Learning Forum

[vkumar]

How can we decide 1st cell height for resolution of log layer in turbulent flow induced by bouncy effect (U is not known because we have kept the value of U=0 initially). 

[Karthik R]

Hello vkumar,

The first cell height is decided by the model you are planning on using for your analysis. If you are planning to use a model with wall functions, your y+ value should be between 30 and 60. If you are going to resolve the viscous sub-layer, your y+ values will have to be close to 1. Depending on these y+ values and your flow Reynolds number, you have to estimate the first layer thickness (y).

• Based on the definition of y+, to evaluate the value of y, you will need U_tau.


• U_tau can be estimated if you know your shear stress value.


• Shear stress can be obtained from the knowledge of skin friction coefficient and velocity scale. Depending on internal or external flow, you should be able to get a correlation for skin friction from any standard textbook. These correlations generally are dependent of the Reynolds number.

I hope this helps.

Best Regards,

Karthik

[vkumar]

Hi Karthik

Thank you for your response. 

But this problem is regarding turbulent natural convection, where we don't need the value of either Re or u since it is not valid, And we know that turbulent natural convection is decided by Rayleigh number and not Reynolds number. Now can you tell me the procedure to determine the value of first cell height in the turbulent natural convection over the plate.

[vkumar]

Hi Karthik

Thank you for your response. 

But this problem is regarding turbulent natural convection, where we don't need the value of either Re or u since it is not valid, And we know that turbulent natural convection is decided by Rayleigh number and not Reynolds number. Now can you tell me the procedure to determine the value of first cell height in the turbulent natural convection over the plate.

[paguado]

Hi vkumar,

Estimating a characteristic velocity in a natural convection-driven problem is difficult and rather inaccurate because even the simplified 2D equations yield a ODE that has to be solved numerically when the self-similarity approach is used. I would rather use a coarser mesh initially to provide a first estimation of y+. With this you can have an idea of the mesh resolution you need close to the wall. A second simulation with a more suitable mesh will give you more reliable information. 

[vkumar]

Hi, paguado,

Thank you very much for your response.

I  was trying hit and trial method for deciding  1st cell height to be in a range of 30

Can you please tell the procedure to resolve viscous sublayer (Y+<5) and log layer(30) at the same time in a single mesh design. 

[paguado]

Hi vkumar,

If you decide to resolve the viscous sublayer (y+ = 1) then you should check where y+ is larger in the precursor mesh you created. Then ensure you use enough mesh resolution in this region. This will probably lead to over-resolved regions in your domain if you keep this y+ with this conservative approach. A more economic approach is to use k-e with enhanced wall functions. This will adapt the production and dissipation terms according to the mesh resolution so it will model those regions where y+ is large and resolve those where your mesh is fine enough to capture the viscous sublayer.