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 of S11 parameter for different mesh element sizes

Please login with a confirmed email address before reporting spam

Hi,

I have achieved two very different S11 parameter plots for the same patch antenna simulation at "extremelly coars" and "normal" element sizes of mesh. I attached these two results as image files to this question. "normal" one is more similar to the measurement results. But how can I be sure which element size is enough for me. Because, the finer element sizes are very heavy for my computer.

Thank you



4 Replies Last Post May 21, 2019, 12:01 p.m. EDT
Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 5 years ago May 20, 2019, 4:37 p.m. EDT
Updated: 5 years ago May 20, 2019, 4:37 p.m. EDT

Well, you can be moderately more confident that the normal mesh is giving better results than the coarser mesh. You might want to also try models with different settings of discretization (linear, quadratic, etc.), or use a slow but stable iterative solver (if you aren't already), like GMRES with the SOR vector preconditioner, along with a relatively fine mesh, and then even consider a higher level of discretization (if you have enough memory), to achieve high accuracy. Regardless, use your finest mesh settings on the patch and its local substrate, but leave the surrounding space/volume at coarser mesh settings, to keep the computational size under control. Finally, also bear in mind that if you are modeling just one simple patch antenna, but are already running out of memory, then you are either using your memory rather wastefully in your model or have a very limited-capability computer. If the latter, you may want to think about upgrading your computer hardware.

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
Well, you can be moderately more confident that the normal mesh is giving better results than the coarser mesh. You might want to also try models with different settings of discretization (linear, quadratic, etc.), or use a slow but stable iterative solver (if you aren't already), like GMRES with the SOR vector preconditioner, along with a relatively fine mesh, and then even consider a higher level of discretization (if you have enough memory), to achieve high accuracy. Regardless, use your finest mesh settings on the patch and its local substrate, but leave the surrounding space/volume at coarser mesh settings, to keep the computational size under control. Finally, also bear in mind that if you are modeling just one simple patch antenna, but are already running out of memory, then you are either using your memory rather wastefully in your model or have a very limited-capability computer. If the latter, you may want to think about upgrading your computer hardware.

Please login with a confirmed email address before reporting spam

Posted: 5 years ago May 21, 2019, 3:19 a.m. EDT

Thank you for your answer Robert.

From your answer, I understand that I have to go to the details of the solver :( . This is bad news for me, because I thought I could simulate antennas by using classical instructions as in Application Gallery (https://www.comsol.com/models).

By the way my computer configuration is as follows: 64 Bit Intel Core i5 4590 CPU @ 3.30 GHz 4 GB RAM (DDR3)

Thank you for your answer Robert. From your answer, I understand that I have to go to the details of the solver :( . This is bad news for me, because I thought I could simulate antennas by using classical instructions as in Application Gallery (https://www.comsol.com/models). By the way my computer configuration is as follows: 64 Bit Intel Core i5 4590 CPU @ 3.30 GHz 4 GB RAM (DDR3)

Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 5 years ago May 21, 2019, 11:35 a.m. EDT
Updated: 5 years ago May 21, 2019, 11:34 a.m. EDT

You're welcome. By the way, I recommend you install a lot more RAM, if you can.

-------------------
Scientific Applications & Research Associates (SARA) Inc.
www.comsol.com/partners-consultants/certified-consultants/sara
You're welcome. By the way, I recommend you install a *lot* more RAM, if you can.

Please login with a confirmed email address before reporting spam

Posted: 5 years ago May 21, 2019, 12:01 p.m. EDT

I'll try. Thanks again.

I'll try. Thanks again.

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.