I have a working Bessler wheel in my simulation !

[AB Hammer]

eavogels

With the rod between them the lower one will always have more weight. I still only looked at it as a Chaos Pendulum, myself.

[eavogels]

I had a hard time to come into this site again but in the mean time I had Stephan's simulation running many times. They all stoped with the thick line pointing to 3 o'clock.
If someone else has another result perhaps wm2d gives different results for different processor speeds?
/Eric

[AquariuZ]

@niente

The simulation is invalid. Whoever did this put an illegal value in the field that specifies the mass of the polygon. Change it to 1.0 kg and it will behave as expected. See screenhot below. The programme should have rejected the value entered, but that is an editing problem not a real fault in the simulation.

Hans von Lieven

I beg to differ Hans. If you input one gram (i.e 0.001 kg) it will still accelerate, be it at a slower rate.

And 1.000e-004 kg equals 0.0001 kg or a 10th of a gram, so this is certainly not an invalid value.
Bet that as it may, anything over one gram eventually stalls the model.

I can find nothing wrong with it frankly, it is completely symmetrical, with standard 1.00 kg weights for the individual components, gravity and air resistance active....

So what then? Program error?

If not...

Puzzling. I wonder if this can be built... 

[hansvonlieven]

@aquariuz


You misunderstand me, I did not say that this was not a valid mathematical expression. The program considers it an illegal value and crashes.
It would appear this field takes only integers.

Hans

[AquariuZ]

@Hans

As a software engineer I can assure you they are using scientific filters to display data which is out of the ordinary range, this is quite common for scientific software. Again if you use a value of 0.001 KG (in decimal range) it still shows the acceleration. The same goes for Kinetic energy and all other variables which can be assigned to the model components. This makes a lot of sense, especially if you need to accuratly measure torque, drag, rotational accelleration and deacceleration. You simply cannot do it in another way because you will be quickly running into 32-bit calculation limitations. Variables like long, double, float can hold only so much accuracy behind the decimal point, so you will need at least represent your numbers scientifically to remain accurate to say around 100 places behind the decimal point.

I have contacted Design Simulation Technologies (makers wm2d) support about this last night, and have sent them the model. I am currently scaling it down from 100s of meters to .5 meter components and see if I can reproduce then as well. The reason this huge model works may even be earth rotation adding to momentum I have no clue.

It?s up to DST to prove their software is not failing here, if they debug the model and find nothing wrong Niente has Something which is definitly not "niente". If they do conclude it is a glitch I cannot possible see why wm2d should be used in the future by anyone unless they fix the specific coding errors which cause this.

Hans, I admire your work and writings, but on this particular point I emphatically disagree with you.