Sum of torque

[EOW]

The problem on the red arm come from the difference of surface S8/S9, so if I put balls inside the part of the red disk (only the part that is inside the container) I can cancel the pressure from S8/S9 (I attrack the balls inside the red disk from springs from the container). The force on the curve part (circle) don't change the torque on the disk. The red disk has always its torque around itself.

[EOW]

First image: if I take this simple device the force F will lost an altitude so a potential energy will be lost. But the disk will turn more and more. The sum of energy is constant or not... ?

Second image: I put balls inside the part of disk that is inside the container. I attrack balls with springs. The disk will have the force F so a torque. But the container don't have any torque. The container has balls +springs.

[EOW]

If I take this device:



The support turns at w0 and the purple disk can turn around itself, at start the angular velocity of the purple disk is 0.
I can adjust the angle alpha like I want before start but the angle is constant after start.

If alpha=0° a point A fixed on the purple disk will be at differents position with time:



If alpha=2 or 3° it's the same but the point moves up/down like that:



And like that:



The angle can be increased more at 20° for example.

So if the point A moves down or up, it's possible to attach a mass, this mass will move up (under gravity) and the purple will turn a little. I win the energy of the rotation of the purple disk and the mass that moves up. THERE IS NO TORQUE ON THE SUPPORT.

[EOW]

The video of the device, I simulated with Ansys, 10 seconds, it's possible to look at the rotation of the disk, at start the disk has no rotation around itself, a small black circle is fixed on the disk to look at the rotation :

https://youtu.be/Pjc4dIf1aWI

If with this device the sum of energy is constant. I can use the last device with the balls+springs but WITHOUT MASS, or a mass as lower as possible. Like that I don't have these problems of inertia. I can recover the torque on the red axis from the pressure and accelerate more and more the device.

[EOW]

I can place the disks like that:

http://s5.postimg.org/i21birqon/image.png

to increase the angular velocity of each disk