Abaqus Convergence Discussion

8/17/2019 Abaqus Convergence Discussion

1/21

Hi all,

I'm performing a mesh convergence test on a simple 3D shell model with 1st and 2nd ordershell elements.I use CA 11.5 on Ubuntu 12.04 with Intel cpu's. I just measure the displacement of a

particular node (magnitude).I have the base geometry in .med format and I keep bisecting the element with Homard,within Salome-Meca 2013-2. As you can see in attached images, 1st and 2nd order elementsdoes not converge to an identical result (in the graphs, tmxy, x is the number of bi-sectioningand y means the order of elements, so tm 42 means a very fine mesh with 2nd orderelements).We tested the same model in Abaqus, and the results converge for 1st and 2nd order ofelements.The simulation is performed to study steady state displacement under harmonic excitationwith DYNA_LINE_HARM. The .comm files for 2 types of elements with base geometry areattached.

Same convergence problem exist also for static and DYNA_LINE_HARM for shell and 3Delements too.

I'll really appreciate any input.

Thanks, bme

Attachments:mesh_convergence.zip , Size: 53.35 KiB, Downloads: 89

Offline

#2 2014-02-05 14:47:31

Johannes_ACKVA Member From: Ingenieurbüro für Mechanik, DERegistered: 2009-11-04Posts: 485

Website

Re: Mesh convergence for 1st and 2 order elements

hello,

before discussing the (more complicated) harmonic result I raccomand to compare thesemodels also in a static and then in a modal analysis. From the static comparision You canknow if the reason is in the stiffness or in the mass.

http://web-code-aster.org/forum2/attachment.php?item=6628http://web-code-aster.org/forum2/attachment.php?item=6628http://web-code-aster.org/forum2/viewtopic.php?pid=44390#p44390http://web-code-aster.org/forum2/viewtopic.php?pid=44390#p44390http://web-code-aster.org/forum2/viewtopic.php?pid=44390#p44390http://web-code-aster.org/forum2/profile.php?id=781http://web-code-aster.org/forum2/profile.php?id=781http://www.code-aster.de/http://www.code-aster.de/http://www.code-aster.de/http://web-code-aster.org/forum2/profile.php?id=781http://web-code-aster.org/forum2/viewtopic.php?pid=44390#p44390http://web-code-aster.org/forum2/attachment.php?item=6628


2/21

If mass is the reason: Is Your shell thick or thin? If thick, the rotational mass(AFFA_CARA_ELEM(..INER_ROTA..) is important which is not considered by allmodelisations.

If stiffn is the reason: MODELISATION='DKT' has no shear stiffness. If You shell is thick,

try if You find a difference using MODELISATION='DST'.

Is You shell curved? In this case DKT and DST, QUAD4-elements must have all nodes in a plan to have correct results.

Did You arrange that GROUP_NO='bc' containing the clamped nodes must be different for both modelistions? In MODELISATION='COQUE_3D' it must also contain the mid-edge-nodes.

Regards,

Johannes_ACKVA

Ingenieurbüro für MechanikD 91717 Wassertrüdingen / Germany

www.code-aster.de Training & Support for NASTRAN andCODE-ASTER

Offline

#3 2014-02-06 21:33:26

bme Member Registered: 2012-05-10Posts: 46


Hi Johannes,Thank you so much for your comments.As I said before, the problem with mesh convergence of 1st and 2nd order elements exist forstatic and also 3D elements. The problem is also worse for 3D elements.

Considering the static model, the problem may be referred to the stiffness of the formulation.The Geometry (attached in .med format) is a curved shell, with a diameter almost 1cm, andthe thickness is 0.2 mm, which makes sense to consider it as a thick shell. I also extruded thismodel to generate 3D model. (using code-asterme1=CREA_MAILLAGE(MAILLAGE=MESH,

COQU_VOLU=_F(NOM='PRESSv',GROUP_MA='press',

http://web-code-aster.org/forum2/viewtopic.php?pid=44417#p44417http://web-code-aster.org/forum2/viewtopic.php?pid=44417#p44417http://web-code-aster.org/forum2/viewtopic.php?pid=44417#p44417http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/viewtopic.php?pid=44417#p44417


3/21

EPAIS=-0.0002,PLAN='INF',),);

The 3D model with 2nd order elements converges to the results of shell model with 2nd orderelements very well, but the 1st order 3D model, after 4 times refinement (8 layers of

elements), it does not converge at all.

Regarding your point about boundary nodes, yes I select all nodes after each refinement andalso converting to 2nd order elements.

Since the problem exist in static models, I have not tried rotational mass, for stiffness problem, since 3D model has the same problem, I have not tried thick shell models.I've attached results and also command and initial meshes of 3d and shell models.

Attachments:

Mesh_convergence2.zip, Size: 64.4 KiB, Downloads: 81

Offline

#4 2014-02-12 00:00:24

Thomas DE SOZA Guru From: EDFRegistered: 2007-11-23

Posts: 2,748


Hi,

There are several things to take into account :

- for the 2D models, you're using a plate model (DKT for 1st order) vs. a shell model(COQUE_3D for 2nd order). Moreover one is made for thin shells (DKT) whereas the otheraccounts for the shear stress in thick shells (COQUE_3D). Maybe you should use DSTinstead of DKT. What about the models you used in Abaqus ?What do the final refinements look like for both 1st and 2nd order models ? Were they thesame in Abaqus when convergence was reached ?

- for the 3D model, is the refinement uniform or mainly concentrated in the shell part ? Thisis not surprising to be unable to have a decent result with only one layer of standard linearelements.

TdS

Offline

http://web-code-aster.org/forum2/attachment.php?item=6635http://web-code-aster.org/forum2/attachment.php?item=6635http://web-code-aster.org/forum2/viewtopic.php?pid=44494#p44494http://web-code-aster.org/forum2/viewtopic.php?pid=44494#p44494http://web-code-aster.org/forum2/viewtopic.php?pid=44494#p44494http://web-code-aster.org/forum2/profile.php?id=24http://web-code-aster.org/forum2/profile.php?id=24http://web-code-aster.org/forum2/profile.php?id=24http://web-code-aster.org/forum2/viewtopic.php?pid=44494#p44494http://web-code-aster.org/forum2/attachment.php?item=6635


4/21

#5 2014-02-13 21:08:57



Hello Thomas,Thanks for your comments.1. I tried with TDS shell elements, and it converged to the 2nd order shell elements very nice!2. I used S3 and STRI65 shell elements in Abaqus for 1st and 2nd order elements,respectively and the meshes are identical for code-aster and Abaqus.3. In each step of mesh refinement, whole model was bisected, i.e., 3dtm 41 consists of 8

layer of 21888 elements in each layer. So the convergence problem for 3D element has not been solved yet.Thanks,

bme

Offline

#6 2014-02-14 13:47:59

Thomas DE SOZA

Guru From: EDFRegistered: 2007-11-23Posts: 2,748


bme wrote:

1. I tried with TDS shell elements, and it converged to the 2nd order shell elements very nice!2. I used S3 and STRI65 shell elements in Abaqus for 1st and 2nd order elements,

respectively and the meshes are identical for code-aster and Abaqus.

Great. S3 is indeed a 1st order thick-shell plate model which should be similar to DST so thismakes sense. The previous element you used, DKT, likely corresponded to Abaqus STRI3.

bme wrote:

3. In each step of mesh refinement, whole model was bisected, i.e., 3dtm 41 consists of 8layer of 21888 elements in each layer. So the convergence problem for 3D element has not

been solved yet.

I just noticed that your material is almost incompressible. This is unlikely you will be able tosolve this case with 1st order PENTA6 given the '3D' model. I'm not familiar with Abaqus

http://web-code-aster.org/forum2/viewtopic.php?pid=44518#p44518http://web-code-aster.org/forum2/viewtopic.php?pid=44518#p44518http://web-code-aster.org/forum2/viewtopic.php?pid=44518#p44518http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/viewtopic.php?pid=44534#p44534http://web-code-aster.org/forum2/viewtopic.php?pid=44534#p44534http://web-code-aster.org/forum2/viewtopic.php?pid=44534#p44534http://web-code-aster.org/forum2/profile.php?id=24http://web-code-aster.org/forum2/profile.php?id=24http://web-code-aster.org/forum2/profile.php?id=24http://web-code-aster.org/forum2/viewtopic.php?pid=44534#p44534http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/viewtopic.php?pid=44518#p44518


5/21

but judging from the documentation I'm pretty sure C3D6 elements would result in the same behaviour.The '3D_OSGS' model in Code_Aster 11.5 handles incompressibility with PENTA6 elements

but at the moment it won't work with DYNA_LINE_HARM (only STAT_NON_LINE).Another possibility would be to switch to TETRA4 elements and use the '3D_INCO_UP'

model.

TdS

Offline

#7 2014-02-15 16:45:32

Nima Member

From: CanadaRegistered: 2009-05-04Posts: 119


Hi,

Very interesting discussion!

So, can COQUE_3D or DST elements be used for simulating nearly-incompressible materials

safely?

Offline

#8 2014-03-18 18:30:48



Hi Thomas,1. I changed the Poisson's ratio to 0.3 (to be away from incompressible conditions), results of1st and 2nd order 3D elements converged.

2. But I could not perform nearly incompressible conditions (i.e., 3D_INCO and3D_INCO_UP) in my simulations. Apparently only STAT_NON_LINE,DYNA_NON_LINE can implement those elements (Ref: R3.06.08, Part 4.3), while I'm usingDYNA_LINE_HARM :-(

http://web-code-aster.org/forum2/viewtopic.php?pid=44549#p44549http://web-code-aster.org/forum2/viewtopic.php?pid=44549#p44549http://web-code-aster.org/forum2/viewtopic.php?pid=44549#p44549http://web-code-aster.org/forum2/profile.php?id=614http://web-code-aster.org/forum2/profile.php?id=614http://web-code-aster.org/forum2/viewtopic.php?pid=44860#p44860http://web-code-aster.org/forum2/viewtopic.php?pid=44860#p44860http://web-code-aster.org/forum2/viewtopic.php?pid=44860#p44860http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/profile.php?id=19275http://web-code-aster.org/forum2/viewtopic.php?pid=44860#p44860http://web-code-aster.org/forum2/profile.php?id=614http://web-code-aster.org/forum2/viewtopic.php?pid=44549#p44549


6/21

3. Nima, at least in my simulations (DYNA_LINE_HARM and MECA_STATIQUE)COQUE_3D and DST converge, for v=0.499.

BME

Offline

#9 2014-03-18 18:43:42

Nima Member From: CanadaRegistered: 2009-05-04Posts: 119


Thanks for updating us.

I use 3D_INCO in DYNA_LINE_TRAN.

http://web-code-aster.org/forum2/viewtopic.php?pid=44861#p44861http://web-code-aster.org/forum2/viewtopic.php?pid=44861#p44861http://web-code-aster.org/forum2/viewtopic.php?pid=44861#p44861http://web-code-aster.org/forum2/profile.php?id=614http://web-code-aster.org/forum2/profile.php?id=614http://web-code-aster.org/forum2/profile.php?id=614http://web-code-aster.org/forum2/viewtopic.php?pid=44861#p44861


7/21

What is "hourglass effect" in finite element

analysis?

How the reduced integration works which give rise to hourglass effect? What are the wayswe counter the hourglass effect?3 Answers

Achilleas Vortselas, Doctorate in Tribology: Wear Modeling25.7k Views • Upvoted by Mukunda Madhava Nath, CAE engineer at General MotorsAchilleas is a Most Viewed Writer in Finite Element Analysis.

What is hourglassing?

It is essentially a spurious deformation mode of a Finite Element Mesh, resulting from theexcitation of zero-energy degrees of freedom. It typically manifests as a patchwork of zig-zagor hourglass like element shapes (Fig.1), where individual elements are severely deformed,while the overall mesh section is undeformed. This happens on hexahedral 3D solid reducedintegration elements and on the respective tetrahedral 3D shell elements and 2D solidelements.

Fig.1 Typical manifestation of the hourglassing effect (LSDYNA), image source:

esocaet.com Hourglassing - ESOCAETWIKIPLUS

Why does hourglassing happen?

I believe the following segment from Getting started with Abaqus manual explains it well:

Consider a single reduced-integration element modeling a small piece of material subjected to pure bending (Fig.2)

https://www.quora.com/profile/Achilleas-Vortselashttps://www.quora.com/profile/Achilleas-Vortselashttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Achilleas-Vortselashttps://www.quora.com/profile/Achilleas-Vortselashttps://www.quora.com/topic/Finite-Element-Analysishttps://www.quora.com/topic/Finite-Element-Analysishttps://www.quora.com/topic/Finite-Element-Analysishttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttps://www.quora.com/profile/Achilleas-Vortselashttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttp://www.esocaet.com/wikiplus/index.php/Hourglassinghttps://www.quora.com/topic/Finite-Element-Analysishttps://www.quora.com/profile/Achilleas-Vortselashttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Achilleas-Vortselas


8/21

Fig. 2 Deformation of a linear element with reduced integration subjected to bending moment M.

Neither of the dotted visualization lines has changed in length, and the angle between themis also unchanged, which means that all components of stress at the element's singleintegration point are zero. This bending mode of deformation is thus a zero-energy mode

because no strain energy is generated by this element distortion. The element is unable toresist this type of deformation since it has no stiffness in this mode. In coarse meshes thiszero-energy mode can propagate through the mesh, producing meaningless results.

The concept of reduced integration elements is explained here [1].

How is hourglassing diagnosed?By optical inspection of the mesh deformation, increasing its scale factor of necessary. Butmost reliably by directly comparing the energy contained in the zero energy modes(hourglass energy) with the internal energy of the system. Hourglass energy should notexceed 5% of internal energy. This is done by plotting the energies from history data.

How is hourglassing addressed?

By inserting an artificial stiffness to the hourglass deformation modes (the default

way utilised in static/quasistatic problems). By inserting an artificial viscosity (preferred for dynamic and high velocity impact problems).

By using fully integrated elements. By refining the mesh.

The basic hourglass control methodologies have been pioneered by Belytschko and Flanagansince 1981 [2]. Details about the application of hourglass control can be found here [3-7].

[1] FAQ: What is reduced integration in the context of finite element analysis?

[2] Ted Belytschko, Jame Shau-Jen Ong, Wing Kam Liu, James M. Kennedy, Hourglasscontrol in linear and nonlinear problems, Computer Methods in Applied Mechanics and Engineering , Volume 43, Issue 3, May 1984, Pages 251-276, ISSN 0045-7825, Hourglasscontrol in linear and nonlinear problems. (Hourglass control in linear and nonlinear

problems) [3] Presentation by J. Day - LSTC (for LSDYNA) PDF [4] LSDYNA Support page Hourglass - LS-DYNA Support [5] Leonard E. Schwer, Samuel W. Key, Thomas A. Pučik, Lee P. Bindeman, An Assessment

of the LS-DYNA Hourglass Formulations via the 3D Patch Test, 5th European LS-DYNAUsers Conference PDF [6] Amit H. Varma, Finite Elements in Elasticity, Purdue University CE-595 lecture notes

(for ABAQUS) PPT [7] ABAQUS/Explicit Advanced topics seminar, lecture slides, p.31 PDF

http://www.twi.co.uk/technical-knowledge/faqs/structural-integrity-faqs/faq-what-is-reduced-integration-in-the-context-of-finite-element-analysis/http://www.twi.co.uk/technical-knowledge/faqs/structural-integrity-faqs/faq-what-is-reduced-integration-in-the-context-of-finite-element-analysis/http://www.twi.co.uk/technical-knowledge/faqs/structural-integrity-faqs/faq-what-is-reduced-integration-in-the-context-of-finite-element-analysis/http://dx.doi.org/10.1016/0045-7825%2884%2990067-7http://dx.doi.org/10.1016/0045-7825%2884%2990067-7http://dx.doi.org/10.1016/0045-7825%2884%2990067-7http://dx.doi.org/10.1016/0045-7825%2884%2990067-7http://www.sciencedirect.com/science/article/pii/0045782584900677http://www.sciencedirect.com/science/article/pii/0045782584900677http://www.sciencedirect.com/science/article/pii/0045782584900677http://www.sciencedirect.com/science/article/pii/0045782584900677http://ftp.lstc.com/anonymous/outgoing/jday/hourglass.pdfhttp://ftp.lstc.com/anonymous/outgoing/jday/hourglass.pdfhttp://ftp.lstc.com/anonymous/outgoing/jday/hourglass.pdfhttp://www.dynasupport.com/howtos/element/hourglasshttp://www.dynasupport.com/howtos/element/hourglasshttp://www.dynasupport.com/howtos/element/hourglasshttp://www.lsdyna.eu/uploads/media/Schwer-18Feb05.pdfhttp://www.lsdyna.eu/uploads/media/Schwer-18Feb05.pdfhttp://www.lsdyna.eu/uploads/media/Schwer-18Feb05.pdfhttps://engineering.purdue.edu/~ahvarma/CE595/CE595%20Section%205.ppthttps://engineering.purdue.edu/~ahvarma/CE595/CE595%20Section%205.ppthttps://engineering.purdue.edu/~ahvarma/CE595/CE595%20Section%205.ppthttp://imechanica.org/files/l2-elements.pdfhttp://imechanica.org/files/l2-elements.pdfhttp://imechanica.org/files/l2-elements.pdfhttp://imechanica.org/files/l2-elements.pdfhttps://engineering.purdue.edu/~ahvarma/CE595/CE595%20Section%205.ppthttp://www.lsdyna.eu/uploads/media/Schwer-18Feb05.pdfhttp://www.dynasupport.com/howtos/element/hourglasshttp://ftp.lstc.com/anonymous/outgoing/jday/hourglass.pdfhttp://www.sciencedirect.com/science/article/pii/0045782584900677http://www.sciencedirect.com/science/article/pii/0045782584900677http://dx.doi.org/10.1016/0045-7825%2884%2990067-7http://dx.doi.org/10.1016/0045-7825%2884%2990067-7http://www.twi.co.uk/technical-knowledge/faqs/structural-integrity-faqs/faq-what-is-reduced-integration-in-the-context-of-finite-element-analysis/


9/21

Written Nov 14, 2013 • View Upvotes • Answer requested by Mukunda Madhava Nath

Related Questions

More Answers Below

What are the most important problems in finite element analysis? Are there any good Android applications that do finite element analysis? Why interpolation methods such as finite element analysis have to satisfy the partition

of unity? Reviews of: Finite Element Analysis How much does a mixed formulation increase the complexity of a finite element

problem?

Adrian Bennett, I believe in God's design for marriage: one man and one woman for life.9.2k Views • Upvoted by Mukunda Madhava Nath, CAE engineer at General MotorsAdrian is a Most Viewed Writer in Finite Element Analysis. Thanks for the A2A

Achilleas Vortselas's answer is excellent. I will not go over again the things he has written,nor could I do a better job of it. But I would like to add something I think is very valuable andanswers the last part of your question which is "What are the ways we counter the hourglasseffect?" As an acknowlegment to the reader, p-FEM is my area of research.

So, first reduced integration is method to prevent something called "locking", in particular"shear locking", in some finite elements which prevents the solution from correctlyconverging. This is basically one mistake introduced to cover another. The best way to

prevent the "hourglass" effect is to avoid the whole problem of shear locking and reducedintegration to begin with. This can be done through using higher order finite elementformulations. Lower order elements have a basis which is not capable of completelyexpressing the structural behavior within the elements. Higher order elements do.

The p-FEM method in contrast to the h-version of the finite element method uses increasing polynomial order of the element basis functions to refine the solution and obtain greateraccuracy in analysis. In the traditional h-version, usually the basis functions are only of thefirst or second order, and increasing the accuracy of the solution is done through meshrefinement. This involves creating new meshes at each step to get a better solution. In the p-version of the finite element method, the mesh is not altered, but the order of the basisfunctions is increased. This is very convenient because the mesh does not have to be changed.The convergence of the p-FEM method has been shown to be superior to the h-version also.Few elements are needed for p-FEM. There are very significant advantages of the p-FEMover the traditional finite element method.

A particular significant advantage of the p-version of FEM over the h-version is theacceptable aspect ratios of the elements. In the h-version there is a severe limitation on the

aspect ratio of the elements. Aspect ratios of several hundred to 1 can be used in the p-version. Even solid elements can be used to represent thin plates and shell elements without

https://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Achilleas-Vortselashttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Achilleas-Vortselashttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=jNiek8TCOO2https://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=jNiek8TCOO2https://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=jNiek8TCOO2https://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis#MoreAnswershttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis#MoreAnswershttps://www.quora.com/What-are-the-most-important-problems-in-finite-element-analysishttps://www.quora.com/What-are-the-most-important-problems-in-finite-element-analysishttps://www.quora.com/Are-there-any-good-Android-applications-that-do-finite-element-analysishttps://www.quora.com/Are-there-any-good-Android-applications-that-do-finite-element-analysishttps://www.quora.com/Why-interpolation-methods-such-as-finite-element-analysis-have-to-satisfy-the-partition-of-unityhttps://www.quora.com/Why-interpolation-methods-such-as-finite-element-analysis-have-to-satisfy-the-partition-of-unityhttps://www.quora.com/Why-interpolation-methods-such-as-finite-element-analysis-have-to-satisfy-the-partition-of-unityhttps://www.quora.com/Why-interpolation-methods-such-as-finite-element-analysis-have-to-satisfy-the-partition-of-unityhttps://www.quora.com/Why-interpolation-methods-such-as-finite-element-analysis-have-to-satisfy-the-partition-of-unityhttps://www.quora.com/unanswered/Reviews-of-Finite-Element-Analysishttps://www.quora.com/unanswered/Reviews-of-Finite-Element-Analysishttps://www.quora.com/unanswered/How-much-does-a-mixed-formulation-increase-the-complexity-of-a-finite-element-problemhttps://www.quora.com/unanswered/How-much-does-a-mixed-formulation-increase-the-complexity-of-a-finite-element-problemhttps://www.quora.com/unanswered/How-much-does-a-mixed-formulation-increase-the-complexity-of-a-finite-element-problemhttps://www.quora.com/unanswered/How-much-does-a-mixed-formulation-increase-the-complexity-of-a-finite-element-problemhttps://www.quora.com/unanswered/How-much-does-a-mixed-formulation-increase-the-complexity-of-a-finite-element-problemhttps://www.quora.com/profile/Adrian-Bennett-3https://www.quora.com/profile/Adrian-Bennett-3https://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Adrian-Bennett-3https://www.quora.com/profile/Adrian-Bennett-3https://www.quora.com/topic/Finite-Element-Analysishttps://www.quora.com/topic/Finite-Element-Analysishttps://www.quora.com/topic/Finite-Element-Analysishttps://www.quora.com/profile/Adrian-Bennett-3https://www.quora.com/topic/Finite-Element-Analysishttps://www.quora.com/profile/Adrian-Bennett-3https://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Adrian-Bennett-3https://www.quora.com/unanswered/How-much-does-a-mixed-formulation-increase-the-complexity-of-a-finite-element-problemhttps://www.quora.com/unanswered/How-much-does-a-mixed-formulation-increase-the-complexity-of-a-finite-element-problemhttps://www.quora.com/unanswered/Reviews-of-Finite-Element-Analysishttps://www.quora.com/Why-interpolation-methods-such-as-finite-element-analysis-have-to-satisfy-the-partition-of-unityhttps://www.quora.com/Why-interpolation-methods-such-as-finite-element-analysis-have-to-satisfy-the-partition-of-unityhttps://www.quora.com/Are-there-any-good-Android-applications-that-do-finite-element-analysishttps://www.quora.com/What-are-the-most-important-problems-in-finite-element-analysishttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis#MoreAnswershttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=jNiek8TCOO2https://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Achilleas-Vortselas


10/21

problems of shear locking.

I have been working with p-FEM and I simply do not have to consider hour-glass effect,reduced integration, or shear locking. With higher order elements, all this becomesunnecessary.

Updated Sep 6, 2014 • View Upvotes • Answer requested by Mukunda Madhava Nath

Brmv Krishna 5.1k ViewsHourglassing is a state of strain, which is free of energy (ZEM: Zero Energy Mode) and canemerge in case of one-point-integrated solid- (hexahedrons) and shell elements

Hourglass modes are mostly caused by:- concentrated loads

- contact (contact force at several nodes )

in LS-DYNA there are 2 possibilities to prevent Hourglassing:

1. using a fully integrated element typedisadvantages: - more computation time- more sensible with respect to large elementdeformations

2. using the automatic stabilisation against this deformation with- *HOURGLASS (input for each part) or

- *CONTROL_HOURGLASS (global control)Written Jun 5, 2014 • View Upvotes

Related Questions

What are some of the reputed and high impact conferences and journals related tofinite element analysis/methods?

What is the best way to learn finite element analysis in LS Dyna?

How does Finite Element Analysis compare and contrast with continuum mechanics? What are the main industries and machinery where FEA (Finite Element Analysis) is

used? Which research labs are working in the intersection of computer-aided design,

computer graphics, and the finite element method?

Where can I find examples of finite element analysis? How does one go about a Finite Element Analysis project?

Finite Element Analysis: Does beam orientation should be same as global co-ordinateof mesh model?

How does finite element analysis help manufacturers to improve product performance?

What are some good books to learn finite element analysis? What is a complete explanation of essential boundary condition and natural boundary

condition in Finite Element Analysis?

What exactly is Finite element analysis?

https://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Adrian-Bennett-3https://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Adrian-Bennett-3https://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=d4kZjk1EVDHhttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=d4kZjk1EVDHhttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=d4kZjk1EVDHhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/profile/Brmv-Krishnahttps://www.quora.com/profile/Brmv-Krishnahttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Brmv-Krishnahttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Brmv-Krishnahttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=dtt3fcd5rPhhttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=dtt3fcd5rPhhttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=dtt3fcd5rPhhttps://www.quora.com/What-are-some-of-the-reputed-and-high-impact-conferences-and-journals-related-to-finite-element-analysis-methodshttps://www.quora.com/What-are-some-of-the-reputed-and-high-impact-conferences-and-journals-related-to-finite-element-analysis-methodshttps://www.quora.com/What-are-some-of-the-reputed-and-high-impact-conferences-and-journals-related-to-finite-element-analysis-methodshttps://www.quora.com/What-are-some-of-the-reputed-and-high-impact-conferences-and-journals-related-to-finite-element-analysis-methodshttps://www.quora.com/What-are-some-of-the-reputed-and-high-impact-conferences-and-journals-related-to-finite-element-analysis-methodshttps://www.quora.com/What-is-the-best-way-to-learn-finite-element-analysis-in-LS-Dynahttps://www.quora.com/What-is-the-best-way-to-learn-finite-element-analysis-in-LS-Dynahttps://www.quora.com/How-does-Finite-Element-Analysis-compare-and-contrast-with-continuum-mechanicshttps://www.quora.com/How-does-Finite-Element-Analysis-compare-and-contrast-with-continuum-mechanicshttps://www.quora.com/What-are-the-main-industries-and-machinery-where-FEA-Finite-Element-Analysis-is-usedhttps://www.quora.com/What-are-the-main-industries-and-machinery-where-FEA-Finite-Element-Analysis-is-usedhttps://www.quora.com/What-are-the-main-industries-and-machinery-where-FEA-Finite-Element-Analysis-is-usedhttps://www.quora.com/What-are-the-main-industries-and-machinery-where-FEA-Finite-Element-Analysis-is-usedhttps://www.quora.com/What-are-the-main-industries-and-machinery-where-FEA-Finite-Element-Analysis-is-usedhttps://www.quora.com/Which-research-labs-are-working-in-the-intersection-of-computer-aided-design-computer-graphics-and-the-finite-element-methodhttps://www.quora.com/Which-research-labs-are-working-in-the-intersection-of-computer-aided-design-computer-graphics-and-the-finite-element-methodhttps://www.quora.com/Which-research-labs-are-working-in-the-intersection-of-computer-aided-design-computer-graphics-and-the-finite-element-methodhttps://www.quora.com/Which-research-labs-are-working-in-the-intersection-of-computer-aided-design-computer-graphics-and-the-finite-element-methodhttps://www.quora.com/Which-research-labs-are-working-in-the-intersection-of-computer-aided-design-computer-graphics-and-the-finite-element-methodhttps://www.quora.com/Where-can-I-find-examples-of-finite-element-analysishttps://www.quora.com/Where-can-I-find-examples-of-finite-element-analysishttps://www.quora.com/How-does-one-go-about-a-Finite-Element-Analysis-projecthttps://www.quora.com/How-does-one-go-about-a-Finite-Element-Analysis-projecthttps://www.quora.com/Finite-Element-Analysis/Does-beam-orientation-should-be-same-as-global-co-ordinate-of-mesh-modelhttps://www.quora.com/Finite-Element-Analysis/Does-beam-orientation-should-be-same-as-global-co-ordinate-of-mesh-modelhttps://www.quora.com/Finite-Element-Analysis/Does-beam-orientation-should-be-same-as-global-co-ordinate-of-mesh-modelhttps://www.quora.com/Finite-Element-Analysis/Does-beam-orientation-should-be-same-as-global-co-ordinate-of-mesh-modelhttps://www.quora.com/Finite-Element-Analysis/Does-beam-orientation-should-be-same-as-global-co-ordinate-of-mesh-modelhttps://www.quora.com/How-does-finite-element-analysis-help-manufacturers-to-improve-product-performancehttps://www.quora.com/How-does-finite-element-analysis-help-manufacturers-to-improve-product-performancehttps://www.quora.com/How-does-finite-element-analysis-help-manufacturers-to-improve-product-performancehttps://www.quora.com/How-does-finite-element-analysis-help-manufacturers-to-improve-product-performancehttps://www.quora.com/How-does-finite-element-analysis-help-manufacturers-to-improve-product-performancehttps://www.quora.com/What-are-some-good-books-to-learn-finite-element-analysishttps://www.quora.com/What-are-some-good-books-to-learn-finite-element-analysishttps://www.quora.com/What-is-a-complete-explanation-of-essential-boundary-condition-and-natural-boundary-condition-in-Finite-Element-Analysishttps://www.quora.com/What-is-a-complete-explanation-of-essential-boundary-condition-and-natural-boundary-condition-in-Finite-Element-Analysishttps://www.quora.com/What-is-a-complete-explanation-of-essential-boundary-condition-and-natural-boundary-condition-in-Finite-Element-Analysishttps://www.quora.com/What-is-a-complete-explanation-of-essential-boundary-condition-and-natural-boundary-condition-in-Finite-Element-Analysishttps://www.quora.com/What-is-a-complete-explanation-of-essential-boundary-condition-and-natural-boundary-condition-in-Finite-Element-Analysishttps://www.quora.com/What-exactly-is-Finite-element-analysishttps://www.quora.com/What-exactly-is-Finite-element-analysishttps://www.quora.com/profile/Brmv-Krishnahttps://www.quora.com/What-exactly-is-Finite-element-analysishttps://www.quora.com/What-is-a-complete-explanation-of-essential-boundary-condition-and-natural-boundary-condition-in-Finite-Element-Analysishttps://www.quora.com/What-is-a-complete-explanation-of-essential-boundary-condition-and-natural-boundary-condition-in-Finite-Element-Analysishttps://www.quora.com/What-are-some-good-books-to-learn-finite-element-analysishttps://www.quora.com/How-does-finite-element-analysis-help-manufacturers-to-improve-product-performancehttps://www.quora.com/How-does-finite-element-analysis-help-manufacturers-to-improve-product-performancehttps://www.quora.com/Finite-Element-Analysis/Does-beam-orientation-should-be-same-as-global-co-ordinate-of-mesh-modelhttps://www.quora.com/Finite-Element-Analysis/Does-beam-orientation-should-be-same-as-global-co-ordinate-of-mesh-modelhttps://www.quora.com/How-does-one-go-about-a-Finite-Element-Analysis-projecthttps://www.quora.com/Where-can-I-find-examples-of-finite-element-analysishttps://www.quora.com/Which-research-labs-are-working-in-the-intersection-of-computer-aided-design-computer-graphics-and-the-finite-element-methodhttps://www.quora.com/Which-research-labs-are-working-in-the-intersection-of-computer-aided-design-computer-graphics-and-the-finite-element-methodhttps://www.quora.com/What-are-the-main-industries-and-machinery-where-FEA-Finite-Element-Analysis-is-usedhttps://www.quora.com/What-are-the-main-industries-and-machinery-where-FEA-Finite-Element-Analysis-is-usedhttps://www.quora.com/How-does-Finite-Element-Analysis-compare-and-contrast-with-continuum-mechanicshttps://www.quora.com/What-is-the-best-way-to-learn-finite-element-analysis-in-LS-Dynahttps://www.quora.com/What-are-some-of-the-reputed-and-high-impact-conferences-and-journals-related-to-finite-element-analysis-methodshttps://www.quora.com/What-are-some-of-the-reputed-and-high-impact-conferences-and-journals-related-to-finite-element-analysis-methodshttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=dtt3fcd5rPhhttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Brmv-Krishnahttps://www.quora.com/profile/Brmv-Krishnahttps://www.quora.com/profile/Mukunda-Madhava-Nathhttps://www.quora.com/api/mobile_expanded_voter_list?type=answer&key=d4kZjk1EVDHhttps://www.quora.com/What-is-hourglass-effect-in-finite-element-analysis/answer/Adrian-Bennett-3


11/21

Finite Element Analysis: What are some really interesting and influential topicsrelated to steel?

Can transfinite numbers be used with the finite element method? What is the best reference book for finite element method?

https://www.quora.com/Finite-Element-Analysis/What-are-some-really-interesting-and-influential-topics-related-to-steelhttps://www.quora.com/Finite-Element-Analysis/What-are-some-really-interesting-and-influential-topics-related-to-steelhttps://www.quora.com/Finite-Element-Analysis/What-are-some-really-interesting-and-influential-topics-related-to-steelhttps://www.quora.com/Finite-Element-Analysis/What-are-some-really-interesting-and-influential-topics-related-to-steelhttps://www.quora.com/Finite-Element-Analysis/What-are-some-really-interesting-and-influential-topics-related-to-steelhttps://www.quora.com/unanswered/Can-transfinite-numbers-be-used-with-the-finite-element-methodhttps://www.quora.com/unanswered/Can-transfinite-numbers-be-used-with-the-finite-element-methodhttps://www.quora.com/What-is-the-best-reference-book-for-finite-element-methodhttps://www.quora.com/What-is-the-best-reference-book-for-finite-element-methodhttps://www.quora.com/What-is-the-best-reference-book-for-finite-element-methodhttps://www.quora.com/unanswered/Can-transfinite-numbers-be-used-with-the-finite-element-methodhttps://www.quora.com/Finite-Element-Analysis/What-are-some-really-interesting-and-influential-topics-related-to-steelhttps://www.quora.com/Finite-Element-Analysis/What-are-some-really-interesting-and-influential-topics-related-to-steel


12/21

Why do I get different results from

Abaqus/CAE as the mesh differs?

My question regards biomedical models and finite element analyses in Abaqus.

Which results should I trust?

Thanks in advance.

Topics

Finite Element Method Computer-Aided Engineering Biomedical Engineering Finite Element Analysis Abaqus

Oct 14, 2014Share

0 / 0

All Answers (8)

Darby Luscher · Los Alamos National Laboratory

Hi Cyrus,

I agree with the response provided by Younes and will expand upon his responseslightly.

There is a discretization error in the finite element solution because of theapproximation of the actual solution field with element shape functions. As youincrease the number of elements in the solution, they provide a more accuraterepresentation of the solution field and the discretization error is reduced.

You can perform a convergence analysis or mesh refinement study by tabulating theresult a specific location versus the characteristic element size for solutions fromdifferent meshes. You can often fit results from this numerical study to an expressionof the type: d(h) = d0 + Ah^p where d is a solution variable (at one specific location)

https://www.researchgate.net/topic/finite_element_methodhttps://www.researchgate.net/topic/computer-aided_engineeringhttps://www.researchgate.net/topic/biomedical_engineeringhttps://www.researchgate.net/topic/finite_element_analysishttps://www.researchgate.net/topic/abaqushttps://www.researchgate.net/profile/Darby_Luscherhttps://www.researchgate.net/profile/Darby_Luscherhttps://www.researchgate.net/profile/Darby_Luscherhttps://www.researchgate.net/profile/Darby_Luscherhttps://www.researchgate.net/topic/abaqushttps://www.researchgate.net/topic/finite_element_analysishttps://www.researchgate.net/topic/biomedical_engineeringhttps://www.researchgate.net/topic/computer-aided_engineeringhttps://www.researchgate.net/topic/finite_element_method


13/21

like displacement, h is element size. d0, A, and p are fitting parameters. If you do this,then d0 is an estimate of the exact numerical solution and d(h)-d0 is an estimate of theerror associated with that variable for the mesh with an element size of h.

Best, DJ Luscher

Oct 14, 2014

Cyrus Ahmadi Toussi · Hakim Sabzevari University

Thak you both ,Mr Nouri and Mr Luscher, it was so helpful.You know Ive modeled 9different dental prosthesis, i choose the seed 0.001 but in some models i get 2 nodes toclose and makes me change the seed ,and when i do that the results arent corresponds

to others.

Oct 14, 2014

Sanan H Khan · Indian Institute of Technology Kanpur

FEM results are always mesh dependent. For better results use finer mesh.

Oct 15, 2014

Marcias J Martinez · Clarkson University

I would recommend that you read the paper by: MacNeal & Harder - StandardProblems FE Accuracy. It shows how standard elements fail to produce the analyticalsolution.

Oct 16, 2014

Pradeep George · New York University Abu Dhabi

Please do a mesh independence study initially.

Oct 16, 2014

https://www.researchgate.net/profile/Cyrus_Ahmadi_Toussihttps://www.researchgate.net/profile/Cyrus_Ahmadi_Toussihttps://www.researchgate.net/profile/Sanan_Khan2https://www.researchgate.net/profile/Sanan_Khan2https://www.researchgate.net/profile/Marcias_Martinezhttps://www.researchgate.net/profile/Marcias_Martinezhttps://www.researchgate.net/profile/Pradeep_George2https://www.researchgate.net/profile/Pradeep_George2https://www.researchgate.net/profile/Hojjat_Badnavahttps://www.researchgate.net/profile/Pradeep_George2https://www.researchgate.net/profile/Marcias_Martinezhttps://www.researchgate.net/profile/Sanan_Khan2https://www.researchgate.net/profile/Cyrus_Ahmadi_Toussihttps://www.researchgate.net/profile/Pradeep_George2https://www.researchgate.net/profile/Marcias_Martinezhttps://www.researchgate.net/profile/Sanan_Khan2https://www.researchgate.net/profile/Cyrus_Ahmadi_Toussi


14/21

Hojjat Badnava · Isfahan University of Technology

A fundamental problem of incorporating material model with softening behavior(such as damage and softening plasticity) in standard continuum models is theinherent mesh sensitivity. This mesh sensitivity goes beyond the standard

discretisation sensitivity of numerical approximation methods for partial differentialequations and is not related to deficiencies in the discretisation methods. Instead, theunderlying reason for this mesh sensitivity is a local change in character of thegoverning partial differential equations. This local change of character of thegoverning set of partial differential equations leads to a loss of well-posedness of theinitial boundary value problem and results in an infinite number of possible solutions.After discretisation, a finite number of solutions results. For a finer discretisation, thenumber of solutions increases, which explains the observed mesh sensitivity. Sincethe observed mesh sensitivity is of a fundamental.

What kind of material model you have used?

https://www.researchgate.net/profile/Hojjat_Badnavahttps://www.researchgate.net/profile/Hojjat_Badnavahttps://www.researchgate.net/profile/Hojjat_Badnava


15/21

convergence and mesh refinement

Classic List Threaded

6 messages Options

zandfub1 Reply | Threaded | More

Jan 26, 2009; 11:00am

convergence and mesh refinement

126 posts

Hi,

i am looking for reasons, why the stresses do not converge with mesh

refinement. (Each calculation converges!)

The model consists of two blocks put on top of each other withdifferent isotropic materials. The lower block is larger than theupper, the upper one has square edges. There is surface to surfacecontact with small sliding and exponential pressure overclosure

between the blocks. The contact surfaces are plane and a high adjustvalue was chosen. The lower one is fixed while the upper one iscompressed on the top perpendicular to the surface. Fixation and loadentry are far away from the contact. It is a static analysis withabaqus 6.8.

Now i refined the hex mesh (C3D8) in each iso direction two fold (sothat the number of elements is 8fold). After that i refined the meshagain 8fold (so the mesh is refined 64fold compared to the initialone), but the stresses in the elements in the area of contact are eachabout doubling with mesh refinement. I looked at the stresses in theintegration points, in the nodes and in the element centroids.Everywhere the same problem.

Has anyone an idea?

Thanks,

Thomas

Remove Ads

Frederic Levesque Reply | Threaded | More Jan 26, 2009; 3:21pm

Rép. : convergence and mesh refinement

http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13772http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13772http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=site_paymenthttp://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=site_paymenthttp://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1066http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1066http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13773http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13773http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13773http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1066http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=site_paymenthttp://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13772http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://void%280%29/http://void%280%29/http://void%280%29/


16/21

20 posts

Hi!

Maybe you should try further refinement. Also, if you look at the theory ofcontact mechanics, the pressure is infinite at the edge of the contact area (if Iunderstand well the case you described). The finite element model will never

be able to return you an infinite pressure. At that point, the model might notconverge very well. Maybe you should look at another physical quantity toassess convergence. A teacher once advised me to use strain energy for acontact mechanics problem. Maybe you should give it a try.

I hope it can help

Fred

--- En date de : Lun, 26.1.09, zandfub1 a écrit :

De: zandfub1 Objet: [Abaqus] convergence and mesh refinementÀ: [hidden email] Date: lundi 26 Janvier 2009, 5 h 00

Hi,i am looking for reasons, why the stresses do not converge with meshrefinement. (Each calculation converges!)

The model consists of two blocks put on top of each other withdifferent isotropic materials. The lower block is larger than theupper, the upper one has square edges. There is surface to surfacecontact with small sliding and exponential pressure overclosure

between the blocks. The contact surfaces are plane and a high adjustvalue was chosen. The lower one is fixed while the upper one is

compressed on the top perpendicular to the surface. Fixation and loadentry are far away from the contact. It is a static analysis withabaqus 6.8.

Now i refined the hex mesh (C3D8) in each iso direction two fold (sothat the number of elements is 8fold). After that i refined the meshagain 8fold (so the mesh is refined 64fold compared to the initialone), but the stresses in the elements in the area of contact are eachabout doubling with mesh refinement. I looked at the stresses in theintegration points, in the nodes and in the element centroids.Everywhere the same problem.

Has anyone an idea?

http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=0http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=1http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=2http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=2http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=2http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1066http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=2http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=1http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13773&i=0


17/21

Thanks,

Thomas

Offrez un compte Flickr Pro à vos amis et à votre famille.http://www.flickr.com/gift/

[Non-text portions of this message have been removed]

zandfub1

Reply | Threaded | More Jan 27, 2009; 8:52am

Re: Rép. : convergence and mesh refinement

126 posts

> the pressure is infinite at the edge of the contact area

Thanks Fred, that was the hint!

Thomas

MikeBrown_CivlEng Reply | Threaded | More Jul 21, 2009; 7:15pm


http://www.flickr.com/gift/http://www.flickr.com/gift/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13776http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13776http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1639http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1639http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13774http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13774http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13774http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1639http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13776http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://www.flickr.com/gift/


18/21

3 posts

In reply to this post by Frederic Levesque

Fred, Thomas, or anybody else,

I'm having a similar problem with my model. I'm studying a liner systemused in water main rehabilitation. I'm modeling a situation where the liner

goes from spanning a section of the cast iron pipe that is intact to a sectionwhere the cast iron is deteriorated (essentially non-existent). I'm applying aninternal pressure to the liner which makes the liner want to wrap around thecast iron section that is intact...

-------- cast iron----------------------- liner^^^^^^^^^^^^^^^^^^^^^^^ applied intenal pressure

I'm not sure if that illustration will turn out properly. Anyway, how will I beable to accurately determine the stresses at "the bend", since in theory they

are infinite. Where do the stresses begin to actually converge to the correctvalue (how far away from the stress concentration)? I know i'll have toinvestigate that myself. I'm just looking for ideas in general.

Thanks in advance.

Mike Brown

--- In [hidden email], Frederic Levesque wrote:

>> Hi!>> Maybe you should try further refinement. Also, if you look at the theory ofcontact mechanics, the pressure is infinite at the edge of the contact area (if Iunderstand well the case you described). The finite element model will never

be able to return you an infinite pressure. At that point, the model might notconverge very well. Maybe you should look at another physical quantity toassess convergence. A teacher once advised me to use strain energy for acontact mechanics problem. Maybe you should give it a try.

>> I hope it can help>> Fred>> --- En date de : Lun, 26.1.09, zandfub1 a écrit :>> De: zandfub1 > Objet: [Abaqus] convergence and mesh refinement> À: [hidden email] > Date: lundi 26 Janvier 2009, 5 h 00

>>

http://abaqus-users.1086179.n5.nabble.com/convergence-and-mesh-refinement-tp13772p13773.htmlhttp://abaqus-users.1086179.n5.nabble.com/convergence-and-mesh-refinement-tp13772p13773.htmlhttp://abaqus-users.1086179.n5.nabble.com/convergence-and-mesh-refinement-tp13772p13773.htmlhttp://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=0http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=0http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=0http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=1http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=1http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=1http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=1639http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=1http://abaqus-users.1086179.n5.nabble.com/user/SendEmail.jtp?type=node&node=13774&i=0http://abaqus-users.1086179.n5.nabble.com/convergence-and-mesh-refinement-tp13772p13773.html


19/21

>>>>> Hi,

>> i am looking for reasons, why the stresses do not converge with mesh> refinement. (Each calculation converges!)>> The model consists of two blocks put on top of each other with> different isotropic materials. The lower block is larger than the> upper, the upper one has square edges. There is surface to surface> contact with small sliding and exponential pressure overclosure> between the blocks. The contact surfaces are plane and a high adjust> value was chosen. The lower one is fixed while the upper one is> compressed on the top perpendicular to the surface. Fixation and load> entry are far away from the contact. It is a static analysis with> abaqus 6.8.>> Now i refined the hex mesh (C3D8) in each iso direction two fold (so> that the number of elements is 8fold). After that i refined the mesh> again 8fold (so the mesh is refined 64fold compared to the initial> one), but the stresses in the elements in the area of contact are each> about doubling with mesh refinement. I looked at the stresses in the> integration points, in the nodes and in the element centroids.> Everywhere the same problem.

>> Has anyone an idea?>> Thanks,>> Thomas>>>>>

>>>>>>>>>>>

> Offrez un compte Flickr Pro à vos amis et à votre famille.> http://www.flickr.com/gift/

http://www.flickr.com/gift/http://www.flickr.com/gift/http://www.flickr.com/gift/http://www.flickr.com/gift/


20/21

>> [Non-text portions of this message have been removed]>... [show rest of quote]

zandfub1 Reply | Threaded | More Jul 22, 2009; 6:58am


126 posts

> I'm just looking for ideas in> general.>

> Thanks in advance.>> Mike Brown

Hi Mike,

if theory says that the stresses are infinite (is this really true?) than itseems that the reality is not described correctly by the FE model if thestresses are not infinite in reality (very wise, isn't it?)On the other hand FE is not a tool to describe reality but to describe thetheoretical model which says (?) that stresses are infinite.

Just my thinkings,

Thomasdukuru Reply | Threaded | More May 29, 2014; 6:00pm


1 post

This post has NOT been accepted by the mailing list yet.

Infinite pressure is possible if your point is infinitely small. force/area. Theonly difference is that infinitely small objects don't exist (to myknowledge :). So even though in real life there is a round to the edge of a

block, the abaqus model only has a node, and since nodes don't have volume,the point is infinitely small in Abaqus.

but i like the thought haha.

regards-Mike

http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13775http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13775http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=5799http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=5799http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=21234http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=21234http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=5799http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=21234http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=5799http://void%280%29/http://void%280%29/http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=reply&node=13775http://abaqus-users.1086179.n5.nabble.com/template/NamlServlet.jtp?macro=user_nodes&user=263http://void%280%29/


21/21

Abaqus Convergence Discussion

Documents