CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   K-Epsilon Model (

sangit September 7, 2004 20:46

K-Epsilon Model

I am a student of TU Munich Germany. And I am a novice in CFD. I am studying the k-epsilon turbulence model and using 2 equation of this model to simulate flow inside a pipe. I am discritisizing the two differential equations of k and epsilon and then having the values at new time step from that. But my problem is that at the wall I am using k=0 but there I donot know the value of epsilon .... i only know variation of epsion perpendicular to wall is zero. So I am putting the varitional derivative perpendicular to the wall is zero and calculating the value parallel to the wall from the epsilon discritisized equation .....

But My problem is that in the epsilon equation there is k in division which is eqaul to zero at wall .... and this is making my result to infinite ....and it is making the entire reults to infinte values ....because the two equations are coupled by k and epsilon of previous step ..... under the circumtances what should I do in the epsilon eqation at the wall .....

And also I have used the turbulent viscosity equal to zero where k is zero(though epsilon may be zero) .... is it correct .....

Any suggesion will be of great help to me .... as I am really confused

Halim Choi September 8, 2004 05:00

Re: K-Epsilon Model
There are two ways to calculate the fluid flow using the turbulence models, one is using the wall function method and the other is calculating all the way to the wall. If you use the wall function method with k-epsilon model(we usually call this the high Reynolds number model), the epsilon at the wall is meaningless and is not specified, and the epsilon value at the first grid point from the wall is specified by Epsilon=CDTQ*K**1.5/(CAPPA*DELTA) where CDTQ=0.1643, CAPPA=0.4187, K is the turbulent kinetic energy and DELTA is the normal distance from the first gid point to the wall. In this case you should also use the wall functions for momentum equations and K-equation. If you uses the low-Reynolds number model, which calculate flow all the way to the wall, there are some slightly different ways to specify the epsilon at the wall and the simplest and usual way is by Epsilon=2*VISNU*K/DELTA**2 where VISNU is the kinematic viscosity, K and DELTA is defined the same as above. I am not sure what kind of turbulence model you are using and I want to recommend to read some turbulence modeling books before coding the program. I hope this helps.

Gook luck

Halim Choi

Abhijit Tilak September 9, 2004 13:19

Re: K-Epsilon Model

What k-e model you are using ? If you use High-Reynolds number approximation your near boundary point must preferably be outside the viscous sub layer i.e. y+ >11.6 In that case de/dn = 0 B.C. is used for epsilon. and the value at near boundary pt. is calculated from the law of the wall. For Low-Re form you need to solve right through the viscous sub-layer. BC's are different then.


All times are GMT -4. The time now is 07:12.