FSI Analysis - High frequency problem?

[thanhvan]

Hello everybody,

I am doing the FSI analysis in 3D model. In my model, the solid boundary will move with a particular freqency and amplitude, when this boundary moves it causes the boundary of the fluid change, so that we have something like a pump.
At the frequence f=100Hz, I got a nice result, i.e the flux of the flow is considerable large. However, when I increase f=7000 Hz (which is the same with my device), I have no flux of the flow (it's too small compare with f=100Hz, which is supposed to be smaller).
I got stuck with it for a long time, and have no idea how to solve it.
I am free to show all of you my model, please help me if you have experience.
Also, contact me anytime at dauthanhvan@gmail.com.
Regards.

 

[pja]

The coupled FSI system will have complex eigenvalues in the same way that a damped structural model will..more than likely 7000 Hz is not near one of these eigenvalues hence the response is small. Also make sure you are resolving the temporal scales well enough..if you are forcing this thing at 7000Hz you should have a coupling timestep(physical) of at least 1/(15*7000*2*Pi)

 

[thanhvan]

Thanks pja,

Sorry that I havent made my question clear. My problem is that the unreasonable results. This is the results of the pump 100HZ create much larger output than the pump 7000Hz. Though the solution all are convergent.
Would you give some more advice?

 

[pja]

And you didn't understand my answer..just because your answer converged doesn't mean that you are resolving all of the necessary temporal scales in the problem with your timestep..what is your timestep? Also what leads you to believe that forcing at 7000Hz should give you a larger response then forcing at 100Hz??

 

[thanhvan]

The simulation is modeled based on the prototype that I made. I also did the timestep smaller than the value you recommended. Thanks pja. Also, I assumed the maximum the value of the flow, and set the time step and mesh size so as to the flow will not skip any mesh size at every time step. I was also recommended to have a mesh size smaller enough so that the sound velocity is catched at each element, however, this will make the model so huge that I could not try.

Tthere is a nozzle which the fluid flows through, and the fluid is incompressible, laminar flow. This is a piezoelectric pump, however, I did not mention about the piezoelectric effect at this time.

Regards,
Thanh Van.