consistency, heat always flows between adjacent parts at a rate proportional to the difference between their temperatures. I mean, I'm not crazy, right? That makes sense to y'all too?

The way it actually seems to work, though, is that conduction is somehow proportional to distance between parts … and not just distance, but distance between centers of mass. If the center of mass of two adjacent parts is close together, heat will flow between those two parts faster than if the centers of mass are farther apart. Even if the parts stay the same and you just move the CoM around.

That's why my propellant tank exploded. Because the second engine I was modding had a higher center of mass in its 3D model than the first one did, heat flowed into the above tank faster. Too fast, in fact, for the tank to radiate the heat away, so it overheated and failed about twenty seconds after liftoff.

Short of just arbitrarily dialing down the heatProduction parameter of that engine, I have no idea how to go about fixing this. I mean, in a sane world the heatProduction of an engine would depend on the engine itself, its thermodynamics and whatnot. But instead, it seems like we have to set it based on how we hope heat flows out of it as its running, which is a function of what it's attached to.

Does anybody have any advice for me at all on this? Cause seriously, I'm this close to just changing all my heatProduction parameters to give my engines a full-throttle equilibrium temperature of boiling water.

[ummwut]

If your engines are heating up like a Mainsail, yeah, turn the heat down. I don't know how deadly re-entry works with heat, but try testing without it.

[EndlessWaves]

Is it the centre of mass or is it the scene origin? If it's the scene origin then you can put it where you want and move the center of mass using the CoMOffset parameter.