Pendulum

DaveStupple

Is it possible to model a simple pendulum?

I had a go, but the element stretches absurdly. Not sure if dynamic response is the right analysis, but mass is not available in the others. Should I be using quasi-static and model the mass as a proper solid?

This is of no practical importance to me at all.

Victor

It requires nonlinear dynamic response analysis. Nonlinear because the restoring force applied by gravity depends on the displacement. Dynamic instead of quasi-static because quasi-static only produces solutions that are in static equilibrium at every time step.

DaveStupple

Your two points make perfect sense - the first one is less obvious to me than the second one (should have been!). I managed to get one swinging last night, but I can't get any to solve today. If I uncheck Automatic Time Steps the solver gives up with a no convergence error. If I have Auto Time Steps, the solver gives up with a time step too small error. If I get it working, I'll try adding dampers to dissipate the energy slowly.

Victor

It's just as well it's not obvious because it's wrong :P Sorry about that! You can have a restoring force that depends on displacement in the form of a linear spring. The reason not to use linear is really that displacements and rotations don't update the stiffness which behaves as if the structure is always in its initial configuration. In this case, that initial configuration has the bob unconstrained in the vertical direction so it would be free to fall downward endlessly.

DaveStupple

I'll have to think on that. But thanks for the clarification. Maybe that is why I got the silly stretching in my early attempts.

SnuggleKittens

This is something I too tried a while ago. I think your problem, as Victor said was that you need a nonlinear dynamic analysis.

disla

Chain Pendulum with Nonlinear dynamic analysis. It also works for low speeds.

VMH

disla, can you share your model?

DaveStupple

Thanks Disla and SnuggleKittens for the two examples - just the sort of experiments I had in mind.

disla

Sure. I have added some springs and extreme mass to be more “interesting”.
Running the model is 2 minutes.

Victor I notice that postprocessing is only using 1 core. ¿ Is it possible to get all cores working setting some variable apropiately?

Thanks in advance.

Victor

The graphics are only single-threaded.

VMH

Thanks disla!

SimCommons

DaveStupple

Some great ideas to play with, thanks everyone. If I create anything interesting I'll post it.