How should Reaction Wheels work?

[Tuareg]

Actually I strongly suspect in this situation, the 'missing' inertia does something weird like move the CoM or is dissipated as heat as the offset mounted RW flexes the craft.

if you go back a couple of posts this is exactly what i wrote if you would apply torque to a mass-less object at the end of it it wouldn't rotate around its com but around the axis the torque applied. the object doesn't just change this behavior with having mass just cos why not or because "all our empirical observations". it has a reason, there is a force compensate for that change and that's where the missing torque is going

- - - Updated - - -

What is not clear to me is why the larger object would rotate around a different axis. Can you elaborate?

what? we are talking about the SAME object rotating around different axes

[Red Iron Crown]

Tuareg, surely you can explain your points without being condescending or dismissive about it. Phrases like "just read the sentences together with some common sense" or "not complicated... " add nothing to the conversation.

what? we are talking about the SAME object rotating around different axes

I was attempting to differentiate between the wheel itself and the larger object to which the wheel is attached. So, why would the larger object to which it is attached rotate around a different axis depending on wheel placement (let's assume a simple vessel with one wheel at CoM and another at an extremity)?

Edited February 19, 2015 by Red Iron Crown

[Tuareg]

Tuareg, surely you can explain your points without being condescending or dismissive about it. Phrases like "just read the sentences together with some common sense" or "not complicated... " add nothing to the conversation.

I was attempting to differentiate between the wheel itself and the larger object to which the wheel is attached. So, why would the larger object to which it is attached rotate around a different axis depending on wheel placement (let's assume a simple vessel with one wheel at CoM and another at an extremity)?

i might dont get your question...

when you apply a torque, though a freefalling object will rotate around its com (its just because its kinetic energy balances its movement out) the rotation center is not the com but where you apply the torque

[Red Iron Crown]

i might dont get your question...

when you apply a torque, though a freefalling object will rotate around its com (its just because its kinetic energy balances its movement out) the rotation center is not the com but where you apply the torque

I don't understand this, it seems contradictory. It will rotate around its CoM but the rotation center is the location of the reaction wheel. How can it be both?

[Tuareg]

I don't understand this, it seems contradictory. It will rotate around its CoM but the rotation center is the location of the reaction wheel. How can it be both?

The basic principle of physics is that we use for every calculations about forces the point where the forces are applied. every differentiation has to be calculated with derived forces applied on an arm equal with the distance. there is no such thing that you apply a torque at point A and you make calculations with the same torque at point B.

[KSK]

i might dont get your question...

when you apply a torque, though a freefalling object will rotate around its com (its just because its kinetic energy balances its movement out) the rotation center is not the com but where you apply the torque

You've just contradicted yourself, unless I'm misunderstanding you. First you say that a freefalling object will rotate about it's CoM, but then you say that the rotation centre is not the CoM but where you apply the torque?

In which case, why do both of your cross shaped example craft rotate about their centres? For the first craft (reaction wheel in the centre) it should, but for the second one (reaction wheel at the end of one 'arm' of the cross) it should not - if what you've just said is correct. You are applying the torque at the end of one arm, therefore it should rotate about that arm.

[Tuareg]

You've just contradicted yourself, unless I'm misunderstanding you. First you say that a freefalling object will rotate about it's CoM, but then you say that the rotation centre is not the CoM but where you apply the torque?

In which case, why do both of your cross shaped example craft rotate about their centres? For the first craft (reaction wheel in the centre) it should, but for the second one (reaction wheel at the end of one 'arm' of the cross) it should not - if what you've just said is correct. You are applying the torque at the end of one arm, therefore it should rotate about that arm.

omg... i've explained it already a couple of times, the rotation center is not where it seems to rotate but where you rotate it aka where you apply the force/torque. there is no any contradiction. we don't use visualization in physical calculations.

the 2 craft rotates around the same center because the objects have their kinetic energy, and that is balancing out the rotation forces of the second craft and thats why it rotates slower because those energies are taken from the torque...

[Red Iron Crown]

Please share some of the calculations you refer to. Not a wiki page, a simple worked example for us. Show that, mathematically, a wheel spun up at the CoM of a larger object adds a differing amount of angular momentum than the same wheel spun up to the same speed some distance from the CoM.

[Tuareg]

Please share some of the calculations you refer to. Not a wiki page, a simple worked example for us. Show that, mathematically, a wheel spun up at the CoM of a larger object adds a differing amount of angular momentum than the same wheel spun up to the same speed some distance from the CoM.

there is no simple calculation to it, it needs inertia matrix and integral calculations. as i've said its far more complex than what you think and there is an entire science dedicated to it. you would need to read about parallel axis theorem... (i didn't link its wiki page )

[Renegrade]

Show that, mathematically, a wheel spun up at the CoM of a larger object adds a differing amount of angular momentum than the same wheel spun up to the same speed some distance from the CoM.

Er, aren't we talking about angular velocity? O.o

[Red Iron Crown]

Er, aren't we talking about angular velocity? O.o

[Tuareg]

If the objects have the same polar moment of inertia (i.e. same mass and mass distribution, to oversimplify) they will have the same angular momentum at the same angular velocity. The distinction matters because momentum is conserved but velocity is not.

"The polar moment of inertia must not be confused with the moment of inertia, which characterizes an object's angular acceleration due to a torque." i start to feel uncomfortable.

ugh, i just asked my gf about the situation (note she is a beautician) and though i had to explain some things she knew the answer. i think its time to quit for me now and let you keep thinking... but instead of high level principles try to apply common sense. and this time dont take it as personal attack, i said it seriously.

[Renegrade]

If the objects have the same polar moment of inertia (i.e. same mass and mass distribution, to oversimplify) they will have the same angular momentum at the same angular velocity. The distinction matters because momentum is conserved but velocity is not.

RIght, angular momentum is the cross product of angular velocity and the moment of inertia (did a mini-review there. It's still foggy but clearer than before). My thoughts here are that the moment of inertia is different, resulting in a different angular velocity for a given angular momentum, and that the moment of inertia has to be different at the edges of an object than on the CoM.

(I'm visualizing the CoM as line in these cases, by the way. The RW is "on" the CoM if it's axle IS that line, and "off" The CoM if it's axle is NOT on that line)

Does this make sense at all, or am I imaging weird things again? (wouldn't be the first time)

[Red Iron Crown]

"The polar moment of inertia must not be confused with the moment of inertia, which characterizes an object's angular acceleration due to a torque." i start to feel uncomfortable.

"In classical mechanics, moment of inertia may also be called mass moment of inertia, rotational inertia, polar moment of inertia, or the angular mass." You're just being pedantic now.

[KSK]

Well I have my answer. This thread hasn't really helped me figure out how reaction wheels work in real life (I think we're largely talking past each other), but I have figured out how they should work in KSP.

1. Leave them as they are

2. Get rid of them entirely. Well almost entirely.

1.

omg... i've explained it already a couple of times, the rotation center is not where it seems to rotate but where you rotate it aka where you apply the force/torque. there is no any contradiction. we don't use visualization in physical calculations. *snip*

(my emphasis added)

Fine, but in that case I don't care about realism in this context. If I can't visualise what's going to happen if I hang a reaction wheel off a spacecraft in a particular place, then I'm far happier abstracting it out. Stick on a 'reaction wheel module' and assume that my engineers have set things up so that everything works as advertised, i.e. I have a certain amount of control authority about each axis. In the same way that I can put a monopropellant tank anywhere on my ship and just assume that the monoprop gets properly routed to my RCS thrusters.

Let the purists write a "Real Wheels" mod if they like but keep stock the way it is.

2. Get rid of the wretched things. Maybe have a small underpowered (compared to current values) reaction wheel part for satellites and possibly let the Rockomax hub have torque values, so folks can add momentum gyros to space stations if they like. Everything else, lets use RCS like we do in real life. Replace the current Reaction Wheels parts with a correspondingly sized Attitude Control Ring. The ring includes a store of monopropellant and four symmetrically placed mini RCS quads around the outside. It's intended for attitude control only and players will probably want to include the current RCS parts in their designs if they intend their ships to do any serious translation manoeuvres.

Edited February 19, 2015 by KSK

[Renegrade]

2. Get rid of the wretched things. Maybe have a small underpowered (compared to current values) reaction wheel part for satellites and possibly let the Rockomax hub have torque values, so folks can add momentum gyros to space stations if they like. Everything else, lets use RCS like we do in real life. Replace the current Reaction Wheels parts with a correspondingly sized Attitude Control Ring. The ring includes a store of monopropellant and four symmetrically placed mini RCS quads around the outside. It's intended for attitude control only and players will probably want to include the current RCS parts in their designs if they intend their ships to do any serious translation manoeuvres.

I'd be fine with reducing them to 10% of current power and reducing RCS prices. The rest of it is just out of academic interest in how it works in real life.

(Actually my KSP tests seem to imply that it doesn't matter where it's placed right now)

[Tuareg]

I'd be fine with reducing them to 10% of current power and reducing RCS prices. The rest of it is just out of academic interest in how it works in real life.

(Actually my KSP tests seem to imply that it doesn't matter where it's placed right now)

there would be a simple way to make them somewhat realistic but in my guess it would be too much for squad based on how sas still works

RW-s can have 2 effects, a weaker force what can only balance the craft in a stable state (but it actually works against controls too) and should be far weaker than the current effect (works like gyro) and one which is the same as current RW-s aka can make the craft rotate, can stop rotation and can keep balanced bigger crafts on the expense of rcs OR extra electricity /there are magnetic ways to desaturate RWs/. the programmatic problem with this is that squad would have to define a giro limit, continuously watch for the torques growing bigger than that and apply the rcs/electricity use based on that.

(this is the system iss is using)

it would make so you can balance a well built craft with very little to no expenses (with generally a continuously rolling gyro) while rotating a heavy craft or balancing a very off craft would cost a lot more (spinning up/down and desaturating the RW). i dont know for casuals how much would it be too complex

but i think simply nerf them to 10% is not the solution

[KSK]

You know it is actually possible to post a suggestion without including a snarky comment about Squad. Try it sometime - it'll make your posts a lot nicer to read.

[CaptainKipard]

12 pages is already a lot to read. Will there be a summary?

[tater]

Yeah, as said before, I think that a saturation system is in order, as well as a nerf. A nerf is needed because I can hold landers at a 45 degree angle, which is very strange, even to the newbies playing this game. There should also be a saturation system in place for 2 reasons:

1) Their fuel (electricity) is easily replenishable. This means that you can use reaction wheels for a long time, wait a bit, and then use them for a long time again.

2) They get rid of the need for RCS. If all of your orientation controls are through reaction wheels, there is no need to have RCS, except for docking. Heck, with good engine control you can DOCK without RCS, which effectively gets rid of one important aspect of the game.

The saturation system doesn't need to be complex; it can be a simple timer that slowly ticks down until you regain power. The reaction wheels NEED to be nerfed though, for the sake of balance.

I agree here 100%.

I feel in the same way about reaction wheel saturation as about limits to engine ignition. It's realistic, but it also requires a playstyle, where you plan all maneuvers in advance, and execute them carefully. I've enjoyed the casual KSP playstyle, where you fly manually, overshoot frequently, and then compensate, and reaction wheel saturation would change that.

On the other hand, reaction wheels are currently so powerful that there are no meaningful choices involving them. Nerf them to around 10 kNm/tonne (which is still around 5x higher than the CMGs on ISS), and you'll have to make real choices between using reaction wheels, RCS, and thrust vectoring.

I think the gameplay would be better, not worse <shrug>. I like having ways things can go wrong, it's often the most fun.

Or place RCS in first research tier, remove reaction wheels from pods and drones (Or make them weaker) and place reaction wheels higher in tech tree.

This would encourage players to use RCS

There should be some small, command-pod sized, low thrust RCS units pretty much right away in the tech tree. RCS is required for rendezvous, and that really needs to be the next step in player training after achieving orbit. Both Gemini and Soyuz/Progress took this path as it creates meaningful OPTIONS for early career players---direct ascent vs kerbin or mun rendezvous, or even multiple lanchs to create the craft (both USA and CCCP considered this option).

[John FX]

Trouble is, even fading saturation isn't "good enough" for some players, and while reaction wheels do get hot, linking their function to heat isn't realistic enough.

You really cannot please everyone all of the time...

RCS is great for lots of things, R-wheels will only rotate you after all, and it's not unheard of for satellites to have no rotational control.

Oh and fixing SAS hold is on the todo list, no need for even more modes

Well myself, rather than have nothing (read 'the same as we already have') I'd have something better but not 'good enough'. If you can't please all the people, even some of the time, please the ones you can...

My vote would be for a non-game ending limit on reaction wheels, having the saturation fade over time is not realistic but it would be good gameplay IMHO

Glad that SAS will be looked at to make it less 'twitchy'. It's needed.

On another note, the discussion on the tweet 'During Squadcast @maxmaps asked the viewers if we should change the way reaction wheels work in #KSP - what's your opinion?' has changed into a deep physics discussion about how reaction wheels work in real life.

Maybe some moderator is around that could split off that discussion to the science labs so this thread could be kept on track?

[sal_vager]

I'm not sure what Maxmaps actually meant by that comment to be honest, reaction wheels should rotate your vessel, whether they saturate or not doesn't change their basic function of providing rotational control.

Maybe he was thinking of reaction wheels adding other abilities? I'd not think that'd be a good idea.

[John FX]

I'm not sure what Maxmaps actually meant by that comment to be honest, reaction wheels should rotate your vessel, whether they saturate or not doesn't change their basic function of providing rotational control.

Maybe he was thinking of reaction wheels adding other abilities? I'd not think that'd be a good idea.

Nor me. I am quite happy with reaction wheels just turning my craft. I assumed he meant things like getting saturated or a reduction/increase in force as those would affect gameplay.

As an aside, there already is a mod that does the 'fading saturation' concept although it seems fairly new and is not totally bug free yet.

http://forum.kerbalspaceprogram.com/threads/111049-Plugin-0-90-%28Semi-%29Saturatable-Reaction-Wheels-v1-1-%2821-Feb%29

[Invader Myk]

Yes, I'm not really understanding what point the video is trying to make. The front craft (with the reaction wheel in the centre) - I get that. It's symmetrical so the rotation axis passes through the com. Reaction wheel spins one way about that axis, ship spins the other way about the same axis. Net angular momentum zero. But I can't figure out how the second craft works at all. How is it still managing to spin about it's centre?

First let us look at a simple lever. To make a lever effective you must have a pivot, something other than the object applying force or the object force is being applied to. On a surface and subject to gravity, the force applied to the pivot is redirected to the surface it sits on and to the opposite side of the lever from where force is applied. The net result is that the lever moves around the pivot affecting the target object and >>apparently<< leaving the pivot unaffected.

Now, when we remove the pivot a lever pushes on itself making a point along the lever the pivot. If the lever is in contact with a surface under the affect of gravity at one end, then that end will become the pivot. Again we see that the pivot >>appears<< unaffected by the forces applied to the lever. However, if that lever is unable to transfer any of the force applied to it to another object, then the point at which force is applied becomes a pivot and a second pivot is created at the lever's center of mass and a third pivot is created equidistant and opposite the the first pivot (where force is applied) from the second pivot (the center of mass). This causes three rotational movements, one around each of the pivots. If all pivots are co-located, then the lever will >>appear<< to rotate around but a single pivot. If the the pivots are not co-located, then the lever will rotate around three points. However, because two of the pivots are in dirrect opposition to each other, they will cause the lever to >>appear<< to rotate slower than the same lever with force applied at its center of mass.

[Renegrade]

There should be some small, command-pod sized, low thrust RCS units pretty much right away in the tech tree. RCS is required for rendezvous, and that really needs to be the next step in player training after achieving orbit. Both Gemini and Soyuz/Progress took this path as it creates meaningful OPTIONS for early career players---direct ascent vs kerbin or mun rendezvous, or even multiple lanchs to create the craft (both USA and CCCP considered this option).

Mercury also had little tiny RCS units too (just in the nose, and just for steering, not translation, at least, not intentional translation). Also Apollo's own units were half the power of KSP's, despite the fact that the CSM/LEM combination was something like 40t in LEO. So we could definitely use some variety in the size of RCS units.

I happen to really like the units in RLA Stockalike - not only does it have tiny ones (good for light manned ships or probes) but also bi-directional ones, a nicer looking linear port, and quads designed to be mounted at a 45-degree angle...

Couple those with reaction wheel nerfs (remove 'em from pods entirely, 10% strength on the dedicated units), and you got yourself a semi-realistic space program without having to go all the way to modelling saturation. (although I think actual, correct saturation would be a great learning tool, it would probably be too much of a 'Y U NO WERK?' issue for a big percentage of the userbase)