Sjack Abeling Gravity Wheel and the Worlds first Weight Power Plant

[mondrasek]

I need more to be convinced that decreasing the time step, let alone setting up air resistance to non-zero is indeed the proper methodology and not just seeking ways to prove the device is non-working.

This need of yours I understand and respect.  And because of it you answered that the sims so far were inconclusive.  I, on the other hand, feel that the testing and my understanding of the sims has conclusively shown that they fail to show a device that works as Abeling describes.  And so I stated, "With all due respect to Omnibus", my differing opinion.

Just as enormous power can be generated in a resonant electrical (or mechanical) system, yet not be extracted as more energy than was initially input, so can the iterative nature of the sim software set up motion where it would not exist in the real world.  The acid test is to apply a load and see if the sim can overcome that.  The quickest small global load (that also mimics real world conditions) is air resistance.  If turning that on dampens your movement to zero, that movement was not a force that could do any work and can be ignored (and likely just a software aberration).  If you can design something that overcomes that simple test, then you may have something worth pursuing.  The Abeling design has not shown any simulated tendency towards producing usable output.  I conclude from this that the design as simulated does not work.

The time step and integration error are debatable.  But you agree that the smaller the time step, the more accurate the sim is supposed to become, right?  And the (relatively large) default setting allows for more real time performance on slower machines?  So that is what I would put in my software for default as well.  It does not mean it is any way optimal.  We both know that smaller is more accurate.

Integration error needs to be set smaller with smaller sized model elements.  This is obvious with how this value also decreases the allowable interference between colliding elements.  So decreasing this value also increases accuracy.

There is always the possibility that a specific and small value of time step and integration error will set up a resonant condition, forcing a competing interference, especially when pinching one moving object (rollers) between two others (wheel slots and guides).  When this happens it is usually obvious and you can change one or both values a small amount to get away from this "resonant" condition.  But I stand by the fact that decreasing both is the proper way to increase accuracy and usually rules out aberrant behaviors caused by gross values.  I also respect your desire to have less subjective information and to draw your own conclusions.

[Cloxxki]

@Dusty
Your replication is going to work, I'm pretty sure. It all makes sence to me now.

I'd recommend : larger diameter weights as they roll the wood.
Heavier weights as you planned.

Maybe if you use rollers (steel or alu as you work with) spacer OVER the bearing to increase the diameter to reduce resistance, you'l also get some spinning of said bearing to work, so they won't be stuttering as much, and it will become one smooth operation.

As I warned on Youtube, I'd consider building a brake first (wedge some under there with a lever) before you do a test of more than 1 revolution. Bolt the frame to the floor.

I feel the excitement, when you take the next step, we'll see a working perpetuum mobile, which will turn into a hightly woodshredder unless it's rpm's are controlled :-)

[Omnibus]

@mondrasek,

I'm trying to understand this. Air resistance aside (we'll consider it zero), I'm getting the same integration message error at the same position of the wheel no matter whether Integration Error is set to 0.01m or to 0.0001m when using the Runge-Kutta method (for this purpose I'm checking the Warning for the Integration Error). Euler's method just explodes and we'll exclude it for now. I don't think the animation step matters either. If you ignore that error, in the case of lower threshold the wheel finds equilibrium while at higher threshold keeps going (all that is under non-ideal conditions -- both frictions on 0.3, elasticity at 0.5, no air resistance). The smaller threshold should be better but it obviously isn't in this case -- error stays the same but at larger thresholds the performance differs from the performance at the lower thresholds.

[mondrasek]

Omni, please post the sim you are currently playing with and I'll take a look.

[Omnibus]

@mondrasek,

Please, see attached.