1. Why is the performance of SHELL181 in vibration (modal) analysis poor?
This element appears to be giving consistently stiffer results. Representative percentage errors
are of the order of 5 ' 8 as the mode number approaches 100 in a structure having nearly 19000 DOF.
Other shell elements ' 43, 63 and 93 are performing better.
The foregoing findings are based on studies for a case study problem involving four ANSYS elements and
2 NASTRAN elements. The findings are summarized in the attached XL sheet.
(From product development) For first order elements, lumped mass approximation usually provides better accuracy.
Apparent "poor" performance of 181, in higher modes, that you have observed is due to its use of a "consistent" mass matrix.
Due to the emphasis on nonlinear analysis, and other reasons, we chose reduced integration as the default. That necessitates
the use of consistent mass terms (you may encounter spurious modes otherwise).
To get the behavior closer to 63/43 in linear elastic applications, pl. use LUMPM,on (and keyopt(3)=2). I am sure you will see good agreement.
You may also notice in your output that we do recommend using full integration, as the material is elastic. We will expand that recommendation
for lower order elements to use lumped mass approximation in a future release.