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 The information provided may be out of date.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.

Setting up solver loop for 2 physics.

Please login with a confirmed email address before reporting spam

Hi,

I have a problem that involves EM wave (frequency around 10^14 Hz) and mechanical movement (frequency 10^6 - 10^8 Hz). Basically the EM wave will give a force on the structure, and the force depends on the current movement/bending of the structure. However, because the frequency of EM wave is much higher than that of mechanical movement, it's ok to calculate the EM wave at each time step as if the mechanical structures are static in time.

Therefore, my plan is to have a time-dependent solver for solid mechanics, and for each time step, I calculate the EM part using RF module's frequency domain solver, with moving mesh (ALE) connecting the solid mechanics and EM parts. I then have an expression that will calculate force from EM field configuration and use it as the load in solid mechanics for the next time step.

My question is how do I set up the solver as described above. How can I have the frequency domain solver at each time step of the time-dependent solver?

Thank you in advanced. My version is 4.0. I wouldn't mind if I really have to use livelink to Matlab or something along that line.


7 Replies Last Post 25 apr 2013, 00:21 GMT-4
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 7 set 2010, 02:48 GMT-4
Hi

a few commens on your model:
1) I'm not sure the air should be part of your SolidMechanics analysis
2) I'm neither not sur how the RF forces at that frequency can act on your srtucture, the structural eigenmodes starts at 0.8E8Hz, while your RF freuqnecy is six orders of magnitude higher, but OK I'm not 100% sure what you are testing anyhow, so I leave that to you
3) if you use a frequency domain solver you are in what I call the modal domain, you define an amplitude and the equations are developped for continuous sinus type excitation of frequency freq (and perhaps a phase factor, in same cases). The resulting forces are averages over a period, this is then what you couple to the other physics
4) When you have several physics, COMSOL cannot always tell with solver to use for which physics, specially when you mix solver types, then you can override the "automatic" choice in the <<main physics node - Advanced settings - Equation form>>: change from Automatic to i.e. frequency domain or static.
5) when you define your physics, and add a new one you should then define the default solvers, you intend to use
6) n the solver tree you might have to set up the sequence manually to get a segregated sequence, all depends on what you are doing

--
Good luck
Ivar
Hi a few commens on your model: 1) I'm not sure the air should be part of your SolidMechanics analysis 2) I'm neither not sur how the RF forces at that frequency can act on your srtucture, the structural eigenmodes starts at 0.8E8Hz, while your RF freuqnecy is six orders of magnitude higher, but OK I'm not 100% sure what you are testing anyhow, so I leave that to you 3) if you use a frequency domain solver you are in what I call the modal domain, you define an amplitude and the equations are developped for continuous sinus type excitation of frequency freq (and perhaps a phase factor, in same cases). The resulting forces are averages over a period, this is then what you couple to the other physics 4) When you have several physics, COMSOL cannot always tell with solver to use for which physics, specially when you mix solver types, then you can override the "automatic" choice in the : change from Automatic to i.e. frequency domain or static. 5) when you define your physics, and add a new one you should then define the default solvers, you intend to use 6) n the solver tree you might have to set up the sequence manually to get a segregated sequence, all depends on what you are doing -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 7 set 2010, 15:07 GMT-4
Dear Ivar,

Thank you for a quick response. I actually forgot to mention one important thing, that is the force from the EM field is independent of phase, so that means the force is static as long as the structure does not move much (ie. the force does not oscillate at the frequency for the EM wave but will only change roughly at the same rate the structure moves physically). This independence of the force on the phase is the justification to use frequency domain for EM wave part. (For more concrete example, if I calculate the force to be 1N in +y direction, that means 1N steady in+y, not 1N*cos(omega t) in +y. )

I tried two solver steps. I set the first solver to be RF frequency domain solver to solve the EM field and I wrote an expression to calculate the force (remember it's phase independent, so it's just a static number with no time-dependence). Then I set the second solver to be time-dependent solid mechanics solver. It is my understanding that if I set the solvers like this, after the 1st solver is done, the 2nd solver will just take the force from the first solver and use the force as a steady force for all time steps, without updating the 1st solver with deformation at each time step, am I right? That is, after 1st solver is done, 2nd solver won't go back to 1st solver at all, am I right?

My goal is to have the time-dependent solver, at each time step, go back and update the RF frequency domain solver with the deformation, and have the RF solver recalculate the EM force (because the field distorts due to the structure's movement), and use this new force for the next time step in the time-dependent solver. How can I do that?

Sorry for the lengthy post but this problem is hard to describe.

Thank you,
Varat Intaraprasonk

Dear Ivar, Thank you for a quick response. I actually forgot to mention one important thing, that is the force from the EM field is independent of phase, so that means the force is static as long as the structure does not move much (ie. the force does not oscillate at the frequency for the EM wave but will only change roughly at the same rate the structure moves physically). This independence of the force on the phase is the justification to use frequency domain for EM wave part. (For more concrete example, if I calculate the force to be 1N in +y direction, that means 1N steady in+y, not 1N*cos(omega t) in +y. ) I tried two solver steps. I set the first solver to be RF frequency domain solver to solve the EM field and I wrote an expression to calculate the force (remember it's phase independent, so it's just a static number with no time-dependence). Then I set the second solver to be time-dependent solid mechanics solver. It is my understanding that if I set the solvers like this, after the 1st solver is done, the 2nd solver will just take the force from the first solver and use the force as a steady force for all time steps, without updating the 1st solver with deformation at each time step, am I right? That is, after 1st solver is done, 2nd solver won't go back to 1st solver at all, am I right? My goal is to have the time-dependent solver, at each time step, go back and update the RF frequency domain solver with the deformation, and have the RF solver recalculate the EM force (because the field distorts due to the structure's movement), and use this new force for the next time step in the time-dependent solver. How can I do that? Sorry for the lengthy post but this problem is hard to describe. Thank you, Varat Intaraprasonk

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 8 set 2010, 01:15 GMT-4
Hi

then I believe we are basically in line and agree, and I would also use your approach, for me this is a compined solving but with a segregation sequence EM - ST, so you should look after the setting of a multiphyiscs segregation solver settings.

I would have to set it all up and test it out mayself, but I have no time now, nor in the coming days, so I can only suggest to take a look at the solver doc, and try to find another example in the library having two coupled physics with a segregated solver example

--
Good luck
Ivar
Hi then I believe we are basically in line and agree, and I would also use your approach, for me this is a compined solving but with a segregation sequence EM - ST, so you should look after the setting of a multiphyiscs segregation solver settings. I would have to set it all up and test it out mayself, but I have no time now, nor in the coming days, so I can only suggest to take a look at the solver doc, and try to find another example in the library having two coupled physics with a segregated solver example -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 19 ott 2011, 04:14 GMT-4
Dear Varat and Ivar,

Did either of you get this working? I have the same dilema but slightly different physics - I am trying to get the distribution of charges (from transport of diluted species) as a result of an electric potential over time, so that the charge distribution updates the electric potential at each time step. Any help will be greatly appreciated!

Regards,

Richard
Dear Varat and Ivar, Did either of you get this working? I have the same dilema but slightly different physics - I am trying to get the distribution of charges (from transport of diluted species) as a result of an electric potential over time, so that the charge distribution updates the electric potential at each time step. Any help will be greatly appreciated! Regards, Richard

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 20 ott 2011, 11:30 GMT-4
I could not get it to work.
I could not get it to work.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 apr 2013, 22:35 GMT-4
hello, i have the same problem, you get it to work now?
hello, i have the same problem, you get it to work now?

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 apr 2013, 00:21 GMT-4
No, I already gave up. If anyone stumbles upon this question again, I already gave up on this problem and won't have a solution. If you have a solution, please post it though because I also want to see it.
No, I already gave up. If anyone stumbles upon this question again, I already gave up on this problem and won't have a solution. If you have a solution, please post it though because I also want to see it.

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.