Just another WM2D wheel

[niente]

Hi all, here's another wheel model, made by just some weighs and springs. Here forces should balance, but giving an initial speed this wheel will rotate, thanks to a tiny centrifugal force. Model is in attachment.
Could it run forever in your opinion?

[Wubbly]

Looks very similar to an MT002.

Here's my simulation of MT002: viewtopic.php?p=167490#167490

I couldn't get MT002 to work, but if you play with the spring constant, you could probably get it to run for a long time before it dies down.

[niente]

I changed a bit weighs and spring parameters, now this model auto-starts rotating and it keeps a nice rotation speed. I set WM2D simulation params to "accurate"...

[Fletcher]

Hi niente .. I took your sim and cleaned it up a little. The parts weren't placed exactly symmetrically around the CoR for example (gives a slight pendulum effect). Then I made the center wheel transparent and turned on View System CoM. You can zoom in close to see this CoM placement for yourself in detail. Then I added some other inputs and outputs so you can play with the sim in different ways. For example I added a Motor which rotates the wheel at 1 rpm for 0.25 secs before switching off. And I added a System Kinetic Energy Calculator so you can see the KE dropping off over time.

The first sim was with Air Resistance at zero (unrealistic). The later pic is with AR on LOW. Have a look a the plots for trends !

Basically the spring attached weights bounce around a bit before settling down. I changed all the components elasticity to 0.95 to minimise deformation impact energy losses. These losses are why even when AR is OFF it still looses energy over time as heat etc.

Anyways have a play with the sim if you feel so inclined. You could turn off the motor completely (0 rpm or 0 secs) but the idea was to give it a certain rotation, then see if it sped up or slowed down.

Try running it without the motor as a comparison to your sim ?

[Wubbly]

This is another take on it. I cranked up the angular velocity of the background wheel from 1 rpm (6 degrees per second) to 90 degrees per second, and decreased the elasticity to 0.5. I tried to match up the other parameters of the build. It dies down as you would expect an MT002 to do.

[niente]

Thanks for your answers!! I'll try your models asap!

The model posted by Wubbly, which is using rails, doesn't takes into account the horizontal movement of weighs, so it's perfectly balanced.

In my thought, it's the lateral movement, due to centrifugal force, that causes that tiny imbalance that makes the wheel to turn. Otherwise forces are perfectly balanced.

In theory, this lateral movement should occur when the weigh is going down, while it shouldn't occur on the other side of the wheel (where the weigh is lifted) because the weigh itself is kept in position by the tiny walls surrounding it (my model is far from perfect, weigh is a little loose). Walls should not be perfectly parallel but they should be narrow where they're attached to the pole and more distant where they terminate (see drawing).
[sorry for my poor English ;-) ]

[Fletcher]

Here's my sim of your sim upgraded to open the throats by 10 degrees etc.

Air Resistance is OFF.

Don't mind the inconsistent constraints warning and just click yes to carry on. It is because a motor is activated.

You may have to play around with the spring lengths as you see fit to get the weights to move under CF's on the down-going side (CCW) as you imagine.

[Bahamus]

It's good

[MrVibrating]

All i've done is raise the accuracy settings of the sim from post #1 to a 'minimum' level; it just sits there, as you'd expect..


It could've been done in a cleaner manner by using slot joints in order to eliminate all collisions, which require the most CPU time to solve accurately..

Fundamentally tho, gravity's constant (it's always 1 G) as is height (1 meter is always 1 meter), and mass (1 kg is always 1 kg), thus we can apply the simple rule that "closed loop trajectories thru static fields yield zero net energy".



The only way to make mechanical energy in a statorless wheel is by generating momentum, which in turn requires an effective violation of Newton's 3rd and 1st laws.. only then are we even making a start on solving the mystery..




ETA: ...another relevant tip - in any situation where collisions are necessary / desirable, they can be accomplished and controlled much more easily via rope slack... set the 'elasticity' of the rope / pulley accordingly, apply as much slack as the travel you need, and you get all the results of a perfect collision without the sim needing to calculate any actual contact between bodies, or, thus, overlap errors..


The main lesson here tho is to raise the accuracy settings - use fixed integration steps, and for most interactions, 1 kHz is a reasonable low baseline frequency, just for rough testing.. You wanna be at 10 kHz + for any interesting results tho.