Atom topic feed | site map | contact | login | Protection des données personnelles | Powered by FluxBB | réalisation artaban

You are not logged in.

- Topics: Active | Unanswered

Pages: **1**

**marco.mueller****Member**- Registered: 2020-01-16
- Posts: 6

Hi / Bonjour,

I'm quite new to code_aster and could obtain good results for models without contact. Now, for assemblies I have to introduce contact and created a benchmark model to get familiar with the process.

I get results, but:

- maximum displacement is 2.5 times higher than theoretic value

- reaction force is just 75 % of applied force (due to applied weak springs?)

- calculation takes quite long (> half hour for this simple model)

- I see some stress noise at my symmetry boundary conditions. Is this due to the linear>quadratic>biquadratic conversion inside code_aster, whereas node groups for boundary conditions were defined in Salome?

I'm glad for any comment / help I can get.

Thanks

Marco

Offline

**mib****Member**- Registered: 2019-11-10
- Posts: 17

marco.mueller wrote:

Hi / Bonjour,

- maximum displacement is 2.5 times higher than theoretic value

- reaction force is just 75 % of applied force (due to applied weak springs?)

is the rate :Max model Dimension / Min model Dimension <= 8 ?

marco.mueller wrote:

- calculation takes quite long (> half hour for this simple model)

it is short... depend from meshes dimensions and p.c. memory, and processor speed, increasing meshes dimension reduce time and precision of computation, decreasing meshes dimension, increase time computation and precision.

Offline

**marco.mueller****Member**- Registered: 2020-01-16
- Posts: 6

As I said, it is a sheet metal. I've attached result pictures.

Anyone who can help me with the combination of 2D elements and contact?

Thanks!

Offline

**mib****Member**- Registered: 2019-11-10
- Posts: 17

marco.mueller wrote:

As I said, it is a sheet metal. I've attached result pictures.

Anyone who can help me with the combination of 2D elements and contact?

Thanks!

is also possible in 3D (see attachment) under the condition of dimensions rate which i said.

BUT

your answer is good, i want understand how is possible in 2D model.

best regards.

Offline

**marco.mueller****Member**- Registered: 2020-01-16
- Posts: 6

I've managed to match reaction and applied force by weaker springs. Also the noise of stress went away by converting the mesh to biquadratic before creating node groups.

But still I'm wondering why it takes 40 minutes to compute 4 load steps for my "small" model: (1 contact, 250 slave elements)

Is it due to the overall size of the contact which spreads over the whole model? Is there a way to limit the region of contact pair detection?

Also and still the maximum displacement is very high.

Offline

**mib****Member**- Registered: 2019-11-10
- Posts: 17

marco.mueller wrote:

I've managed to match reaction and applied force by weaker springs. Also the noise of stress went away by converting the mesh to biquadratic before creating node groups.

But still I'm wondering why it takes 40 minutes to compute 4 load steps for my "small" model: (1 contact, 250 slave elements)

Is it due to the overall size of the contact which spreads over the whole model? Is there a way to limit the region of contact pair detection?

Also and still the maximum displacement is very high.

This is the case on 2D

very well !!!

in 3D model i observed which :

1) applying force like BC the solver goes mad, applying pressure like BC the solver thinks very good

2) increase the number of DOF the time of computation was longer

3) a long time of meshes creation on salome corresponded to a long time of computation in code aster

4) studying the documentation abut the solved i understand that PETSC solver reduce very much the time of computation ( but in the common version available there isn't the correlate library )

*Last edited by mib (2020-01-20 13:28:27)*

Offline

**mecour****Member**- From: Ostrava (Czech)
- Registered: 2011-04-04
- Posts: 131

Hello,

few question to your model:

Why do you apply discrete elements to whole structure called Blech, I really don't understand the meaning of that. Does this substitute the friction in contact. Because without any constraint in z direction the "blech" structure fall down somewhere. Also the stiffness of discrete elements is very small.

Why do you use all Blech mesh as master contact face. I think that only small part can participate in contact.

mecour

Offline

**marco.mueller****Member**- Registered: 2020-01-16
- Posts: 6

the discrete elements apply a "weak spring" to the structure, which is necessary for discrete contact. Otherwise no static solution would possible. There may be a smarter way to do this, but for this model it works.

I take the whole face because of the simple nature of the model. I also tested a smaller master contact face without any advantage for the performance. Thats btw what is stated in the documentation, too.

Coming back to my progress: I've found that my theoretic reference included a failure. Compared with a model that uses equivalent boundary conditions to replace the contact, I've found similar displacement values.

So finally I'm looking for a better performance for my model! Anyone who can help out here?

For those that would just like to have a look at my (very basic) contact settings (keep in mind, I'm using COQUE-3D and STAT_NON_LINE):

Kontakt = DEFI_CONTACT(FORMULATION='DISCRETE',

MODELE=FEModell,

ZONE=_F(CARA_ELEM=Eigensch,

DIST_COQUE='OUI',

GROUP_MA_ESCL=('Auflage_Kontakt', ),

GROUP_MA_MAIT=('Blech_Last_Kontakt', )))

Offline

Pages: **1**