My Algodoo design spins endlessly each time I re-build it. Advice please.

[justabil]

I have been tinkering with Algodoo for years and this is the first time I have encountered this situation. I understand that simulation applications have flaws and can give false results. I am hoping for guidance on how better to test it.

The design is EXTREMELY basic/simple and fits with most of the prominent "clues" Bessler left behind (and especially one of his prominent drawings). I have rebuilt the model from scratch about a dozen times, each time different sized components.

If I give the wheel a slight push, it rolls forward, then back, and then settles into a resting position after about a dozen back and forth partial diminishing rotations. If I give it a stronger push, it spins continuously without any decline in "Angular momentum" or "Angular Velocity". It has run overnight without any decline.

I get the same behavior when pushing the wheel in either direction.

QUESTION: The wheel's "Angular Momentum" and "Angular Velocity" is "pulsating". It has a consistent "high" and "low" for these two measurements. On one of the external wheel drawings (see top of this forum page) there is a "hammer" like apparatus on the outside of the wheel. Until now, I had thought this was to slow the wheel (keep it from going too fast). But now I am thinking it was used to even out the high and low energy into one even rotation. Any thoughts?

QUESTION: To test what is going on, I paused the simulation and attached a small weight to the bottom of the side going up. I resumed simulation and removed the weight once it reached the top of the rotation. The wheel still spins endlessly but now at a reduced velocity and momentum. I'm guessing this fact is fatal to my design hopes.(right?).

Note: If I leave the added weight to the wheel, it still continues to rotate endlessly with a reduced momentum.

Thanks.

[Fletcher]

Around here that's known as a zero runner i.e. maintains its kinetic energy levels for no apparent reason - what you have to do is first identify where you think the energy is coming from to achieve that [assuming you have losses ON in your sim], so that it makes sense - if you don't have a quick & dirty means of adding energy losses, like air friction, then add a small load to the circumference like adding a small jockey wheel pushing down on the rim by a spring - this will drain some of the wheels energy approximating system losses - air drag is much simpler.

The thing about a sim to be valid is that it should demonstrate two things.

1. the ability to accelerate [not coast].

2. still accelerate under a known load however small initially - you should be able to increase load until it doesn't accelerate.

P.S. I don't use that program but I assume they are all fairly similar.

[justabil]

Thanks Fletcher.

That's the advice/insight I was looking for.

It helps knowing that "zero runners" are not uncommon.

In my experience, air drag in Algodoo causes strange things to happen. A simple balanced lever will go "out of balance" when air drag is on. I guess that says something about the software's reliability.

Thanks again.

[triplock]

Having used Algodoo for a couple of years now I soon realised that, after a couple of false dawns, that it is relatively easy to simulate devices that appear to show net gain.

I can, if I so wished, construct a very simple set up and make it revolve for 24 hours.

I have no doubt that the OP has used the hinge function, and maybe as part of a levered set up. The problems with the hinges is that they stretch away from the point of fixing. This induces an unrealistic resetting of the back wheel, thus inducing rotation towards the hinge.

It's even worse if you use a combination of levers and spring. You can get 100 rpm + !!!!

Basically, any sim, especially Algodoo are only ever any good for just checking the motion of component parts, nothing more. The only way to verify results on the computer is to build a prototype.


Chris

[justabil]

Thanks Chris.

Yep. I know about the Algodoo hinge issue. If you put a hinged component that overlaps slightly with a component with the same collision level, the thing will take off like a rocket!

Both you and Fletcher have convinced me a working prototype is the only sure thing. Prior to that, I want to work out the math and geometry of it.

-Bill