[NicoCFD]

Hi all,

I'm using Fluent double precision to simulate the airflow around a 3D Wing. I'm trying to estimate lift and drag coefficients for different angles of attack at Mach 0.7. I am using Spalart-Allmaras and SST turbulence models (I compare the two)

For now, I'm doing a grid convergence study by changing the wall spacing. I am using hybrid meshes that I created with Pointwise. I have structured blocks in the vicinity of the wing and in the wake, the outer part of my mesh is unstructured. I have results on my first meshes for average Y plus values of 22, 13 and 6, using the density based implicit pseudo-transient solver. However, as I try to run my simulation for lower Y plus values, I can't get converged results.

I've tried lowering the pseudo-time step, then disabling the pseudo-transient feature, by using an explicit relaxation factor of 0.25 and CFL number of 0.25, I've been able to perform a few 1st order iterations and lower the residuals to 10^-2. But when I switch to second order, I can't get the solution to converge, even by reducing the CFL number to values as low as 0.01 with an explicit relaxation factor of 0.1.

Has anyone encountered a similar problem, or knows what to do?
I believe my convergence issues are due to the high aspect ratio cells at the wall. Is there a way to refine the near wall cells in the tangential directions only to lower the aspect ratio?

Thank you in advance!

[sajad]

Originally Posted by NicoCFD

Hi all,

I'm using Fluent double precision to simulate the airflow around a 3D Wing. I'm trying to estimate lift and drag coefficients for different angles of attack at Mach 0.7. I am using Spalart-Allmaras and SST turbulence models (I compare the two)

For now, I'm doing a grid convergence study by changing the wall spacing. I am using hybrid meshes that I created with Pointwise. I have structured blocks in the vicinity of the wing and in the wake, the outer part of my mesh is unstructured. I have results on my first meshes for average Y plus values of 22, 13 and 6, using the density based implicit pseudo-transient solver. However, as I try to run my simulation for lower Y plus values, I can't get converged results.

I've tried lowering the pseudo-time step, then disabling the pseudo-transient feature, by using an explicit relaxation factor of 0.25 and CFL number of 0.25, I've been able to perform a few 1st order iterations and lower the residuals to 10^-2. But when I switch to second order, I can't get the solution to converge, even by reducing the CFL number to values as low as 0.01 with an explicit relaxation factor of 0.1.

Has anyone encountered a similar problem, or knows what to do?
I believe my convergence issues are due to the high aspect ratio cells at the wall. Is there a way to refine the near wall cells in the tangential directions only to lower the aspect ratio?

Thank you in advance!

Hi NicoCFD
I am working on the same problem (3d rotor blade) and facing same problem.when i decrease first cell width to lower yplus values i can't get converged residuals. in fact residuals doesn't go lower than e-2 and remains constant . I can decrease residual values by increasing first cell width but the yplus values increases to 600 that is unacceptable. i wonder if you've found the solution.
Regards