Wrong Q factor

Please login with a confirmed email address before reporting spam

I am simulating 1/4 of a SiN membrane suspended with 4 thin arms and clamped to the surrounding substrate, the membrane is very thin (100nm) but very wide and long (high aspect ratio managed using Swept function in the mesh) and the entire structure has an isotropic damping equal to 1/Q where Q is set to 100.

Performing an eigenfrequency study, I am expecting to find for all the modes a value of Q (that COMSOL evaluates automatically) equal to the value I set (100). Unfortunately, for the fundamental mode this doesn't happen; precisely, the Q of the fundamental mode is sometimes higher (120,130,150) and sometimes smaller (70, 80) than 100. In addition, every time I launch a new simulation run (even though I do not change nothing) the value of Q (and consequently the Real and Imaginary parts of the eigenfrequency) changes.

Is it normal? Is it a big issue?



4 Replies Last Post Apr 13, 2026, 3:47 a.m. EDT
Acculution ApS Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 4 days ago Apr 9, 2026, 8:58 a.m. EDT

Not an expert in the field, but had a quick look. The geometry is very thin relative to its other dimensions, and that might lead to some ill-conditioning? Would it be an idea to model with shells instead?

-------------------
René Christensen, PhD
Acculution ApS
www.acculution.com
info@acculution.com
Not an expert in the field, but had a quick look. The geometry is very thin relative to its other dimensions, and that might lead to some ill-conditioning? Would it be an idea to model with shells instead?

Please login with a confirmed email address before reporting spam

Posted: 4 days ago Apr 9, 2026, 10:51 a.m. EDT

Not an expert in the field, but had a quick look. The geometry is very thin relative to its other dimensions, and that might lead to some ill-conditioning? Would it be an idea to model with shells instead?

I cannot use Shell due to other contraints. And actually I also noticed that if I add a pre-stress to my membrane, the value of Q is completely wrong for all the modes.

>Not an expert in the field, but had a quick look. The geometry is very thin relative to its other dimensions, and that might lead to some ill-conditioning? Would it be an idea to model with shells instead? I cannot use Shell due to other contraints. And actually I also noticed that if I add a pre-stress to my membrane, the value of Q is completely wrong for all the modes.

Henrik Sönnerlind COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 3 days ago Apr 10, 2026, 12:28 p.m. EDT

Réne is correct about the assumption of ill-conditioning. You can improve things by using a better mesh. Avoid those highly distorted elements, and use a finer mesh in general.

But there is one fundamental problem. You use a continuity condition between two partially overlapping boundaries. With the current mesh only small fractions of elements overlap in that region. That makes the connection completely unpredictable. I would guess that it acts as a hinge, and that may not be the intent. You need to have at least one full element on the destination side overlapping the source side in order to transfer a moment.

-------------------
Henrik Sönnerlind
COMSOL
Réne is correct about the assumption of ill-conditioning. You can improve things by using a better mesh. Avoid those highly distorted elements, and use a finer mesh in general. But there is one fundamental problem. You use a continuity condition between two partially overlapping boundaries. With the current mesh only small fractions of elements overlap in that region. That makes the connection completely unpredictable. I would guess that it acts as a hinge, and that may not be the intent. You need to have at least one full element on the destination side overlapping the source side in order to transfer a moment.

Please login with a confirmed email address before reporting spam

Posted: 2 hours ago Apr 13, 2026, 3:47 a.m. EDT

Réne is correct about the assumption of ill-conditioning. You can improve things by using a better mesh. Avoid those highly distorted elements, and use a finer mesh in general.

But there is one fundamental problem. You use a continuity condition between two partially overlapping boundaries. With the current mesh only small fractions of elements overlap in that region. That makes the connection completely unpredictable. I would guess that it acts as a hinge, and that may not be the intent. You need to have at least one full element on the destination side overlapping the source side in order to transfer a moment.

I followed your suggestions and tried a much finer mesh and I also extended the overlapping surface so that there are more than just 1 mesh domain in the overlap region (as showed in the picture). Unfortunately, this changes didn't give any improvements ( eigenfrequency of fundamental mode and Q change at each run).

Do you think that using an Union instead of an Assembly (as I am doing now) could give improvements? I am open to other suggestions. Thank you

>Réne is correct about the assumption of ill-conditioning. You can improve things by using a better mesh. Avoid those highly distorted elements, and use a finer mesh in general. > >But there is one fundamental problem. You use a continuity condition between two partially overlapping boundaries. With the current mesh only small fractions of elements overlap in that region. That makes the connection completely unpredictable. I would guess that it acts as a hinge, and that may not be the intent. You need to have at least one full element on the destination side overlapping the source side in order to transfer a moment. I followed your suggestions and tried a much finer mesh and I also extended the overlapping surface so that there are more than just 1 mesh domain in the overlap region (as showed in the picture). Unfortunately, this changes didn't give any improvements ( eigenfrequency of fundamental mode and Q change at each run). Do you think that using an Union instead of an Assembly (as I am doing now) could give improvements? I am open to other suggestions. Thank you

Reply

Please read the discussion forum rules before posting.

Please log in to post a reply.

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.