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 The information provided may be out of date.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

Hi,

I have used structural mechanics to simulate an indenting sphere on a half space polymer film. I am changing the elastic modulus of the polymer film to go from hard material to soft material.
I am trying to reach a certain force using weak constraints. When I use the elastic moduli from some GPa to around 80 MPa for indentation of 100- 400 nm, very well I can use linear elastic physics without selecting the box "consider geometry nonlinearity".

However, as long as I choose the elastic modulus 60 MPa or less with indention between 500 nm to 600 nm (I want to reach to a certain force and as the material is soft indentation is high), the simulation fails. It does not stop but a big triangular max pick appears at around 500 nm. However if I turn on "consider nonlinear geometry " and change the poission ratios, it works. but then I reach to a force at much lower indentation that I cannot compare the results with the previous calculations.

I would like to know why do that happens? and how can I fix it?


Yours,
Marjan

11 Replies Last Post 25 févr. 2013, 12:22 UTC−5
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. 2012, 15:08 UTC−5
Hi

what about telling us slightly more, i.e. size of sphere compared to your 0.5um indentation, thickness of layer pinched, what kind of change to nu ?
It's difficult to give you any clear int so long one cannot better understand de details here ;)

--
Good luck
Ivar
Hi what about telling us slightly more, i.e. size of sphere compared to your 0.5um indentation, thickness of layer pinched, what kind of change to nu ? It's difficult to give you any clear int so long one cannot better understand de details here ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 1 févr. 2012, 16:39 UTC−5
Hi Dear Ivar,

Thanks for your reply. In fact I am using axisymmetry geometry and I have a very stiff shell (Something like 1/8 of a half sphere) pressing against a block of material. The material height is 100 micron and in R direction, I have used perfectly match layer mesh elements. Apart from that pressing force, I have a boundry force that acts between the shell and polymer. That is about 11.98 MPa and pulls the surface up. That is a relatively large force that acts on some nm lenght scale. something like figure 3 in this page

rsta.royalsocietypublishing.org/content/366/1870/1627.full

The difference is that my particle is really pressing the surface and this pressure is deforming the soft surface upward.

Well I start from nu of 0.33 for 3 GPa polymer and come down to 0.4999 for 20 MPa. I wish to go evenmore down. The softer the film gets the more it deviates from Hertz model. Right now I consider nonlinear geometry for polymer film but not for the stiff shell . If I do not change the nu for my film, comsol cannot solve my problem
Thanks again,
Marjan


Hi

what about telling us slightly more, i.e. size of sphere compared to your 0.5um indentation, thickness of layer pinched, what kind of change to nu ?
It's difficult to give you any clear int so long one cannot better understand de details here ;)

--
Good luck
Ivar
[/QUOTE

Hi Dear Ivar, Thanks for your reply. In fact I am using axisymmetry geometry and I have a very stiff shell (Something like 1/8 of a half sphere) pressing against a block of material. The material height is 100 micron and in R direction, I have used perfectly match layer mesh elements. Apart from that pressing force, I have a boundry force that acts between the shell and polymer. That is about 11.98 MPa and pulls the surface up. That is a relatively large force that acts on some nm lenght scale. something like figure 3 in this page http://rsta.royalsocietypublishing.org/content/366/1870/1627.full The difference is that my particle is really pressing the surface and this pressure is deforming the soft surface upward. Well I start from nu of 0.33 for 3 GPa polymer and come down to 0.4999 for 20 MPa. I wish to go evenmore down. The softer the film gets the more it deviates from Hertz model. Right now I consider nonlinear geometry for polymer film but not for the stiff shell . If I do not change the nu for my film, comsol cannot solve my problem Thanks again, Marjan [QUOTE] Hi what about telling us slightly more, i.e. size of sphere compared to your 0.5um indentation, thickness of layer pinched, what kind of change to nu ? It's difficult to give you any clear int so long one cannot better understand de details here ;) -- Good luck Ivar [/QUOTE

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 21 févr. 2013, 11:07 UTC−5
Hi Marjan,

Are you doing a time-dependent study or stationary study?
Thanks a lot.
Hi Marjan, Are you doing a time-dependent study or stationary study? Thanks a lot.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 21 févr. 2013, 12:06 UTC−5
....Both.

--
Marjan
....Both. -- Marjan

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 févr. 2013, 08:47 UTC−5
Hi Marjan,

Thanks for your reply, I saw other post from you.www.uk.comsol.com/community/forums/general/thread/34/
I'm simulating contact between tip and sample, but it is difficult to converge doing time dependent study. I use small time step and the discretization of displacement is larger than contact. As there is no inertial contact, the inertial pressure should be zero right? More over, have you met the 'warning: New force constraints node detected' when tip indents the sample? I appreciate if you could give me some advice. Thanks a lot.

Regards,
Wenqi Deng
Hi Marjan, Thanks for your reply, I saw other post from you.http://www.uk.comsol.com/community/forums/general/thread/34/ I'm simulating contact between tip and sample, but it is difficult to converge doing time dependent study. I use small time step and the discretization of displacement is larger than contact. As there is no inertial contact, the inertial pressure should be zero right? More over, have you met the 'warning: New force constraints node detected' when tip indents the sample? I appreciate if you could give me some advice. Thanks a lot. Regards, Wenqi Deng

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 févr. 2013, 10:19 UTC−5
Hi Wenqi,

What are you using for your time dependent tip movement control? Parametric sweep or continuation. I would say that you would better use continuation.

What is the material that you are indenting? How much is the elastic modulus?

I would suggest that you first try to solve your problem and get convergence when your tip is in contact with your sample surface and then extend it to a case that the tip is not in contact. About the initial pressure: The initial pressure is the pressure that can be 2 orders of magnitude larger than the maximum pressure that occurs in contact area. It helps convergence and depending on your material elastic modulus you can estimate that.

You may also need to change some parameters is solver but first please tell me what is your elastic module and if you are moving a boundary or your whole indenter.

I have also faced the error you have faced but it can have different meanings. I can only give my comments if I know your boundary conditions and see your mesh. It can be due to both.

I hope it can help you :).

--
Marjan
Hi Wenqi, What are you using for your time dependent tip movement control? Parametric sweep or continuation. I would say that you would better use continuation. What is the material that you are indenting? How much is the elastic modulus? I would suggest that you first try to solve your problem and get convergence when your tip is in contact with your sample surface and then extend it to a case that the tip is not in contact. About the initial pressure: The initial pressure is the pressure that can be 2 orders of magnitude larger than the maximum pressure that occurs in contact area. It helps convergence and depending on your material elastic modulus you can estimate that. You may also need to change some parameters is solver but first please tell me what is your elastic module and if you are moving a boundary or your whole indenter. I have also faced the error you have faced but it can have different meanings. I can only give my comments if I know your boundary conditions and see your mesh. It can be due to both. I hope it can help you :). -- Marjan

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 févr. 2013, 11:28 UTC−5
Hi Marjan,

The tip movement is controlled by a sinusoidal force(which depends on time variable 't'), it should be the contiuation you mentioned?

The sample elastic modulus is 3e9[Pa] (PPMA). Both the tip and sample are defined as linear elastic model. I define tip surface and sample surface as contact pair and use tetrahedral mesh(whose minimum element size is 1[nm]). And bottom of the sample is fixed.

Thanks. Looking forward to more advice.

Regards,

Wenqi Deng

Hi Marjan, The tip movement is controlled by a sinusoidal force(which depends on time variable 't'), it should be the contiuation you mentioned? The sample elastic modulus is 3e9[Pa] (PPMA). Both the tip and sample are defined as linear elastic model. I define tip surface and sample surface as contact pair and use tetrahedral mesh(whose minimum element size is 1[nm]). And bottom of the sample is fixed. Thanks. Looking forward to more advice. Regards, Wenqi Deng

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 févr. 2013, 12:12 UTC−5
Hi Wenqi,

You can give values to " t" in continuation of parametric sweep. Use continuation. Please see the attached file. The second point is check that size of your mesh elements on the PMMA is 1.5 times smaller than the mesh size of your sphere. Check 2 times as well.

For penalty factor you should not have problem if you set the initial pressure about 1000 Pa or even larger.

So your simulation is 3d. Is it true that your system is only fixed at the bottom of your sample or support onto which the indenter indents and is everything else free? Then you should reconsider your model. Comsol faces problem to solve it.

What is the theory you want to compare your simulations with? I used Hertz model. As it is a model for half infinite surface, I used infinite element domains in definition part of comsol (see the attached file please) and decreased the number of mesh elements a lot. But my support is divided to several domains . The model became also very exact. It is as if you had a half space surface.

I did the reverse of what you did. I controlled displacement and with week constraints, I calculated reaction forces. Do you want to see the indentation? It is also interesting :).

Regards from Germany,
Marjan
Hi Wenqi, You can give values to " t" in continuation of parametric sweep. Use continuation. Please see the attached file. The second point is check that size of your mesh elements on the PMMA is 1.5 times smaller than the mesh size of your sphere. Check 2 times as well. For penalty factor you should not have problem if you set the initial pressure about 1000 Pa or even larger. So your simulation is 3d. Is it true that your system is only fixed at the bottom of your sample or support onto which the indenter indents and is everything else free? Then you should reconsider your model. Comsol faces problem to solve it. What is the theory you want to compare your simulations with? I used Hertz model. As it is a model for half infinite surface, I used infinite element domains in definition part of comsol (see the attached file please) and decreased the number of mesh elements a lot. But my support is divided to several domains . The model became also very exact. It is as if you had a half space surface. I did the reverse of what you did. I controlled displacement and with week constraints, I calculated reaction forces. Do you want to see the indentation? It is also interesting :). Regards from Germany, Marjan


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 févr. 2013, 12:27 UTC−5
..... See the attached new file for the continuation time dependent solver

--
Marjan
..... See the attached new file for the continuation time dependent solver -- Marjan


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 févr. 2013, 09:37 UTC−5
Hi Marjan,

Thanks for your reply. I've been thinking about my model. I apply a dynamic load on the tip and fix the bottom of my sample, it is not a complicated model, I cannnot figure out why comsol faces problem to solve it? Did you come across this problem or have you found out the source of this problem?

Regards,

Wenqi
Hi Marjan, Thanks for your reply. I've been thinking about my model. I apply a dynamic load on the tip and fix the bottom of my sample, it is not a complicated model, I cannnot figure out why comsol faces problem to solve it? Did you come across this problem or have you found out the source of this problem? Regards, Wenqi

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 févr. 2013, 12:22 UTC−5
Dear Wenqi

.... you do not have enough boundary conditions. It is too much free everywhere. It is difficult for comsol to solve it. you should think of putting somehow more boundary condition. try to solve such a problem for a hard rigid surface both analytically and with FEM. Then you can understand which assumptions you want to use and you immediately understand that for solving the problem on hard surfaces you need some boundary conditions. Can you share only your geometry with me? I can give you my comments.

--
Marjan
Dear Wenqi .... you do not have enough boundary conditions. It is too much free everywhere. It is difficult for comsol to solve it. you should think of putting somehow more boundary condition. try to solve such a problem for a hard rigid surface both analytically and with FEM. Then you can understand which assumptions you want to use and you immediately understand that for solving the problem on hard surfaces you need some boundary conditions. Can you share only your geometry with me? I can give you my comments. -- Marjan

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.