Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

geometric nonlinearity

Please login with a confirmed email address before reporting spam

When should I check the option of "geometric nonlinearity" ? I understand it is used for large deformations but what is meant by large deformation? What should the magnitude of a deformation be inorder to be classified as large deformation?

I am trying to model an artery and am applying some pressure to expand it. Do I need to select the geometric nonlinearity in my case?

Thanks for the help!

Sincerely,
Sirisha

11 Replies Last Post 20 sept. 2013, 09:40 UTC−4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 janv. 2011, 21:30 UTC−5
Do you get a solution without enabling it?

The solver should return an error fairly early if the only problem is geometry nonlinearity. In that case, just check the box and re-run.

If you are still not getting a solution with nonlinearity enabled, I would recommend experimenting with solver settings.

It is likely for large deformations that you are going to need an iterative solve with a pre-conditioner.
Do you get a solution without enabling it? The solver should return an error fairly early if the only problem is geometry nonlinearity. In that case, just check the box and re-run. If you are still not getting a solution with nonlinearity enabled, I would recommend experimenting with solver settings. It is likely for large deformations that you are going to need an iterative solve with a pre-conditioner.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 29 janv. 2011, 23:11 UTC−5
I am getting a solution with the nonlinearity "unchecked' but I still don't understand when I am supposed to "check" the geometric nonlinearity. My problem is nonlinear but I don't know the functionality of geometric nonlinearity.

From what I read in the manual, geometric nonlinearity is the same as enabling large deformations. I am still not sure in what case should I use large deformations/geometric nonlinearity.

Thanks for the help.
Sirisha

PS: I am having convergence issues with this option checked.
I am getting a solution with the nonlinearity "unchecked' but I still don't understand when I am supposed to "check" the geometric nonlinearity. My problem is nonlinear but I don't know the functionality of geometric nonlinearity. From what I read in the manual, geometric nonlinearity is the same as enabling large deformations. I am still not sure in what case should I use large deformations/geometric nonlinearity. Thanks for the help. Sirisha PS: I am having convergence issues with this option checked.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30 janv. 2011, 03:50 UTC−5
Hi

the large deformation changes the equations used for the stress-strain relationship (see the doc), this adds a cross derivative term (note: large deformations are incompatibile with Antisymmetry BC conditions, be warned).

The simplest way around is to make a run without, and one with, and compare the results.

But if you are in structural or fluide-structure interaction FSI, and in artery analysis, I would rather expect to have it turned on (which means slower solving nad more calculations)

--
Good luck
Ivar
Hi the large deformation changes the equations used for the stress-strain relationship (see the doc), this adds a cross derivative term (note: large deformations are incompatibile with Antisymmetry BC conditions, be warned). The simplest way around is to make a run without, and one with, and compare the results. But if you are in structural or fluide-structure interaction FSI, and in artery analysis, I would rather expect to have it turned on (which means slower solving nad more calculations) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30 janv. 2011, 16:50 UTC−5
Thanks Ivar!

I think I have an idea now .......it needs to be turned on for my case but the problem is that if I turn on the geometric nonlinearity, the problem does not converge. There may be other things that I have to try to make the problem converge.

Thanks once again,
Sirisha
Thanks Ivar! I think I have an idea now .......it needs to be turned on for my case but the problem is that if I turn on the geometric nonlinearity, the problem does not converge. There may be other things that I have to try to make the problem converge. Thanks once again, Sirisha

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30 janv. 2011, 17:20 UTC−5
Hi

if you have a fuid flow case, often the default initial conditions of "0" are giving trouble, you could set some average velocity (parabolic profile) and pressure drop (i.e Poiseille) to help the system to start. then if you use only velocity, define an absolute pressure somewhere on a point, to fix the gauge pressure

--
Good luck
Ivar
Hi if you have a fuid flow case, often the default initial conditions of "0" are giving trouble, you could set some average velocity (parabolic profile) and pressure drop (i.e Poiseille) to help the system to start. then if you use only velocity, define an absolute pressure somewhere on a point, to fix the gauge pressure -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 30 janv. 2011, 23:27 UTC−5
I will use a fluid flow but that's only after I have a successful run with the current case. In the current problem, all I have is a hollow pipe (this is my solid mechanics module) and I am applying a pressure on the internal boundaries. So its just a pressure applied hollow pipe that I expect to expand radially and contract longitudinally. Unfortunately the problem is not converging when I use a hyperelastic model. It also does not converge when I use a linear elastic model with the nonlinear geometry checked.

Thanks,
Sirisha
I will use a fluid flow but that's only after I have a successful run with the current case. In the current problem, all I have is a hollow pipe (this is my solid mechanics module) and I am applying a pressure on the internal boundaries. So its just a pressure applied hollow pipe that I expect to expand radially and contract longitudinally. Unfortunately the problem is not converging when I use a hyperelastic model. It also does not converge when I use a linear elastic model with the nonlinear geometry checked. Thanks, Sirisha

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 31 janv. 2011, 15:58 UTC−5
I am wondering if someone can tell me why the attached problem is not converging. I am not very familiar with the solid mechanics part of COMSOL so I am not sure what the problem is missing that will help it converge. Are my boundary conditions wrong?

Thanks for the help.

Sincerely,
Sirisha
I am wondering if someone can tell me why the attached problem is not converging. I am not very familiar with the solid mechanics part of COMSOL so I am not sure what the problem is missing that will help it converge. Are my boundary conditions wrong? Thanks for the help. Sincerely, Sirisha


Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 févr. 2011, 03:17 UTC−5
Hi

I believe its because you "hit" it too hard: you start your simulation at rest, no stress, all "0" and in the next instant you apply 4 bars pressure.
That is tough for the poor solver ;)
Try a continuation sweep with a parameter and ramp your load in 10 or more steps.

Then by forcing one "point " fixed + 1 symmetry plane you create singularities, I would rather propose to block, by opposed pairs, all 8 points (or at last the 4 externals) on your symmetry plane with y=0 and the two vertical ones with z=0 (Prescribed constraints).

Or you do it fancier, by creating a coupling variable and resolve it to have the central average section center to remain on Y=0,Z=0 and you force no twist (Rx)

Finally if your case is fully symmetric, you could use 3 symmetry planes, which reduces by 4 the number of elements, and blocks the system on axis

--
Good luck
Ivar
Hi I believe its because you "hit" it too hard: you start your simulation at rest, no stress, all "0" and in the next instant you apply 4 bars pressure. That is tough for the poor solver ;) Try a continuation sweep with a parameter and ramp your load in 10 or more steps. Then by forcing one "point " fixed + 1 symmetry plane you create singularities, I would rather propose to block, by opposed pairs, all 8 points (or at last the 4 externals) on your symmetry plane with y=0 and the two vertical ones with z=0 (Prescribed constraints). Or you do it fancier, by creating a coupling variable and resolve it to have the central average section center to remain on Y=0,Z=0 and you force no twist (Rx) Finally if your case is fully symmetric, you could use 3 symmetry planes, which reduces by 4 the number of elements, and blocks the system on axis -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 févr. 2011, 16:54 UTC−5
He he :) I though COMSOL does some magic for me and things will work out when I have the right physics :)

Thanks for the valuable knowledge.

I am trying the parametric sweep now and its running. I am doing 0:5:40000Pa. So I expect it to get done in 2-3 days :) Before that I tried with a pressure of 2Pa instead of 40,000 and it worked!

I also tried the 4 externals with the prescribed displacement of y=0 and z=0(vertical points) and the problem converges faster (provided the pressure is still 2Pa). But the result I am seeing with this scenario is not what I want. When I apply the pressure, I want the inlet face to stay where it is while the rest of the pipe expands radially and contracts longitudinally. I am not sure how to simulate this. I did the fixed constraint based on what I saw in a COMSOL tutorial. If you have any ideas about how to achieve this result then please let me know.

Also...........you wrote that "Or you do it fancier, by creating a coupling variable and resolve it to have the central average section center to remain on Y=0,Z=0 and you force no twist (Rx)". I don't know how to create a coupling variable and I am looking through the user manual but am not able to locate any information on this. Can you please provide me with more hints or directions on where I can locate more information on how to create coupling variables.

I tried incorporating symmetry (in which case I consider only half the geometry of what I attached earlier) but I am running into "failed to find a solution" error. I am not sure how to resolve this.

Thanks so much for the help.

Sincerely,
Sirisha
He he :) I though COMSOL does some magic for me and things will work out when I have the right physics :) Thanks for the valuable knowledge. I am trying the parametric sweep now and its running. I am doing 0:5:40000Pa. So I expect it to get done in 2-3 days :) Before that I tried with a pressure of 2Pa instead of 40,000 and it worked! I also tried the 4 externals with the prescribed displacement of y=0 and z=0(vertical points) and the problem converges faster (provided the pressure is still 2Pa). But the result I am seeing with this scenario is not what I want. When I apply the pressure, I want the inlet face to stay where it is while the rest of the pipe expands radially and contracts longitudinally. I am not sure how to simulate this. I did the fixed constraint based on what I saw in a COMSOL tutorial. If you have any ideas about how to achieve this result then please let me know. Also...........you wrote that "Or you do it fancier, by creating a coupling variable and resolve it to have the central average section center to remain on Y=0,Z=0 and you force no twist (Rx)". I don't know how to create a coupling variable and I am looking through the user manual but am not able to locate any information on this. Can you please provide me with more hints or directions on where I can locate more information on how to create coupling variables. I tried incorporating symmetry (in which case I consider only half the geometry of what I attached earlier) but I am running into "failed to find a solution" error. I am not sure how to resolve this. Thanks so much for the help. Sincerely, Sirisha

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 févr. 2011, 01:39 UTC−5
Hi

0:5:40000 is really small steps probably a 40000*range(0,0.1,1) would do

You can also fully fix the inlet tube section, but then it cannot expand, if you just block the external points so they can expand radially your system will solve and remain stable, without inlet shape artefact's (over-constrained surface)

It in the Define subnode you can find them, make very simple examples to begin with

full symmetry is possible (oups check it in the doc for your hyper elastic material there might be an incompatibility there, just as "large deformations are mostly incompatible with antisymmetric boundary conditions, but not symmetric)

Again take a simple case to start with

--
Good luck
Ivar
Hi 0:5:40000 is really small steps probably a 40000*range(0,0.1,1) would do You can also fully fix the inlet tube section, but then it cannot expand, if you just block the external points so they can expand radially your system will solve and remain stable, without inlet shape artefact's (over-constrained surface) It in the Define subnode you can find them, make very simple examples to begin with full symmetry is possible (oups check it in the doc for your hyper elastic material there might be an incompatibility there, just as "large deformations are mostly incompatible with antisymmetric boundary conditions, but not symmetric) Again take a simple case to start with -- Good luck Ivar

Ramamoorthy Sunil Kumar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 sept. 2013, 09:40 UTC−4
This function is not available in 4.3b version.
Solutions obtained with and without this function vary significantly. I dont how to locate this option in 4.3b version
This function is not available in 4.3b version. Solutions obtained with and without this function vary significantly. I dont how to locate this option in 4.3b version

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.