The Bessler Wheel, mystery solved.

[AquariuZ]

Try this out:  Start and stop the sim.  Then reset.  Your GPE values for the Crossbar and Pendulum are should now read -263.689J and -35.424J respectively.  These are the values that WM2D is calculating they have based on their mass and position in the Y plane relative to an absolute zero.  You drew these objects with their individual Centers of Gravity below the Y axis zero line, so the values are negative.

We could compare these initial starting numbers to the ending numbers after the sim is run to see how much GPE each gains or looses.  But I prefer to do it a bit differently.  I like to modify the equations for the GPE measurement boxes to effectively "zero" them.  It is like calibrating the sim so that those measurements start at, or near, a zero value.  To do that I ADD 263.689 to the equation for the GPE of the Crossbar, and ADD 35.424 to the equation for the GPE of the Pendulum.  Now if you start, stop, and reset the sim you will see that the starting GPE values are very small.  You can increase the precision of the sim and get more significant digits to zero it even further if you like.

Once the sim is zeroed in this way you can see the exact numbers I reported after it runs and stops on it's own.

The Center of Gravity (same as Center of Mass) of the Pendulum is falling much, much further than the Center of Gravity of the Crossbar is rising.  The result sum of the SYSTEM of the two is an overall lowering of the CoG.

Now I have a fairly decent model running for a while - and you are right. Damn Newton.
Back and forth, back and forth losing momentum slowly but surely. Still moving after ten minutes,  it takes a really long time to find equilibrium, but that's beside the point.

Gonna try one more thing, just to be sure.

Thanks mondrasek, you do make sense.

AZ

FWIW: my current model attached. Huge dimensions are needed because of WM limitations.

[phaedrus]

I tried to replicate Gurangax's idea of the Bessler wheel.  Didn't work.  I'm open for any input to make it work.


https://www.youtube.com/watch?v=Q0-WHDsGNjE


https://www.youtube.com/watch?v=g4VjYddOL4w


https://www.youtube.com/watch?v=aBAGelJkBZ0


Dusty

Agree with AquariuZ.  Awesome work you did!  Are you the one Gurangax told to construct a real working model?
Anyway, I would like to comment on the device in the 2nd video (v=g4VjYddOL4w), although I wouldn't be surprised if you no longer have that version available for modifications at this point since you have constructed another version shown in your 3rd video.

This thing you have constructed works pretty much the way it is supposed to.  The trouble is, the sliding-part (what Gurangax refers to as a "crossbar"), that needs to be much heavier than the part that flies out with the weights attached to it.  The lever part is supposed to be a lighter part, that uses leverage to move the heavier sliding part from below the central axis to above it.  So one thing you need to do is make the lever longer, less massive, but you also need to make the part of the lever that is on the other side of the pivot shorter so you get more leverage.  Obviously this will mean the sliding part will not move so far as now, but if you can get it to move just enough so the total center of mass of the slider plus the lever moves up above the axis of the "wheel" as the lever comes down, then it seems to me it should work.  The way you have constructed it, it looks like the lever is MUCH more massive then the slider.  What is amazing to me is that theoretically, it doesn't even matter how far above the center of the axis you move the center of gravity.  If you can just move it enough, then the whole thing will become top heavy and should try to rotate.  Although obviously it will rotate more forcefully, the higher you can get the center of gravity above the axis each time.

Also note that in the video of the .wm2d simulation (https://www.youtube.com/watch?v=Ddg0kUSJMk0), he shows the lever as starting out in a totally vertical position.  My only problem with that was, you would need to come up with some kind of mechanism that would keep that lever in the vertical position (up against the side of the crossbar) from the time it fell down onto it, until it was once more in the up-cycle (that is, 180° after it fell down).  I don't believe you want the lever to be able to start swinging away from the crossbar until it is all the way up on the top again.

[Dusty]

Yes, phaedrus, I contacted gurangax about replicating and he said yes.


Thanks for your input, I'll try out what you said.  The way I built the device I can easily swap out parts and re-adjust the lever arm.


Dusty

[AquariuZ]

My only problem with that was, you would need to come up with some kind of mechanism that would keep that lever in the vertical position (up against the side of the crossbar) from the time it fell down onto it, until it was once more in the up-cycle (that is, 180° after it fell down).  I don't believe you want the lever to be able to start swinging away from the crossbar until it is all the way up on the top again.

...and that's the problem.

AZ

[MarkE]

...and that's the problem.

AZ

In PMMs, overcoming the "sticky spot" never seems to get solved.