Hi,
I try to solve 2D channel flow. But I have some problems. Before explaining the problems, I mention my settings.

I use D2Q9 lattice. Also I use LBGK.

I use one way bounce back scheme. (on north and south walls)
For example south boundary
f(2,i,0)=f(4,i,0)
f(5,i,0)=f(7,i,0)
f(6,i,0)=f(8,i,0)
 I give velocity BC at west domain with ZouHe scheme. Al so I give fully developed velocity profile.
uy(0,j)=1.5d0uo(1.d0(4.d0/h**2)((jdy)(h/2.d0))*2)
rhow=(f(0,0,j)+f(2,0,j)+f(4,0,j)+2.d0(f(3,0,j)+f(6,0,j)+f(7,0,j)))/(1.d0uy(0,j))
f(1,0,j)=f(3,0,j)+2.d0rhowuy(0,j)/3.d0
f(5,0,j)=f(7,0,j)+((f(4,0,j)f(2,0,j))/2.d0)+rhowuy(0,j)/6.d0
f(8,0,j)=f(6,0,j)((f(4,0,j)f(2,0,j))/2.d0)+rhowuy(0,j)/6.d0
 As a outlet BC, I give zero gradient velocity (first order)
f(3,n,j)=f(3,n1,j)
f(6,n,j)=f(6,n1,j)
f(7,n,j)=f(7,n1,j)
 For calculating the equilibrium distribution function,
feq= wirho(1+3(eiu)+4.5(eiu)*21.5(uu))
 My process sequence is;
As a initial calculate feq for initial condition, and f=feq,
And iterations begin. Collision step, streaming step, BC (velocity, bounce back and outflow), calculatong rho, u and v for new iteration.
My problem is, although flow is fully developed, veloctity is decreased from that I gave. Also density is decreased. So I lose mass.
How can I solve this problem? Please help me
Daralcan