[maruthamuthu_venkatraman]

Hello Foamers,

Iam trying to predict the vortex shedding for Cylinder at Re 1.2E+05 using Low Re KE model Launder and sharma .

Observed Results :

1. Drag coefficient is not at all oscillating in a regular pattern even after 20 Shedding periods. Finally it gives me a constant value of 0.2.

2. Lift coeffcient starts to oscillate in the begining and then it becomes constant with a value of 1.5 after 6 vortex shedding periods.

3. Yplus is less than 1 , All turbulent quantities are set to 1e-10 for cylinder walls. Symmetry BC is used on the top and bottom domain.
Outlet is set to constant with 0 for pressure and all others are set to zero gradient.

4. k and e values are 0.35 and 310 at inlet.

5. Grid is fine enough to capture this frequency.


If anybody has Successfully predicted the shedding frequency at Re 1e+05 , Then shed some light upon this problem.

Residual for last time step.

Time = 74.2625

Courant Number mean: 0.0140374 max: 0.288887
DICPCG: Solving for p, Initial residual = 0.00921648, Final residual = 8.75815e-05, No Iterations 55
time step continuity errors : sum local = 6.25117e-11, global = 8.42844e-13, cumulative = -3.73129e-13
DICPCG: Solving for p, Initial residual = 0.00491658, Final residual = 4.72697e-05, No Iterations 57
time step continuity errors : sum local = 3.37288e-11, global = 8.02879e-13, cumulative = 4.2975e-13
DICPCG: Solving for p, Initial residual = 0.00082418, Final residual = 7.74048e-06, No Iterations 60
time step continuity errors : sum local = 5.51504e-12, global = 1.09791e-12, cumulative = 1.52766e-12
DILUPBiCG: Solving for epsilon, Initial residual = 3.74769e-06, Final residual = 3.74769e-06, No Iterations 0
DILUPBiCG: Solving for k, Initial residual = 5.534e-06, Final residual = 5.534e-06, No Iterations 0


Thanks

[sachin]

hi
I havent done simulations in this area but a small doubt in this simulations because your Reynolds no. is too high. Does Low Re k-eps model handle it?
can you try the solutions with high Re k-eps model...just a suggestions

Sachin

[maruthamuthu_venkatraman]

Hello Sachin,
Thanks for your Comments. ACtually i need to resolove the turbulent boundary layers compeletely with y+ less than 1. I realize that in OPENFOAM all High RE model are working with wall functions.

I dont know how to resolve turb boundary layers up to wall with High Re models in OpenFoam. If you know any tricks let me know...

Regards

[hjasak]

We have at least 3 low-Re turbulence models with integration to the wall - please have a look at what is implemented. I also think somebody spoke about this recently in the Forum.

Hrv

[maruthamuthu_venkatraman]

Hello Hrvoje,

I have gonethrough that thread how to incorporate the near wall treatment for low Reke models. That example he gave for Low Re ke model, Launder and sharma. I have followed it as it is.

i changed the default zero gradient for the walls and set the fixed values for k and e in the order of 1e-10.

But if you mean That i should read the thread how to impliment the Near wall treatment for High Re flows then I will go through it.

Shall i follow the same method for Realizable ke model as i do for low Re ke model in the BC settings.

i.e. setting a fixed value (very low ) for all turbulence quantities at walls as posted by someone... Clarify !

Thanks

[anothr_acc]

Hi,

I found that second order closure turbulence models were required for separation to occur at the correct angle for sub-critical Reynolds number flows around Re=33k. I am investigating OpenFOAM at the moment for exactly this application; I used Fluent, I'm afraid, for previous work, with the Reynolds Stress Model, resolving through the boundary layer, with time steps able to give more than 50 increments per shedding cycle.

BR,
Mark.

[sven]

I was also working on vortex shedding around a circular cylinder at re=90k. Even with a second order turbulence model (I used the standard k epsilon model and a especially to vortex shedding modified version of this model) i did not succeed in getting the correct angel of separation. Furthermore, I did not get a constant signal of CD and CL (drag and lift coefficients) over time. There was always a lot of noise in these signals. I would appreciate if you have any suggestions to me or if we could exchange our experiences! Have a good one.

Sven

[ivan_cozza]

Hi Foamers,
I'm doing some CFD on a 2D cylinder at Re = 1.4x10^5, rhoPisoFoam solver, k-Omega SST turbulence model, with wall function for mut (mutSpalartAllmarasWallFunction) and omega, and fixedValue k = 1x10^12.

I0m able to obtain stable CD and CL for very low values of Co (0.2), the separation angle and St frequency are captured with an error of 20% with respect to experiments, and the average CD value are around 0.2 (from the experiments, this must be order of 1).

Can anyone share some info on how to obtain better results?

Thanks!