[CXI]

Dear all,

I am a new user to OpenFoam, so please forgive me, should I post something silly.

Recently I have been trying to validate my case by using flow around cylinder as reference, however, I cannot seem to get any reasonable values for Cd, neither for low nor high Reynolds numbers. FYI, the way I fix Re is simply by varying velocity (from the formula Re=l*v/nu).

For this simulation I use a cylinder with diameter and length of 40mm.

The Reynolds number is fixed at ~90, with nu being 0.000015 and calculated velocity of 0.03375 m/s.
Naturally, I have also edited forceCoeffs file with the appropriate values.

Additionally, I am using simpleFoam with kOmegaSST turbulence model.

If someone could look at this case, I would really appreciate it. Also, do I need to change turbulent viscosity (nut) values, to obtain the correct results?

Finally, please find the case files below.

http://www.filedropper.com/90tar

Thank you for your help!


EDIT: I managed to increase the precision of my results by applying a different (presumably correct) turbulentKE value, however, I have another question now.

How do you calculate the turbulent length scale?

Also, are the formulas below correct, when used with simpleFoam?

turbulentOmega = k / epsilon

k = 1.5 * (UI)^2

I = 0.16 Re ^-1/8.

epsilon = cmu*k^1.5/l (where l is turbulent length scale)

[wouterremmerie]

Hi CXI,

interesting problem!

This post also deals with those quetions. Perhaps you could look at:
- your wall function settings: are you using these? And what are your Y+ results? The posts states that for small Y+ values you should not use any wall function at all. Does it make a difference if you turn off the wall functions (if you have these)?
- perhaps the safer choice for calculations is k-epsilon, as it may be less sensitive to these settings?
- small value of epsilon instead of 0?
- have you tried adding layers to the cilinder, using snappyHexMesh? under addLayerControl?


Which settings are you using in your fvSchemes file? First or second order interpolation? upwind, central differencing, ...? Not an expert on this, but those can cause errors as well.


this link gives some more info on when to use wall functions. For example:
"Wall functions should only be used for 30 < y+ < 100-300, which means that the first cell is in the log layer and not in the sublayer. If you want to use kOmega_SST without a wall function, y+ should be around 1. "



This link states that initial field settings are used for the first time step, but should be slave over time to what you specify at the inlet.

This link then, gives more information on how to specify boundaries at the inlet. For example, instead of using reynolds number, which is sometimes difficult to determine objectively, you could use turbulence intensity values and others.


Looking forward to your and other peoples feedback!

[davibarreira]

Just complementing the response given by Wouter.

The flow past a circular cylinder may seems as a simple case to validate, but it is actually quite complex. Some papers published the results trying to validate k-wSST for flow past a cylinder at various reynolds, but the results are not very good (I actually tried myself). I'm posting here my results using three different RANS models, so perhaps it saves you some time. The main problem is in the intermediate values of Reynolds, in which the boundary layer is partially laminar and partially turbulent.

Regarding the boundary conditions. For external flow, you usually use turbulence intensity of something around 0.5% ~ 1%, then you calculate your k, epsilon and omega. The length scale is 0.07*L, in which L is the characteristic length (in this case, L is the diameter of your cylinder).

Let me know if you make any progress, cause Im still working on this case.

[app1e]

I think maybe you can try laminar flow first