KSP vs IRL reaction wheels

[Lukaszenko]

I know real life reaction wheels saturate, but I've also been hearing that KSP ones are way overpowered and perhaps perform some other dubious physics tricks.

What are the real differences? How much overpowered are they? Could they be used IRL to stabilize an EVE ascending spaceship, for example?

If the differences between KSP and IRL are large, is there a mod to make them behave more realistically?

I'm basically trying to figure out if I should continue using reaction wheels as I have been, install said mod (if it exists), or ditch reaction wheels altogether and use RCS for realism.

I understand Newton's laws and all that, but when it comes to angular momentum and gyros and gimbals, it's still a bit tricky for me to wrap my head around how it all should work.

[AngusJimiKeith]

Unfortunately, I don't have any specifics sitting in front of me right now, but:

-IRL wheels are much less powerful. They would not turn a whole rocket on ascent or be much use in atmosphere.

-IRL wheels are extremely expensive, probably moreso than the equivalent RCS.

-IRL wheels are notoriously unreliable. They are very often the first thing to fail on spacecraft that carry them. We lost Kepler to bad wheels, and Hubble's have been replaced several times over the years.

[RidingTheFlow]

RL reaction wheels also accumulate rotation momentum over time (because they are not 100% friction less and not 100% effective) - eventually this will have to be cancelled by other means, for example conventional RCS - otherwise reaction wheel will stop working by reaching maximum operating rotation speed even to keep spacecraft at zero rotation.

I think this one of most important bits KSP misses, the fact that it should be pretty much impossible to "kill rotation" for extended period of time by reaction wheel only (but hey, stock KSP kills rotation even without reaction wheel, simply by time acceleration - so I guess this will have to be fixed first if any)

Edited July 17, 2014 by RidingTheFlow

[biohazard15]

KSP "reaction wheels" is not reaction wheels. They're CMGs (control moment gyroscopes) - http://en.wikipedia.org/wiki/Control_Moment_Gyro

[RidingTheFlow]

CMG suffer from eventual momentum saturation in same way though. Eventually you will have to rotate your gyro as fast as it can just to keep your craft still due to odd momentum accumulated.

There simply no 100% perfect reaction-less control in the real world.

[CaptRobau]

Completely realistic reaction wheels would be pretty hard to do in fly-by-the-seat-of-your-pants KSP. If I understand RWs correctly, they spool up for a while and then they point to where you want. That's no issue if you have computer sending commands (yaw 5 degrees, roll 10 degrees, etc.) but wouldn't work if you want to have WASD controls.

[Hattivat]

AFAIK there are three main differences:

1. In real life reaction wheels store momentum, they don't make it disappear. What this means is that if some external source (like unbalanced propulsion system) is for example applying constant spin to your craft, the reaction wheel can only stop some amount of it. At some point it will saturate, and once it has saturated, you need to use an equal amount of opposite external force (like RCS) to discharge the momentum to make the reaction wheel usable again.

2. In real life reaction wheels are an extremely expensive, specialized equipment, in KSP they are an integral part of every rocket.

3. Real life reaction wheels are small and relatively weak.

Long story short: IRL reaction wheels are pretty much useless when your craft is under any sort of acceleration, ie. when its engines are working. They are only really useful for changing orientation when your craft is drifting, like a space station or a space telescope.

So the answer to the question "Could they be used IRL to stabilize an EVE ascending spaceship?" is a definite no to my knowledge.

As for the mods: There is a mod package called Realism Overhaul that does what you describe ( http://forum.kerbalspaceprogram.com/threads/84689-0-23-5-Realism-Overhaul-ROv6-0Alpha4-Modlist-for-RSS-Update-13-July-2014 ), but it changes many other things too, so it's a pretty big commitment. Personally I love it, but the experience is very different from the stock game, so it's definitely not for everyone. If you decide to try it, I recommend doing so on a separate, second installation of the game; this way you can always come back to stock KSP if you decide that total realism is not for you.

You can also simply challenge yourself to restrict your usage of reaction wheels in a stock game, it could be an interesting way to spice things up. In any case, there is no reason to feel bad about using reaction wheels, it is a conscious design decision on part of Squad to have reaction wheels work like magic. It is to make flying rockets by hand manageable, keep in mind to no one really does that in real life.

Edited July 17, 2014 by Hattivat

[AngusJimiKeith]

Completely realistic reaction wheels would be pretty hard to do in fly-by-the-seat-of-your-pants KSP. If I understand RWs correctly, they spool up for a while and then they point to where you want. That's no issue if you have computer sending commands (yaw 5 degrees, roll 10 degrees, etc.) but wouldn't work if you want to have WASD controls.

This is also true, which brings up another point: how much disbelief can you suspend? I personally suspend my disbelief for the KSP wheels...because, to me, it would be less fun with realistic ones.

[RidingTheFlow]

AFAIK there are three main differences:

1. In real life reaction wheels store momentum, they don't make it disappear. What this means is that if some external source (like unbalanced propulsion system) is for example applying constant spin to your craft, the reaction wheel can only stop some amount of it. At some point it will saturate, and once it has saturated, you need to use an equal amount of opposite external force (like RCS) to discharge the momentum to make the reaction wheel usable again.

To add to this, plenty of external momentum in RL comes from external sources KSP does not model:

- atmospheric turbulence (yes, in RL atmosphere has no really clear boundary so over couple of days these stray air molecules could make spacecraft slowly turning even in pretty high orbit)

- solar wind (and light pressure). Yes, light from the sun actually exerts force on things it strikes. Over weeks of interplanetary flight it will be enough to incur spin on the spacecraft

Eventually this momentum will saturate reaction wheel (or whatever gyro/inertial system is used) and spacecraft will not be able to stop turning. Carrying a bit of RCS to cancel it out is the only option - but even RCS will be slowly used up.

In essence, there is should be a continuously drained resource just to make spacecraft stable as time passed. Its not possible to make spacecraft which stays stationary during infinite amount of time without expending some reaction mass (at least with technologies we currently posess).

[Sky_walker]

This is also true, which brings up another point: how much disbelief can you suspend? I personally suspend my disbelief for the KSP wheels...because, to me, it would be less fun with realistic ones.

Would it really be so much worse with lower power and limited "momentum" resource? They'd still be more effective per ton than RCS.

[RidingTheFlow]

Personally I don't see why it would be worse to promote more usage of RCS (as is it now it's barely usable for docking only). I've tried stock rebalance and quite liked what they've done there, I don't think it makes stock game experience any worse.

But its one of these things if ever be addressed, only after "scope completion" phase is finished, if ever... After all, current system is here and it works (even if imperfectly).

Edited July 17, 2014 by RidingTheFlow

[AngusJimiKeith]

Would it really be so much worse with lower power and limited "momentum" resource? They'd still be more effective per ton than RCS.

No, but I'm not going to go out of my way to change it. I'm also addressing some of the IRL differences that you didn't: delayed reaction would honestly just be a pain from a gameplay perspective, for example, and I'm not onboard the "random failures" bandwagon either.

[toril]

also with real cmg's you can't just stack them up, or plant them on the ship everywhere and expect to get a stronger rotational force, as they would rip themselves apart fighting each other unless they are aligned correctly.

[Lukaszenko]

So I tried playing with no reaction wheels for more "realism", and I learned that indeed suspension of realism is sometimes necessary in the interest of fun. But, at least now I know precisely why.

IRL they also spend countless man-hours optimizing trajectories and burns and timings, and when they finally perform them they know exactly when, where, and how long. In KSP, sometimes you make a burn, you overshoot, you turn around, then left and then right and then your're close enough....and I just spent half my ship's mass worth of RCS

[MalfunctionM1Ke]

Let me leave this here

[worir4]

There was a Scott Manly video that I watched on YouTube. I can't remember the name but it showed how to RW did stuff against the laws of physics such as rotating the craft at its center of mass. You can try this by putting a reaction wheel in between two symmetrical fuel tanks. Make sure one is empty then put it in orbit. You can actually move your craft by transferring fuel to the tanks every time you rotate once. In real life this is impossible.

[Dodgey]

The idea of reaction wheel saturation as a game mechanic came up once before and the happy medium that we came to was that the reaction wheels would accumulate momentum over time as they were used. Once they accumulated too much they would stop working untill they could bleed off the momentum, sort of like waste heat in the KSPI mod. This would mean that you still have reactionless motion however you cannot use it for an infinet amount of time in a single go.

[UberFuber]

To add to this, plenty of external momentum in RL comes from external sources KSP does not model:

- atmospheric turbulence (yes, in RL atmosphere has no really clear boundary so over couple of days these stray air molecules could make spacecraft slowly turning even in pretty high orbit)

- solar wind (and light pressure). Yes, light from the sun actually exerts force on things it strikes. Over weeks of interplanetary flight it will be enough to incur spin on the spacecraft

Eventually this momentum will saturate reaction wheel (or whatever gyro/inertial system is used) and spacecraft will not be able to stop turning. Carrying a bit of RCS to cancel it out is the only option - but even RCS will be slowly used up.

In essence, there is should be a continuously drained resource just to make spacecraft stable as time passed. Its not possible to make spacecraft which stays stationary during infinite amount of time without expending some reaction mass (at least with technologies we currently posess).

There are other methods to "bleed off" saturated reaction wheel/CMG that does not require using expendable resources.

http://en.wikipedia.org/wiki/Control_moment_gyroscope#International_Space_Station

One method is called the gravity gradient torque. Basically, if you have a really long spaceship, or at least large enough that there's a "gravity gradient" across it, you can position the spaceship such that there's sufficient tidal force trying to lock a spaceship in a fixed orientation. Then, you can desaturate the CMG while being locked that way.

[cpast]

There was a Scott Manly video that I watched on YouTube. I can't remember the name but it showed how to RW did stuff against the laws of physics such as rotating the craft at its center of mass. You can try this by putting a reaction wheel in between two symmetrical fuel tanks. Make sure one is empty then put it in orbit. You can actually move your craft by transferring fuel to the tanks every time you rotate once. In real life this is impossible.

I don't think that's the fault of reaction wheels, though, at least not if they do what you say (rotate the craft at its center of mass). Real reaction wheels *do* rotate the craft about the center of mass; anything else would violate conservation of momentum (as it would change the velocity of the center of mass without ejecting anything). Not sure what the method is precisely, but real reaction wheels can't move the craft precisely because they *do* rotate the craft about the center of mass. The real question is "what in KSP moves the center of mass without using some form of reaction"; *that* causes reactionless motion.

For instance, if the method is something like "rotate so the full tank is behind, pump fuel from back to front, rotate so the full tank is in the back, repeat", the issue is that transferring fuel from back tank to front tank would really shift the craft backwards while keeping the center of mass in the same place. Not sure if that's actually what makes this "drive", but it seems to fit the description you gave; this type of scheme works if reaction wheels rotate about the CoM and fuel transfers don't keep the CoM in the same place (shifting the craft). Propulsion appears to happen on the rotation, but the real propulsion happens when you transfer fuel (the craft may appear to move during rotation, but the important motion is the center of mass, which would only happen when you transfer fuel). A similar scheme (you just rotate to put full tank in front and transfer fuel from front to back) works if fuel transfers shift the craft to keep CoM in the same place but reaction wheels rotate about the wheel. Here, propulsion appears to happen when fuel is transferred (shifting the craft), but actual propulsion happens on the rotation (when CoM moves).

Edited July 22, 2014 by cpast

[JWag]

Somewhere I read that vernier engines were coming in 0.24. To me, these are small monoprop or biprop RCS thrusters capable of very deep throttling, producing tiny amounts of impulse for attitude and translational trim. I got very excited, as I thought we'd be able to more easily play without CMGs.

The part that was released is a fixed-throttle LFO engine linked to the RCS controls, with 12 times the thrust of the RCS quads. D'oh.

It doesn't spoil the fun or anything, but it would've been nice to have some proper verniers so we could go CMG-less.

[cpast]

Somewhere I read that vernier engines were coming in 0.24. To me, these are small monoprop or biprop RCS thrusters capable of very deep throttling, producing tiny amounts of impulse for attitude and translational trim. I got very excited, as I thought we'd be able to more easily play without CMGs.

The part that was released is a fixed-throttle LFO engine linked to the RCS controls, with 12 times the thrust of the RCS quads. D'oh.

It doesn't spoil the fun or anything, but it would've been nice to have some proper verniers so we could go CMG-less.

Vernier thruster means "small relative to some other engine"; you seem to think of it as small relative to the main RCS thrusters (it's not necessarily very throttleable; in the context you're thinking of, fine control is because it's much weaker than the main RCS, so controlling how long you fire it for gives fine control), but it can also be small relative to the main engines (e.g. on the Atlas rocket), which seem to be what KSP's are based on.

[NecroBones]

There was a Scott Manly video that I watched on YouTube. I can't remember the name but it showed how to RW did stuff against the laws of physics such as rotating the craft at its center of mass. You can try this by putting a reaction wheel in between two symmetrical fuel tanks. Make sure one is empty then put it in orbit. You can actually move your craft by transferring fuel to the tanks every time you rotate once. In real life this is impossible.

The video was about KSP's way of cheating with regards to inertia, in that it doesn't simulate a momentum shift accompanying the pumping of fuel from one tank to another. Scott was able to "walk" the ship through space because the CoM shifting that KSP does when pumping fuel, only applies as a "center of rotation" and "camera focus", rather than physically moving the spacecraft around the CoM (That is, the CoM moves in space relative to the universe, while the fuselage's position doesn't, when in fact it should be the other way around).

This is the video:

[Renegrade]

fine control is because it's much weaker than the main RCS, so controlling how long you fire it for gives fine control)

Speaking of that, I'd love to see something like RLA Stockalike's tiny RCS quads added to the core game...

[Motokid600]

Aren't the RW's shaped wrong as well? There all flat cylinders so wouldn't such a RW only be good for rotational movements? Pitch seems impossible. I try to picture what it would look like inside a KSP reaction wheel. And I can only imagine the "wheel" capable of only one, rotational axis.

[cpast]

Aren't the RW's shaped wrong as well? There all flat cylinders so wouldn't such a RW only be good for rotational movements? Pitch seems impossible. I try to picture what it would look like inside a KSP reaction wheel. And I can only imagine the "wheel" capable of only one, rotational axis.

Think "heavy smaller wheel on the side".