Saturable Reaction Wheels

[alecdacyczyn]

The Problem

I know that this has been brought up before by the physics purists...

The way that the reaction wheels currently work is grossly unrealistic. In the real world, inertial attitude control systems impart torque on spacecraft by spinning internal masses on the opposite direction (there's a little more to it with CMGs, but let's not complicate things). In doing so, they store angular momentum and have limits on how much they can store. There must be a way to eventually 'bleed off' that momentum or else that will become ineffective.

In KSP reaction wheels don't work that way. They seem to magically grab onto the aether and apply their rated torque to the spacecraft for as long as the electricity holds out; and there is never a reckoning. Players have been long known to exploit this to build landers that can keep themselves upright on steep slopes that would have otherwise causes their top-heavy ship to tip over and to indefinitely counteract unbalanced aerodynamic forces.

The Solution

The coding to rectify this would actually be very simple. Each reaction wheel, and device which contains one, would have a added state variable to represent the stored angular momentum (in either Euler-angle style or quaternion style). The stored vector will obviously need to be relative to the universal coordinate system, not the ship's orientation. Whenever the part receives a signal to apply torque, either from user input (WASDQE) or from SAS, it will attempt to increase the internal stored momentum in the opposite direction and then, if this would not cause it to go over its saturation limit, it will impart torque on the vessel as normal. Parts currently have a torque rating (in kN∙m, or something) and would be given a new Angular Momentum Saturation rating (in kN∙m∙s). I think a figure of about 5x the torque rating feels about right (that is to say, it would take about five seconds of making you ship spin faster and faster before it saturates).

There will only be a couple GUI functions needed. The user will of course need an indicator. Some small graphical or numerical thingy next to the SAS light would be great. It would show the total magnitude of AM stored in all reaction wheels throughout the craft. Holding the mouse over this indicator would show a breakdown of AM for the yaw, pitch and roll axes with regards to the crafts current orientation. Also a 'Spin Down' button would be incredibility handy (clicking on the indicator thingy would be intuitive, maybe a keyboard button too). The Spin Down button would directly signal all active reaction wheels apply a torque in the same rotational direction that as their AM vector. This would, through the normal physics, transfer AM from the reaction wheels to the rest of the ship (or to the ground if the ship happens to be landed and stable enough to not fall over). In addition to the Spin Down button on the NavBall instrument cluster, there should be a button on the menu that pops up when you click on the specific part (Inline RW, Control Module, etc) what would activate the function in that part alone while decoupling it from any SAS signals (uses for this below).

Implications

Scenario 1: In space, not spinning, with stored AM in you reaction wheels, SAS off, RCS off. Pressing the Spin Down button causes you ship to begin spinning as the AM is drained from your reaction wheels. You are now spinning like crazy. In panic, you hit the SAS. The SAS works to bring you to a stop and effectively puts that momentum right back into the reaction wheels where is started.

Scenario 2: In space, not spinning, with stored AM in you reaction wheels, SAS on, RCS off. Pressing the Spin Down button doesn't cause much visible to happen. As AM is drained from the RWs, your ship begins to turn a tiny bit before the SAS detects it and dutifully tries to cancel it out with a signal to the RWs, precisely undoing the Spin Down effects. Not much happens but a loss of electric charge. You may see you ship quiver a bit as the two systems fight each other.

Scenario 3: In space, not spinning, with stored AM in you reaction wheels, SAS on, RCS on this time. When RCS is turned on the signal from the SAS are routed to the RCS thrusters instead of the reaction wheels. So when you press the Spin Down button, and directly send a signal to the RWs, they will do their thing and impart torque on the craft. But this time the SAS will sense the change and fire the RCS thrusters to compensate. You bleed of excess angular momentum at the expense of monopropellent (without craziness!).

Scenario 4: In a lander approaching the surface of the Mun, coming down onto a 45deg slope, reaction wheels not near saturation, SAS on, RCS off. You make contact (or at least on of you lander legs does). SAS does it's thing and uses the huge reaction wheel you brought to keep you mostly upright. But the AM indicator immediately begins climbing. Soon, it reaches saturation, your ship looses torque, and the inexorable happens: you fall over. [Perhaps it should be setup such that the loss of torque happens gradually starting around 80% saturation. But in the not-increase-saturation direction it remains 100% effective (or even better).]

Scenario 5: In a lander approaching the surface of the Mun, coming down onto a sensible 15deg slope, reaction wheels not near saturation, SAS on, RCS off. You make contact. The SAS acts to dampen the rotational jolt that happens due to your imperfect landing and keeps you from falling over patheticly. The AM indicator climbs a little before you toggle off SAS and settle down gently. Once all lander legs are firmly on the surface you hit the Spin Down button. The reaction wheels transfer their stored AM to the craft. The lander is sufficiently stable and gravity is high enough that this doesn't flip you over (it's only 20% of rated torque). Whether SAS is on or not doesn't matter as you will not move significantly. You bleed off excess stored AM to the Mun through your landing legs.

Scenario 6: You dock at a space station, non-zero AM in you reaction wheels. Once you are docked the two systems effectively become one. Hitting the Spin down would cause the reaction wheels in your small ship AND the big station to dissipate momentum and cause the whole conglomeration to spin. You don't want that. So, while SAS in turned on station wide, you click on your ships reaction wheel (and/or command module, etc) and click the 'Spin Down' button in its little dialogue thing. AM is transferred to the combined ship/station and SAS responds. But because SAS signals are decoupled from specific parts while using the part specific dialogue thing, the other active reaction wheels, namely the station's huge one, do all of the work to compensate. Your ship has just transferred its excess AM to the station. (If your ship has multiple RW devices then it may be nessissary to Toggle Torque on any that you don't want absorbing it instead.)

Scenario 7: Great, now my station's RW is nearing saturation. What now?! Firing RCS thrusters like in Scenario 3 would work. But a better solution would be to stick pairs of xenon thrusters at the far reaches of your station pointing such that the translational effects cancel out as they impart (very weak) torque on the structure. Run these thrusters while SAS is turned on and (assuming you're running the right pair) the reaction wheels will (slowly) wind down while keeping the station steady. You send your excess angular momentum into space in the most mass efficient way possible.

Misc. Scenarios: You're flying under power through an atmosphere using control surfaces or through space using a gimbaled rocket engine, stored AM in your reaction wheels, SAS on. Pressing the Spin Down button imparts torque on the craft. The SAS sees this exactly as it would torque from any other source and manipulates the control surfaces or gimbals to compensate.

Conclusion

The coding changes to implement this would be small and tidy and effect only the Reaction Wheel module (and a trivial bit of the GUI). Everything else, all of the above complex behaviours, will couple through and result from the physics engine (no spaghetti code flying everywhere!). It would take little time at all. I actually looked at doing this myself as a plugin. However, after much searching, suitably detailed API documentation to tell me if this is even possible through a plugin, overriding such basic functionality, remains elusive. So I'm posting this as a Suggestion.

---------------------------------------------------------------

[EDIT: Reworked after forum responce. (Post from second page copied here).]

Yikes. Message received. In a rocket science game about delta-Vs, mass ratios, and orbital mechanics the concept of paying back your debt to Newton might boggle the poor little noobs' minds....

Everyone seems to agree that reaction wheels ought to be nerfed in some way. But the problem with simply knocking down the torque rating is the there IS a legitimate use for strong short-term torque. We need it for turning heavy ships in preparation for maneuvers, to stabilize rough landings, etc. It is not desirable to make everything sluggish (think of the last time you launched a probe and forgot to include a reaction wheel in the booster system. augh!). Down-rating is not the solution. The problem is people using their ability to produce high torque indefinitely to do things that are comically impossible. What is needed is some sort of time integral limit behavior.

I think I know how I can make everyone happy. Change the Reaction Wheel module to have a saturation limit, but also an 'aether-magical' angular momentum dissipation rate. For example:

Inline Reaction Wheel

Torque: 20 kN∙m

Saturation: 100 kN∙m∙s

Dissipation: 5 kN∙m (kilonewton meter seconds... per second)

So the system "remembers" how much angular momentum is being stored it it, but also "forgets" at a rate of 5 units/second and quickly returns to zero. At that rate you would NEVER even notice any difference during normal activities. However, suppose you land on a steep slope and need 13kN∙m to keep you lander upright. It would take 100/(13-5)=12.5 seconds for it to reach saturation. After it runs out of 'oomph' the output doesn't suddenly drop to zero; the Dissipation rate makes room at a rate of 5kN∙m. This would be the effective torque at saturation (along that vector, normal full power in the other direction). And once the demands on the reaction wheels are removed, the saturation level will return to zero after only 20 seconds. You would never need to think about bleeding off AM. And therefore, there is no need for a 'Spin Down' button of anything similar. If you are flying a ship that is slightly out of balance and needs only 4kN∙m to keep it stable then you could do so indefinitely because AM would be dissipated faster then it would build up. You wouldn't have to worry about it.

The exact figures would needs some tweaking and balancing; but you get the picture. I think this is an ideal fix. If Squad implemented this tomorrow and didn't tell anyone an experienced player could go for weeks and never notice the change.. until they tried to use the reaction wheel kludge to force something that otherwise wouldn't work.

Thoughts?

Edited May 2, 2014 by alecdacyczyn

[Luis]

Scenario 7: Great, now my station's RW is nearing saturation. What now?! Firing RCS thrusters like in Scenario 3 would work. But a better solution would be to stick pairs of xenon thrusters at the far reaches of your station pointing such that the translational effects cancel out as they impart (very weak) torque on the structure. Run these thrusters while SAS is turned on and (assuming you're running the right pair) the reaction wheels will (slowly) wind down while keeping the station steady. You send your excess angular momentum into space in the most mass efficient way possible.

How does the ISS shed angular momentum now? Are they really burning propellant to unwind their gyros every few days or weeks?

Edit: This report says that the ISS chooses its attitude so that the net torque over a whole orbit is zero. So they normally only need to burn propellant for major attitude changes to accommodate docking.

Edited April 28, 2014 by Luis
added some info

[alecdacyczyn]

How does the ISS shed angular momentum now? Are they really burning propellant to unwind their gyros every few days or weeks?

The ISS is designed and operated to avoid building up much angular momentum in the first place. But when they do need to shed some, usually after docking maneuvers, they use little thrusters (RCS).

[Kasuha]

Yes, KSP reaction wheels are unrealistic and I used to think that they should be "fixed" too. But KSP is a game. Not only it is meant to be easier to handle than real machinery (let's admit it, almost everything in KSP is unrealistic in certain way) but it has also its bugs and imperfections. Such as phantom torque applied on ships if they move at less than 750 m/s through space. Or imperfections such as all rotational momentum magically disappearing as soon as we enter time warp. I would hate if I had to take corrective action every now and then just to keep my station straight because my reaction wheels become saturated by phantom torque in a few minutes needed to dock.

Also I think realistic implementation would not be as simple as you mention. Because rotating reaction wheels don't provide just torque in certain direction, they come with gyroscopic effects as well.

[alecdacyczyn]

...rotating reaction wheels don't provide just torque in certain direction, they come with gyroscopic effects as well.

Sure, it doesn't need to be that realistic. But management of angular momentum is a key part of the operation of any real-world spacecraft.

And keep in mind that five seconds of full torque is a lot of AM. I don't think you'd ever have to worry about it in normal spaceflight. This would mainly only effect things like launches, landings, and reentries. And during most spaceflight you are using gimbaled engines which have the ability to impart torque of free. Perhaps the advantage of Advanced SAS over regular SAS is that ASAS will automatically, opportunistically manage it for you in the background during engine burns.

[xrayfishx]

This would be very confusing for new players and since SQUAD wants to keep them happy (screw the experienced ones), this will never happen in stock KSP.

[Red Iron Crown]

I'm an experienced player, and I think this adds unneeded complexity even though it increases realism. Lower torque ratings on the reaction wheels might be a good idea, I basically only use RCS for translation now.

[Sirrobert]

This is a GAME

Adding something that makes it more difficult to enjoy the GAME just for the purpose of realism is a horrible thing to do

[vexx32]

It would be a bit too in-depth for the stock game, I think, but it'd be awesome to see in a mod.

[BudgetHedgehog]

Yeah, I'm not fond of this idea. I get that it increases realism which I'm always for, but there comes a point where you really shouldn't have to worry about every single thing because it's there and realistic - that point is bleeding out monoprop to stop a ship spinning wildly during spin-downs. If that happens and I run out of monoprop, I'd have to wait until the next transfer window or whatever and that's not something I enjoy.

[TMS]

Just nerf the smaller reaction wheels. They've needed it since 0.21 and the 0.22 SAS rebalance.

[alecdacyczyn]

Yikes. Message received. In a rocket science game about delta-Vs, mass ratios, and orbital mechanics the concept of paying back your debt to Newton might boggle the poor little noobs' minds....

Everyone seems to agree that reaction wheels ought to be nerfed in some way. But the problem with simply knocking down the torque rating is the there IS a legitimate use for strong short-term torque. We need it for turning heavy ships in preparation for maneuvers, to stabilize rough landings, etc. It is not desirable to make everything sluggish (think of the last time you launched a probe and forgot to include a reaction wheel in the booster system. augh!). Down-rating is not the solution. The problem is people using their ability to produce high torque indefinitely to do things that are comically impossible. What is needed is some sort of time integral limit behavior.

I think I know how I can make everyone happy. Change the Reaction Wheel module to have a saturation limit, but also an 'aether-magical' angular momentum dissipation rate. For example:

Inline Reaction Wheel

Torque: 20 kN∙m

Saturation: 100 kN∙m∙s

Dissipation: 5 kN∙m (kilonewton meter seconds... per second)

So the system "remembers" how much angular momentum is being stored it it, but also "forgets" at a rate of 5 units/second and quickly returns to zero. At that rate you would NEVER even notice any difference during normal activities. However, suppose you land on a steep slope and need 13kN∙m to keep you lander upright. It would take 100/(13-5)=12.5 seconds for it to reach saturation. After it runs out of 'oomph' the output doesn't suddenly drop to zero; the Dissipation rate makes room at a rate of 5kN∙m. This would be the effective torque at saturation (along that vector, normal full power in the other direction). And once the demands on the reaction wheels are removed, the saturation level will return to zero after only 20 seconds. You would never need to think about bleeding off AM. And therefore, there is no need for a 'Spin Down' button of anything similar. If you are flying a ship that is slightly out of balance and needs only 4kN∙m to keep it stable then you could do so indefinitely because AM would be dissipated faster then it would build up. You wouldn't have to worry about it.

The exact figures would needs some tweaking and balancing; but you get the picture. I think this is an ideal fix. If Squad implemented this tomorrow and didn't tell anyone an experienced player could go for weeks and never notice the change.. until they tried to use the reaction wheel kludge to force something that otherwise wouldn't work.

Thoughts?

[EDIT: Oh, and the realism purists could produce a trivial mod that sets the dissipation rates on everything to zero. Then you would have to bleed of builtup AM manually with RCS, etc.]

Edited April 29, 2014 by alecdacyczyn
buy -> but

[Luis]

That actually sounds like an excellent compromise. I've been thinking about this and I had come to the conclusion that free, unlimited, magical torque did feel a bit too cheaty to me. I was planning to have a personal 'rule' for my next career save that RCS should be used for all turning manouvers. But I like this better. We get high power torque for short term manouvers and low power torque for very minor station keeping. But not enough torque to magically hold a lander on a hillside forever. If this was added to the stock game, I'd be quite happy. If it was created as a mod, I'd install it on any save that was using the other realism mods, such as FAR and Deadly Reentry.

[Sirrobert]

Now THAT, sounds amazing.

It's amazing how such a small addition can turn a suggestion from bad to great. I never even thought of that as an option

[Red Iron Crown]

Now THAT, sounds amazing.

It's amazing how such a small addition can turn a suggestion from bad to great. I never even thought of that as an option

++ Great revision. Makes the reaction wheels less exploitable without adding more micromanagement. I would reduce the dissipation rate a bit, but that's a minor detail.

It needs to remember which direction the AM it's storing came from, a subsequent torque in the opposite direction should reduce the stored AM.

[lajoswinkler]

It would be far to complex to implement the first suggestion, but the second one is a nice compromise.

[KASASpace]

So, stored torque?

That gets rejuvenated in the system.

Like this: It stores AM and uses it up once used. It then regains some of it back when the RWs are used to stop spinning in that direction.

[Dodgey]

I like that idea, it is a nice happy medium.

[PlonioFludrasco]

What about an overheating-like mechanic?

I mean, when you are trying to stop your spinning ship using RWs, a "saturation" bar appears near a RW icon in staging (remember early versions of KSP?) and starts increasing, like overheating on engines. If you stop applying torque on the ship before the bar fills, it will desaturate slowly somehow (maybe draining monopropelleant or for free but more slowly) untill it's empty. If you saturate completely the RW module, you will be unable to use the module until you desaturate it manually, for example clicking a button on the part menu (or in another way, but balanced to encourage the player to not saturate the module).

I think that's not very complicated (both to implement and being understood by new players), and nerfs quite well the actual RWs.

I know real RWs doesn't works that way, but that's a game!

EDIT: With that, you don't have to add a new resource like "Stored Angular Momentum" , that can also get new players disorientated.

Instead, it works with already present game mechanics.

Edited April 29, 2014 by PlonioFludrasco

[KASASpace]

Now that's an idea!

[tntristan12]

This is a brilliant and well thought out idea, but something I feel would be better suited as a mod or difficulty setting.

[Killerdude8]

Realism for the sake of Realism is what kills games for good, I didn't realize you hated KSP OP...

Reaction wheels are the way they are for Gameplay reasons, wouldn't be much of a game when you have no fuel left to do anything 'cause it was spent on not spinning.

KSP is not a Sim.

It is a game.(quite a fun one at that.)

[Sirrobert]

Realism for the sake of Realism is what kills games for good, I didn't realize you hated KSP OP...

Reaction wheels are the way they are for Gameplay reasons, wouldn't be much of a game when you have no fuel left to do anything 'cause it was spent on not spinning.

KSP is not a Sim.

It is a game.(quite a fun one at that.)

On the second page there is an alternate suggestion that does work.

That's what we're aproving off

[Guest]

I read a couple of complaints about a lander "magically' using sas to hold a lander on a hillside and I have a question. If the electricity is available from solar panels and batteries or RTG's and the gravity is sufficiently low enough, like on Gilly, wouldn't the lander actually hold it's position on the hillside? If not why not?

[Red Iron Crown]

I read a couple of complaints about a lander "magically' using sas to hold a lander on a hillside and I have a question. If the electricity is available from solar panels and batteries or RTG's and the gravity is sufficiently low enough, like on Gilly, wouldn't the lander actually hold it's position on the hillside? If not why not?

IRL, to hold the ship in position the reaction wheel would have to spin faster and faster until it reached its RPM limit, at which point it would fail either by bearing failure or the wheel being torn apart centrifugally. The RPM limit is the saturation the OP is suggesting should be simulated.