Sjack Abeling Gravity Wheel and the Worlds first Weight Power Plant

[3decimal14]

New simulation about the Abeling wheel from the
ab-az-cm-2.wm2d file.

I removed the motor and applied just a force onto the wheel for 400 frames to speed it up.

http://www.youtube.com/watch?v=rM8SmU9pjcI

But also this does not work.
Just elliptical pathes in a wheel do not work to get a gravity wheel working.

This animation has a force applied for the first 400 frames, to help speed it up.
After the 400 frames it must work by itsself and you see, how it slows down and then stops

Hi, don´t you need more than 2 balls/weights in the wheel simulation?
To me it looks difficult to get more weight on one side than the other.

Just my thoughts.

[Omnibus]

Talking about strange behaviour...  Why is this going the wrong side???

Now, this is the winner so far.

[mondrasek]

@mondrasek,

How do you explain the rigid joint problem? Also, do you think parallel processing would improve matters and maybe somehow the object-oriented programming would be a palliative solution? Sorry to get into these software issues and detract from the discussion at hand but do you know what language was used to program this and any other details you've come across? Just curious.

Stefan's revelation of the issues with rigid joints was news to me.  Thanks for the hint Stefan!  I haven't played around with it yet, but my guess is this:  In WM2D we are working with three degrees of freedom, linear translation in X, linear translation in Y, and rotation around Z.  With the rigid joint you constrain all three degrees of freedom at one location.  With a pin joint you constrain only the X and Y translation.  Now I believe you can calculate the X and Y translation within one cycle of an iteration.  But you would need a second cycle (different formula) to calculate the rotation.  So, which do you calculate first, and what position of the other (current or future) do you use when doing so?  Similar to a spring, you have two different equations to solve for the rigid joint, and each is dependant on the other.  So you must approximate one and induce extra error in the other.  Unless the time step and integration errors are small, this can cause an error larger than the scale of the models can handle and the sim to blow up.

I'm not a (current) computer programmer, so I don't know how this could be programmed other than the iterative method that is simplest to use that I would use if doing it by hand (unimaginable).  I am also just a casual user of the program and have no details about how it was written.

M.

[Omnibus]

But, see, Stefan suggested to use double pin joint instead of rigid joint and that also constrains the three degrees of freedom while fixing the error.

[Cherryman]

Oke, forget Abeling, i went back to my basic design,

Look at this one, i can bring back the ball up to a height, and i can lift a ball from that height over the middle point..     

Now find a way to combine it..  And at higher speeds it would be easier, then you can skip the low pickup