Discussion Forum

Hi

there a few issues I do not fully catch in your model.

a) you set the domain to fixed, then why fix the bottom side and why model the "fixed domain at all ? Personally I would not use a domain fixed region as often the "flexibility" of the contact area is used to analyse and improve the convergence, a fixed domain is in principle "infinitely stiff", no ?
b) I would reduce the model, at least the large "fixed region and cut it at x=1.2 or there around
c) you are in 2D, so its an long (1m by default) strip you are simulating, or is it a 2d_axi you meant ?
d) Why use the "Parametric Sweep, and not "just" a Stationary - Extension - Continuation ?


Then the issue prescribed displacement or imposed boundary load. If you look carefully when you impose a prescribed displacement X=0 on the "symmetry boundary, the symmetry BC is flagged "overwritten" and is no longer active, these two are not compatible, it's one or the other, and these are not strictly the same either (check the underlying equations)

To get the displacement forced in Y you must also give a restriction on the motion in Y, set to a multiple of Parameter "Param" something like: "(0.2[mm]-0.32[mm])*Param" and have Param = range(0,0.1,1)

And use a scale factor for the deformation of "1" to see the contact appearing at the right place, on my 4.2 its solved in a few seconds like that

--
Good luck
Ivar

Hey Ivar,

First of all ,Thanks a lot for taking out time to address my issues:). Also I would like to say I am quite new to structural Issues.I have been into Aerodynamics( CFD and Flow simulation) for some years and i am just trying to get into a new field:) Some of my queries may sound amateur but I hope you will understand it:)

a)
I agree with your proposal and initially I has started with using just a constraint on bottom boundary of this curved surface. But at same time when i had consulted the COMSOL support team,They recommended me to use a fixed constrained for the entire solid domain. They wrote me "In structural mechanics mode you have to constrain your model in the space". This is the reason that I fixed domain in my model.but This time, i have replaced these domain constraints !

b)
I accept this mistake of over using the resources .But this is due to reason that i was trying to find results by varying radius of curved surface too.

c)
I am trying to simulate a 2d-Axisymmetric case. In principle ,I have to do this problem in 3D.but since i am new to COMSOL so i thought it would be better to get a good base with 2d-Axisymmetric case first.

d)
Frankly Speaking I do not understand when you say just a ' Stationary - Extension - Continuation'. As stated above,I had contacted COMSOL Support regarding this problem some weeks back and They had also given me the idea of "Parametric Sweep'" for contact problems. I have still used a Parametric sweep in files that are attached with this message.



And well,I have tried to implement the Prescribed Displacement following your guidelines.I have tried to do it in two ways.

i)
Applying a Prescribed Displacement " (0.2[mm]-0.32[mm])*para " in Y direction on the edge of Chip( I have removed the symmetry BC from this end now). and setting X displacement=0. . In this case ,The stress distribution seems good enough but the chip does not come into a complete contact with the base at the centre part.
( see attached file [Comsol_case1_convexbase__prdisp_r20l1.5th20.mph])

ii)
In this case,I have come up with an Expression " (-0.32[mm]+((20.2[mm]-sqrt((20[mm]*20[mm])-X^2))))*para " for Displacement in Y direction ( Using equation of Circle and Radius R =20mm) but now, I have applied the displacement in lower boundary of the chip.Here,I have defined two edges with Symmetry BC .With this, I obtain a good approximation of Deformation and the chip seems to go down and fit perfect on the curved base.But the Stress Distribution is no where near any realistic results.
( see attached file [Comsol_case1_convexbase__prdisp_CircExpression_r20l1.5th20.mph])


Can you please have a look and point out the mistakes and errors.

Best Regards,

Irfan

Dear Ivar !!!


Please give me some suggestions:(

Regards,

Irfan

Hi,

I have a question:

I have a sphere in contact with a thin film. I want to move the sphere vertically for 0.5 micrometer. I have considered the top part of my sphere which is at 4 to move until 3.5 micrometer. Until now I was using a parameter movement like
-(4e-6-(0.1e-6*para) with the range (4e-6, -0.1e-6, 3.5e-6). it did not work. On the other hand now with [4e-6-3.5e-6]*para with the range(4.0e-6,-0.1,3.5e-6) it is working however it even works much better with range [0, 0.1,1] . I cannot understand what is happening. Please explain for me a little bit what is parameter sweep doing,

I am looking forward to your help
Marjan

If the initial coordinate of the top of the sphere is at 4 and you want to move it to 3.5 microns you can apply a displacement of 0.5e-6*para and vary para from 0 to 1, for example range(0,0.1,1.0). The first value (para=0) will give no displacement and the last one (para=1) will move the top of the sphere to its final location.

Nagi Elabbasi
Veryst Engineering

Hi Dear Nagi,

Thanks. I have 2 other questions that Iwould be very happy if you answer me.

1- Now that I have my sphere on the thin film, I consider them as pair. When are they pair and when roller?

2- When do we normally use coupling? When we have two different physics, for example? Do we use coupling when we have only one physics?

Thanks again,
Marjan

Hi Dear Nagi,

Thanks. I have 2 other questions that Iwould be very happy if you answer me.

1- Now that I have my sphere on the thin film, I consider them as pair. When are they pair and when roller?

2- When do we normally use coupling? When we have two different physics, for example? Do we use coupling when we have only one physics?

Thanks again,
Marjan

Dear Marjan,

My apologies for delay in response.
I just saw the message today.

well I am sure The help from Mr. Nagi has clarified the your issues regarding Prescribed displacement and Range.
Now about When to use Contact Pair or Roller .

Roller just works like a roller Constraint in Mechanics i.e. zero displacement in the normal direction to the boundary, but the boundary is free to move in tangential direction.Whereas ,for a Contact pair two surfaces cannot penetrate each other under deformation .
I am also trying to simulate something similar now i.e. to fix thin film on a sphere by prescribed displacement. ;)

About the question for coupling, I think we need to have two different physics. because coupling works in a way just as a source destination pair:) [ This is my personal viewpoint, so plz. dont hesitate to ask other about it :) ]

Regards and Good Luck,

Irfan

Hi Dear Irfan,

Thanks. Just one more question. Do you also have large deformation?If yes, which theory are you using as an analytical solution to verify your simulation. I want to check the stress distribution. However it seems that linear elastic models like Hertz models do not work for me. I need a theory that fits hyperelastic material.
Any idea that helps me would be really great,

Yours,
Marjan

Hi Dear Irfan,

Thanks. Just one more question. Do you also have large deformation?If yes, which theory are you using as an analytical solution to verify your simulation. I want to check the stress distribution. However it seems that linear elastic models like Hertz models do not work for me. I need a theory that fits hyperelastic material.
Any idea that helps me would be really great,

Yours,
Marjan

Dear Marjan,

well yes i have a large deformation . in case of my study in and using Sx=E*th/ 2R where E is Elasticity modulus , th the thickness and R is radius of curved surface.
Besides are you considering the ' Geometric non linearity' option in Linear Elastic Model settings?

Regards,

PS : my apologies again for a delayed response.

Hi Dear Irfan,

Thanks a lot. Well as I am working with Silicon rubber, I am not that much interested in linear elastic material. However I tested the system with linear elastic material including nonlinearity as well. Now I am performing the simulation with hyperelastic material, neo hookean formalism.
I googled and I found that as long as deformation is less than 20%, I can use neo hookean formalism. But how should I consider that ? 20% deformation with respect to the total size of slave or the deformation in mesh element size? I am looking forward to your answer.

Yours,
Marjan