Menu

Discussion Forum

New Discussion
Filter

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 The information provided may be out of date.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.

Does size of Geometry affect the solution ?

Posted Jul 6, 2015, 8:46 p.m. EDT Low-Frequency Electromagnetics, RF & Microwave Engineering, Geometry, Mesh, Modeling Tools & Definitions, Parameters, Variables, & Functions Version 4.3 5 Replies

Ashish Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

I am looking for electric field distribution in a coaxial waveguide. Now if my waveguide is of order of mm range, the electric field pattern is uniform and matches the analytical solution for TEM mode. Now if I reduce the dimension to um(10^-6 m) range, i am seeing the anomaly. I am attaching the figures for the same. I am not sure what am I missing ?
I am operating at 1 GHz frequency. On increasing the frequency, the anomaly decreases.
For both cases I have used "Extremely Fine" element size option.
5 Replies Last Post Jul 8, 2015, 10:15 a.m. EDT
Lasse Murtomäki
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 7, 2015, 2:35 a.m. EDT
Converting the problem dimensionless would remove the need of scaling geometry. Of course if the size is of the order of the wavelength of the signal situation changes, but you are not there yet.

You have used the same mesh size for both cases which means that their number of elements is very different?
Converting the problem dimensionless would remove the need of scaling geometry. Of course if the size is of the order of the wavelength of the signal situation changes, but you are not there yet. You have used the same mesh size for both cases which means that their number of elements is very different?
Ashish Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 7, 2015, 2:58 p.m. EDT
Could you suggest to make dimensionless. I am using 1 GHz as frequency which means 0.3m would be the wavelength. Now as I am going smaller and smaller it should be much smaller to the wavelength.
Regarding the meshing, i have even tried "Extremely Fine" as Element Size and Refine as 2 to increase the number of elements, still the same issue exist for the um(10^-6) ranged coaxial cable.
The inner cable is of radius 0.5*n and outer one as 3.5*n where n = 10^-6.
Could you suggest to make dimensionless. I am using 1 GHz as frequency which means 0.3m would be the wavelength. Now as I am going smaller and smaller it should be much smaller to the wavelength. Regarding the meshing, i have even tried "Extremely Fine" as Element Size and Refine as 2 to increase the number of elements, still the same issue exist for the um(10^-6) ranged coaxial cable. The inner cable is of radius 0.5*n and outer one as 3.5*n where n = 10^-6.
Ashish Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 7, 2015, 4:51 p.m. EDT
Now even operating the coaxial cable (The inner cable is of radius 0.5*n and outer one as 3.5*n) with n = 10^-3 at frequency at 1 MHz is also giving the same problem.
Reducing the frequency increases the wavelength hence the ratio of physical size and wavelength decreases. Is there any specific ratio which need to be followed to avoid any this anomaly ?
Now even operating the coaxial cable (The inner cable is of radius 0.5*n and outer one as 3.5*n) with n = 10^-3 at frequency at 1 MHz is also giving the same problem. Reducing the frequency increases the wavelength hence the ratio of physical size and wavelength decreases. Is there any specific ratio which need to be followed to avoid any this anomaly ?
Lasse Murtomäki
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 8, 2015, 1:29 a.m. EDT
You could scale the dimensions by, e.g. the cable radius. What is the equation you simulate?
You could scale the dimensions by, e.g. the cable radius. What is the equation you simulate?
Ashish Certified Consultant
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 9 years ago Jul 8, 2015, 10:15 a.m. EDT
I am using the Physics as "Electromagnetic Waves, Frequency Domain" and "Mode Analysis" as Preliminary Study.
The standard equation which is solved is wave equation:
delX (mu_r^ -1 )(del X E) - k^2(epi_r - j sigma/( omega*epio) E = 0

But the real issue which I am facing is that for a particular combination of dimension of waveguide and frequency of operation I am not getting field patterns similar to theoretical ones.

Outer Radius = 3.43*n
Inner Radius = 0.5*n
Dielectric Constant of material is 11
Conductors are Perfect Electric Conductors.

Till now if for two cases I am seeing discrepancy:
1. Waveguide is of the order of um( n = 10^-6) and Frequency = 1 GHz
2. Waveguide is of the order of mm( n = 10^-3) and frequency = 1 MHz

The ratio of dimension vs wavelength is same in both cases. I am not sure which limitations am I hitting ?
I am using the Physics as "Electromagnetic Waves, Frequency Domain" and "Mode Analysis" as Preliminary Study. The standard equation which is solved is wave equation: delX (mu_r^ -1 )(del X E) - k^2(epi_r - j sigma/( omega*epio) E = 0 But the real issue which I am facing is that for a particular combination of dimension of waveguide and frequency of operation I am not getting field patterns similar to theoretical ones. Outer Radius = 3.43*n Inner Radius = 0.5*n Dielectric Constant of material is 11 Conductors are Perfect Electric Conductors. Till now if for two cases I am seeing discrepancy: 1. Waveguide is of the order of um( n = 10^-6) and Frequency = 1 GHz 2. Waveguide is of the order of mm( n = 10^-3) and frequency = 1 MHz The ratio of dimension vs wavelength is same in both cases. I am not sure which limitations am I hitting ?

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.

Suggested Content