Elmer Discussion Forum

Dear all,

I am a new user of Elmer and I am really impressed by it;s capabilities. I would like to do a model according to this link:

http://www.caesarsystems.co.uk/NAFEMS_b ... s/le3.html

I am aware of Elmer ShellSolver. I have read the ElmerSolver documentation and unfortunately I didn't find any hints how to apply nodal forces (as in benchmark in link above) with use of ShellSolver. Could you please point me in right direction how this model can be done in Elmer?

Thank you for your help.
Best regards,
Mdudek

In most of the tutorials and test cases loads are applied with boundary conditions, however you can target nodes instead of a boundary element. This example loads node 1 with 1000000 Newtons


Boundary Condition 1
Target Nodes(1) = 1
Force 2 = 100000.
End

Hi

This is probably one of the test cases. It is called "Shell_OpenHemisphere" under fem/tests.

-Peter

Thanks, it is not exactly the same test case but it is an example on how it would be approached.

I worked through the NAEFEMS case, Elmer gets a 0.182 displacement with this mesh. The NAFEAMS target was 0.185

Hi

Great work Kevin! Do you know how the NAFEMS reference solution is defined? After all, it is pretty old. It would be interesting to know the accurate solution. With such a coarse mesh the reference cannot be one?

-Peter

It is described in this document which I have not reviewed yet,
https://www.nafems.org/publications/res ... ter/r0006/

Reference solution
This is a test recommended by the National Agency for Finite Element Methods and Standards (U.K.): Test LE3 from NAFEMS publication TNSB, Rev. 3, “The Standard NAFEMS Benchmarks,” October 1990.

Thank you for your responses. I have managed to simulate this model thanks to your helpful post. I also found dr Gupta tutorial which explain the syntax of shell solver. I left the link, maybe somebody would find it useful:

https://www.researchgate.net/publicatio ... _is_also_a

Is it possible to simulate such shell with quadratic elements? As far as I noticed Elmer doesn't support QUAD 8 elements in shell solver, but will it be possible with QUAD 9?

And I would like also to ask question: Why U3 = 0 on AC edge is added as boundary condition? If I remove this constraint (according to the NAFEMS benchmark the displacements on the edge AC are free) the result are incorrect (please see attached files). Also point E should be fixed in every direction according to the same benchmark from this site:

https://abaqus-docs.mit.edu/2017/Englis ... -c-le3.htm

Could you please point me why I obtain incorrect results? Thank you for your help.

Regards,
Mdudek

Without restraining Z along AC it will not obtain the published answer, I tried it both ways in ABAQUS, NASTRAN, and Elmer same result. If you look at the actual ABAQUS input files on the attached link they restrain Z at the point of the load. If Z is not restraint the half symmetry model rotates out of plane.

Actually that is not true. I managed to obtain correct result without locking AC edge with Ansys and Code_Aster and friend of mine got correct results with use of Abaqus (please see attached screenshots) so probably model in Elmer is incorrect.