[Scarpafico]

Hey,
I’m working on an external flow around a small submarine for quite some time now. I tried a lot of different settings and read many papers and posts here. I must say: a great forum!
Unfortunately I’m stuck now and need some help. My Problem:
I have experimental date which I cannot match! 64N was measured. Ansys CFX was easily able to get as close as 68N with SST. The closest I could get is 74N drag with Spalart-Almaras and 82N with komegaSST.
I optimized my geometry a few times and got good convergence in most of the cases. I tried initialization with potentialFoam or higher fvSchemes from the beginning with not much improvement. I used the motorBike-case as reference and included some higher fvSchemes after some time. Some of my settings/results for an example case:

• OpenFOAM 2.3.1
• Inlet water velocity 2m/s
• komegaSST with k=0,0001588 and omega 0,0382
• Y+ 12 … 22 (small areas with y+=1)

Please find the full information (fvSchemes, fvSolution, CheckMesh, p, U, omega, residual and force plot etc.) attached. Unfortunately I cannot post a geometry plot, but imagine a small torpedo-shaped submarine 3,5m long and with a diameter of maybe 0,5m.

I hope someone out there can give me a hint, how to get my drag force closer to the measured 64N!

PS: I calculated omega backwards with the cfd-Online-Tools by using TuL = 0,3m (10% of the body length was recommended somewhere here in the forum) and k=0,0001588. The simulations show a final omega of roughly 2000 … 5000 and k=2.8e-6 … 0.0381 in ParaFoam. Could that be the problem?

Cheers,
Christian

[joshm]

Could it be due to your y+ falling in the buffer region? You'reusing the nutkWallFunction which is for y+ in the region 30~500 or so.

[Scarpafico]

Thanks for the hint. I started with Y+ ~ 1 in my first calculations. I'm at 30 to 50 now, but this calculation needs still some more time to converge! However, due to flow speed at the tip and the stern (aft), small faces and geometry curvature I still have areas with Y+ close to 5! Should I expect a larger impact from these small areas?

I will try to increase the boundary layer further without getting to much non-orthogonality

[Scarpafico]

I managed to increase Y+ to the range of 40 to157. However the bow and some minor areas in the back are still around 14. I will try to fix this next week.
The drag force decreased to 62N, which would be fine when the pressure force wouldn’t be twice as high as the viscous force. Actually it should be the other way around!

Compared to the example case above I didn’t use cellMDLimited this time:

 
divSchemes
{
/* default Gauss linear;
div(phi,U) bounded Gauss upwind;
div(phi,omega) bounded Gauss upwind;
div(phi,k) bounded Gauss upwind;*/
//changed after 4500 iterations:
div(phi,nuTilda) Gauss linearUpwind default;
div(phi,U) Gauss linearUpwind grad(U);
div(phi,k) Gauss linearUpwind default;
div(phi,omega) Gauss linearUpwind default;
 
div((nuEff*dev(T(grad(U))))) Gauss linear;

and used limited 1.0 instead of 0.5

 
laplacianSchemes
{
//default Gauss linear corrected;
//changed after 4500 
default Gauss linear limited 1.0;
}

Can someone please help me to get out of this mess?!