CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Modified turbulent viscosity (

Christoph Oehy April 18, 2000 03:54

Modified turbulent viscosity
Hi, I'm working with horizontally stratified two-phase flows and I'm using the standard k-epsilon model. As I need to take into account the stratification, I'm trying to modify the turbulent viscosity depending on the local Richardson number as it was proposed by Rodi (1980). Has anyone already done this? Is the VIS(INODE,IPHASE) only the laminar, turbulent or the effective viscosity? (I thought in turbulence modelling it becomes the turbulent viscosity). Can I use a conventionally central differencing scheme to compute the local Richardson number? Thanks, Chris

Li Xinfeng April 18, 2000 08:26

Re: Modified turbulent viscosity
I think the vis(inode,iphase) is the laminar viscosity. and the turbulent viscosity is defined by this method. vist=Cmiu*desitiy*k*k/epsilon. I don't understand you second question well,why not be able to use the convertionally central differencing scheme? Hope I can give you some help. Best regards

Andreas Abdon April 20, 2000 03:19

Re: Modified turbulent viscosity
Hi Christoph,

I suspect VIS(INODE,IPHASE) to be the effective viscosity but I suggest that you print out this variable for a turbulent run to make sure... Sometimes for high Re flows the laminar viscosity is neglected so that the effective viscosity is identical to the turbulent viscosity. I don't know if this is the case for the high Re k-eps in CFX.

Regards Andreas

ps. please let me know if VIS is not the effective viscosity!

Athar Zaidi April 23, 2000 22:08

Re: Modified turbulent viscosity

In USRVIS, variable VISN(INODE,IPHASE) stores the effective viscosity in case of turbulent flows. VISN(INODE,IPHASE) = VISL+VIST ,where VISL is the molecular viscosity and VIST is the turbulent viscosity. CFX4.2 calculates VIST on fly and doesn`t stores it, however in solving the transport equations of k and epsilon you have the terms like (mu+mu_T/sigma), CFX actually subtracts VIST from VISN to get mu and since solver has already calculated VIST it divides this quantity by sigma to solve the transport equation. I have tried this and it works for Newtonian fluid flow , for non-Newtonian fluid flow where VISL is also a function of shear rates whole new strategy has to be adopted using user scalars for k and epsilon or omega ( depending which model you are using). We have successfully implemented this strategy for low-Reynolds-number k-w model for modeling non-Newtonian blood flows, both for steady state and transient flows. I hope this will help you . regards Athar

LI Xinfeng April 27, 2000 22:06

Re: Modified turbulent viscosity
I ask the technical support.he told me the vis is efficient.

All times are GMT -4. The time now is 05:34.