Rotating Station Stability

[1ArmedScissor]

Hi, I've built a station that is meant to rotate around one axis. To my understanding, once that rotation is achieved, sas is turned off and no other thrust is applied, the station should continue to rotate along that axis. However, in 100k orbit around Kerbin, the station starts tilting once i stop accelerating the rotation. Any clues what I might be missing?

 

 

Persistent Rotation will work for a while, but can lead to very sudden unprovoked shifts in the axis.

I have tried removing different parts. The only thing I can get to rotate without instability is a single-file rocket.

Edited August 24, 2017 by 1ArmedScissor

[WanderingKid]

Rounding errors, offset CoM, part #'s, clipping being erratic... Lots of stuff.  Pics please.

Also, side note, once it goes 'on rails' when you switch away from it, the spinning will stop.  Also if you time warp.

[Snark]

Also, depending on how your station is shaped and what axis you're trying to spin it on, you could be running into this:

https://en.wikipedia.org/wiki/Tennis_racket_theorem

Also, how rigid is your station?  Even a small amount of bending could throw off rotation and result in a chaotic system.

[Yoochem]

This mod might help if you want to maintain your spin throughout warp:

 

[WanderingKid]

The docking port and claw attachment off the center of spin (assuming the blue arms are the traveling arms) will slightly offset Center of Mass because it's not matched on the other side.  With CoM slightly off, and SAS not countering, you will get imbalance eventually, particularly with a good sized fuel tank there.

To add additional minor influences, on another axis there's something hanging off the large engine's gas tank that looks like a tiny cylinder of something with a docking port on the end.

KSP and rounding errors being what they are, if you're not perfect, don't spin.

 

[1ArmedScissor]

Those side elements might not help, but even without them it always starts to lose stability. What I haven't tried is a setup where most of the weight is outside on the rotating "ring". Not knowing much about physics, it felt more natural to leave the weight in the middle for stability (and cheaper to accelerate and decelerate into rotation I assumed).

Thanks for all your help so far though, much appreciated!

Do you have any good examples for fully rotating stations?

[OHara]

Stations that look like wheels are stable in rotation.  KerbalX is a nice place to find examples

Your picture is helpful.  Your station looks a lot like the tennis-racket that @Snark suspected above, with the main engine being the handle.

On 8/23/2017 at 0:47 PM, 1ArmedScissor said:

To my understanding, once that rotation is achieved, sas is turned off and no other thrust is applied, the station should continue to rotate along that axis.

That is not true in general, only for rotation around the axis with the highest 'moment of inertia', that is, the axis that puts the mass as far form the rotation axis as possible, like a wagon wheel rotating on its axis.

A tennis racket is stable when rotated as you would if you swing it as if to use it as an axe.  Flipping it as a frying pan is quickly unstable, as Snark's link says.  

Spinning it around its handle is stable for short times only---the tennis racket would have the same momentum but with lower kinetic energy if it could shift its rotation axis to move more like an axe, and as it flexes and loses energy in creaking joints it eventually does rotate like an axe.

Assuming you want your station to rotate about the axis of its main engine, try re-arranging so the heavy things are closer to the plane perpendicular to that axis, moving the engine to the center, and moving anything else you can to make it more like a wheel with the center of mass near its center.

Edited August 26, 2017 by OHara

[qzgy]

Try removing your random monoprop storage duct-taped to the hub and the isru unit, and making your station longer (increase the length between capsule and engine). Then try spinning it about the long axis (capsule to engine). Your CoM is not directly through the axis of rotation at the moment, so you will experience instability in rotation.

On 8/25/2017 at 4:26 PM, OHara said:

A tennis racket is stable when rotated as you would if you swing it as if to use it as an axe.  Flipping it as a frying pan is quickly unstable, as Snark's link says.  

Spinning it around its handle is stable for short times only---the tennis racket would have the same momentum but with lower kinetic energy if it could shift its rotation axis to move more like an axe, and as it flexes and loses energy in creaking joints it eventually does rotate like an axe.

Uhh, spinning around the handle is stable. Like permanently, AFAIK. It doesn't have a tendency to be unstable in that axis.

Also momentum and kinetic energy have the same basic units - mass and velocity. Explain how keeping one the same while not affecting the other one works? (momentum p=mv, kinetic energy KE=1/2mv²)

[Dafni]

On 24.8.2017 at 0:48 AM, Snark said:

Also, depending on how your station is shaped and what axis you're trying to spin it on, you could be running into this:

https://en.wikipedia.org/wiki/Tennis_racket_theorem

I was thinking about that too @Snark but do you know if KSP is smart enough for this?

[OHara]

3 hours ago, qzgy said:

spinning around the handle

is stable for a perfectly rigid tennis racket. But if the rotation is the tiniest bit off the axis of symmetry, the rotation axis drifts through the body. If the the body is not perfectly rigid, it flexes and loses energy and eventually (after maybe an hour) spins like an axe.  The energy E=0.5mv² can go down while conserving momentum because the 'v' is different for different parts of the rotating body.

The 19th-century classical mechanics way of working out the details is to use the constant-of-motion angular momentum L = I ω where I is the moment of inertia, and then energy E=0.5 I ω²,  The moment of inertia depends on which axis the body rotates about, so it can drift to rotating about the axis with the large I, have a slower rotation rate ω, and lower energy E, while keeping the same angular momentum L. 

There's video of a demonstration in the space shuttle skylab.

Edited August 29, 2017 by OHara

[Snark]

17 hours ago, Dafni said:

I was thinking about that too @Snark but do you know if KSP is smart enough for this?

I'd phrase it more like this:  I suspect that KSP is probably not smart enough not to do this. 

So-called "effects" like this one (or, for example, "Oberth effect") aren't actual, real "effects" that need special code to model them.  Rather, they're simply natural consequences of Newton's laws.

And Newton's laws are really darn simple.  The calculations aren't necessarily quick or cheap; if you're dealing with a large collection of objects all interacting with each other, the number of interactions goes through the roof and it takes a whole lot of computations.  But the math is really simple, and it's basically a matter of repeatedly applying very simple equations to lots of individual components.

So, basically any physics engine is going to produce behavior like this, if it accurately models moments of inertia.  In other words:  I'm guessing that KSP does, in fact, demonstrate this effect-- not because any Squad dev explicitly modeled it, but just because they're sitting on top of the Unity physics engine, which is presumably decent at modeling the interactions of rigid objects colliding, rotating, etc.

Of course, that's just a guess, I haven't tested this myself in-game.  Perhaps some enterprising individual would care to perform the experiment? 

[Hotaru]

PS. Not my video, to be clear. Just showing that it'd been done.

Edited August 30, 2017 by Hotaru