Turbulence model parameters and equations

Maximus91 · October 19, 2012, 11:02

Dear CFD users,

Right now I´m working on a CFD computation for external flow around a very streamlined car.
I have some questions regarding the turbulence parameters. I´ll first specify what formulas I used to calculate some parameters. FYI, I use FineOpen of Numeca. I will either use the SA or Mentor-SST k –omega (Extended wall) model.

Description of the stream: A free steam around a car driving in the open i.e. on the road. So it is an external flow. The free stream velocity $V_{\infty}$ is 25 m/s.

There are a couple parameters I have to specify in the CFD software program which im not complete sure about, being:

Boundary conditions:
for the k-omega model: k and $\varepsilon$
for the Spalart-Allmaras model: turbulent viscosity

Initial conditions:
for the k-omega model: k and $\varepsilon$
for the Spalart-Allmaras model: ratio $\frac{\tilde{\nu}}{\nu}$ (taken as 1)
So it comes down to calculating/guestimating these parameters. I used the following formulas (all in SI units) to calculate these parameters:

$U_{\tau}=\sqrt{\frac{\tau_{wall}}{\rho}}=\sqrt{0.5*V_{ref}^2*C_f}$ (From Numeca FineOpen manual)
$C_f=\frac{0.027}{Re_x^{1/7}}$ (Also from FineOpen manual)
$Re_x=\frac{\rho_{ref}*V_{ref}*L_{ref}}{\mu}$ with $\mu$ being the dynamic (molecular) viscosity
$k=\frac{\tau_{wall}}{\rho}*\frac{1}{C_{\mu}}$ with $C_{\mu}=0.09$ and $\frac{\tau_{wall}}{\rho}=0.5*V_{ref}^2*C_f$
$\varepsilon=C_{\mu}*(\frac{\mu_t}{\mu})^{-1}*\frac{\rho_{ref}*k^2}{\mu}$ I took $\frac{\mu_t}{\mu}=1$ (Numeca specifies in the manual that a typical value for this ratio is 1 for external flows). So before I can calculate $\varepsilon$ , I first calculate k.

Turbulent (eddy) viscosity $\mu_t=C_{\mu}*\rho*\frac{k^2}{\varepsilon}$ (So before calculating this parameter, I first calculate k, then $\varepsilon$ and then $\mu_t$ )

Using the above mentioned formulas, I was able to calculate all the needed parameters. Still I’m not sure if these formulas are correct or maybe are not valid in every case.

However, when I just assume these formulas are correct I calculated k, $\varepsilon=3075$ and $\mu_t$ . This yielded
k = 0.718, $\varepsilon=3075$ and $\mu_t=1.85*10^-5$ . Do these values make sense in anyway, or are they completely out of range?

The actual reason I’m posting this question is to validate our computation. One year ago, a group performed the same computation for the exact same car. This yielded a drag force of about 40N. The problem we are having now is that the drag force is in the order of 10^4 Newton. Obviously this does not make sense. Right now the mesh I runned in FineOpen contains about 18 million cells with refinements near the car itself, behind the car, beneath the car etc. etc.

So what I’m specifically asking:
1) Can someone verify the method I use, especially the formulas I use
2) What may cause the problem of a way too high drag value?
3) Do the values of k and epsilon make sense?

I know these are a lot of questions, but any help is much appreciated.
Thank you in advance,

Maximus

Maximus91 · October 24, 2012, 14:20

Anyone any suggestions?

October 19, 2012, 11:02	Turbulence model parameters and equations	#1
Maximus91 New Member Maarten Join Date: Sep 2012 Posts: 3 Rep Power: 14	Dear CFD users, Right now I´m working on a CFD computation for external flow around a very streamlined car. I have some questions regarding the turbulence parameters. I´ll first specify what formulas I used to calculate some parameters. FYI, I use FineOpen of Numeca. I will either use the SA or Mentor-SST k –omega (Extended wall) model. Description of the stream: A free steam around a car driving in the open i.e. on the road. So it is an external flow. The free stream velocity $V_{\infty}$ is 25 m/s. There are a couple parameters I have to specify in the CFD software program which im not complete sure about, being: Boundary conditions: for the k-omega model: k and $\varepsilon$ for the Spalart-Allmaras model: turbulent viscosity Initial conditions: for the k-omega model: k and $\varepsilon$ for the Spalart-Allmaras model: ratio $\frac{\tilde{\nu}}{\nu}$ (taken as 1) So it comes down to calculating/guestimating these parameters. I used the following formulas (all in SI units) to calculate these parameters: $U_{\tau}=\sqrt{\frac{\tau_{wall}}{\rho}}=\sqrt{0.5V_{ref}^2C_f}$ (From Numeca FineOpen manual) $C_f=\frac{0.027}{Re_x^{1/7}}$ (Also from FineOpen manual) $Re_x=\frac{\rho_{ref}V_{ref}L_{ref}}{\mu}$ with $\mu$ being the dynamic (molecular) viscosity $k=\frac{\tau_{wall}}{\rho}\frac{1}{C_{\mu}}$ with $C_{\mu}=0.09$ and $\frac{\tau_{wall}}{\rho}=0.5V_{ref}^2C_f$ $\varepsilon=C_{\mu}(\frac{\mu_t}{\mu})^{-1}\frac{\rho_{ref}k^2}{\mu}$ I took $\frac{\mu_t}{\mu}=1$ (Numeca specifies in the manual that a typical value for this ratio is 1 for external flows). So before I can calculate $\varepsilon$ , I first calculate k. Turbulent (eddy) viscosity $\mu_t=C_{\mu}\rho\frac{k^2}{\varepsilon}$ (So before calculating this parameter, I first calculate k, then $\varepsilon$ and then $\mu_t$ ) Using the above mentioned formulas, I was able to calculate all the needed parameters. Still I’m not sure if these formulas are correct or maybe are not valid in every case. However, when I just assume these formulas are correct I calculated k, $\varepsilon=3075$ and $\mu_t$ . This yielded k = 0.718, $\varepsilon=3075$ and $\mu_t=1.85*10^-5$ . Do these values make sense in anyway, or are they completely out of range? The actual reason I’m posting this question is to validate our computation. One year ago, a group performed the same computation for the exact same car. This yielded a drag force of about 40N. The problem we are having now is that the drag force is in the order of 10^4 Newton. Obviously this does not make sense. Right now the mesh I runned in FineOpen contains about 18 million cells with refinements near the car itself, behind the car, beneath the car etc. etc. So what I’m specifically asking: 1) Can someone verify the method I use, especially the formulas I use 2) What may cause the problem of a way too high drag value? 3) Do the values of k and epsilon make sense? I know these are a lot of questions, but any help is much appreciated. Thank you in advance, Maximus

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Is this understanding of turbulence models correct?	3kha	Main CFD Forum	3	January 31, 2011 22:31
Eddy Viscosity Model Parameters	Tristan	CFX	5	January 20, 2010 16:25
Reynolds transport, turbulence model, etc	Beginner	Main CFD Forum	1	January 7, 2009 06:36
some questions about SA turbulence model?	J. Jiang	Main CFD Forum	1	June 11, 2006 21:02
ANSYS/FLOTRAN	Mark Snyder	Main CFD Forum	7	January 17, 2000 17:32

October 24, 2012, 14:20		#2
Maximus91 New Member Maarten Join Date: Sep 2012 Posts: 3 Rep Power: 14	Anyone any suggestions?