This is my suggestion for a Gravity Generator inspired by Milkovic and Marjanovic. Since I lack the skill and equipment for building one myself I post this concept with, what I hope an explanation, to this group. I made a Lego model which hopefully make it a little bit more comprehensible.

A DC motor at (e) rotates the two pendulums (d-h) and (f-g).

These are synchronized 180 degrees apart so that when (g) is pointing up, (h) is pointing down and vice versa. This way all horizontal centrifugal forces are cancelled out leaving only vertical centrifugal forces.

Both pendulums are moving clockwise. This will then initiate an oscillating movement of the (d-e-f) beam.

This oscillating movement is rotating the (b) wheel back and forth with the frequency of the pendulums.

The (a-b-c) beam is securely fastened to the ground to harvest the gravity power.

At (a) there is a generator which moves clockwise and at (c) there is a generator moving counter clockwise. This is made possible by utilizing a one directional "bicycle" type of gear. For the movement to be somewhat continuos, flywheels at (a) and  (c) are needed. The generators are then connected i parallel to get the output power.

In theory the maximum torque at (e) is M = 2 * w^2 * m * r * l, where w is the rotation speed (rad/s), m is the mass of (h) and (g), r is the radius of the pendulum (d-h or f-g) and l is the length of the momentum axis (d-e or d-f). The minimum torque is of course 0 (when the pendulums are horizontal).

Average torque at optimal load is then M=w^2 * m * r * l, which gives us the the maximum output power of  P = M * f, where f is the frequency of oscillation. f = w / (2*pi) gives P = ( M * w ) / (2 * pi )

Maximum output power is then, P = ( w^3 * m * r * l ) / ( 2 * pi ). This however assumes no movement of the beam (which will not rotate the (b) wheel, so if we allow a 30 degree movement the P will be multiplied by cos(30 deg), ie. 0,86.

As an example if ; m=1kg, r=0.4m, l=0.5m, f=5Hz (w=31.4 rad/s) will give P=848 W.

The initial energy needed to put each pendulum in motion is E= 0.5 * w^2 * r^2 * m, ie. totally E=158 J.

Since the movement of the beam make each oscillation longer than 2*r*pi, we have to compensate by adding energy for each oscillation. This added power can be calculated as: P(in) = ( w^2 * m * l * sin(30deg) )/ (4 * pi ). This can be elaborated upon if needed.

From these formulas we can deduce that theoretically, in the long run
COP= 2 * w * r / sin(angle), in this case COP=50.

This however is in a completely frictionless environment so in reality we will not achieve this. It is important to remember though that the efficiency is directly correlated  to the speed of the pendulum.

Please correct me if I've done any mistakes since I've only put a few hours into this and would very much like your thoughts on it.

Alan

thanks for your response.

I've browsed through a few designs as well. The reason for choosing the design with a motor for input power and generators for output is simplicity and measurability. Maybe I'm not in the right forum, then please redirect me.

Also I have not seen many designs with double pendulums where mechanical stability is enhanced by canceling out horizontal centrifugal forces and leaving the forces affecting the momentum beam with torque.

Also very few designs utilize the enhanced efficiency of rotating the pendulums/unbalanced wheels 360 degrees with hundreds of rpm. This I would like to see since speed dramatically enhance output power (by power of 3).

My aim is to help with the design of a practical machine in kW range and theoretically a machine sizing 1m*0.5m*0.5m should be able to do that with pendulums rotating at 500 rpm.

Either I missed something trivial or I have not succeeded to explain my line of thought the way I wanted.

The function of the motor is to force the two pendulums into rotation, "synchronized" but 180 degrees apart. This is essential, as is, that this part of the system (motor and both pendulums) is separated from the creation of output momentum. Chaos is not wanted...

As long as the momentum beam is secured in one position the force needed from the motor to keep rotating the pendulums is only to overcome friction.

There is of course the energy needed to set the pendulums in motion at the beginning which depends on the desired speed of rotation. This energy is only needed at start though.

When the momentum beam is then released it will start to oscillate with the rotation of the pendulums and energy is needed from the motor to make up for the longer path for the pendulum weights (to keep the same rotation speed). This energy is however tiny compared to the momentum created from the centrifugal force of the pendulums.

Fast rotation of the pendulums will give large torque at the pivot point of the beam. The output power will be a function of this torque and the frequency of oscillation as described in earlier post.

The power is harnessed by one generator for each direction of oscillation with the help of flywheels. Of course the size of the wheels and generators have to be tuned to the speed of rotation and the losses from friction.

Theoretically the output potential should only be limited by mechanical issues (bearings, material strength etc) as the rotation speed, weights and length of pendulums and momentum beams could be altered as needed.

I hope this makes  it a little clearer, please do correct me if you see any flaws in my thinking.

The distance is the movement of the pivot for the pendulum.  What I noticed is that when the pivot is allowed to move a longer distance I needed to apply my input force over a longer time period, the less it moved the shorter time period I had to apply force.  IIRC it has something to do with angular momentum.

That makes sense since kinetic energy in the pendulum is calculated

E=2 * (pi*r*w)^2 * m

So if Energy is preserved (no friction) and and r (radius of movement) is increased, then w (angular speed) is decreased. To keep constant angular speed (rpm) you'll need to add energy to compensate for increased r. I guess also losses from friction could be increased by a larger r since the weight have a longer path to travel each oscillation.