[Gabriel Roudi]

Hello,

I am trying to simulate flow in a synthetic jet actuator. Inlet velocity of the cavity is made by a vibrating membrane with an approximate frequency 1000 Hz (makes suction and ejection). Outlet velocity is generating outside of the nozzle which changes periodically over time. I have chosen below Physics for the simulation however the exit velocity is not matching with the experimental results. I would be grateful about your advises.

Coupled Flow, Ideal Gas, Implicit Unsteady, K/Epsilon Turbulence, RANS, Axisymetric, moving boundary condition (morphing-velocity grid) in the inlet, polygonal fine mesh

best regards,
Gabriel

[me3840]

You need to give more information. What does the mesh look like? What is the timestep? How did you treat the membrane condition? How far off are the results? Is the membrane producing what the experiment did?

[Gabriel Roudi]

Hi,

- The mesh is attached for your attention.
- Time step depends to the membrane frequency and is obtained from below formula according to the literature.
Delta t = 1/(fr*1000)
if frequency is 32 --- time step=0.00003125

- I used moving boundary condition (Morphing - Velocity grid) to treat the membrane.

- The exit velocity which i need to get is close to 3.2 m/s. However I get different velocity with different setup.
Coupled flow - different CFL No. different velocity exit (0.7 to 6 m/s)
Coupled flow - different time step different velocity exit
Coupled flow - different Max Inner Iteration different velocity exit
Segregated flow - totally different velocity exit (12 m/s)

- Yes, the velocity field seems to be logical.

I am confused which flow physics or time setup shall i utilize. I would be grateful if you could advice.

best regards,
Gabriel

[me3840]

Your mesh needs serious work. You can't have that many prisms and have them all collapse like that. You need to have prisms ratioed correctly so that you have a good y+ value, good resolution through the boundary layer, and a good transition ratio to core cells. With your various volume refinements, you likely need to split your walls up so you can define different prism settings on each wall. Your volume growth rates are very large, you might need to smooth them out, but this seems pretty low Re too.

How well is the moving membrane emulating the experimental version?

[Gabriel Roudi]

How are the meshes now?

[Gabriel Roudi]

Moving membrane deflection is simulated with the attached formula. I set in the User Field Function.

[me3840]

That is much better, but I still think the prisms are too thick on the bottom portion of that wall.

That equation is experimentally derived or? How do you know it represents the physical test well?

[Gabriel Roudi]

There is only maximum deflection measurement available in center of the membrane during process. The rest I utilized the mentioned equations for deflection input.