Hi @NickK, welcome to engineering. This will require a lot of engineering, since is a very high mass that is spinning at a very high speed, despite the fact that acceleration is not a issue. I think you need to provide more information regarding the arrangement and shape/volume/diameter of wheels when they rotate (are they coaxial), and surface roughness, because that will play in the aerodynamic drag you mention. Other information will most certainly be required, but the drag is probably one of your main losses that the motor would need to compensate for.

Hi @Nmech, thank you for your feedback. I have adjusted the parameters; hope that was able to clarify the question.

100kg (or rather, 3000kg) at 1m wide spinning at 3000RPM is friggin' dangerous. And on a 3 meter long shaft at that. Likely to explode from being out of balance or vibration or flex due throwing it out of balance. If you're asking simple questions like how to calculate the bare basic torque requirements and are not sure whether a flywheel is needed or not (it's not clear why you think you need a flywheel especially when your load is already a flywheel) then you might be in way over your head. Get professional, experienced assistance. This is like a power plant steam turbine mass and speeds.

What @DKNguyen said. As an example of what a much smaller mass can do, I know of an accident where a large railway diesel engine was being tested (by professional engineers.) The flywheel came loose from the crankshaft with the engine running at about 2500 RPM. It went through the side of the engine, through the brick wall of the test building, across about 1km of railway sidings, through several fences and two more brick walls, and finally stopped inside the kitchen of someone's house. Make this device at your own risk!

@alephzero And this one is vertical so no one standing anywhere for a few km around it is safe if run above ground, or god forbid on an elevated floor.

Thank you @DKNguyen for your valid concerns. I should probably point out that this will not been done in open fields but at a fully enclosed lab whose walls will be reinforced with layers of thick steel plates for such situations and all test analysis will be observed remotely through computer screens.

@NickK If you're in a lab you should get someone who knows what they are doing. This sounds like a project with a budget of hundreds of thousands of dollars anyways.

Well you would need to use $\tau=I\alpha$ and select a desired target RPM and how long you're willing to wait. Then add in torque due to wind resistance which is beyond me. Probably just some experimentally derived coefficient of drag and drag torque scales to the squared of speed.

@DKNguyen Will do that; thanks for your feedback.

@NickK You really don't have any idea what you are proposing to do here. Unless you have a project budget of the order of \$10m to \$100m, just forget about it. My day job involves testing jet engine rotors, so I do have some idea - but what you are proposing is 100 times bigger than anything I have ever worked with.

@DKNguyen The tangential speed of the OP's disk rims is about Mach 0.5. Unless they are proposing to run the device in a vacuum chamber, the motor power to generate the artificial tornado that will be created is of the order of megawatts. (And those megawatts will all end up as heat, somewhere in the system....)

@alephzero Yeah, I calculated that too but decided it wasn't even worth mentioning since a propeller weighs nowhere near as much and a steam turbine matched things better.

Holy carp! +3t at 50Hz. I'm struggling with how the magnetic bearing component could work with the mass involved.

@StainlessSteelRat Sounds like OP wants to build this except that this one is twice the diameter, 33% heavier and spins twice as fast: furukawa.co.jp/en/release/2015/kenkai_150415.html It's does not use an ordinary magnetic bearing, it uses a superconducting magnetic bearing.

does it need a flywheel! This thing will have inertia like a freight train. You are designing a very edge-condition device. A good engineer would take an estimate of running load but would expect that only a proper test will tell you how much power is actually needed to keep such a large device spinning.

I have edited the problem statement