[seadmiral]

After searching the forum, I have seen relatively limited discussion regarding validation of pressure distributions via simpleFOAM with experimental data.

I have a converged simpleFOAM result for 2D, incompressible, laminar, and attached flow about a thick airfoil. Using ptot as total pressure at the inlet, I have defined cp,

cp = 2(p* - ptot*)/(U0^2)

where * denotes density-normalized pressure values from simpleFOAM, p*=p/rho, ptot*=ptot/rho, and U0 is the velocity at the inlet.

Plotting these coefficients gives a puzzling non-smooth profile (Fig. attached).

Has anyone else observed this for a converged case? Any suggestions are much appreciated.

[seadmiral]

Thought I'd reply to my own post in case anyone else needs help in the future...

I found my surface geometry to be way too coarse, especially near the leading edge. Once I reverted to B-Splines in my gmsh setup to refine the initial STL file, I had success with snappyHexMesh in building the 2D grid. Then with a simple RANS (k-omega) turbulence model I was able to very closely match experimental pressure profiles.

The key is surface refinement. It is probably overkill as I haven't run any grid sensitivity study yet, but I used 1000 points around the leading edge. My STL file for a simple airfoil is 64MB...but I have good results

[sail]

Originally Posted by seadmiral

Thought I'd reply to my own post in case anyone else needs help in the future...

I found my surface geometry to be way too coarse, especially near the leading edge. Once I reverted to B-Splines in my gmsh setup to refine the initial STL file, I had success with snappyHexMesh in building the 2D grid. Then with a simple RANS (k-omega) turbulence model I was able to very closely match experimental pressure profiles.

The key is surface refinement. It is probably overkill as I haven't run any grid sensitivity study yet, but I used 1000 points around the leading edge. My STL file for a simple airfoil is 64MB...but I have good results

sorry, I've just seen your post after you've resolved your problem.

but for future reference, about 100-150 points in the cordwise direction should suffice, given that they are clustered near the LE.

best regards.

[seadmiral]

Again with the hope of solving someone's problem in the future...I certainly have learned heaps from this forum.

Attached are the laminar simpleFoam & RANS results, compared to test data. There are certainly some more tweaks but the preliminary result is not bad.

@sail: Perhaps turbulence was a bigger deal here than I initially gave credit. However, the mesh is certainly much improved from the first plot.

Thanks for your tip, I'll be sure to try it out.

[s.m]

Originally Posted by seadmiral

Thought I'd reply to my own post in case anyone else needs help in the future...

I found my surface geometry to be way too coarse, especially near the leading edge. Once I reverted to B-Splines in my gmsh setup to refine the initial STL file, I had success with snappyHexMesh in building the 2D grid. Then with a simple RANS (k-omega) turbulence model I was able to very closely match experimental pressure profiles.

The key is surface refinement. It is probably overkill as I haven't run any grid sensitivity study yet, but I used 1000 points around the leading edge. My STL file for a simple airfoil is 64MB...but I have good results

Hi,
i have a smilar problem, would you please guide me?
i am working on multi element airfoils, my pressureCoeffs figures for 3 element separately are in the attachment. As you see the pressureCoeffs has flactuation in three of them, whould you please explain more that how should i refine my STL file to have a smooth pressureCoeffs figure.
Thank you very much
your answer may help me a lot.