k-epsilon problem: non-physical turbulent viscosity production in coaxial free-jet
What are some unexpected producers of turbulent viscosity when using k-epsilon?
In our simulation, concentrically aligned free-jets are simulated for comparison with experimental data. In this figure, turbulent viscosity measured on the domain's centerline is shown in red, the axial velocity measured on the domain's centerline is shown in green, and the experimental axial velocity data which we seek to match is shown in orange. These curves are overlaid on top of the domain which is shown in blue. http://lh4.ggpht.com/_PqzKw7xCW0A/TE...c0/s800/tv.jpg Why would k-epsilon produce the large amount of turbulent viscosity starting at about 40% of the length of the domain? Grid dependency in the direction of flow has been eliminated. Further simulation details:
|
k, epsilon, rho plot
Below are plotted the constituent quantities comprising the turbulent viscosity in our simulation. The turbulent viscosity is defined as follows: http://www.cfd-online.com/W/images/m...7849438c8e.png From initial inspection of the plots, it looks like k is at fault. Why would k suddenly jump at X/D = 8 ? Any suggestions? http://lh4.ggpht.com/_PqzKw7xCW0A/TE...lots_bitte.jpg |
All times are GMT -4. The time now is 06:52. |