I'm gonna share again the design i came up with, not the first who came up with it tho.

Advantage of this system is it doesn't need a ferrofluid.

There is a better way to do this with much less water in useless bottom part, but principle is the same.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

imagine a circular pipe, bottom half 4 times the diameter of the upper,

there is a pump inside the system,

in short, centrifugal force will be 4 x greater in upper part despite the

fact that volume of water is 4 times less.

this is due to venturi effect (continuity principle), water in narrower part

speeds up proportionally to how much narrower pipe is (

https://youtu.be/UJ3-Zm1wbIQ?t=97 )

and centrifugal force is proportional to velocity squared.

let's say 5 liters is enclosed in the tube, since upper half is 1/4 diameter,

there are 4 liters in bottom half and 1 liter in upper half.

let's say pump is pushing bottom part water at 1m/s, this means water

in upper part moves at 4m/s.

let's say radius of the circle is 0.5m

let's calculate the centrifugal force

http://www.calctool.org/CALC/phys/newtonian/centrifugalfor 4 liters at 1m/s force is 8N or 0.8kg.

for 1 liter at 4m/s force is 32N or 3.2kg.

or constant 2.4kg, but since only at the center force is pure y component it is 1/2 or 1.2kg average up

and

for 4 liters at 2.5m/s force is 50N or 5kg.

for 1 liter at 10m/s force is 200N or 20kg.

or constant 1/2 x 15kg = 7.5kg up

etc

of course increase of velocity means more friction and very low friction tube

must be used, like electropolished stainless steel, but principle is clear and should be tested.