Can someone confirm this in FEMM?

[broli]

Here are the Bx results of my simulations. Geographical setup is identical of yours. Red parts act as North poles, blue ones as south poles I asume.
How do you calculate your COP? Integrate Bx over distance?

Yes that graph is correct. The way you calculate the cop is indeed by integration. you integrate from a certain value -x to the maxima, this is your gain. Then you integrate from the maxima to 0, this is your loss. You can divide them by each other to get the COP.

[lumen]

I believe Robbie's simulation is correct, and if you actually calculate the work as (force x distance) you would also find it to be equal or zero gain.

Broli , your graph shows an instant change in direction and I cannot see how this could happen.
I may mount some actual magnets in my setup anyway and check the data but the entire setup is very simple and has likely been tried several times already.

From previous testing of similar setups, I can say Robbie's setup is very close to what I would expect.

[robbie47]

Here's the FEMM script I used for those who want to play with it.
It generates all three magnets and then steps the moving magnet from left to right.
Output is written to text files with numbers separated by a comma, so spreadsheet program can import them.
I use FEMM v4.2

Just unzip the file and load the file as *.lua file and its runs.

[broli]

lumen, I see the wrong assumption I made. I assumed that force is in direction relation with the B-field. But this is wrong (it would have been correct if it was a magnetic monopole), the force is in relation with the derivative of the B-field. It occurred to me when I realized a constant uniform B-field causes no force on a magnet. So at the maxima point the derivative is 0 meaning the field is near uniform there and thus the force would be 0 N. So my last graph is incorrect and robbie4's has the correct relation of force.

I have taken the derivative of the math equation and integrated it from -infinity to +infinity. The result I get is a big fat 0. So that's not good  . This thread can almost retire now. My last request would be asking robbie to integrate his first graph. From some negative or positive point to 0 and share the result.

[0c]

@broli,

I am aware of a simple pendulum-like experiment with some similarity to your configuration, that does seem to show a gain, at least in one direction. But when the pendulum swings the opposite direction, there is a loss. It's nowhere near a 450% gain, only a few percent, but this is not a simulation. Maybe you guys can help each other.

The video: http://www.youtube.com/watch?v=SP_o1_jBUSM

Discussion: http://www.fizzx.com/viewtopic.php?t=415

@Broli, before you abandon this thread, please look at the experiment I referenced and comment on it.