[Guava Wang]

Dear all,

Now my simulation model is that 10mm cyclinder with 20µm little radial clearance. I want to calculate the fluid is through the clearance condition.

-which turbulent model and wall function is suitable to use?
- how can i check mesh size in layer position with y plus in a reasonable range?

hope someone can help me. thank you in advance

[FMDenaro]

Originally Posted by Guava Wang

Dear all,

Now my simulation model is that 10mm cyclinder with 20µm little radial clearance. I want to calculate the fluid is through the clearance condition.

-which turbulent model and wall function is suitable to use?
- how can i check mesh size in layer position with y plus in a reasonable range?

hope someone can help me. thank you in advance

I think that in microfluidic you have more relevant effects by viscosity and capillarity than worry about turbulence. I don't think you have a Re number sufficiently high to substain turbulent structures....

[Guava Wang]

Originally Posted by FMDenaro

I think that in microfluidic you have more relevant effects by viscosity and capillarity than worry about turbulence. I don't think you have a Re number sufficiently high to substain turbulent structures....

hi FMDenaro,

In the radial clearance, the velocity is 30m/s, which is turbulent structure. now my main issue is how to adapt the mesh size in reasonable range to assure the calculation accurate.

have a nice day
guava

[FMDenaro]

Originally Posted by Guava Wang

hi FMDenaro,

In the radial clearance, the velocity is 30m/s, which is turbulent structure. now my main issue is how to adapt the mesh size in reasonable range to assure the calculation accurate.

have a nice day
guava

Hi,
I am not an expert about microfluidic, I just see from some estimation, velocity and lenght of 20µm turn out to produce a small Re number. What about the axial velocity? I doubt that the energy production can balance the energy dissipation to substain turbulent structures and produce inertial range in the energy spectra... but I could be in error

[flotus1]

What kind of fluid are you simulating? How high is the viscosity?

Estimating Re with an approximate value of air (1.5 e-5 m^2 s^-1) yields Re=40.

So your flow is definitely laminar (assuming the fluid you use doesn't have a lower viscosity than air) since the critical Re is in the range of 2300 for a circular pipe.

[Guava Wang]

Originally Posted by flotus1

What kind of fluid are you simulating? How high is the viscosity?

Estimating Re with an approximate value of air (1.5 e-5 m^2 s^-1) yields Re=40.

So your flow is definitely laminar (assuming the fluid you use doesn't have a lower viscosity than air) since the critical Re is in the range of 2300 for a circular pipe.

hi flotus1,
viscosity is 2.24e-07 m^2 s^-1, and i have check the Re number, which is larger than 2300 in the clearance.
I can't upload the model as a picture, maybe the problem didn't give a clear description. my proposal for this case is to know the leakage with piston through the clearance 20 µm.

have a nice day
guava

[flotus1]

Ok sorry for the detour, now lets get back to your initial question.

Since Re is still pretty low, the computational cost of fully resolving the boundary layer without a wall-function should not be too high.
Try to get the Y+ values below 1 to achieve this with a small growth factor (1.05-1.1) in wall-normal direction.

Thanks to the low Re again, the choice of the turbulence model will not be too important in this simulation. Any of the basic k-epsilon formulations should yield reasonable results.

[FMDenaro]

Originally Posted by flotus1

Ok sorry for the detour, now lets get back to your initial question.

Since Re is still pretty low, the computational cost of fully resolving the boundary layer without a wall-function should not be too high.
Try to get the Y+ values below 1 to achieve this with a small growth factor (1.05-1.1) in wall-normal direction.

Thanks to the low Re again, the choice of the turbulence model will not be too important in this simulation. Any of the basic k-epsilon formulations should yield reasonable results.

I agree...what's more, your Re_tau is quite low such as a DNS is realizable

[Guava Wang]

thanks. finally i adapt K-E model with scalable wall function, that can give good result.