LIAISON_OBLIQUE fails, if called twice for a node?

GFric · 2021-02-18 13:58:03

Hello,

I'm trying to support a simple 8-noded cube. For a side wall, I want horizontal support, for the bottom face support in all directions.

AFFE_CHAR_MECA( ...
    LIAISON_OBLIQUE = (
        # left face fixes global DX - works
        _F(GROUP_NO=('gNdLeft',), ANGL_NAUT=(0, 0), DX=0,),
        # front face fixes global DY  - works
        _F(GROUP_NO=('gNdFront',), ANGL_NAUT=(90, 0), DX=0,),
        
        # bottom face global DZ=0 - works
        _F(GROUP_NO=('gNdBottom',), ANGL_NAUT=(0, 90), DX=0,), # DY=0, DZ=0,),
        
        # now, try to also fix the bottom face in global DX,DY directions -> FAILS
        # any of these lines crash MECA_STATIQUE with EXECUTION_CODE_ASTER_EXIT_4647=0
        # _F(GROUP_NO=('gNdBottom',), ANGL_NAUT=(0, 90), DY=0,), # DY=0, DZ=0,),
        # _F(GROUP_NO=('gNdBottom',), ANGL_NAUT=(0, 90), DZ=0,), # DY=0, DZ=0,),
    ),

I think the problem is, that the nodes that are both, side and bottom, get two liaisons applied, that "fight" against each other.

Background: In real life, the side faces are spline shaped walls that I want to support 'normal to the face' and optionally also 'transversal to the normal'. I'm calculating the normals for each node individually and try to fix each node. Is there a better was to do this?

Attached is a very simple project file.

mf · 2021-02-18 14:47:50

Hello,

try SANS_GROUP_MA ('without_group_mesh', literally) or SANS_GROUP_NO ('without_group_nodes') in your BC definitions, it should work. Your assumption is correct, in principle it applies to all BCs. So you should decide which BCs you want to define in the way you did (perhaps the bottom one) and then use SANS_GROUP_XX in the others that have common nodes with the bottom,

Mario.

Last edited by mf (2021-02-18 14:53:16)

GFric · 2021-02-22 14:11:05

Thank you for your help.
This way I can fix the walls OR the floor. But I want the floors to be fixed horizontally, the bottom vertically and the ounter bottom edges BOTH horizontally AND vertically. Sort of "adding/accumulating a DOF".

mf · 2021-02-22 17:11:33

Hello,

could you post your example?

Mario.

GFric · Yesterday 08:34:36

I'm so sorry. I forgot to attach the example.
See attached.

Attachments:
support-addition.zip, Size: 26.24 KiB, Downloads: 3

mf · Yesterday 13:18:42

Hello,

unfortunately, I am currently on a only-SM machine right now, for CA I'd have to launch my bigger machine, which is not in use this week. Am I correct in the assumption that you want to fix these groups? So it is not clear to me why you want to use LIAISON_OBLIQUE. A simple DXi=0 in DDL_IMPO would not work on this mesh (cant open it in SM...unfortunately)? Are the walls of of this cube separate somehow in this mesh?

Mario.

Last edited by mf (Yesterday 16:04:11)

GFric · Yesterday 14:19:48

My real situation is way more complex. See the image attached.
I want to fix the outer faces normal to their face direction plus the bottom face in normal and transversal direction. At the common edges, both supports should add.

Attachments:
support_faces.png, Size: 19.04 KiB, Downloads: 8

#1 2021-02-18 13:58:03

LIAISON_OBLIQUE fails, if called twice for a node?

#2 2021-02-18 14:47:50

Re: LIAISON_OBLIQUE fails, if called twice for a node?

#3 2021-02-22 14:11:05

Re: LIAISON_OBLIQUE fails, if called twice for a node?

#4 2021-02-22 17:11:33

Re: LIAISON_OBLIQUE fails, if called twice for a node?

#5 Yesterday 08:34:36

Re: LIAISON_OBLIQUE fails, if called twice for a node?

#6 Yesterday 13:18:42

Re: LIAISON_OBLIQUE fails, if called twice for a node?

#7 Yesterday 14:19:48

Re: LIAISON_OBLIQUE fails, if called twice for a node?

Board footer