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.

Different results for axisymmetric and 2D geometries

Please login with a confirmed email address before reporting spam

Hi
I am simulating the heating of a lipossome by a focused laser beam in the 3.5 version. I have only just began to simulate the heating of the aquous medium surrounding the liposome. The module is the General Heat Transfer with transient analysis.
I am entering a heat source equation into a subdomain defined by a rectangule 2mm high and 0.2mm wide. The subdomain is defined as water using the materials library. Opacity is defined as transparent. Boundary conditions are set to insolation except the top one which is set to heat flux. Time is set to 1 second.
I used 2 different geometries: axisymmetric and 2D. Both have the same mesh size, equations, constants and inital conditions yet they present VERY different temperature results after running the simulation. The axisymmetric shows a temperature increase of 9ºC and the 2D shows an increase of over 50ºC! Does anyone knows why is this happening?
Thank you

13 Replies Last Post 10 févr. 2011, 07:34 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 20 janv. 2011, 09:13 UTC−5
Hi

could it be that the 2D is per meter depth ,while the 2D axio is per "loop length" = 2*pi*r

have you checked your units and their coherence, that usually tells where an how to adapt the GUI fields and when to correct for loop length,respectively tru depth/thickness

--
Good luck
Ivar
Hi could it be that the 2D is per meter depth ,while the 2D axio is per "loop length" = 2*pi*r have you checked your units and their coherence, that usually tells where an how to adapt the GUI fields and when to correct for loop length,respectively tru depth/thickness -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 21 janv. 2011, 06:07 UTC−5
Why does it surprise you, that the results of a 2D-model differ to them of a 2D-axisymmetric model (with the "same geometry")?

At least, the equations are not the same. -> x,y and r,z
Why does it surprise you, that the results of a 2D-model differ to them of a 2D-axisymmetric model (with the "same geometry")? At least, the equations are not the same. -> x,y and r,z

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 21 janv. 2011, 06:42 UTC−5
Hi

I see we have the same suspicion ;)

--
Good luck
Ivar
Hi I see we have the same suspicion ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2 févr. 2011, 12:02 UTC−5
Thanks for the responses
I did consider the distinction between the two kinds of geometry and the transformations that have to be performed to the equations from one case to the other. But that does not justify the great increase of temperature that occurs in the 2D case, since the heat source equation provides with the same amount of energy in both cases.
By "loop length" I believe you mean the "thickness" fields in the geometric properties on the postprocessing menu. They are indeed on the default value of 1m and I tried to change on the 2D to nm and um scales without much success. In the 2D axio I did the same in the boundary tab on the "geometric properties" box, but I am not sure if that is correct. Also tried inserting 2*pi*r in the same field, and still it did not change the resuslts.
I am not sure what you mean by coherence in the units.
Cheers
Thanks for the responses I did consider the distinction between the two kinds of geometry and the transformations that have to be performed to the equations from one case to the other. But that does not justify the great increase of temperature that occurs in the 2D case, since the heat source equation provides with the same amount of energy in both cases. By "loop length" I believe you mean the "thickness" fields in the geometric properties on the postprocessing menu. They are indeed on the default value of 1m and I tried to change on the 2D to nm and um scales without much success. In the 2D axio I did the same in the boundary tab on the "geometric properties" box, but I am not sure if that is correct. Also tried inserting 2*pi*r in the same field, and still it did not change the resuslts. I am not sure what you mean by coherence in the units. Cheers

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2011, 05:21 UTC−5
Hi,

I still don't quite get it.

May be you could explain it more detailed. How is it possible to model a "desired" geometry in two ways, 2D AND 2D-axisymmetric, which should result in the same imagined 3D geometry? 2D means no changes in one cartesian direction (du/dz = 0 -> infinite elongated 3D-geometry in z-direction), 2D means no change in the direction of phi (du/dphi =0 -> finite geometry). This leads to two different heat transfer PDEs, therefore you are not using the same equations for 2D and 2D-axisymmetric! When I'm doing some quick checks, I easily get difference of about 50%.

best regards
Hi, I still don't quite get it. May be you could explain it more detailed. How is it possible to model a "desired" geometry in two ways, 2D AND 2D-axisymmetric, which should result in the same imagined 3D geometry? 2D means no changes in one cartesian direction (du/dz = 0 -> infinite elongated 3D-geometry in z-direction), 2D means no change in the direction of phi (du/dphi =0 -> finite geometry). This leads to two different heat transfer PDEs, therefore you are not using the same equations for 2D and 2D-axisymmetric! When I'm doing some quick checks, I easily get difference of about 50%. best regards

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 3 févr. 2011, 05:55 UTC−5
Hi

sorry for not being clear, you are right in 2D you must check your "thickness" and in 2D-axi the "thickness" is replaced by the "loop length" which equals 2*pi*r

ie. an integration in 2D-axi over the surface area of the integrand 2*pi*r gives you the total volume of your part, as it would be if in 3D. If the integrand is "1" you get the section area of your 2D-axi cut view

Use the "units" to check the coherene of the dimensions you compare

Be sure you get the differences between 2D and 2D axi, the 3D representation is quite different, the latter "loops around an axis, the thickness is not constant, and the resulting 3D topology is radically different

--
Good luck
Ivar
Hi sorry for not being clear, you are right in 2D you must check your "thickness" and in 2D-axi the "thickness" is replaced by the "loop length" which equals 2*pi*r ie. an integration in 2D-axi over the surface area of the integrand 2*pi*r gives you the total volume of your part, as it would be if in 3D. If the integrand is "1" you get the section area of your 2D-axi cut view Use the "units" to check the coherene of the dimensions you compare Be sure you get the differences between 2D and 2D axi, the 3D representation is quite different, the latter "loops around an axis, the thickness is not constant, and the resulting 3D topology is radically different -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2011, 07:08 UTC−5
Hi
I see what you mean when you mention that each case would not result in the same geometry in 3D. For that I am more inclined to use the 2D axisymetric to create a 3D geometry for my case.
I created the two cases for comparison, I did expect some difference in the results for the reasons you pointed out. I was just surprised by the huge difference, since notwithstanding the different PDEs one might get in each case, in both cases we have the same heat source and medium. Physically it doesn't make much sense to me why one segment of volume heats to a much greater extent than the other simply because we are using different geometries to solve our equations.
Regards
Hi I see what you mean when you mention that each case would not result in the same geometry in 3D. For that I am more inclined to use the 2D axisymetric to create a 3D geometry for my case. I created the two cases for comparison, I did expect some difference in the results for the reasons you pointed out. I was just surprised by the huge difference, since notwithstanding the different PDEs one might get in each case, in both cases we have the same heat source and medium. Physically it doesn't make much sense to me why one segment of volume heats to a much greater extent than the other simply because we are using different geometries to solve our equations. Regards

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2011, 08:16 UTC−5
Doesn't make sense? Why not? :)

If you want, you can check it analytically. Have a look at "Conduction of heat in solids" by Carslaw and Jaeger. I'm pretty sure, there you will find examples for comparison. I guess, Jaeger and Carslaw would had loved Comsol, if they had ;)

kind regards
Doesn't make sense? Why not? :) If you want, you can check it analytically. Have a look at "Conduction of heat in solids" by Carslaw and Jaeger. I'm pretty sure, there you will find examples for comparison. I guess, Jaeger and Carslaw would had loved Comsol, if they had ;) kind regards

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2011, 13:01 UTC−5
I will indeed check Carslaw and Jaeger again :) I'm going to work on the 2Daxisymmetric since it's more suitable to create the 3D model and the temperature results are fairly close to those obtained empirically. I hadn't thought of the thickness and loop length issue yet, that will be important in the next phase. Thank you both for your assistance
Regards
I will indeed check Carslaw and Jaeger again :) I'm going to work on the 2Daxisymmetric since it's more suitable to create the 3D model and the temperature results are fairly close to those obtained empirically. I hadn't thought of the thickness and loop length issue yet, that will be important in the next phase. Thank you both for your assistance Regards

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 3 févr. 2011, 14:19 UTC−5
Hi

the 3D version of a circle in 2D versus 2D-axis is basically saying: what are the difference betweeen a cylinder and a thorus, (or a churro and a doughnut ;), no ?

--
Good luck
Ivar
Hi the 3D version of a circle in 2D versus 2D-axis is basically saying: what are the difference betweeen a cylinder and a thorus, (or a churro and a doughnut ;), no ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 4 févr. 2011, 07:30 UTC−5
Yes, and to be more precise: a very very very long churro! Therefore I would recommend to choose 2D mode instead of 2D-axi :)

Comsol meets snacks - FEM is going to be easy ... ;)
Yes, and to be more precise: a very very very long churro! Therefore I would recommend to choose 2D mode instead of 2D-axi :) Comsol meets snacks - FEM is going to be easy ... ;)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 févr. 2011, 06:25 UTC−5
Hi
Everyday objects are useful when doing 3D modelling ;) My circle is centered at r = 0 so it gives out a sphere in 3D. But I'll do the same for a 2d geometry for comparisson.
Hi Everyday objects are useful when doing 3D modelling ;) My circle is centered at r = 0 so it gives out a sphere in 3D. But I'll do the same for a 2d geometry for comparisson.

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 10 févr. 2011, 07:34 UTC−5
Hi

do not forget that in 2D-axi you should have no geoemtry to the left for r<0, everything should end at most "on axis" <0,1> direction

--
Good luck
Ivar
Hi do not forget that in 2D-axi you should have no geoemtry to the left for r

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.