Alternating between tension and compression

Ummm, something is weird.
I’m doing a simple stress analysis of a surface under the self-weight.
I’m using membrane elements instead of shell elements to neglect bending moments.
I think the analysis was converged correctly, but the principal stresses draw radial bands, and between the bands, compression and tension forces show up alternately.
I don’t know I explained the issue well, but the below attached image shows the problem.

The same band appear when I used shell elements instead. And there is no big difference between linear/nonlinear analyses.
Please help! :pray:
kiwi3d (155.5 KB)

The membrane element has a preset prestress, which I assume is not intended for your example. Furthermore, I would still recommend using the shell model. If the deflection is little, also the shell model will try to carry the loads by membrane forces if possible. In addition, the real structure will also have a small proportion of bending stiffness.
Increasing the polynomial degree and the accuracy of the analysis avoids the wavyness.

Hi Anna!! thank you for the reply. Yeah, this analysis is to confirm if the shell is membrane-stress governed.
And yes, I didn’t notice prestress is set to non-zero value by default.
I have been thinking the best number of polynomial degree is always 3… I’ll try higher degree.

Thank you!

yeah, p=3 is sufficient for displacements and bending moments. It may perform worse for normal forces and shear forces.

1 Like

A quick follow-up…

I decreased the equilibrium accuracy to 1.0E-14 and the result was refined a lot.

Thank you!