# Calculate statistics of Turbulent channel flow with LES

 Register Blogs Members List Search Today's Posts Mark Forums Read

 May 10, 2016, 12:26 Calculate statistics of Turbulent channel flow with LES #1 New Member   Sungjin Won Join Date: Dec 2012 Location: Seoul/Republic of Korea Posts: 15 Rep Power: 12 Hi, I want to learn about some basic information about calculating the statistics in turbulent channel flow. I use Reynolds number based on centerline velocity and all of variables are normalized by channel centerline velocity and channel half width. (I develop my simulation from laminar with zero mean random perturbation. centerline velocity is defined in the poiseuille flow in laminar) Also I conserve the mass flow rate(not pressure gradient constant), so the pressure gradient is changed every iteration. The reference what I refer is Kim and Moin in 1987. They used Re_c = 3360 but my case 4150. The reason why I use 4150 is came from mass conservation. As flow developing, the velocity profile highly developed near the wall. Because of the mass conservation, the centerline velocity which is divided by characteristic velocity should be decreased as 0.77... In my case, dynamic SGS model derive U+ at the center about 20 not 18. I think there is no special error in present code. (high u_rms and low v & w_rms are calculated comparing to ref) My questions are 1. If I use pressure gradient constant scheme, can I conserve the channel centerline velocity divided by characteristic velocity as 1.0? 2. If I apply SGS model, is there a special treatment needed to calculate the friction velocity?[/SIZE][/SIZE]

 May 10, 2016, 12:55 #2 Senior Member   Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,402 Rep Power: 68 1) fixing pressure gradient is a way but does not define the choice of the non-dimensional quantity. If you want to work with the non-dimensional form and you fix the pressure gradient to -1, then the velocity is computed directly in terms of u+ and is not 1 at the centerline 2) what do you mean in your question? the dynamic SGS model would provide zero contribution near the wall (if your grid is enough refined)

 May 10, 2016, 23:31 #3 New Member   Sungjin Won Join Date: Dec 2012 Location: Seoul/Republic of Korea Posts: 15 Rep Power: 12 1) Thank you denaro. It helped me a lot 2) Actually, I know that eddy viscosity should be zero at the wall. However, some discrepancy with reference makes me doubt about my code for statistics. So I want to confirm that my code is good for calculating statistics. I think I need debugging again. Thank you Denaro Kind regards