It's good to have come this far but we have reached a point where Sjack must disclose his contraption already. Some numerous old farts can't even decipher Besslers wheels after decades of experience so I doubt we will see the light in here using wm2D  . I do agree on using the tools and technology of current times. It would be stupid to not use any hand drill for instance because Bessler didn't  . Or to not use computer simulation to optimize the device quickly.

For the last time:  Decreasing the Animation Step and Integration Error increases the accuracy of the sim!  As both of those factors approach zero, the sim approaches reality.  That is how these things work.   It is similar to integrating the area under a curve.  The smaller the quadrangles used to approximate the area under a curve, the better the accuracy.  As the quadrangles approach infinity (time step approaching zero), the result approaches the exact area.

You can blow up any sim by raising the Integration Error and Animation Step.

The effect of the Integration Error is related to the size/mass of the object you are simulating.  So large wheels do fine with larger values.  This wheel is smaller and requires smaller values.  So there is not a correct default value for each sim.  The program allows for you to change these values based upon your need (gross visualization vs. exact analysis), the size of objects, and your CPU speed and patience.

In this case the Integration Error and Animation Step are just too high.  Since the sim process is iterative, it must calculate the positional change and forces due to one object at a time.  If it gets to object #7 (for example) and this causes a huge overlap with #1 that was calculated previously, you get crap results.  So you need smaller values as you increase the number of components as well.

As was stated several times, when you are narrowing in on a specific action or see something unexpected, always decrease the size of the Animation Step and Integration Error to increase sim accuracy and check your results.

M.

It's hard for me to say this but you are completely wrong.

It's good to have come this far but we have reached a point where Sjack must disclose his contraption already. Some numerous old farts can't even decipher Besslers wheels after decades of experience so I doubt we will see the light in here using wm2D  . I do agree on using the tools and technology of current times. It would be stupid to not use any hand drill for instance because Bessler didn't  . Or to not use computer simulation to optimize the device quickly.

Who are you calling an old fart? Just how old are you?

@mondrasek,

Like I said, I’m still not convinced that in this particular case introducing finer accuracy steps is beneficial and that the finer it is the more it approaches reality. Case in point – rotation becomes un-physically slow, to mention an example. Let alone that at finer accuracy limits constraints are ignored. I understand where you’re coming from and indeed you’re right regarding the integrating the are under a curve. Unfortunately here we have integrations and solving difference equations in real time in a serial fashion at that. As you correctly pointed out earlier one of these serial steps may take longer (especially if more iterative cycles are to be carried out to reach the accuracy limit) than some other step and the result is out of sequence calculation. That it’s a sequence problem may be seen when you put very crude limits, just for the heck of it. You’ll see this disjointing visually. I’m still not sure the solution isn’t a matter of optimization rather than getting into the accuracy limits.

I also addressed the mass/size argument pointing out that a model with the exactly same size of balls and wheel performs differently. So, this (mass/size impact on running the algorithm) also isn’t something we should take for granted.

Therefore, as I already said, the problems don’t seem straightforward on both sides of the issue (both on the pro and on the con side). We really need to know what exact methods (Monte-Carlo, Runge Kutta, finite (boundary) elements or what have you) are used and exactly how they are applied. Otherwise the story will be similar to what’s being done in Quantum Mechanics – we’ll have to see it experimentally and then adjust the parameters of the wm2d simulation to appear it’s a great engineering software that “always works” (as they usually refer to QM when “proving” its usefulness in this way, which is almost all they do).

Some more input..

http://www.dumpert.nl/mediabase/478071/bcc5ec2a/fietswiel_vergeet_zwaartekracht.html

It's a gyroscope.

A very large gyroscope.

Cool vid, thanks

It's hard for me to say this but you are completely wrong.

No he's not.

Peace

Never mind

Let me mention once again. The vertical you've drawn is to the left of the vertical passing through the constantly changing its position center of mass of the system. It is that second vertical that you should use in setting up your inequality.That's a remark in addition to the difference-in-constraints remark and to what Stefan said.

Incorrect. When calculating torque in a system such as this the only thing of relevance is the distance between the centre of gravity of the individual weight to the fulcrum measured along the x-axis. The centre of gravity of the entire device is a resultant of all masses present.

The whole idea of having the weights run on a excenter is as old as the hills. There are hundreds of variations that have been tried, none if them have worked for very obvious reasons.

You can use all kinds of fancy mathematics to show that there is a surplus like that idiot that wrote the paper for Milkovic. In the end only the correct mathematical procedure will give correct results.

Hans von Lieven