Discussion Forum

Hi

a quality factor is linked to damping and by default there is none (or perhps some numerical damping physics dependent) so the Q factor will essentially depend on what you defne in the "damping" material property term.

What are you using ?

--
Good luck
Ivar

Hi Ivar,

What do you mean by damping? If you mean whether I use lossy material, I don't use any. Materials that I use all have real valued properties.

Hi

then how can you defin a "Q value" of a resonance if you do not have any "damping" ?

A FEM pogramme will give you the eigenfrequencies depending on (for an isoptropic marial) E, nu, rho (Young modulus, Poisson and density) and the geometrical shape (hence I.. inertia tensor) nothing more, the true Q factor of such resonnaces is not defined, the peaks are infinitely long and narrow (to the numerical damping limit).

So if you need a true Q factor you must define some damping.

However, with 4.1 you have access to the mass participation factors which gives you the relative energies of the different modes w.r.t the three spatial directions x,y,z (unfortunately these are not resolved for the three rotations) the way the mass participation factors are defined are that the sum of the square of each mode "mass participation" factor (for all modes) sum up to the total mass (per direction, and for the rotations it would be summing to the total inertia in the respective directions). Check the doc and the eigenfrequency solver sub-node to select the correct normalisation

--
Good luck
Ivar

Thanks for your answer.

The Q-factor values that I have found are calculated by Comsol itself with this parameter, Qfact_rfw. Why do you think that they are not the true Q-factors? What am I missing here? How should I be calculating them in order to obtain the true ones?

Hi

sorry for the confusion, I'm not saying that COMSOL is not calculating correctly.

If you feed in correct damping values (or in rf corret complex indexes) then the damping is present and the Q factors as defined are there. Depending, on the physics though the material "damping" properties are not always that well known. But for RF and optics this is rather the case.

Pls note that in RF there are different ways to enter the material properties, and if you operate with material index of refraction there is a particular sign convention that COMSOl uses, while there are 2 different common ones in the litterature, just check that you use the same with the same sign for the complex index value.

Again nothing wrong, it's only that one need to know that there are different representations and chack that we all (and COMSOL) uses the same. This applies also for PZT material tensor properties, COMSOL follows the IEEE convention, that is different from the traitional structural convention.

--
Good luck
Ivar

Hi Ivar,

Thanks for your quick reply.

In my model I don't use any material that have complex material properties. The refractive indices are all defined in real values. So I obtain real-valued eigenfrequencies, but even in this case Comsol can provide Q-factors which I have found them rather small but this is another thing. What way should I follow in order to calculate Q-factors? and also have another thing to ask that in which boundary configuration I can obtain the correct Q-factor values?

Elif

What problem are you modeling? You said you're using PMLs so you might have some energy lost through radiation. In If you have no other losses, the Q-value you see is due to radiation.

Hi,

It is a 3D cavity modelling. I am trying to find the correct boundary conditions because as I change the boundaries, q-factor values change as well.

In that case you have a closed metal surface around the resonance cavity. The "right" boundary conditions depend on how you like to model your cavity. Many times it is sufficient to model the outer conductor as PEC and the dielectric as a perfect dielectric (i.e. zero conductivity). In this ideal case, you just set the outer boundaries to PEC. But then there is of course no interest in finding the Q-value. So how do you plan to model the materials? I get a feeling that you haven't understood how the Q-value depends on the material properties?

Hi
Actually i've encountered similar problem when modeling 3D cavity in comsol.
After weeks of such comsol simulations my conlusion are:
1. I dont think one should trust the Qfact_rfw given by comsol, it is no clear what it means and what are the source of the loses in general and specificaly for loss-free materials and PEC boundary conditions. It also tends to change with: meshing, air gap between the cavity and the bounadries and etc.
2. to me it seems that the best way to simulate cavity and loses is
A. not in comsol.
B. in case you are using comsol use: large air gap, PML domain after it and PEC boundary conditions at the end of the PML layer.


Elif, Jones and others- I would appriciate your feedback and opinion.

Hi

I have some problms to follow fully, but I agree one shhould always be suspicious and verify all results even from COMSOL. Now Q factors, for me, are only energy loss related, and the way FEM is set up (hence also COMSOL as a FEM tool) is that you have some numerical "loss de to the solving and number representation" mostly very small (provided your mesh is OK), you have then the material loss and the model / physics simplification which might introduce losses.

If you are sure you have a perfectly balance energy value for your physics of your model it remains the material loss effects, but most material data are reported "without losses".

Hence missing/incomplete input, => not really reliable output

--
Good luck
Ivar

can you explain in simple project what is your mean ? or introduce an example that in it solved a example ? i dont know how simulate a cantilever , 6 eigenfrequencies and q factor of that ,
thank you