Unexpected influence of boundary conditions on buckling modes

filjan · 2021-10-24 13:41:37

Hi the Code_Aster Community!

I obtained unexpected results from simple linear buckling analysis - I've noticed 2 points for discussion:

1. Depending on imposing rotation DOF on boundary edges I got a lot of artificial, unexpected buckling first multipliers (and modes). At last 8th buckling modes with restricted rotations corresponds to the first with free rotations (see attached pictures and snippets from .mess file below)! How to explain this first 7 unreal buckling modes?

2. Whole model is symmetrical to the XY plane, and the first real buckling mode should also be symmetric, but on picture you can find the bigger positive DZ displacement than in negative direction. What can be source of this discrepancy?

In the attachment you can find pictures, .comm file and .rmed mesh and .mess file

A)

First minimum buckling multipliers with free rotations (with COQUE_NCOU=5)
------------------------------------------------------------------------
Calcul modal : Méthode d'itération simultanée
Méthode de Sorensen
numéro charge critique norme d'erreur
1 6.07851E+00 3.42566E-10
2 7.54621E+00 2.42043E-10
3 7.65019E+00 1.62089E-10
4 1.00785E+01 1.03446E-10
5 1.04750E+01 1.15403E-10
6 1.13024E+01 6.11925E-11
7 1.25364E+01 8.58257E-11
8 1.36543E+01 5.32246E-11
9 1.42141E+01 4.96179E-11
10 1.47606E+01 3.89146E-11
Norme d'erreur moyenne : 1.25432E-10
------------------------------------------------------------------------

B)
First minimum buckling multipliers with imposed zero rotations on boundary edges (with COQUE_NCOU=5)

------------------------------------------------------------------------
Calcul modal : Méthode d'itération simultanée
Méthode de Sorensen
numéro charge critique norme d'erreur
1 -6.09552E+00 2.48475E-13
2 -5.81936E+00 3.22682E-14
3 -4.53943E+00 2.11344E-14
4 -3.49609E+00 2.73799E-14
5 3.48710E+00 2.97868E-14
6 4.53262E+00 2.53528E-14
7 5.81460E+00 4.86050E-14
8 6.03323E+00 1.60771E-11
9 6.07375E+00 4.76493E-13
10 6.82361E+00 1.92509E-13
Norme d'erreur moyenne : 1.71791E-12
------------------------------------------------------------------------

Best regards
filjan

Last edited by filjan (2021-10-24 13:43:23)

Attachments:
unexpected_buckling.tar.xz, Size: 129.22 KiB, Downloads: 157

jeanpierreaubry · 2021-10-24 16:10:35

hello

in the vicinity of group "force" the boundary conditions do not represent a symmetry in any one of the two conditions

jean pierre aubry

filjan · 2021-10-24 17:22:49

jeanpierreaubry wrote:

hello
in the vicinity of group "force" the boundary conditions do not represent a symmetry in any one of the two conditions
jean pierre aubry

Of course - symmetry boundaries wasn't purpose here - I mean discrepancy in results in symmetrical areas mesh (see picture).
Boundary conditions shouldn't introduce any eccentricity in buckling modes, but mode occured uneven.

Attachments:
Screenshot from 2021-10-24 18-17-58.png, Size: 45.26 KiB, Downloads: 149

jeanpierreaubry · 2021-10-25 07:55:16

please post her a med 4.0 file so i can run the problem on 15.2 version of code_aster

filjan · 2021-10-25 15:55:49

jeanpierreaubry wrote:

please post her a med 4.0 file so i can run the problem on 15.2 version of code_aster

Here you are. Attached mesh was converted into MED 4.0.0 from 4.1.0 in salome_meca, because I did in gmsh and used C_A 15.4 then.

Best regards
Filjan

Attachments:
frame_corner_40MED.med, Size: 316.5 KiB, Downloads: 157

filjan · 2021-10-25 15:59:04

and unrolled .geo from gmsh in case

Attachments:
B_2_frame_corner.geo_unrolled, Size: 7.43 KiB, Downloads: 159

jeanpierreaubry · 2021-10-26 13:56:59

hello

with this

qu9_mesh = DEFI_GROUP(
	reuse = qu9_mesh,
	MAILLAGE = qu9_mesh,
	CREA_GROUP_NO = (
		_F(TOUT_GROUP_MA = 'OUI',),
		_F(
			NOM='symtop',
			NOEUD=(
				'N13','N427','N428','N429','N14','N430','N431','N432','N15',
				'N29','N804','N803','N802','N20','N869','N868','N867','N30',
			),
		),
	),
);

restr[0]  = AFFE_CHAR_MECA(
	MODELE = model,
	DDL_IMPO = (
		_F(
			GROUP_NO = 'restraint',
			DX = 0.0, DRX = 0.0,
			DY = 0.0, DRY = 0.0,
			DZ = 0.0, DRZ = 0.0,
		),
		_F(
			GROUP_NO = 'force',
			DX = 0.0, #DRX = 0.0,
			DRY = 0.0,DRZ = 0.0,
			#DZ = 0.0, #DRZ = 0.0,
		), 
		_F(
			GROUP_NO = 'symtop',
			DX = 0.0, #DRX = 0.0,
			DRY = 0.0,DRZ = 0.0,
			#DZ = 0.0, #DRZ = 0.0,
		), 
	),
);

these boundary on 'force' and 'symptop' depict a true symmetry about yOz plane
i get good result for the quadratic mesh on results_0

by the way i think a linear mesh with DKT is good enough for your problem

jean pierre aubry

#1 2021-10-24 13:41:37

Unexpected influence of boundary conditions on buckling modes

#2 2021-10-24 16:10:35

Re: Unexpected influence of boundary conditions on buckling modes

#3 2021-10-24 17:22:49

Re: Unexpected influence of boundary conditions on buckling modes

#4 2021-10-25 07:55:16

Re: Unexpected influence of boundary conditions on buckling modes

#5 2021-10-25 15:55:49

Re: Unexpected influence of boundary conditions on buckling modes

#6 2021-10-25 15:59:04

Re: Unexpected influence of boundary conditions on buckling modes

#7 2021-10-26 13:56:59

Re: Unexpected influence of boundary conditions on buckling modes

Board footer