How should Reaction Wheels work?

[panzer1b]

Right now, i would like to see some sort of larger reaction wheels for larger ships.

Realistic or not, i actually like the way the wheels work. Perhaps make them consume a little more power, or have some heat production at max power, or some cooldown to teh amount of torque they can give in a time frame, but plz dont lower their torque. All that would so is force us to use even more parts in any ship (large craft require 20-50 of them to turn at all). Right now, without any dedicated MK-2, MK-3, or 3.5m reaction wheels, capital ships require massive amounts of them as is.

[Tuareg]

In theory, very little. In practice, a lot. A momentum wheel needs relatively large mass and high speed, with speed variance not being very important. A reaction wheel needs high precision control of the wheel at much lower speeds, to control how much torque is being added. It is extremely difficult to build a device that satisfies the requirements for both, so we don't, generally speaking.

well, ISS has 4 little momentumgyros to stabilise it, in the meantime it also use those gyros to keep the station rotating in sync with its orbit (and yes, it has to desaturate but just because of the rotation effect) the gyros about the size of a human torso while the ISS is HUGE compared to them...so yes they are used both way in practice.

what we have to clear up i think, if reaction wheels will be changed, saturation is a problem only if you keep spinning the ship faster and faster for so long that the wheels reach their maximum speed. then they should eat rcs for desaturation to be able to keep accelerate the rotation (negative acceleration is acceleration too, so if you want to stop a rotating craft it would have the same problem). while keeping a craft stable (landed or not) needs a different effect from the same wheels but it should be weaker than the active acceleration effect. if the angular force exceeds the passive (gyro) effect, the active can still help to stabilise it, but then saturation comes again.

my personal view is that this would be way too complex for a game like ksp (they cant even manage a proper sas sadly)

Edited February 15, 2015 by Tuareg

[rkman]

Imo reaction wheels should 'cheat' as little as possible (preferably not at all) wrt how much electrical energy they consume to produce torque. If nothing else, the amount of cheating should not differ to much between different reaction wheels.

Also a cause of the 'jittering' of a small craft when its SAS is on hold pro/retro, +-normal or +-radial, might be that the reaction wheel is simply overpowered for the craft (aside from SAS pid issues).

I don't think we need reaction wheels so powerful that it can make a landed craft balance on two legs.

Personally i don't think saturation adds much to the gameplay, and i certainly do not see the point of making reaction wheels realistic wrt saturation, while not making them realistic wrt the amount of torque they produce.

I don't fancy having to spend 30 mins turning my capital ship because they got nerfed.

You don't have to. Use RCS, that's what it is for.

If the idea is that we should not need to use RCS, then why is it even in the game?

[sherkaner]

I'm absolutely in favor of nerfed reaction wheels overall, to encourage more RCS usage.

[SaturnV]

They're too powerful now, but I don't know if We should nerf it. A big lander can stands on his edge at 45 degree? Not IRL, but looks very kerbal.

If we are nerfing that, then please keep it enough to turn an interplanetery ship with acceptable speed (and with acceptable amount of RWs, like 3~5). Spending 10 minutes turning your ship towards prograde is NOT fun, KSP is a game before a simulator.

I‘d like to see a slower buildup speed of RWs, for better gameplay. Currently it feels like instant-response control with few RWs, which looks very unrealistc.

As for the saturation system, players need 3 more indicators to check the rotation speed for each RW. That will be tedious.

[SorryDave]

Maybe they should just consume a great deal of electricity.

RCS thrusters are horrible for stabilizing a large vehicle. They are just okay for turning a vehicle even if only for a limited number of times. When used to stabilize they turn on and off as they try to counter harmless rocking and consume a lot of fuel. They can be tough to setup on multistage rockets and can translate your vehicle as well as rotate it.

Large tall pointed rockets can easily consume a lot of fuel on its way to orbit, wether or not stabilization was required.

[Jouni]

If we are nerfing that, then please keep it enough to turn an interplanetery ship with acceptable speed (and with acceptable amount of RWs, like 3~5). Spending 10 minutes turning your ship towards prograde is NOT fun, KSP is a game before a simulator.

You can always use RCS for turning. That's why the game has those huge monopropellant tanks. RCS is better for huge ships anyway, because it causes less wobble than reaction wheels.

RCS thrusters are horrible for stabilizing a large vehicle. They are just okay for turning a vehicle even if only for a limited number of times. When used to stabilize they turn on and off as they try to counter harmless rocking and consume a lot of fuel. They can be tough to setup on multistage rockets and can translate your vehicle as well as rotate it.

You're using SAS in a wrong way. With large ships, a better idea is to kill most of the rotation manually, and then switch SAS on a few times for a couple of seconds at a time. It's not an autopilot, just a reactive stabilization system based on a rather simple controller.

Rockets don't need reaction wheels or RCS at all, except for the last stage just before the circularization burn. Thrust vectoring provides more than enough control authority during the ascent.

[Renegrade]

I think they should be nerfed, at the moment they are the best solution almost always.

^ This is the core problem here.

I'd just like for them to actually hold a position. I know a plane is capable of holding say, a 20 degree pitch up attitude, but if I stop tapping S while SAS is engaged, it slowly sinks back down. So I end up just going tap tap tap tap tap tap tap tap tap tap tap tap.. granted, this may be due to FAR's atmosphere, but with 1.0 air acting like air, it's not unreasonable to assume it won't behave the same way.

That's a problem with SAS, not with reaction wheels. And those are two separate, and distinct things (thanks to Past Squad for introducing a massive nomenclature error).

I think for stock, simply reducing the torque for any given reaction wheel to about 1/10th (or less) what it is now would work well. That would put more of an emphasis on RCS. Saturable wheels would be cool but I think a sufficient nerf would reduce their importance enough that it wouldn't really matter.

That's exactly what my Horrible Nerf does (actually I used that as a middle point for the 1.25m wheel (20->2), the 0.625m wheel I set to 0.25 (1/20th) and the large to 8 (roughly 1/4)). I coupled that with cost reductions to RCS (currently, a quad of quads and the 1.25m RCS tank costs about as much as a Skipper...), RCS-only becomes very attractive, especially on shorter missions. For longer missions, hybrid designs reduce RCS consumption (toggle RCS on to turn quickly, toggle it off once on the correct vector and let the RW/thrust vectoring stabilize the burn). Small probes can get away with just reaction wheels, but meh...we don't have tiny-size quads anyhow.

Actually I'd love to see the RCS thruster pairs/quads/45-degree-quads/tiny versions etc all imported from RLA Stockalike to stock. They look very stock-ish and are extremely handy.

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

Horrible Nerf removes them from pods. Haven't touched probe cores yet. However, that would only be temporary RCS use...

[Stormweaver]

I'm in favour of nerfing the torque values significantly. While giving RCS a use in stock would be great on it's own, we're also getting the ISP fix and overhauled aerodynamics in 1.0. With planes actually making a degree of sense and the RCS thrusters having significantly less thrust in atmosphere, reaction wheels in aircraft are going to stick out like a sore thumb as they currently are.

[SorryDave]

With large ships, a better idea is to kill most of the rotation manually, and then switch SAS on a few times for a couple of seconds at a time...

...Rockets don't need reaction wheels or RCS at all, except for the last stage just before the circularization burn. Thrust vectoring provides more than enough control authority during the ascent.

I haven't found this to be true, if you can turn a ship manually it is probably because the command module has a reaction wheel and can provide torque. I generally only add reaction wheels for craft that must change orientation quickly, like for docking.

Thrust vectoring alone does not help keep a rocket on a true course during accent or on a 90 degree heading without a lot of corrections. Fly without toggling "t" for a waste of fuel.

Stability assist is not autopilot. But a good reason to use stability assist is to use pilot skills to land a vehicle without any horizontal velocity, or to change inclination while keeping a circular orbit.

[Jouni]

Thrust vectoring alone does not help keep a rocket on a true course during accent or on a 90 degree heading without a lot of corrections. Fly without toggling "t" for a waste of fuel.

I stopped adding reaction wheels and RCS to launch vehicles in 0.24, because I found them completely unnecessary. There were two reasons:

1. I learned to fly better. I used to steer too aggressively, requiring a lot of control authority to compensate. Over time, I learned to rely on inertia and gentle maneuvers.
   
2. I built rockets instead of pancakes. The further away the engines are from the center of mass, the more torque you get. A single Skipper placed 5 m away from the center of mass is roughly equivalent to two large ASAS modules. With maximal thrust vectoring, the forward component of the thrust is reduced by about 0.1 kN.
   

[Renegrade]

I stopped adding reaction wheels and RCS to launch vehicles in 0.24, because I found them completely unnecessary.

Agreed. For launch vehicles in stock and FAR, you really only need thrust vectoring.

I actually just finished some Munar flybys in my Horrible Nerf (+FAR) with nothing but thrust vectoring (T45s and 909s). Why? I'm playing with 10% science, and HN removes command pod reaction wheels, and at 10%, I wasn't about to waste points on moving tailfins.

[NoMrBond]

Reaction Wheels should work best at the CoM (also their force should apply at their location), RCS should work best away from the CoM

Reaction Wheels should build up a waste resource (say Momentum?) which slowly decays, however;

• The amount of Torque available from the RW should go down as it becomes saturated
  
• The power draw of the RW should increase as Momentum saturation increases
  
• The RW should continuously draw power until the Waste/Momentum is depleted
  

The amount of force they provide should be looked at too, RW's should be about keeping a vessel steady, not providing free breakdancing capabilities

[Tuareg]

I stopped adding reaction wheels and RCS to launch vehicles in 0.24, because I found them completely unnecessary. There were two reasons:

1. I learned to fly better. I used to steer too aggressively, requiring a lot of control authority to compensate. Over time, I learned to rely on inertia and gentle maneuvers.
   
2. I built rockets instead of pancakes. The further away the engines are from the center of mass, the more torque you get. A single Skipper placed 5 m away from the center of mass is roughly equivalent to two large ASAS modules. With maximal thrust vectoring, the forward component of the thrust is reduced by about 0.1 kN.
   

its all right if u stopped using them but casuals need it. 90% of the players will never launch a rocket like you do... i want flying saucepans with candles and i expect them to work because its a game. i wanna strap 2 rovers with dockingstations plus scienslabs and solar towers and so on to one single rocket and send it to the mun because i want to play on the mun and for me to launch a rocket for every bit of these is boring and i want that craft to turn when i want it to turn even when the engines are off... and i want that rocket to land on the mun.

I want this to fly because i dont want to send 10 rockets to bring them to the mun and i dont want to build a giant copy of the saturn V either when i can send this stuff up with a simple 2 stage monster...

Javascript is disabled. View full album

so ppl who comes just to PLAY and have fun needs these extra controls

Edited February 17, 2015 by Tuareg

[Red Iron Crown]

Reaction Wheels should work best at the CoM (also their force should apply at their location), RCS should work best away from the CoM

Real life reaction wheels don't work any better at CoM, the torque they impart is the same regardless of location. KSP RWs apply their force at their location, build a flexy ship with wheels at the ends and you can see the effect clearly. And RCS does work best away from the CoM, due to leverage (it applies a force rather than a torque).

[SorryDave]

its all right if u stopped using them but casuals need it...

so ppl who comes just to PLAY and have fun needs these extra controls

I agree with what you said but I would add that Non-casual players like myself need it as well.

Maybe it isn't needed for launching small rockets or simple missions like fly-bys. For more advanced missions like base building and orbital construction reaction wheels really help.

Docking with only RSC thrusters on a heavy payload is a real pain, as soon as you take your finger off a key, you need to turn off the thrusters to avoid the game trying to stop you from the rotation you started. They are best used for translation

Designing a vehicle to land on the moon with the cockpit reaction wheels disabled using only vectored thrust might require a good pilot, but it isn't good rocket design. Orbiting the moon alone with only vectored engines for control can be done if you aren't trying for any precision.

There are better challenges in this game than tapping away WWWSWSSSW.

[Jouni]

Maybe it isn't needed for launching small rockets or simple missions like fly-bys. For more advanced missions like base building and orbital construction reaction wheels really help.

Docking with only RSC thrusters on a heavy payload is a real pain, as soon as you take your finger off a key, you need to turn off the thrusters to avoid the game trying to stop you from the rotation you started. They are best used for translation

You're getting it wrong. Reaction wheels are good for small and medium ships, especially if the command pod is strong enough to rotate the ship. Larger ships are more efficient and dock better with just RCS. Similarly, small rockets may use reaction wheels or fins for steering, while large rockets save a lot of mass by using thrust vectoring.

[Invader Myk]

Reaction wheels and SAS should not be intrinsically linked. The SAS should be an on-board computer (OBC) that translates the pilot's commands to the reaction control system. Reaction wheels, like RCS thrusters, engine gimbals, and wing flaps, are part of the reaction control system.

There should be an option to activate or de-activate reaction wheels just like there is an option for the RCS thrusters.

Giving a command to change the attitude on one axis should not have the SAS/OBC forget how it was maintaining the attitude on the other two axes.

If a saturation system is not used, then reaction wheels could be balanced by giving them a maximum angular momentum that would disable a reaction wheel from increasing the speed in a direction that has reached the RW's maximum AM. Input from the pilot would still change the target heading, but the RW would not be used again until the input is changed. This would enable reaction wheels to quickly reorient small craft but make reorientation take more time as a craft's mass increases.

The SAS/OBC should try to orient the ship to the heading where the last input was received. Instantly stopping rotation, like how SAS operated in early versions, was too unrealistic even for the intent of the game's design. Similarly, the continuous changing of the target heading is equally unrealistic for the intent of the game's design.

In relation to use on atmospheric craft, the SAS/OBC should try to maintain the angle above the horizon instead of slowly re-orienting the craft towards its direction of travel. This will allow an aerodynamically stable craft to maintain altitude. If the problem of continuously changing the target heading is fixed, this should be taken care of by that same fix.

The SAS/OBC should not give commands to parts that physically can not affect rotation. Such as, flaps and winglets while in vacuum and the gimbals of throttled down or shutdown engines.

SAS/OBC adjusted throttling should be implemented for multiple active engines.

I like the idea of reaction wheel saturation, but I feel it is more of an advanced feature that should appear in mods instead of the core game. That being said, the core game would need to programmed in a way that would allow for this mechanic to be added.

edit:

Posted then went back to removing eccentricity from my station's orbit. I performed a prograde burn then told Jeb to prepare for a retrograde burn. How does he do this? Does he spin up the reaction wheels just enough to get the ship turning, then spin them the other direction to stop rotation close to a retrograde orientation? Absolutely not, this is Jeb and he does everything as much as possible for as long as possible. The station starts careening wildly towards retrograde. Quickly the retrograde marker is seen on the navball, but Jeb doesn't slow down the ship's rotation. He gives those wheels all the juice they'll take up to the instant that the ship is facing retrograde. Then Jeb slams the reaction wheels into reverse stripping the gears, cracking the axle, and shattering the wheel. Hundreds of pieces of shrapnel tear though the ship and crew leaving a cloud metallic confetti in orbit, if reaction wheels were any where near realistic. Instead the rotation slows and finally stops about 90^o beyond retrograde. So Jeb keeps spinning those wheels until once again the ship is heading retrograde before he starts them countering the rotation. This continues for about five cycles before Jeb gets the ship steady.

Edited February 18, 2015 by Invader Myk
Jeb's adrenaline addiction

[SorryDave]

You're getting it wrong. Reaction wheels are good for small and medium ships, especially if the command pod is strong enough to rotate the ship.

I'm not getting this all wrong, but I do appreciate your insight and taking the time to elaborate on my opinion.

In fact I agree with you completely that reaction wheels are great for small medium craft and vectored engines are only good for steering during accent (large or small).

Even for large craft reaction wheels help when spinning/banking to orientate parts and do so without burning fuel. I think a large craft with RCS thrusters and a reaction wheel would be better than with just RCS.

For heavy ships like in your links, I like to use action keys and LV-1/LV-1R engines instead of the Vernor. Sure they are heavier at .09 instead of .08 per 12 thrust but they have better ISP and in my opinion look really spacey on big rockets. The thrust is also changeable preset and with the throttle allowing better control.

I tend to use the Vernor since their introduction just for translation. The monopropllant thrusters are a waste of fuel on ships that big, in my opinion.

With all respect sir, this isn't all wrong, just my humble opinion.

[Renegrade]

Real life reaction wheels don't work any better at CoM, the torque they impart is the same regardless of location.

That's not right - pick up a baseball bat, and try to turn it from it's center of mass with a torque - now try to turn it from the handle. Same principle. That's a torque. Reaction wheels/CMGs make a torque around their main axis (well, CMGs can have multiple axis but let's not get too complicated here)..

(when I say turn it, I don't mean roll it along it's long axis, naturally. Although that does illustrate that the CoM is really a line when you're talking about applying a torque...)

KSP RWs apply their force at their location, build a flexy ship with wheels at the ends and you can see the effect clearly. And RCS does work best away from the CoM, due to leverage (it applies a force rather than a torque).

This bit's right though.

• The RW should continuously draw power until the Waste/Momentum is depleted
  

That's a good point. Real reaction wheels draw power when they have some momentum stored..and CMGs always draw power.

The amount of force they provide should be looked at too, RW's should be about keeping a vessel steady, not providing free breakdancing capabilities

Why? The smallest wheel is only providing about 3,300 ft-lb of torque, or about ten 5L V8 Mustang engines worth. That isn't a lot, is it? I forgot the sarcasm tags, didn't I?

For heavy ships like in your links, I like to use action keys and LV-1/LV-1R engines instead of the Vernor. Sure they are heavier at .09 instead of .08 per 12 thrust but they have better ISP and in my opinion look really spacey on big rockets. The thrust is also changeable preset and with the throttle allowing better control.

FYI the Vernor is #lolmassless, it's lighter than LV-1 series. Why? I have no idea. Oh well, maybe after 1.0, all drives under 500kg will be massless and we can build RCS systems out of 48-7Ses :/

[Crzyrndm]

That's not right - pick up a baseball bat ... Same principle.

Not quite. Spacecraft aren't fighting gravity to stay at the same altitude, the baseball bat is. I'm struggling to think of an earth analogy because it would have to rotate about the CoM while being torqued from any point about a vertical axis (no gravity), but RIC's statement should be atleast approximately right (I can't be bothered doing the math right now -.-)

---

So the discussion here and the suggestions here may have sparked my interest in how this could actually play inside of KSP. Some experimentation quickly put the first iteration of the idea on a slippery slope, not only would timewarp pose some issues, but KSP damps rotation above a certain rate (read: you slow down without any input). So as suggested, stored momentum decaying over time is not only sounds better for gameplay, you don't have to fight KSP much over it either.

This is the prototype I've been playing around with for those that are interested (NOTE: No promises over the validity of the maths involved. This is back of the scratchpad type stuff. E: Have since found several minor holes, invalid math is guaranteed ). Reaction wheels store momentum in world space coordinates, the greater the amount stored, the less torque is available along aligned axes. The inflight tweakables allow the torque storage and bleed rates to be adjusted (for getting an idea of how things feel). Bleed Rate is a % of the maximum part torque being removed every second, Max Scale controls how much momentum can be stored (and thus the rate at which available torque is reduced).

I did make some small changes to the suggested implementation (torque fades away instead of cutting down when you hit the limit being the main one)

Edited February 18, 2015 by Crzyrndm

[Superfluous J]

Not quite. Spacecraft aren't fighting gravity to stay at the same altitude, the baseball bat is. I'm struggling to think of an earth analogy because it would have to rotate about the CoM while being torqued from any point about a vertical axis (no gravity), but RIC's statement should be atleast approximately right (I can't be bothered doing the math right now -.-)

Decades ago while I was in school, we used a flywheel (Literally a bicycle wheel with handles for an axle, so you could hold it by the center and get it spinning very fast) and an office chair. Sit in the chair, have someone spin up the flywheel and hand it to you. Then futz about with and and see what happens.

Not perfect but it does show you how you still rotate around COM even if you hold the flywheel way out at arm's length.

[Jouni]

Even for large craft reaction wheels help when spinning/banking to orientate parts and do so without burning fuel. I think a large craft with RCS thrusters and a reaction wheel would be better than with just RCS.

As ships grow larger, reaction wheels and RCS thrusters scale up in a different way.

Let's assume that we have a small ship with enough torque for angular acceleration of 1 rad/s^2. We build a large ship that's 4x longer, roughly similar in shape, and uses the same fraction of its mass for steering. Because the ship is 4x larger, it's 64x heavier and its radius of gyration is 4x longer than in the small ship.

If the large ship uses reaction wheels for steering, it has 64x more of them than in the small ship, generating 64x more torque. Unfortunately the moment of inertia of the ship is 1024x higher, reducing the angular acceleration to 0.0625 rad/s^2. The large ship is therefore 16x less maneuverable than the small ship.

If we use RCS thrusters instead, we have 64x more of them in the large ship, but we can also place them 4x farther away from the center of mass than in the small ship. As the torque is 256x higher, the ship is capable of angular acceleration of 0.25 rad/s^2, making it 4x less maneuverable than the small ship but 4x more maneuverable than the same ship using reaction wheels.

For any given mission profile and ship shape, there is a critical size of the ship, at which reaction wheels and RCS thrusters are equally efficient. If the ship is smaller than that, you need less mass for steering if you use reaction wheels instead of RCS thrusters. In ships larger than that, any level of maneuverability can be achieved with less mass by using RCS thrusters and monopropellant instead of reaction wheels.

For heavy ships like in your links, I like to use action keys and LV-1/LV-1R engines instead of the Vernor. Sure they are heavier at .09 instead of .08 per 12 thrust but they have better ISP and in my opinion look really spacey on big rockets. The thrust is also changeable preset and with the throttle allowing better control.

The problem in using any LFO engines (including the Vernor) for steering is fuel flow. It's much harder to get it right than with RCS thrusters, especially if you already have a lot of fuel tanks and fuel lines in the ship. The O-10 might be a better choice, if you can spare enough action groups for toggling them.

[Red Iron Crown]

That's not right - pick up a baseball bat, and try to turn it from it's center of mass with a torque - now try to turn it from the handle. Same principle. That's a torque. Reaction wheels/CMGs make a torque around their main axis (well, CMGs can have multiple axis but let's not get too complicated here)..

That difference is due not to the torque applied being different, but the different polar moment of inertia when rotating it on a different axis. If you suspend the bat on a string so the axis is always the same, the same torque applied will create the same rotation regardless of where along its length it is applied.

[Tuareg]

Real life reaction wheels don't work any better at CoM, the torque they impart is the same regardless of location. KSP RWs apply their force at their location, build a flexy ship with wheels at the ends and you can see the effect clearly. And RCS does work best away from the CoM, due to leverage (it applies a force rather than a torque).

lol. physics. The moment of inertial is minimum around an axis that passes through the CoM, so it's "harder" to get the object to rotate around an off-CoM axis. also, if there isnt a fixed axis (the center of rotation isnt pinned to an infinitely heavy object) the object u wanna rotate off-centre will move away on a spiral trajectory as the centripetal forces are off-centre too, aka its far better to place the RW into CM. its an other question that KSP (and unity) doesnt calculate these forces fight.