Reaction wheels guidelines/recommendations

[steuben]

My google-fu is weak.

Are there any recommendations for how much torque per tonne is recommended?

You could use an OKTO to turn 100 tonnes. But it would take a long time. And an Advanced Reaction Wheel Module, Large in 300kg of craft would flip it around like a tiddlywink in a tornado.

The trick is finding the happy medium.

I think I could do it from basic physics... but high school is long ago and far away. and slightly complicated by some of what I have found says that distance from CoM doesn't change the effectiveness of the modules. Unless I'm understanding some of the posts wrong.

Edited August 21, 2015 by steuben

[Wolfox]

Sounds like you might like mod Tweakable Everything, which allow you to adjust torque for reaction wheel parts, separately for different axis and even in flight!

[CalculusWarrior]

Tweakable Everything actually shows you the torque values, meaning you could do the calculations, based upon the mass of the craft. Personally, I navigate by 'feel', and put small reaction wheels on big probes, medium reaction wheels on unwieldy 1 metre ships, and the big reaction wheels on very large vessels (like big space tugs).

[ajburges]

BTW: torque per tonne is much less useful than torque per moment of inertia. That is angular acceleration.

It is truly a personal preference. Folks have different tolerances.

Too little and turns take too long and can be overpowered by engine torque. For example: my heavy space plane lifter takes over 10 seconds to turn 90° without thrust. Worse, I need to do the manuver without SAS as it can't correctly make the turn. The feedback mechanism does not consider angular momentum! That lack of consideration also means SAS can't control a ship with too much torque.

For ships with a higher moment of inertia, you may want to use reaction wheels as a damper and rely on RCS for attitude control. Large ships tend to wobble. Having enough reaction wheel torque to turn them quick trends to induce oscelations.

[Wolfox]

I think that SQUAD balanced those for us pretty nicely already. Just use the same diameter like most of the rocked, up to 2-3 lenghts of the longest tank in the given diameter, and you will be just fine.

EDIT:

Of course, situation is different for spaceplanes, stations with many parts connected by wobbly clampotrons or various aviation applications like take-off or VTOL support.

[Snark]

...some of what I have found says that distance from CoM doesn't change the effectiveness of the modules.

That's correct, it doesn't matter where on your craft you place the modules-- they're equally effective anywhere.

(The only case in which placement would matter would be if you had a very big, wobbly, floppy craft that's prone to bending, and you might care where you apply the torque based on how it makes things wobble. But for your case that's probably not an issue.)

Generally just doing seat-of-the-pants, as folks have already posted in this thread, works pretty well.

[Geschosskopf]

I think that SQUAD balanced those for us pretty nicely already. Just use the same diameter like most of the rocked, up to 2-3 lenghts of the longest tank in the given diameter, and you will be just fine..

Agreed. A 1.25m reaction wheel is good for a stack of about 2x FLT-800 tanks, a 2.5m wheel is good for about 2 orange tanks, etc. (although you need a mod to get 3.5m and bigger wheels). This means that if your rocket has multiple stages of different diameters, you need the appropriately sized wheel for each stage. Now, if you've got an asparagus lifter, you really only need 1 set of wheels for its central stack because the side stacks will go before you need to do much turning, and that will be in the air anyway so you can use tail fins for that.

You can also use RCS or Vernor engines instead of RCS. I typically make huge SSTO rocket lifters just because, and these are so long that 1 wheel won't do them very well when they need to line up for the circularization burn. So I usually give them just enough RCS for that 1 maneuver, or use Vernors.

[cantab]

Mass alone is less important than moment of inertia. In laymans terms a long skinny ship is harder to turn than a short fat one weighing the same.

Also I'd say keep in mind reaction wheels aren't always the best control method. For launchers I find the engine gimbals give plenty of control by themselves. Long skinny things are good candidates to use RCS or Vernors. Aeroplanes can and IMHO should use control surfaces. And small probes are generally happy with the weak reaction wheel in an OKTO or HECS core and have no need for additional torque.

[Laie]

As a rule of thumb, I go for 1 unit of torque for every 2t of vessel. Sometimes twice as much or only half, depending on circumstances and whatnot.

Hint: distributing the reaction wheels all over the vessel rather than lumping them together in one place will mitigate bending and wobble (this is not a hard and fast rule, there are exceptions).

[steuben]

I'm designing against in orbit/low surface g, with no aero effects.

I'm looking for a Fermi approximation, for example a dv of 5k/s will get you to orbit from the surface of Kerbin.

So based on the ARWM-Lg and 2 orange tanks ... puts it at roughly .8 kNm per 1 t.

After that, it is a question of getting the bird aloft and testing how it responds.

...

After some quick testing maybe 1 kNm per tonne.

...

Working a bit with hyperedit, rcs build aid, and the craft designs. I'm going with 1 kNm per tonne as a first approximation.

Edited August 21, 2015 by steuben