density growth with time

Hi freinds,
i have written code for solving flow inside 2D channel(235*23).BC:uniform constant
velocity at inlet,fully developed at outlet(by zou-He method).It works and give parabolic profile
so that (maximum velocity at outlet/U0) reach 1.48 and remain constant with time. But
if I trace the density(rho) of any arbitrary node it continusely growth with time and never becomes steady!!
isn’t it abnormal? what’s the problem?(tau=0.7)

I checked the mass flux=sum(rho(x,y)*u(x,y)) over y, it decrease at the length of channel! but I can’t find any reason for this. BC for walls of channel bounce back :(f(5,2,6,1)=f(7,4,8,3)). at outlet zou-He for velocity with u(nx,y)=u(nx-1,y) & v(nx,y)=v(nx-1,y)
inlet velocity U0=0.1 (low reynolds). Does anyone have a simillar problem? Or give me a nanalytical solution to campare with my result??.

Thanks.

Have a look here, esp. section 3 and 4 C.