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

You are not logged in.

#1 2014-06-24 16:46:00

pafortin
Member
Registered: 2014-05-06
Posts: 23

Error after using LIAISON_RBE3

Hello,

I am using Code_Aster (v11.5 on Debian) to model the deformations and efforts on a container ship. To this day, my company would use Nastran and I am studying the possibility of using CA instead. For my tests, I am using only a slice of an "empty" container ship (without any container box) under dummy loadings (gravity, pressure, nodal forces). So far everything went rather well (thanks to the help of this forum and to Jean-Pierre Aubry's book).
I then tested the same slice this time with container boxes. The Nastran methodology of my company was to model the container boxes by discrete masses linked to the hull by RBE3 elements. Luckily for me, CA has the LIAISON_RBE3 operator which does exactly the same job as the original RBE3 of Nastran. After a few experimentations, I managed to get my syntax correct. Unfortunately, I get an error related to the factorisation which I do not understand at all :

<EXCEPTION> <CALCULEL6_15>

L'élément diagonal u( 1 , 1 ) de la factorisation est nul.
la solution et les estimations d'erreurs ne peuvent être calculées.

This error appeared only when I added the RBE3 modelization so I assume it is linked to the usage of LIAISON_RBE3 but I have nothing to confirm that.
I am rather new to FEM so I don't even know what step of the solver is causing this problem. The mistake is probably on my side but I have no clue what I need to modify.
I tried to look around if anyone had had the same problem but I haven't found anything.

Has anyone ever seen this error or knows how to work around it ?
Any help or suggestion is welcome !

My input and mess files are attached.

Best regards,

Pierre-Arthur Fortin


Attachments:
cont_ship_tranche.zip, Size: 87.62 KiB, Downloads: 193

Offline

#2 2014-06-24 20:21:35

keeswouters
Member
From: kuringen
Registered: 2007-12-01
Posts: 144
Website

Re: Error after using LIAISON_RBE3

Hoi Pierre-Arthur

I used the RBE3 command once for applying a torque to a solid and I had to use some trick on the master node.
Because a 3D node will only carry translations, no rotations, you can add a ''mass and/or stiffness'' element to it to get the rotational dofs:

<code>
N_stmass=AFFE_CARA_ELEM(MODELE=CylModel,
                        INFO=2,
                         DISCRET=(_F(CARA='M_TR_D_N',
                                     GROUP_NO='Nmaster',
                                     VALE=[0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.00,0.00,0.00]),
                                 _F(CARA='K_TR_D_N',
                                    GROUP_NO='Nmaster',
                                    VALE=[0.0,0.0,0.0,0.0,0.0,0.0]),),);

nodeRBE3=AFFE_CHAR_MECA(MODELE=CylModel,
                      LIAISON_RBE3=_F(GROUP_NO_MAIT='Nmaster',
                                      DDL_MAIT=('DX','DY','DZ','DRX','DRY','DRZ',),
                                      #DDL_MAIT=('DRZ',),
                                      GROUP_NO_ESCL='Nbot',
                                      DDL_ESCL='DX-DY-DZ',
                                      COEF_ESCL=1,),);
</code>

I did have a glance at your mess file and I am not sure whether you applied this ''trick'' or not and whether it has to be used in your case.
Your code being eg (line 2503 of the mess file):

                             NOEUD_MAIT='N35701',
                              GROUP_NO_ESCL=(
                                             'R99245_0',
                                             'R99245_1',
                                             ),
                              DDL_MAIT=(
                                        'DX','DY','DZ','DRX','DRY','DRZ',),
                              DDL_ESCL=(
                                        'DX-DY-DZ',
                                        'DX-DY-DZ',
                                        'DX-DY-DZ',
                                        'DX-DY-DZ',

<guess>: I ''think'' you did not apply this ''trick''. From the two master nodes the 3 rotations cannot be determined. </guess>
Is your slave group always fit (no co-linear nodes) to determine the eg DDL_MAIT=(DX','DY','DZ','DRX','DRY','DRZ',) from the  DDL_ESCL=..?
I must say the error message does not indicate such an error.

Last edited by keeswouters (2014-06-24 20:33:44)


kind regards - kees
--
I a parallel univers the laws of mechanics may be different.

Offline

#3 2014-06-25 11:06:58

pafortin
Member
Registered: 2014-05-06
Posts: 23

Re: Error after using LIAISON_RBE3

Hello Kees,

Thank you for your answer.

I think I am already applying your trick in my case as each master node corresponds to a mass M_TR_D_N element (with proper mass and inertia values).
In your example, the mass and stiffness values are set to zero. Would it work with a different value ?
Your trick seems to suggest that efforts can be applied through a RBE3 element only if the master node has been affected with a mass or stiffness property (even if the value is zero). Am I understanding you correctly ?

Anyway and in order to identify the error, I suppressed all the rotation dofs from all my master nodes so that the RBE3 carries only translation information.
Unfortunately the problem remains the same, I get exactly the same error.

I agree with you, the error message is unfortunately not very clear in this case.

Any other idea why the usage of RBE3 generates this factorisation related error ?

Thank you for your help.

Regards,

Pierre-Arthur

Offline

#4 2014-06-25 19:32:04

keeswouters
Member
From: kuringen
Registered: 2007-12-01
Posts: 144
Website

Re: Error after using LIAISON_RBE3

Hoi Pierre-Arthur

pafortin wrote:

I think I am already applying your trick in my case as each master node corresponds to a mass M_TR_D_N element (with proper mass and inertia values).

Okee, I looked into your mail and comm files and after some searching I see that N35701 is coupled to P100317 in group CP100317 that carries a mass and inertia. Is this coupling alright?

pafortin wrote:

In your example, the mass and stiffness values are set to zero. Would it work with a different value ?
Your trick seems to suggest that efforts can be applied through a RBE3 element only if the master node has been affected with a mass or stiffness property (even if the value is zero). Am I understanding you correctly ?

If it works for zero, it should work with any (positive) value.
I would assume that only a mass or only a stiffness attached to a node would be enough (as long as the element has rotational dofs if you need them).

Can you reproduce the error with a small model?
Or just adding one RBE3 constraint (and possibly a mass to the master node directly) and see what happens?

Last edited by keeswouters (2014-06-25 19:47:07)


kind regards - kees
--
I a parallel univers the laws of mechanics may be different.

Offline

#5 2014-06-27 17:14:11

RichardS
Member
From: Munich, Germany
Registered: 2010-09-28
Posts: 549
Website

Re: Error after using LIAISON_RBE3

Hello,
I think this

Is your slave group always fit (no co-linear nodes) to determine the eg DDL_MAIT=(DX','DY','DZ','DRX','DRY','DRZ',) from the  DDL_ESCL=..?

goes in the right direction.
I don't really know how NASTRAN defines the RBE3 connections, or how you converted them into the equivalent CA definitions, but I guess this
conversion is somehow faulty or some information is lost during the process.

If you replace all the 'DY' or 'DX-DY' dofs from your slave node connections by 'DX-DY-DZ' it works (somehow, but I guess the initial values are intended).

It could be that the order of the nodes inside the node groups is different from that of NASTRAN after exporting/importing the mesh file
and thus the RBE3 definition has to be adapted accordingly.

Best regards,
Richard


Richard Szoeke-Schuller
Product Management
www.simscale.com
We are hiring! https://simscale-jobs.personio.de/?language=en#all

Offline

#6 2018-05-16 13:11:44

FilippoP
Member
Registered: 2018-05-15
Posts: 6

Re: Error after using LIAISON_RBE3

Hello!

Does anyone solved this topic? If so, how?

Offline