Closed AndreWeiner closed 3 years ago
Hi @JihooKang-KOR,
a couple of minor comments on the latest commits:
# increase dpi and use latex rendering; include once at the beginning of each notebook
import matplotlib as mpl
mpl.rcParams['figure.dpi'] = 160
mpl.rc('text', usetex=True)
...
# typical figure with one subplot
fig, ax = plt.subplots(1, 1, figsize=(8, 3))
ax.plot(..., label=r"$c_f$")
ax.set_xlabel(r"$\tilde{x}$")
...
# tight layout for saving as PDF
plt.savefig("output/example.pdf", bbox_inches="tight")
Good work so far!
Best, Andre
Hi @JihooKang-KOR, I just finished checking the latest commits. Really great job! Everything nicely scripted and documented.
Regarding the results, I was surprised that the kink in the friction coefficient for small x-values did not disappear when switching off the wall functions. If you open the results with ParaView, you can see oscillations in the p and U fields. The kink in the shear stress is very likely related to those oscillations.
I tested a couple of changes in the settings and found sensitivity to the divergence scheme for U (entry grad(phi,U) bounded Gauss linearUpwind grad(U);
in fvSchemes). I would like to investigate this issue further before moving on to the wall modeling.
Could you test how the following divergence schemes affect the friction coefficient and the oscillations?
// to compare
div(phi,U) bounded Gauss linearUpwind grad(U);
div(phi,U) bounded Gauss upwind;
div(phi,U) bounded Gauss limitedLinear 1;
div(phi,U) bounded Gauss linear;
To reduce the investigated settings, I suggest the following:
Regarding the implementation, I suggest adding another Allrun file to turbulentFlatPlate_noWallFunc, possibly called Allrun.divSchemes, to test the different divergence schemes.
Best, Andre
Dear Jihoo,
thanks for the update on the different divergence schemes. It might be that the oscillations arise from the omega solution. I found similar results in some of the original NASA test cases for the k-omega-SST model. We'll evaluate the results in more detail during our next meeting.
Next, I would like you to test how the Spalart-Almaras (SA) model performs. The SA model is simpler (only one additional transport equation) and might be easier at the beginning for our undertaking. The following two resources will help:
Here are the main steps to modify the setup:
Please, test the setup with and without wall function, and let me know if issues come up. Again, great job so far!
Best, Andre
Hi @JihooKang-KOR,
to get started with this project, I suggest the following:
If questions come up, feel free to use this issue to discuss them.
Best, Andre