dependency of tau in velocity

Hi freinds.
I have written a simple program for 2D channel flow and apply uniform velocity U0=0.05 at inlet,fully developed at outlet :u(nx)=u(nx-1) according
to zou-He BC. bounce back for walls. Nx=235,Ny=35. I have choose a specified tau for simulation,So all parametres will be knownand there is not need to enter Reynolds no.
(with these parameters Re=U0*Ny/nu will be 6.9) but I have an strange behavior in velocity of centre line (U/U0) at the length
of channel(theorically it shoud be 1.5 in fully developed)
if i choose 0.6<tau<1 U/U0 decrease at the length.U(nx,ny/2)/U0<1.5

if i choose a tau :tau>1 U/U increase at the length.U(nx,ny/2)/U0>1.5
at aout tau~1 this will be 1.5
Have any one obsereved this dependency of u/U0 in tau??
Is thete an optimum tau in LBM simulation?(tau=1)

It seems that nobody have such an experience??

Could you show us a plot of your velocity profile? I guess that the boundary conditions are not correctly implemented. What happens to the average density in the simulation? Do you reach a steady state?

Hi Timm
I don’t know how to insert my plots in this forum.Can I have your personal email an send them by email?
avarage density increase with time. the centerline velocity(for example in the ase tau=0.8): (u/U0) strart form1at x=0(inlet) ,reaches 1.5 at around x=20 and then slowly decrease in length to reach 1.47 at outlet(x=235). This problem has confused me and i dont know how to do?

I have sent you a private message