[Yuby]

Hi FOAMers!

I am running a turbulent case with k -epsilon model in a wind tunnel. However, I don't know how to calculate epsilon for a external aerodynamic flow. Other threads link to http://www.esi-cfd.com/faq/index.php...156&artlang=en, however, this gives me unrealistic values for the turbulent dissipation due to the small value of dynamic viscosity. What can I do?

I have found other expressions but in every one appears one geometry parameter (usually related with the mixing length) but I am looking for the inlet boundary condition of epsilon.

And, another question: some tutorials use the same value of the inlet in order to specify the wallfunction value of epsilon in their geometry. When this is correct?
My geometry is a rotating disc, if you need to know.

Thank you very much in advance!

[Blanco]

Hi,

I think you could use this tool to estimate eps value at your inlet, but in any case it is linked to k and I, and you should have done some assumption about turbulence intensity at your inlet...

http://www.cfd-online.com/Tools/turbulence.php

For the second question, usually the eps value at a wall boundary is used just to initialize the sim, unless you would use a fixedValue but I don't think its your case since you mentioned wallfunctions. If you are going to use wall functions then the initial value is used only to initialize the sim. It usual practice to help the convergence by using an eps value quite high at the wall, calculated using wilcox or menter formula for omega as an example, and then deriving eps.

Best

[Yuby]

Thank you for your anser Andrea!

This tool is useful, but it has the same problem, an epsilon too big for a small cinematic viscosity. This tool is based on the formulas of http://www.esi-cfd.com/faq/index.php...156&artlang=en.

Anyway, does anybody know where that formulas come from? In the link there is no reference... and I haven't found them in any paper

[Blanco]

Just for reference...what is your cinematic viscosity? It must be noted that also turbulent viscosity comes into play in eps evaluation, and you have to assume a viscosity ratio...what have you assumed? I found the link you followed pretty useful for my external aerodyn sims in the past, even if I was using k-omega model insted of k-eps. I used intensity equal 0.5% and beta 0.1, with these I get consistent results with experiment.
There is also a nasa paper on selecting boundary conditions properly for external aerodyn sims, but I don't have the reference here now...

In any case, those formulas are coming from the k-eps model and k-omega models, you can find a brief description also on cfd-online. Usually some coefficients are model constants, as 0.09 = cmu in keps model. The link in the tool page should refer to used formulas.


Best

[Yuby]

Thank you Andrea.

I have a turbulence intensity of 5% (maybe too high for a flow around a frisbee, what do you think?) a beta equal to 0.1 and a cinematic viscosity of 1e-05. It's not exact (1.5e-5 is the cinematic viscosity of the air, if I remember properly) but it's because I am using pi-Buckingham and adimensionalization.

Thank you for your help indeed. It's for my thesis, and I have been facing this problem for days.

[Blanco]

You're welcome Ruben.

Yes I think that 5% it's a little bit too high for an external aerodynamic simulation, I would suggest you to move to 1% or 0,5% better (look also at suggested values at the bottom of my linked page, in the previous post).

1,5e-5 m^2/s it's correct cinematic viscosity for air at ambient temp, since beta is 0,1 you will obtain nut equal to 1,5e-6 m^2/s in your results, which is ok (1e-6 m^2/s in your case). Eps value won't be high but that's correct because we are considering the eps value on the inlet boundary, and eps will be low here (freestream undisturbed flow ahead of the object). Eps will be much higher at the wall boundary, you can use menter-wilcox formula to estimate it.

Do you have experimental data to compare with? What is your Reynolds number?

Best

[Yuby]

I have moved my turbulence intensity to 0.5%, and the tool gives me this values:
k=0.024346140000000002 m²/s²
epsilon=53.346107960964 m²/s³
omega=24346.139999999996 Hz

Do you think that they are good and real values? My Reynolds number is 254800.

That's the problem, I haven't found any experimental data, but in this simulation case http://www.sciencedirect.com/science...77705814006328, they say that hydraulic diameter for their frisbee is 0.1 m (Is related to the mixing lenght in external flows?).

0.1 m in that paper is the radius of the frisbee, so I don't know why they choose that number to be the hydraulic diameter.

I am pleased

[Yuby]

Maybe this is the NASA paper you were talking about:

http://ntrs.nasa.gov/archive/nasa/ca...0070035069.pdf

[Blanco]

Yes That's exactly the paper I was talking about. I found it very useful

[Blanco]

Originally Posted by Yuby

I have moved my turbulence intensity to 0.5%, and the tool gives me this values:
k=0.024346140000000002 m²/s²
epsilon=53.346107960964 m²/s³
omega=24346.139999999996 Hz

Do you think that they are good and real values? My Reynolds number is 254800.

That's the problem, I haven't found any experimental data, but in this simulation case http://www.sciencedirect.com/science...77705814006328, they say that hydraulic diameter for their frisbee is 0.1 m (Is related to the mixing lenght in external flows?).

0.1 m in that paper is the radius of the frisbee, so I don't know why they choose that number to be the hydraulic diameter.

I am pleased

Hi Ruben,

I didn't have time to read the paper you cited, I hope I'll have time soon. In any case yes the number you wrote are correct if you consider nu = 1e-5 m^2/s. By the way, why are you using 1e-5 instead of 1.54e-5 m^2/s for air? That value is representative of air at 220 K, nearly -50°C... Note that turbulence parameters (and others) change drastically with kinematic viscosity, therefore small changes can be important (another example, if dynamic viscosity is kept constant, going with nu from 1.54e-5 to 1e-5 means that density of air is changing from 1.185 kg/m^3 to...1.7 kg/m^3...and that's a LOT for air).

Best

[Yuby]

Hi,

I am using that value of cinematic viscosity because I am using Pi-Buckingham theorem.
I am working with my reference case, and I will adimensionate it and apply for other cases.

Do you think that Pi-Buckingham can't be used in turbulence cases?

Thank you for your help