That darn wobbling

[JeeF]

Hey guys,

I've been trying for weeks now to create a propeller engine from scratch and I've come pretty close to a good design.

It can lift a CM, that's how I've been testing optimal wing angle. However, in order for it to be able to be attached to an airplane and actually drive it forwards, I need to make it spin REALLY fast.

Spinning really fast is not the problem... I've done that. The problem is the WOBBLING!

Doesn't matter what I do, what parts I use, how precisely dead-center I install these parts, after a certain rotation speed it all goes to hell.

Here's a video showing the design and the problem:

I've tried many different approaches, angles, parts, sizes... you name it.

I've installed a previous prototype in a big airplane I've made, 4 engines, and it can drive the whole heavy airplane forward at about 25m/s solid steady at about 30% throttle.

Anything above that, yes the speed increases but the wobbling begins and the more throttle I give the worse it becomes till it all explodes.

Shame KSP doesn't allow us to connect both sides of a part to another part, otherwise I'd could double the "firmness" of the design by putting rotatrons on both ends and connecting both ends to the same wing, having the engine spinning in the center of the wing instead of the front.

Any help is appreciated!

Edited September 9, 2015 by DuoDex
removed profanity

[RidingTheFlow]

It looks like physics engine limitation, e.g. lack of integration on rotational component which leads to simplification and lack (Edit: diminishing) of certain "real life" effects like spin-stabilization.

Sadly unlikely anything can be done about it since physics is part of Unity internals I beleive. Maybe new Unity will make it better.

Edited September 8, 2015 by RidingTheFlow

[EladDv]

spin stabilization is a thing in KSP just spin a craft and see how it's more stable- the problem with this configuration is joint stiffness-the joints are not stiff enough to keep it centered in one position and start to slide every which way as you can see when the speeds are lower the spinning is good but when you ramp up the speeds physics steps are no longer keeping up and are making ever increasing approximations- not a KSP problem but a unity one

[The_Rocketeer]

Shame KSP doesn't allow us to connect both sides of a part to another part, otherwise I'd could double the "firmness" of the design by putting rotatrons on both ends and connecting both ends to the same wing, having the engine spinning in the center of the wing instead of the front.

I'm not sure what mods you're using - is the rotor itself a modded part or something u've fabricated stock?

[Yasmy]

So how does KSP model the moments of inertia of a craft? Is each part treated as a point particle, a solid body, a shell, or as something part shape-specific? I guess we could do some rotational acceleration tests of full and empty fuel tanks and other parts to find out, but someone must know already. For a large vessel, a point-particle model should work just fine, but it could behave poorly for some corner cases. Without knowing how rotational inertia works in KSP, it is hard to make a good physics based argument.

One thing you could try is to change the physics max time step, I think. I could be wrong. I don't know much about this. But if you can slow down the integrator, and still have stability issues, then maybe it's a design flaw, and not something like buildup of numerical errors. But this is highly speculative.

You seem to lose stability when you crank up the throttle to max. Does it become unstable if you leave it at half throttle? It entirely possible that your stability is a function of your angular velocity. You could be unstable when your angular velocity passes some natural resonant frequency in the craft. (Top-loading washing machines operate in the opposite manner: Below a certain speed, they are unstable and so they are kept locked on axis. Above a certain speed, they become self-centering: the basin is unlocked and allowed to float on springs, so that uneven loads do not cause the machine to wobble or break. This works because the force is out of phase with the displacement above the resonant frequency, and the basin is driven back towards the center.)

Edited September 8, 2015 by Yasmy

[JeeF]

I'm not sure what mods you're using - is the rotor itself a modded part or something u've fabricated stock?

Rotatron, from Infernal Robotics.

- - - Updated - - -

So how does KSP model the moments of inertia of a craft? Is each part treated as a point particle, a solid body, a shell, or as something part shape-specific? I guess we could do some rotational acceleration tests of full and empty fuel tanks and other parts to find out, but someone must know already. For a large vessel, a point-particle model should work just fine, but it could behave poorly for some corner cases. Without knowing how rotational inertia works in KSP, it is hard to make a good physics based argument.

One thing you could try is to change the physics max time step, I think. I could be wrong. I don't know much about this. But if you can slow down the integrator, and still have stability issues, then maybe it's a design flaw, and not something like buildup of numerical errors. But this is highly speculative.

You seem to lose stability when you crank up the throttle to max. Does it become unstable if you leave it at half throttle? It entirely possible that your stability is a function of your angular velocity. You could be unstable when your angular velocity passes some natural resonant frequency in the craft. (Top-loading washing machines operate in the opposite manner: Below a certain speed, they are unstable and so they are kept locked on axis. Above a certain speed, they become self-centering: the basin is unlocked and allowed to float on springs, so that uneven loads do not cause the machine to wobble or break. This works because the force is out of phase with the displacement above the resonant frequency, and the basin is driven back towards the center.)

One interesting thing is: If I remove the wings, I can spin that thing to a kazillion rpms, and it's rock stable. So the wobbling is being caused by the wing drag somehow.

At first I thought maybe the exhaust from the turbines was causing turbulence on the wings, but even placing the engines really far away still causes the wobbling.

I think I may try with FAR installed. Currently I don't have it installed due to a very annoying bug that was inverting my engine's _____ (forgot the word) at certain speeds. FAR may kill the wobbling. Will post results later.

[Merandix]

How about larger lifting surfaces with lower rpm? I think it's the joints on the rotor becoming unstable due to the high forces on them...

[Stoney3K]

Is it still unstable if you only try to lift up vertically and not apply any cyclic (tilt)? You could do that by angling the wings a little more in the VAB so you have more collective pitch.

It looks like your wobble starts when you move the craft out of its plane of rotation to move forward, which causes a lot of stress on the joints where the wings are attached. This causes the wings to flex, which results in your instability.

From the looks of it, the physics is correct. You need to swivel your rotor axis to move, not apply torque to it with a reaction wheel.

[xtoro]

Have you tried installing Kerbal Joint Reinforcement? It might help by preventing joints from bending...

[The_Rocketeer]

What if you add another free-coasting rotor part in front of the propellor and attach it to a parallel support with struts?

What you currently have is like the wheel on a car, attached at one end. What I'm suggesting is like a wheel on a bike, attached at both ends to an external framework.

[windows_x_seven]

Most likely game limitations. Try using FAR.

Edited September 9, 2015 by windows_x_seven

[JeeF]

How about larger lifting surfaces with lower rpm? I think it's the joints on the rotor becoming unstable due to the high forces on them...

That works but would look damn ugly and would be impossible to achieve any decent speed. I'm looking for realism, 9 thousand RPM and speeds over 200m/s on the airplane.

Is it still unstable if you only try to lift up vertically and not apply any cyclic (tilt)? You could do that by angling the wings a little more in the VAB so you have more collective pitch.

It looks like your wobble starts when you move the craft out of its plane of rotation to move forward, which causes a lot of stress on the joints where the wings are attached. This causes the wings to flex, which results in your instability.

From the looks of it, the physics is correct. You need to swivel your rotor axis to move, not apply torque to it with a reaction wheel.

Wasn't even touching the keyboard, just climbing straight up and increasing throttle.

Have you tried installing Kerbal Joint Reinforcement? It might help by preventing joints from bending...

Yep, using KJR.

What if you add another free-coasting rotor part in front of the propellor and attach it to a parallel support with struts?

What you currently have is like the wheel on a car, attached at one end. What I'm suggesting is like a wheel on a bike, attached at both ends to an external framework.

Tried that, the whole structure wobbles.