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

You are not logged in.

#1 2012-05-31 17:52:08

jeanpierreaubry
Ex-Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 2,894
Website

[SOLVED]problem with GROT_GDEP and LIAISON_SOLID

hello

i have a model that uses POU_D_T and COQUE_3D and LIAISON_SOLIDE to join them
the beam is perpendiclar to the beam
when performing a STAT_NON_LIN with COMP_INCR, RELATION=ELAS and DEFORMATION=GROT_GDEP
i get pretty similar results for displacement compared to a MECA_STATIQUE, 808 mm compared to 865 for static (on a span of 14000)

and the results for stresses around the LIAISON_SOLIDE are enormous, dadiating to similar vales awy from this area

is ths a normal behavior while using LIAISON_SOLIDE

enclose is a view of SIEQ_NOEU in the plates

jean pierre aubry

Last edited by jeanpierreaubry (2012-08-11 10:50:49)


Attachments:
view1.png, Size: 92.95 KiB, Downloads: 191

Offline

#2 2012-05-31 22:25:08

Thomas DE SOZA
Guru
From: EDF
Registered: 2007-11-23
Posts: 2,773

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

Hi Jean-Pierre,

LIAISON_SOLIDE is not made for large displacements. Since the results seem to be similar between MECA_STATIQUE and SNL, this should be OK but beware.

For the stress anomaly, this is a common problem with multi-point constraints such as this one. It may happen because the rigid body condition imposed by LIAISON_SOLIDE causes the structure to be too stiff locally.

TdS

Offline

#3 2012-06-01 06:20:12

jeanpierreaubry
Ex-Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 2,894
Website

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

hello Thomas

thanks a lot for the ansver
it relieves some some stresses of my designer's mind

if i understand well you clain that as long as the displacements are similar in a MECA_STATIQUE and a STAT_NON_LIN
the stress results from the MECA_STATIQUE are meaningful ?
that is my my point of view but your opinion is aprreciated

however is there a way to tackle this kind of configuration in large deformation and large rotation ?
would it give a better result with PETIT_REAC in the strategic area ?

jean pierre aubry

Offline

#4 2012-06-01 08:49:31

Thomas DE SOZA
Guru
From: EDF
Registered: 2007-11-23
Posts: 2,773

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

jeanpierreaubry wrote:

if i understand well you clain that as long as the displacements are similar in a MECA_STATIQUE and a STAT_NON_LIN
the stress results from the MECA_STATIQUE are meaningful ?
that is my my point of view but your opinion is aprreciated

I was a little wrong : I did not mean that exactly. If both computations are done with elastic properties and if the displacements between the two are small and identical then yes the results from MECA_STATIQUE are OK.
The added point is that displacements must be small for MECA_STATIQUE to give accurate results. This is the main assumption made in this operator.

jeanpierreaubry wrote:

however is there a way to tackle this kind of configuration in large deformation and large rotation ?
would it give a better result with PETIT_REAC in the strategic area ?

GROT_GDEP is fine for the plate, for the beam only one model currently supports large displacements (POU_D_TGM).
The main difficulty is that most of the multi-point constraints do not support large displacements : this is the case for LIAISON_SOLIDE, LIAISON_MAIL for example.
In fact there are very few loads that can depend on the displacements : contact is one them.

PETIT_REAC will not help since it is better to use GROT_GDEP when available and since it does not act on these loadings.

Circumventing this usually requires a lot of contortions or imagination.

TdS

Offline

#5 2012-06-01 10:30:58

jeanpierreaubry
Ex-Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 2,894
Website

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

Thomas DE SOZA wrote:

GROT_GDEP is fine for the plate, for the beam only one model currently supports large displacements (POU_D_TGM).

howwever  test case 'ssnl103' uses 'POU_D_T_GD' and looks fine?
load conditions are similar to mine, gravity and wind that will follow on the deformed geometry
which in case of a cantilever geometry may actually reduce the moment at the fixed end  induced by the loads
and thus making for a somewhat safer structure in practice

Thomas DE SOZA wrote:

Circumventing this usually requires a lot of contortions or imagination.

i will try my best

Offline

#6 2012-06-01 12:09:01

Thomas DE SOZA
Guru
From: EDF
Registered: 2007-11-23
Posts: 2,773

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

jeanpierreaubry wrote:
Thomas DE SOZA wrote:

GROT_GDEP is fine for the plate, for the beam only one model currently supports large displacements (POU_D_TGM).

howwever  test case 'ssnl103' uses 'POU_D_T_GD' and looks fine?
load conditions are similar to mine, gravity and wind that will follow on the deformed geometry

Sorry, you're right. I Completely forgot about it (don't like this element :-p).

TdS

Offline

#7 2012-06-01 12:53:53

jeanpierreaubry
Ex-Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 2,894
Website

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

Thomas DE SOZA wrote:

I Completely forgot about it (don't like this element :-p).

pourquoi ?
deli de faciès !

Offline

#8 2012-08-10 08:56:30

toltec
Member
Registered: 2009-01-04
Posts: 92

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

Greetings,

We need to do a STAT_NON_LIN, ELAS (preferably VMIS_ISOT_TRAC) ,GROT_GDEP analyses of a mechanism with stiffness (car suspension). I have done this many times using Abaqus, so am familiar with the convergence challenges. I see from this thread that this will be impossible using Aster because despite having the ability to solve significantly non-linear problems:

a) I will not be able to connect my model as most LIAISON_* connections are unavailable for large displacements (I have seen the high stresses mentioned by Jean-Pierre)

b) Discrete elements should not be used for large displacements (they run using GROT_GDEP, but with warnings)

Given that the above two modelling techniques are central to building a mechanism model:

1) Does EDF have plans to develop LIAISON_* connections to allow for large displacements? If not, what are the contortions and imaginings that Thomas alludes to? This limitation concerns me as I use LIAISON_SOLIDE to connect GROT_GDEP models that aren't mechanisms... results may therefore be suspect.

2) Similarly, are there plans to "upgrade" the discrete elements to GROT_GDEP?

Regards

Gary


Ubuntu 12.04
Intel(R) Core(TM)2 Quad CPU    Q9400  @ 2.66GHz

Offline

#9 2012-08-11 10:26:33

Thomas DE SOZA
Guru
From: EDF
Registered: 2007-11-23
Posts: 2,773

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

Hi,

Building connectors that support large displacements is a difficult task. I can't tell you if it will be on the roadmap.

However I'd like to underline something : there's a difference between the high point-wise stresses and the ability to deal with large displacements when using connectors.
The high stresses come directly from the way these connectors are built and may very well be present even if the connectors did support large displacements.
The fact that they don't support large displacements is because the linear relations applied to the displacements are not updated during the computation. This eases the programming since the matrix profile reamins the same. Changing the profile can be done as this is already the case for contact.

Usually the high point-wise stresses are discarded because we're not interested in the results near the connector.

So the results you obtain with connectors and large displacements may still be correct. It's all the more true if the large displacements occur far from the connectors. Usually using large displacements with for example LIAISON_SOLIDE leads to an expansion of the rigidified part of the structure rather than point-wise higher stresses.

TdS

Offline

#10 2012-08-13 09:55:23

toltec
Member
Registered: 2009-01-04
Posts: 92

Re: [SOLVED]problem with GROT_GDEP and LIAISON_SOLID

Hello Thomas,

Thanks for your reply. I'm currently building a number of simple models to get some understanding of how potential hinge set-ups work with various STAT_NON_LIN options. I'm sure I'll end up posting some questions on the forum, if not I'll post my example models which may help others.

Cheers

Gary


Ubuntu 12.04
Intel(R) Core(TM)2 Quad CPU    Q9400  @ 2.66GHz

Offline

Board footer