Modal Analysis & Mesh Size

[aadilnoor]

Hello,

I ran some model analysis on a bracket and used AutoGEM to generate the mesh. The results were as follows:

First run with AutoGEM – 67 hz, 94 hz, 114 hz, 146 hz for the 1st through the 4th modes respectively.

After changing the mesh to 1 cm max segment length, the modes increased for the identical model to the following:

75 hz, 116 hz, 125 hz, 169 hz for the 1st through the 4th modes.

What are your thoughts on why this would increase? Also is it safer to say that the 75 hz is more accurate than the 67 hz since it is a tighter mesh? Wouldn't the length and width of the bracket be the deciding factor of what it's modes would be, not the size of the mesh?

Thanks,
Aadil

 

[ninjaneer777]

The mesh size will be a contributing factor up to a point. The point where increasing mesh density changes the results little is where it stops mattering.

Usually with Mechanica increasing the Polynomial order of the mesh will allow the results to converge. What convergence method are you using, Quick Check, Single-Pass Adaptive or Multi-Pass Adaptive? If you are using Quick Check or a Multi-Pass Adadptive with a low maximum polynomial order the model probably isn't converging which might be why you are seeing changes when you increase your mesh density.

 

[kan123]

In general, predictions of natural frequency (especially the first mode) should be relatively insensitive to mesh density. Your difference of >10% is unexpected unless the first mesh was too coarse to represent the real geometry or mode shape.

 

[aadilnoor]

Hey Thanks guys for the feedback. The analysis was done in Single-Pass. I played around with the mesh density and found that it did level off. There were some anomalies due to the odd mounting as well as the weighted links. Since this was a ECM bracket that mounts to a large generator engine, the information extracted from this analysis was enough to confirm that we are sufficiently away from the specific frequencies in question.

Thanks again!

 

[Twoballcane]

Hmmm...did you corelate to any hand calcs?

Tobalcane
"If you avoid failure, you also avoid success."