Computer Model. What am I missing?

[swedluap]

I'm sure most of you are familiar with the CAD program called "Interactive Physics". http://www.interactivephysics.com/ . I started using this program a couple months ago. I have been, for fun, playing with some theories that I have thought up over the past 15 or so years. I started out by creating some of the typical Perpetual Motion Machines of yesteryear, just to watch them fail for myself, of course, they did fail, quickly. I then started modifying them and adding some of my theories, trying out some of my ideas, and somehow I was able to create a machine that actually works, on the computer anyway. It starts, speeds up and runs contiuously under its own weight. It is very slow, so I doubt it would have any useful application in real life. I thought that maybe friction was not being taken into account, so I added resistance to it, and it still works. Has a working computer model ever been made? If so has that model been built only to fail in real life? What am I overlooking here? Please help me make this machine fail so I can stop wasting my time, or I might be forced to actually build a real life model, and I just don't have the time!

 

[Integral]

A improperly implemented numerical method can "fail to conserve energy" ie a pendulum which continues to increase its amplitude of oscillation. A frequent breakdown of numeric methods will result in oscillatory motion which is essentially noise.

 

[swedluap]

But it would eventually stop, right?

 

[swedluap]

I have even tried kick starting it to run the wrong way, it takes a while but it does stop and reverse to run continuously in the other direction. I still don't see what I am missing here.

 

[Integral]

It can well be that various errors (round off and errors inherent in the methods used) have swamped your real solution. This is not uncommon and is a reason that people doing serious modeling work can rarely use off the shelf products, you need to understand the methods used and how the inherent errors (all numeric methods have them) effect your solution. Sometimes the noise accumulates so your solution blows up other times it will fall into meaningless oscillations.

You may want to take a smaller time step, see what happens. Depending on the method this can help a lot.

 

[swedluap]

I have experienced both the "blow up" and the "meaningless oscillations" in previous models. I have just started a new test, I have added a small weight to one side to see if my model can overcome this flaw. I will let you know what happens, but it is running for now. I will be gone for a while, check tomorrow, and thanks for your help.

 

[swedluap]

Nevermind, I got my head out of my underside. I figured it out. Thanks.