Distant support/load and COQUE_3D in STAT_NON_LINE + GROT_GDEP

k_zurawski

From U4.81.04:
Si le mot clé GROUP_MA est renseigné, FK (option FORC_NODA) est calculé uniquement sur les éléments
demandés puis assemblé. Le résultat est différent d’un calcul global sur tout le domaine puis réduit aux
éléments demandés. La méthode implantée permet de mesurer la réaction d’un morceau de modèle sur un
autre.
So, it should be possible to limit by element group, but when I changed GROUP_NO to GROUP_MA I'm getting this error:
<EXCEPTION> <CALCUL_38>

le calcul de l'option : FORC_NODA
n'est possible pour aucun des types d'éléments du LIGREL. Si le LIGREL repose sur un MAILLAGE_P, ceci est tout à fait possible localement donc on ne s'arrête plus en erreur fatale (la vérification globale a été désactivée, pensez à vérifiez si cela est attendu)

What are LIGREL elements and MAILLAGE_P?

jeanpierreaubry

quick answer
in FORC_NODA NODA means nodal (on nodes) so the operation is only possible on nodes

What are LIGREL elements and MAILLAGE_P?

this is a typical code_aster message not understandable to laymen like us
i think LIGREL as something to do with an internal LIbrary of ELements
MAILLAGE_P in do not know at all

k_zurawski

jeanpierreaubry I tried both ways, first with GROUP_NO, second GROUP_MA. In both cases I ended up with error. I don't want to export results to Excel, it's possible to plot in Paraview/Paravis, but either I need to subtract reactions from forces in Paravis or restrict forces to nodal/mesh group. It's so simple in other software...

jeanpierreaubry

better share the .mess file

k_zurawski

jeanpierreaubry this is for different, but similar study.

fg-example-02.txt

74kB

jeanpierreaubry

GROUP_MA 'LOAD' contains SEG3 elements that are connected to some plates elements
FORC_NODA in LOAD depends also on these plates and cannot be computed independently
so the drop the keyword GROUP_MA from CALC_CHAMP
and put it in the IMPR_RESU to restrict the output to that group

another point
i made a simpler example of you want to do
one with COQUE_3D the other one with 2 3D elements in the thickness
displacement and stress vary by about 5% and (as i expected) elapsed time for the problem is around 7 times less

k_zurawski

Hello. I restricted the result for FORC_NODA to nodal group 'LOAD'. Load is 10kN/m, T beam is 2m long, so at the last step I should have 20kN in total. I'm getting 2 results of _7251N for DZ and I'm not able to change direction of the arrows using Glyph filter. Something is still off. Could you look into it? I can't find the cause of the problem with the values.

There are very small vectors/arrows between those two:

The goal is to sum the applied forces to 'LOAD' group and plot sums over time step. I can do it in Excel, but this is not the way I should do it.

Additionally, I got few messages about local coordinate systems are not compatible to obtain some results (line 1275 in mess file). I might have made a mistake, with VECT in AFFE_CARA_ELEM, but from my understanding this should be ok.

I uploaded the files to:
https://we.tl/t-fEUxIGdzsr

jeanpierreaubry

what are the result if you print in the .resu ascii file
i cannot give you any advice regarding the use of Paravis, i have given up using such a complicated tool while Gmsh gives me all what i want in a straightforward matter

k_zurawski

jeanpierreaubry Is it possible to display a sum in GMSH? When I calculated it by hand I'm missing 0,8kN.

k_zurawski

I share your opinion on Paraview's complexity, but after a while I got used to it. I don't want to copy data across few programs, so I decided to use what comes with Salome Meca package. But this is the same as in case of C_A - learning curve is steep. In both cases It's worth the time, because there's no other such a complete FEA Open-Source package "on the market".
I know that there's a lot of things you can do with C_A, but despite the fact I have comm files prepared, it takes me twice the time I need to prepare new, similar simulation in other software. The process of obtaining results is very complex and requires knowledge of many dependencies. Like in case of that plot. You might think that you just choose DX and plot it over time. For some odd reason FORC_NODALE contains forces and reactions and this makes it difficult to plot only applied forces. I even don't know how to make a sum of those nodal forces from FORCE_ARETE to extract total load for that plot.

...back to the subject. I'll check the results in *.resu file and I'll get back with some answers, hopefully.

k_zurawski

I checked results for last timestep. For DZ I'm getting the same values like in Paraview's viewport (+7251) for two nodes at the edges of nodal group 'LOAD' and 199 values -33,333... If I multiply 199*(-33,333)=-6633,33 Which is nowhere near 20kN. Since this is FORCE_ARETE and there are 201 nodes on that edge, there should be ~ 0.1kN at every node (except the ones at both ends. I don't understand how it works...

fg-example-02.txt

311kB

jeanpierreaubry

i do not understand how it should be possible to plot a sum in a graphical view
Gmsh or any other

as far as i know the only way to check the validity of the applied forces is to compute the sum of REAC_NODA on the ground support with POST_RELEVE_T
and here there is some difficulties with kinematic constraint like LIAISON_SOLIDE

on Gmsh we can plot individual reaction vector on these nodes as vector
this is explained in my book

k_zurawski

jeanpierreaubry There's filter called calculator you can use to do different calculations, but first I must have applied forces only to a group of nodes to be able to sum them. If result FORC_NODALE contained only forces applied, it would be fairly easy.
If I have 20kN applied gradually across all time steps (step 1 - 10%, step 2 - 20% etc.) and it's really impossible to get the values form every time step I must give up and use Excel instead. But it should be somehow implemented, it's a few clicks in Femap...

jeanpierreaubry

k_zurawski

I must give up and use Excel

i have not used that software for the last 20 years
i use Gnumeric that is Free Software!

k_zurawski

I'd definitely use Paraview to do that, not some other external software...
...if only I could do

k_zurawski

To sum up, there is a way to calculate total force applied by using CALC_TABLE U4.33.03

« Previous Page