Welcome to the forums. Please post in English or French.

You are not logged in.

#1 2022-08-15 18:12:37

jlucas
Member
Registered: 2021-12-15
Posts: 33

= Buckling of a tube under twist (non-linear) calculix vs code-aster =

Hello,

I'm trying to perform a simulation to identify buckling of a tube under a twist moment. So far I did not manage to achieve this with code-aster, mainly, because I could not get right the formalism to apply large rotations. I present bellow my progress starting with the achievement using calculix and then each of my failed tries with code-aster. Any help/comment is welcome.

* Geometry of the tube:
  radius (m): 0.030
  thickness (m): 0.002
  lenght (m): 0.950
* Material properties (carbon steel):
  E (N/m^2): 207000000000.0
  nu: 0.290
* Boundary conditions:
  Fixed in one side;
  Twist moment is applied on the other side (0 - 6. rad).
  All other degrees of freedom except twist are supposed fixed on the loaded face.

The mesh is made of solid 1st order tetrahedral, so very accurate results are not to be expected.
The main objective at present stage, is only to check qualitatively that I the simulation is giving some sensible results.

Offline

#2 2022-08-15 18:22:31

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== CalculiX. ==
Enclosed are the simulation files, load path and an animation of the deformation.

(Note: The input files from CalculiX are compatible with Abaqus, so it should be easy to adapt and run the simulation using this software (in case someone wants to try)).

The loading is applied by connecting all dofs of a reference node with a surface using a *COUPLING/KINEMATIC card (see manual for reference).

****
*COUPLING, REF NODE=7430,SURFACE=sLoad,CONSTRAINT NAME=CN1
*KINEMATIC
1,2,3
****

Output files are converted to vtu in order to be viewed with paraview. Load path shows clearly stability loss associated with buckling and the following animation shows the result of the deformation. Qualitatively as expected.


Attachments:
buckling_tube_calculix.gif, Size: 192.08 KiB, Downloads: 58

Offline

#3 2022-08-15 18:23:35

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== CalculiX Files ==


Attachments:
calculix.zip, Size: 646.61 KiB, Downloads: 32

Offline

#4 2022-08-15 18:25:45

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== Code-Aster: Try A ==

The loading is applied via a discrete element connected with LIAISON_SOLID to the nodes on the face where twist is to be applied. Deformation is set as PETIT, so not really expected to give valid results.


Attachments:
buckling_tube_code_aster_tryA.gif, Size: 318.02 KiB, Downloads: 43

Offline

#5 2022-08-15 18:27:12

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== Code-Aster: Try A. Files ==


Attachments:
tryA.zip, Size: 757.22 KiB, Downloads: 26

Offline

#6 2022-08-15 18:28:35

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== Code-Aster: Try B ==
Mainly the same as Try A, but setting the deformation as "GROT_GDEP".

Looking at the load path and the deformation obviously something is wrong. Help is welcome and appreciated.

The doc U4.44.01 (AFFE_CHAR_MECA) (section 3.6) specifies that to use LIAISON_SOLIDE with large deformation hypothesis, the loading needs to be specified as TYPE_CHARGE='SUIV'. But by specifying such leads to the following error:

║ Erreur utilisateur :   
║   Le chargement contient des relations cinématiques LIAISON_SOLIDE qui sont non-linéaires
║ lorsque l'on utilise EXCIT / TYPE_CHARGE='SUIV'.   
║   Mais ce cas n'est pas traité car il y a au moins un noeud qui porte les degrés de liberté
║ DRX, DRY et DRZ.

So TYPE_CHARGE='SUIV' is not compatible with a rotation?


Attachments:
buckling_tube_code_aster_tryB.gif, Size: 124.4 KiB, Downloads: 39

Offline

#7 2022-08-15 18:30:06

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== Code-Aster: Try B. Files ==


Attachments:
tryB.zip, Size: 567.86 KiB, Downloads: 23

Offline

#8 2022-08-15 18:32:27

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== Code-Aster: Try C ==

Trying a "rotational" boundary condition on all nodes as in example ssnv507 (V6.04.507: Rotation d'une inclusion rigide avec X- FEM)

DTH=6.0;

tournex = FORMULE(VALE='sqrt(X*X+Y*Y)*cos(atan2(Y,X)+DTH*INST)-X',
                  DTH=DTH,
                  NOM_PARA=['X', 'Y', 'INST'],)

tourney = FORMULE(VALE='sqrt(X*X+Y*Y)*sin(atan2(Y,X)+DTH*INST)-Y',
                  DTH=DTH,
                  NOM_PARA=['X', 'Y', 'INST'],)



load=AFFE_CHAR_MECA_F( MODELE=model,
                       DDL_IMPO=(_F(GROUP_MA='sLoad',
                                    DX=tournex,DY=tourney)),
);
 

Results are still not right. See load path and deformation. Also, I don't know with this formulation how to specify in addtion that all other degrees of freedom except twist are fixed.


Attachments:
buckling_tube_code_aster_tryC.gif, Size: 181.76 KiB, Downloads: 49

Offline

#9 2022-08-15 18:33:37

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

== Code-Aster. Try C. Files. ==


Attachments:
tryC.zip, Size: 548.26 KiB, Downloads: 32

Offline

#10 2022-08-17 08:35:12

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,860

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

hello

i think you are loosing track with try C

at first lets forget the critical buckling load and concentrate on the reaction forces

if you were using a much coarser mesh, a few tens times less elements !
you could apply nodal forces on the perimeter at the top to mimic  the torsion couple
first you could find that the results are similar with a coarser mesh
second you could find (as i did) that the results are the same as with LIAISON_SOLIDE
which then cannot be considered at fault

just simply the calculation of reaction is completely erroneous with GROT_GDEP
and this seems to be the case whatever the element type 1D, 2D, 3D or whatever the load as i found out from other studies

and if the reactions are wrong one can wonder if the stress are not wrong either

i made several post here about that with no reaction from EDF team

are we at fault in our way of putting the problem together?
or is code_aster at fault?
it would be nice to have an answer

jean pierre aubry


consider reading my book
freely available here https://framabook.org/beginning-with-code_aster/

Offline

#11 2022-08-17 13:50:08

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,860

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

attached is a simplified similar problem
same overall geometry
only plate elements with just only  240 elements
torsion couple is imitated with only 4 forces at each quadrant
so do not look too much at result around the load area
the general behavior is very similar to jlucas mesh with 22567 elements
applied torsional couple 20000
results are

with GROT_GDEP
sum of reaction 7000
maximum tangential displacement 0.0088
VMIS near fixed points 1.2e9

with PETIT
sum of reaction 20000
maximum tangential displacement 0.024
VMIS near fixed points 3.43e9

analytical
sum of reaction 20000
maximum tangential displacement 0.0224
VMIS near fixed points 1.89e9

comments
reaction with GROT_GDEP are wrong  0.35 times the applied load
displacement with GROT_GDEP are 0.36 times the one with PETIT
VMIS is not so far out from analytical

what to conclude about that????


Attachments:
plate_model.zip, Size: 7.04 KiB, Downloads: 26

consider reading my book
freely available here https://framabook.org/beginning-with-code_aster/

Offline

#12 2022-08-19 10:08:55

jlucas
Member
Registered: 2021-12-15
Posts: 33

Re: = Buckling of a tube under twist (non-linear) calculix vs code-aster =

Thank you Jean Pierre Aubry for looking more deeply into this and for simplifying the problem. It might help to grasp it easily.

I totally agree that the silence on this topic from development team is quite uncomfortable. (I guess most members might be on Holiday?)

I post here some selected links with related topics:

Challenge: Buckling of a tube under twisting moment...
(can not offer any other prize than personal gratification and honor mention in present post!)

Steel tube under twist moment. Large deformations. (code-aster v15.4)

Buckling of a cylinder under a twist moment.

Also might help:
LIAISON_SOLIDE with TYPE_CHARGE='SUIV' and GROT_GDEP

Rotation of a cylinder

DEFORMATION='GROT_GDEP’ and FORC_NODA moment equilibrium

Offline