[SOLVED] PRE_SIGM use for Preload

jlpaladini · 2016-12-30 20:13:37

Hi, My name's Jorge. I want to use the command "PRE_SIGM" to preload a bolt in a joint but I can not do it. How to create a "cham_elem"?
Thanks and sorry for my english.
Jorge

Attachments:
pret.comm, Size: 2.77 KiB, Downloads: 541

jeanpierreaubry · 2016-12-31 14:24:31

hello

without the mesh it is very difficult to tell

jean pierre aubry

jlpaladini · 2017-01-02 12:41:12

Hello, Thanks. I attach the file .hdf

jlpaladini · 2017-01-02 12:43:56

Hello, Thanks. I attach the file .med

Last edited by jlpaladini (2017-01-02 12:44:11)

Attachments:
Compound_Mesh_3.med, Size: 477.4 KiB, Downloads: 464

jeanpierreaubry · 2017-01-02 14:24:15

hello

try this

prett=CREA_CHAMP(
	TYPE_CHAM='ELGA_SIEF_R', 
	OPERATION='AFFE',
	MODELE=mode,
	AFFE=_F(
		GROUP_MA='GrMesh_2_Edges',
		NOM_CMP=('N','VY','VZ','MT','MFY','MFZ',),
		VALE=(10.0, 0, 0, 0, 0, 0,),
	),
	INFO=2,
	PROL_ZERO='OUI',
);

jean pierre aubry

jlpaladini · 2017-01-02 15:05:51

Thank you jean pierre aubry, this is a solution!!
best regards, Jorge

Pisolino · 2020-01-02 17:32:43

Hi all,
i've tried to follow your advice to model the pre-load of a bolt in an example very similar to the one attached before. However in the very first iteration my bolt gets elongated by the value of the stress/modulus. What i wanted to do however was to say to CA that the linear element @start has already a stress "inside" at that fixed elongation.
Am i doing something wrong ?

For example i tried to set a negative PRE_SIGM value but then the simulation does not converge in the first iteration.

Thank you in advance,

Andrea

miib · 2021-11-24 17:32:06

Pisolino wrote:

Hi all,
i've tried to follow your advice to model the pre-load of a bolt in an example very similar to the one attached before. However in the very first iteration my bolt gets elongated by the value of the stress/modulus. What i wanted to do however was to say to CA that the linear element @start has already a stress "inside" at that fixed elongation.
Am i doing something wrong ?
For example i tried to set a negative PRE_SIGM value but then the simulation does not converge in the first iteration.

Thank you in advance,
Andrea

i have a similar problem :

using only PRE_SIGM with out other loads it works (only for a limited range of values)

using only PRE_SIGM with other loads it doesn't work, there isn't its computation

What i don't understand ???

:-)

Last edited by miib (2021-11-24 17:34:03)

miib · 2021-11-24 22:19:10

solved

attached an example :-)

Attachments:
pre_sigm_by_proji.tar.gz, Size: 65.96 KiB, Downloads: 158

miib · 2021-11-28 17:01:54

how to close two plates with a bolt in 3D

is the procedure right ???

if you want try to run the attached :-)

Attachments:
plates_colosed_with_a_bolt.zip, Size: 309.58 KiB, Downloads: 161

jeanpierreaubry · 2021-11-29 14:19:29

hello miib

i had a look at your problem
and i have several questions

why DYNA_NON_LINE i do not see any dynamics in this problem?
why is not there any ramp to the load?
why a COULOMB contact?
why is there a 'model' and a 'modelX'

more generally what are you trying to do ?

jean pierre aubry

Last edited by jeanpierreaubry (2021-11-29 14:34:36)

miib · 2021-11-29 20:48:29

Hi jean pierre

jeanpierreaubry wrote:

why DYNA_NON_LINE i do not see any dynamics in this problem?

because i have several DOF free ( not constrained ) and in this case, using static simulation, the matrice will be singular, DYNA with simplicity bypass this "problem".

jeanpierreaubry wrote:

why is not there any ramp to the load?

because using in PRE_SIGMA, the "voice" VALE_F, i obtain an error that tell me which there is an inconsistency between the function's values and same points of the bodies

jeanpierreaubry wrote:

hello miib
why a COULOMB contact?

it is a refuse of my precedent studies when i used a real thread in screw and nut , ( the value of preload depend by friction coefficient :-D)

jeanpierreaubry wrote:

hello miib
why is there a 'model' and a 'modelX'

because i observed that PRE_SIGM doesn't work in the model where there is others loads, but it works using only its (without others load), so i created two models equal ( the only difference is the names) where in the first i applied all the loads with out pre_sigm , and in the second model i applied only pre_sigm, and after with proji_champ i combine the two effects, this work good.

jeanpierreaubry wrote:

more generally what are you trying to do ?

i'm trying to close two plates with a bolt using 3D elements.

:-)

Last edited by miib (2021-11-29 21:02:34)

jeanpierreaubry · 2021-11-30 06:35:40

i will reply a bit later to the previous points

but
looking at the results of your analysis
it seems to me they are not realistic
did you had a look at the results just simply displacement ?

there is also a warning in the contact treatment
<A> <CONTACT2_15>
that may explain the strange results

jeanpierreaubry · 2021-11-30 09:06:46

enclosed is a modified command file that solves the problem with

adding another load
adding ramps to both loads
using one single model, the one named 'model'

in addition i have used a somehow more realistic material with simple ELAS properties

in the end this gives some more realistic displacements than your file

i still believe that a STAT_NON_LINE with soft springs would handle the problem much better
using DYNA_NON_LINE for the only reason that it solves a STATIC problem does not seems a sound engineer practice

this holds true for some other experiment you tried

if anything is not clear ask

Attachments:
closing-modified.comm, Size: 7.79 KiB, Downloads: 234

miib · 2021-11-30 20:18:18

jeanpierreaubry wrote:

enclosed is a modified command file that solves the problem with
adding another load
adding ramps to both loads
using one single model, the one named 'model'
in addition i have used a somehow more realistic material with simple ELAS properties
in the end this gives some more realistic displacements than your file
i still believe that a STAT_NON_LINE with soft springs would handle the problem much better
using DYNA_NON_LINE for the only reason that it solves a STATIC problem does not seems a sound engineer practice
this holds true for some other experiment you tried
if anything is not clear ask

Thanks :-)

the problem remains

miib wrote:

Hi jean pierre
jeanpierreaubry wrote:
why DYNA_NON_LINE i do not see any dynamics in this problem?
because i have several DOF free ( not constrained ) and in this case, using static simulation, the matrice will be singular, DYNA with simplicity bypass this "problem".

consider that I am self-taught,

attached the error message, someone hacked me the version of code_aster ??? :-)

Last edited by miib (2021-11-30 20:20:03)

Attachments:
message.comm, Size: 143.64 KiB, Downloads: 137

miib · 2021-12-01 06:34:32

miib wrote:

someone hacked me the version of code_aster ??? :-)

mah!!!

using DYNA it works :-)

Attachments:
message - dyna.comm, Size: 104.04 KiB, Downloads: 154

jeanpierreaubry · 2021-12-01 09:07:27

i ran the problem on my own 14.4 version and it fails while it runs well on 15.2 version

the most probable reason is that one

!-------------------------------------------------------------------------------------------------!
! <A> <CONTACT2_15> !
! !
! Contact discrete formulation. !
! The contact zones number 2 and number 4 have 20 nodes common to their surfaces slaves. That can !
! sometimes lead to a singular contact matrix. !
! !
! If computation had suddenly failed, check the definition of your contact surfaces or !
! inform one of key words SANS_NOEUD/SANS_GROUP_NO/SANS_MAILLE/SANS_GROUP_MA. !
! !
! !
! This is a warning. If you do not understand the meaning of this !
! warning, you can obtain unexpected results! !
!-------------------------------------------------------------------------------------------------!

it concerns surfaces 'sv' and 'si'
you should create a group of these common nodes and use SANS_GROUP_NO as stated in the message

you should always read completely the message file when things go wrong!!

anyhow enclosed is a .comm file that creates this GROUP_NO (at the beginning) and runs on 14.4

once this done you should have a look at the results
to my opinion they are physically meaningless
try to find out why!!

miib · 2021-12-01 20:14:11

jeanpierreaubry wrote:

i ran the problem on my own 14.4 version and it fails while it runs well on 15.2 version

ok :-)

jeanpierreaubry wrote:

the most probable reason is that one
!-------------------------------------------------------------------------------------------------!
! <A> <CONTACT2_15> !
! !
! Contact discrete formulation. !
! The contact zones number 2 and number 4 have 20 nodes common to their surfaces slaves. That can !
! sometimes lead to a singular contact matrix. !
! !
! If computation had suddenly failed, check the definition of your contact surfaces or !
! inform one of key words SANS_NOEUD/SANS_GROUP_NO/SANS_MAILLE/SANS_GROUP_MA. !
! !
! !
! This is a warning. If you do not understand the meaning of this !
! warning, you can obtain unexpected results! !
!-------------------------------------------------------------------------------------------------!

solved se attached, the problem in my version 14.4 remains (is the same with 13.6) :-)

jeanpierreaubry wrote:

to my opinion they are physically meaningless
try to find out why!!

right, time step is not realistic :-)

Attachments:
message_contact_alarm_solved.comm, Size: 140.94 KiB, Downloads: 147

jeanpierreaubry · 2021-12-02 08:18:44

you will never get any results if you keep not applying the modifications i suggest

when i speak of physically meaningless
i mean that the calculated displacement are strange to my eyes
you cannot have an opinion about a result you do not obtain

and you do not have a problem of time step but a problem of not applying the SANS_GROUP_NO in the contact

once you get a result we may talk of the many other things that are not really good in your study

miib · 2021-12-02 20:19:10

jeanpierreaubry wrote:

you will never get any results if you keep not applying the modifications i suggest
when i speak of physically meaningless
i mean that the calculated displacement are strange to my eyes
you cannot have an opinion about a result you do not obtain
and you do not have a problem of time step but a problem of not applying the SANS_GROUP_NO in the contact
once you get a result we may talk of the many other things that are not really good in your study

fais-tu de la politique ? :-D
toutefois
il n'en reste pas moins qu'avec STAT_NON_LINE ça ne marche pas alors qu'avec DYNA_NON_LINE ça marche et voilà les résultats (voir pièces jointes)

Sur le sens physique peut-être ok : en réalité le boulon est en traction, dans la simulation et en compression mais je m'intéresse à la fermeture des plaques, et non le boulon en traction.

:-)

Attachments:
after_the_second_timestep_go_crazy.png, Size: 326.68 KiB, Downloads: 128

jeanpierreaubry · 2021-12-03 08:22:47

eh bien oui ça ne marche pas
en autre avec PRE_SIGM qui semble ne pas être admis avec DYNA_NON_LINE
c'est écrit dans U4.44.01 au chapitre PRE_SIGM
on ne peut pas détourner les opérateurs de ce pour quoi ils sont fait
cependant le fichier que j'avais joint tourne avec des résultats très cureiux par rapport à ce que je crois être la réalité physique du problème
l'iamge que vous joigné montre un maillage horrible ! ou je me trompe ?

Last edited by jeanpierreaubry (2021-12-03 08:26:32)

scs · 2021-12-03 11:51:22

refer attachment. few months back i have done boltnut study.

Attachments:
0001.jpg, Size: 470.47 KiB, Downloads: 135

miib · 2021-12-03 19:01:12

jeanpierreaubry wrote:

eh bien oui ça ne marche pas
en autre avec PRE_SIGM qui semble ne pas être admis avec DYNA_NON_LINE
c'est écrit dans U4.44.01 au chapitre PRE_SIGM
on ne peut pas détourner les opérateurs de ce pour quoi ils sont fait
cependant le fichier que j'avais joint tourne avec des résultats très cureiux par rapport à ce que je crois être la réalité physique du problème
l'iamge que vous joigné montre un maillage horrible ! ou je me trompe ?

oui, les maillages ne conviennent pas à un calcul précis mais ils conviennent à un calcul rapide (démonstration)

Considérant que votre schéma de calcul avec STAT_NON_LINE dans mon software ne fonctionne pas, mais ne fonctionne qu'avec DYNA_NON_LINE, d'après votre expérience il est permis d'utiliser PROJI_CHAMP sur PRE_SIGM, je pensais au principe de superposition des effets.

:-)

jeanpierreaubry · 2021-12-04 09:27:20

mon expérience me dirait de ne pas dévoyer les opérateurs
le fait qu'un opérateur (ici DYNA_NON_LINE) donne un résultat n'est en aucun cas une preuve de la validité de la démarche !
de fait je ne sais pas trop bien interpréter la remarque de la doc à propos de PRE_SIGM

pour ma part j'ajouterais au maillage quelques ressorts assez mous pour bloquer les mouvement de corps rigides des corps qui en ont
ce n'est pas la mer à boire et c'est la solution généralement admise

certains softwares ajoutent ces blocages d'eux mêmes d'une manière pas très transparente
code_aster ne fait jamais cela à votre place, a vous de le faire

miib · 2021-12-04 11:09:45

jeanpierreaubry wrote:

mon expérience me dirait de ne pas dévoyer les opérateurs
le fait qu'un opérateur (ici DYNA_NON_LINE) donne un résultat n'est en aucun cas une preuve de la validité de la démarche !
de fait je ne sais pas trop bien interpréter la remarque de la doc à propos de PRE_SIGM

ok, je m'engage à ne pas faire un usage abusif des opérateurs :-)

jeanpierreaubry wrote:

pour ma part j'ajouterais au maillage quelques ressorts assez mous pour bloquer les mouvement de corps rigides des corps qui en ont
ce n'est pas la mer à boire et c'est la solution généralement admise

bon, je ne vous demande pas de perdre du temps à m'apprendre comment le faire, mais gentiment je vous demande un link (pdf, etc) pour apprendre à le faire avec code_aster, en d'autres termes gentiment je vous demande du matériel d'étude. Je n'ai PAS lu votre livre en entier, n'est-ce pas ? (si je ne comprends pas, je demanderai des informations ici) :-)

jeanpierreaubry wrote:

certains softwares ajoutent ces blocages d'eux mêmes d'une manière pas très transparente
code_aster ne fait jamais cela à votre place, a vous de le faire

oui, par exemple, j'ai vu qu'Autodesk Inventor le fait dans son simulateur intégré (je suis un CAD designer).

:-)

#1 2016-12-30 20:13:37

[SOLVED] PRE_SIGM use for Preload

#2 2016-12-31 14:24:31

Re: [SOLVED] PRE_SIGM use for Preload

#3 2017-01-02 12:41:12

Re: [SOLVED] PRE_SIGM use for Preload

#4 2017-01-02 12:43:56

Re: [SOLVED] PRE_SIGM use for Preload

#5 2017-01-02 14:24:15

Re: [SOLVED] PRE_SIGM use for Preload

#6 2017-01-02 15:05:51

Re: [SOLVED] PRE_SIGM use for Preload

#7 2020-01-02 17:32:43

Re: [SOLVED] PRE_SIGM use for Preload

#8 2021-11-24 17:32:06

Re: [SOLVED] PRE_SIGM use for Preload

#9 2021-11-24 22:19:10

Re: [SOLVED] PRE_SIGM use for Preload

#10 2021-11-28 17:01:54

Re: [SOLVED] PRE_SIGM use for Preload

#11 2021-11-29 14:19:29

Re: [SOLVED] PRE_SIGM use for Preload

#12 2021-11-29 20:48:29

Re: [SOLVED] PRE_SIGM use for Preload

#13 2021-11-30 06:35:40

Re: [SOLVED] PRE_SIGM use for Preload

#14 2021-11-30 09:06:46

Re: [SOLVED] PRE_SIGM use for Preload

#15 2021-11-30 20:18:18

Re: [SOLVED] PRE_SIGM use for Preload

#16 2021-12-01 06:34:32

Re: [SOLVED] PRE_SIGM use for Preload

#17 2021-12-01 09:07:27

Re: [SOLVED] PRE_SIGM use for Preload

#18 2021-12-01 20:14:11

Re: [SOLVED] PRE_SIGM use for Preload

#19 2021-12-02 08:18:44

Re: [SOLVED] PRE_SIGM use for Preload

#20 2021-12-02 20:19:10

Re: [SOLVED] PRE_SIGM use for Preload

#21 2021-12-03 08:22:47

Re: [SOLVED] PRE_SIGM use for Preload

#22 2021-12-03 11:51:22

Re: [SOLVED] PRE_SIGM use for Preload

#23 2021-12-03 19:01:12

Re: [SOLVED] PRE_SIGM use for Preload

#24 2021-12-04 09:27:20

Re: [SOLVED] PRE_SIGM use for Preload

#25 2021-12-04 11:09:45

Re: [SOLVED] PRE_SIGM use for Preload

Board footer