Learning Forum

[Grant Stidham]

Hello, I am modeling a flat belt drive system. The idea is to stretch a belt around two pulleys using a displacement joint load, and then to rotationally accelerate the fixed pulley to drive the belt. I am getting the following errors in my transient structural analysis.

This is after cancelling a solution that was running for too long^

Note: In the simulation, there is a moment that is applied to the mobile pulley at the exact instant that the fixed pulley begins to accelerate. This is so the pulley has something to rotate against and to create tension in the belt.

The following are the analysis settings used:

And the load settings:

The following is an image from a simulation that successfully displaced the pulley 60 mm to the right and stretched the belt accordingly. Distances after this seem to be troublesome for convergence (belt is seen passing through big pulley, which obviously cannot happen)

Any help or advice is much appreciated. Thanks!

[Aniket]

just a quick question for last image, are you using actual scale for deformation or is it something else? Can you confirm?

-Aniket

Forum Rules & Guidelines

[Grant Stidham]

Yes that is true scale

[Aniket]

 

Do you know the actual distance between the two pulleys? One thing you can do is apply frictionless support (preferably displacement) so that the belt does not move laterally till it is fully held on two pulleys by only tension in the belt.

I would use a single load step till let us say 60 mm. and then another load step for tensioning the belt from here. You need to play with the number of substeps in these two load steps. Once the two load steps converge well, then you can remove (deactivate) the displacement applied on one lateral wall of the belt in the third load step, and in the fourth possibly apply either moment or rotational acceleration to the driving pulley. I do not understand the reason to apply both, can you explain that?

-Aniket

Forum Rules & Guidelines

 

[Grant Stidham]

The moment is applied to the mobile pulley, and the acceleration is applied to the fixed (driving) pulley. This is so that more tension is created in the belt to ensure an accurate simulation of the flat belt operation. The driving pulley has something to turn against as soon as it rotates, simulating a real life resistance at first acceleration. That's the idea there.

I'll try using less steps like you suggested, that might be one of the issues. The thing is, if the pulley displaces 75 mm (how far I want it to displace ultimately, 60 mm was just a trial), but only in the span of one or two seconds, I'm worried the simulation won't be able to handle a deflection that large in that short of time. Let me know if that's valid logic, but I'll try it out regardless.

Also, for "equally-spaced data points", the number that is chosen here depends on the number of total substeps, correct? I was told messing with this number can make the simulation much faster so that when I do play around with the different time step options, I won't be wasting too much time. Thanks