Tips on shortening mechanism run time

[ptc-4631445]

Hi -- I am having issues with long run times in mechanism dynamics. I am driving my mechanism with a velocity servo motor at 1750 rpm. There are 15 bodies and about 15 connections. The dynamics analysis runs quickly when I drop the motor speed to 2880 deg/sec (480 rpm), but when I try to run at 960 rpm, the program runs fast at first, but then slows down to a crawl at about 55%, and finally stops. On the "Mechanism Settings" I have the "Graphical display during run" un-checked, I do have some parts in assemblies not fully constrained. These are round parts where I have not defined their rotation. Is that a problem? I am trying to make only 1 revolution, for the sake of the graphs, For instance, 1200 rpm would have a duration time of .05 seconds. I can live with the slower rpm's, but feel the real rpm would be more accurate.

Any suggestions would be appreciated.  Thanks

[ChrisKaswer]

You will need to define joints on all your parts (along with density/mass) AND you will need to make sure you constrain the motion of each body (lock them down if you don't need to consider them in the solution). If you leave it up to the solver to determine unconstrained bodies positions/velocities, etc., you'll chew-up compute time.

Hope this helps,

Chris

[ptc-4631445]

Thanks for the tips. What I can't understand is why the program runs great at low rpm's, but can't handle the high rpm's. If I had errors in my connections, etc.,wouldn't that show up at any rpm? Is the issue with the short duration time? When I attempt do reduce the frames, it automatically jumps to a higher number.

[ChrisKaswer]

My guess is that it's because the solver has a more difficult time calculating where the unconstrained bodies want to move given higher accelerations - don't forget, this is what a dynamic analysis is trying to solve for - the movement of all the bodies in your model under whatever input conditions you've set-up. If you run the study at low speeds, it should have an easier time calculating positions, velocities and accels of all parts...even when they're not constrained. My first rule of thumb is always treat your set-ups as disciplined as possible, and you will never have an issue. This means assign mass, proper constraints and create joints formally (only between two bodies) to every part.

[ptc-4631445]

Thanks. I can run this mechanism at 480 rpm with a run time of about 15 seconds. But at an rpm of twice that, it runs rapidly and finally stalls out at about 33%. I am quite sure I have everything constrained properly. I went back over an assembly with some partially constrained components,such as screws, and even constrained their rotation. Didn't seem to help. Attached is a file showing the graph of a 6DOF moment at 480 rpm.

[SweetPeasHub]

Your results seem to be diverging, Try not forcing the RPM to be constant, rather set it as an initial condition and let it coast after that.  This is to stop putting energy into the model. You should also try adding at least a small amount of damping in all 6 dof.  This will probably mean using a bushing.  Look that up in help.  I usually experiment with my damping to find the right amount - increase until you see its obvious effect, then decrease until then influence is acceptable.

[SweetPeasHub]

This variability in time-step (casing time to simulate variation) is normal for any dynamic simulation.  Teaching how to construct models that increase the time-step (reduce sim. time) will be difficult to do in this format as it involves most aspects regarding how the model is constructed, and the time-step is quite specific to each model.  In essence it is a magic art, only gained by experience and hard to communicate.  I will however, cover a few basics.

Model is realistic in every way possible. mass, speeds, etc.

Damping is present to prevent oscillations outside the frequencies of interest, but not in the frequency of interest.

Plot a bunch of measures on joints, contacts, etc...to find out where the short time step is originating, what configuration is the model in... what just happened when the time step drops.

[ptc-4631445]

Thanks for the help. This obviously quite complicated. I am trying to do a balancing of this mechanism. I have attached a graph of a resulting moment around a 6DOF connection. A single revolution is .125 seconds, but I am getting what appears to be 2 cycles instead of 1. I am showing the length & rate. There seem to be more than 51 frames. Notice how the frames are closer together at the peaks. Is this what you refer to as a short time step? The run time at this speed is about 15 seconds, even with the graphics on to verify a single revolution.

I have done this balancing before, and had to settle for using the lower rpm's. A balance is a balance at any rpm, just more accurate at the real rpm.