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 The information provided may be out of date.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.

Biasing a capacitive membrane (CMUT) and applying a superimposed AC signal

Please login with a confirmed email address before reporting spam

Hi everyone,

I am trying to perform a simulation in which I first electrostatically bias a membrane, and then apply an AC signal to the biased membrane, which should in turn cause the membrane to vibrate and radiate a pressure wave into a surrounding fluid medium (water). I have been able to create simulations that statically bias the membrane using electrostatics, plain strain, and ALE (similar to the ALE cantilever example). I have also been able to create a simulation that applies a pressure loading to the membrane which causes it to radiate waves into a fluid half-space. However, coupling these two scenarios seems much more complex.

The ALE cantilever beam describes a method involving a femssr wrapper which should effectively do everything that I need (take a biased solution and superimpose a small-signal AC voltage). Following this process, I get somewhat expected results. Specifically, there is a phenomenon known as spring-softening which essentially implies that as you apply higher bias to a capacitive membrane, its center frequency will shift lower. I observe this effect using the femssr wrapper. Also, when I compare my results using water and air as the fluid medium, I see a pronounced fluid loading effect that shifts the water-loaded membrane to a lower center frequency, as expected. However, I have noticed that the frequency response in both water and air are very low bandwidth (high-Q) systems. I would somewhat expect this in air, but water should be more "damped" and I do not observe this using the femssr method. On a related note, I do observe the high-Q in air, low-Q in water effect when I run my more basic harmonic analysis with no electrostatic bias, so I believe something is wrong with this femssr method.

I have also tried a multi-step approach in the Comsol GUI. First, I bias the membrane as usual. Then, I define a new plain-strain application and a new electrostatic application, and use the biasing simulation results as my linearization point for a harmonic analysis. Thus far, this has only given me results similar to a typical harmonic analysis, in that I do not observe spring-softening. But it is possible that I have not coupled the two simulations together correctly.

In summation, I really just want to know the basic approach to accomplish what I am requesting--to bias a membrane electrostatically, and then superimpose an AC signal such that it couples to the structural domain and causes pressure waves to be radiated. From reading through the various threads, it seems that other people are having similar problems, but as of yet I haven't found a definitive solution. Any advice or recommendations would be appreciated. Thanks!

6 Replies Last Post 2011年12月2日 GMT-5 01:41
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 2010年9月11日 GMT-4 07:55
Hi

from what I read it's also how I approach such cases (apart that for me I call the change of stiffness of bars or membrane for "stress-stiffening") you run one case and use the result as initial conditions for the next run on the next physics set, in a somewhat manual segregated way.

In V4 there are very powerful ways of combining the solvers, but I'm still learining how to add your own detailed solver sequences correctly ;)

For the lack of structural damping due to the fluid it is probably a missing variable coupling the physics (do you need a complex variable to get phase information through, perhaps ?)

Finally running a (damped?) modal (harmonic) case upon the loaded system, telling COMSOL to linearise at the starting point is another thing I havnt tried extensively yet (apart for the structural buckling analysis where it's set up "by default". Perhaps a setup to use as example for you.

In anycase, pls keep us informed, we are many out here wanting to hear how such simulations works for the very many ways of coupling all physics ;)

--
Good luck
Ivar
Hi from what I read it's also how I approach such cases (apart that for me I call the change of stiffness of bars or membrane for "stress-stiffening") you run one case and use the result as initial conditions for the next run on the next physics set, in a somewhat manual segregated way. In V4 there are very powerful ways of combining the solvers, but I'm still learining how to add your own detailed solver sequences correctly ;) For the lack of structural damping due to the fluid it is probably a missing variable coupling the physics (do you need a complex variable to get phase information through, perhaps ?) Finally running a (damped?) modal (harmonic) case upon the loaded system, telling COMSOL to linearise at the starting point is another thing I havnt tried extensively yet (apart for the structural buckling analysis where it's set up "by default". Perhaps a setup to use as example for you. In anycase, pls keep us informed, we are many out here wanting to hear how such simulations works for the very many ways of coupling all physics ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年9月15日 GMT-4 14:23
For those interested, I received a response from Comsol support. It turns out that superimposing an AC signal with a DC bias can only be performed using the femssr wrapper right now. However, the femssr wrapper is not optimized for coupling with all modules, which may explain why it is behaving strangely when coupled with acoustics.

They said that Comsol 4.1 (due out in the fall) should have increased functionality regarding this problem. So in other words, I'm out of luck for now...

Mike
For those interested, I received a response from Comsol support. It turns out that superimposing an AC signal with a DC bias can only be performed using the femssr wrapper right now. However, the femssr wrapper is not optimized for coupling with all modules, which may explain why it is behaving strangely when coupled with acoustics. They said that Comsol 4.1 (due out in the fall) should have increased functionality regarding this problem. So in other words, I'm out of luck for now... Mike

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年5月23日 GMT-4 14:28
Hi

i would be glad if maybe you could help me cause i think it connected to this topic

Attached a simple magnet model

what i want to do is to define:

magnetic field:
Hy = 10*sin(2*pi*freq*t)[A/m]

and to see the result in time domain maybe by 1D graph or even to see the streamlines changing in time

is it possible?

how should i define this magnetic field to act like this on this magnet?

Best regards,
Arye
Hi i would be glad if maybe you could help me cause i think it connected to this topic Attached a simple magnet model what i want to do is to define: magnetic field: Hy = 10*sin(2*pi*freq*t)[A/m] and to see the result in time domain maybe by 1D graph or even to see the streamlines changing in time is it possible? how should i define this magnetic field to act like this on this magnet? Best regards, Arye

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年5月23日 GMT-4 14:30
attached
attached


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 2011年5月24日 GMT-4 01:47
Hi

to model a magnet you need to leave some "space" or "vacuum" or "air" for the field lines to loop around, you cannot "just" model a single magnet, so start enclose your magnet in a sphere 3-4 times greater than the magnet and give it a sigma of 0[S/m] and epsilonr = 1

Then you should study the difference between a harmonic solver and a transient solver case, as I'm not sure what you are really looking after, for the harmonic development you define only the amplitude and phase (in addition to the frequency) and solves much quicker, this is what one uses for a developed time analysis, while the "transient" is mostly there for a study of a "transient" (not repetitive) effect. By the way the time by default in Comsol is "t[s]"

--
Good luck
Ivar
Hi to model a magnet you need to leave some "space" or "vacuum" or "air" for the field lines to loop around, you cannot "just" model a single magnet, so start enclose your magnet in a sphere 3-4 times greater than the magnet and give it a sigma of 0[S/m] and epsilonr = 1 Then you should study the difference between a harmonic solver and a transient solver case, as I'm not sure what you are really looking after, for the harmonic development you define only the amplitude and phase (in addition to the frequency) and solves much quicker, this is what one uses for a developed time analysis, while the "transient" is mostly there for a study of a "transient" (not repetitive) effect. By the way the time by default in Comsol is "t[s]" -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月2日 GMT-5 01:41
hey do you mind attatching your file?
hey do you mind attatching your file?

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.