Problem with channel flow


I have written a 2D LB code to simulate channel flow. I am using the following conditions:

  1. Bounce-back at the top and bottom surfaces.
  2. Periodic condition along the channel.
  3. To generate a uniform pressure difference, I am using the ‘body force method’ as given in Sauro Succi’s book.

I initialize the flow every where with a uniform velocity. The flow is not developing, and I am unable to get a parabolic profile. Re < 50, and I am running it for t = 10000 (lattice units).

Could some one please help me?

Thanks and best wishes,


what does it mean that the flow is not developing? Do you see any change at all? Can you upload a few snapshots of the velocity profile at different times?


Hi Timm,

Thanks for the reply. I see that the no-slip is being obeyed at the walls, but the it doesn’t fully develop into the parabolic profile. I do not see any change beyond this point. I am including a snap shot after the running it for t = 10000.

Thank you.

Best, SK


I’m assuming from your description that the problem geometry is symmetrical around the middle of the channel? If so, all your results should be symmetrical around the middle of the channel, which is not the case, judging by the image you provided. I’d say that there must be a bug hidden in your code somewhere.



I agree, there must be a bug in the code. Unfortunately, Erlend is right, and the bug may be anywhere. From my experience, one of the most probable points to start is the collision (incl. equilibrium computation) and the forcing.
What happens, if you start a simulation with constant, non-zero velocity but without force?


Hi Erlend and Timm,

Thanks for your replies.

I have been on the look for the bug, and unfortunately still have not found it. I get more or less the same profile if I have both the forcing and an inlet velocity or just inlet velocity. I did try with a parabolic profile to begin with and the solution remains the same after running the code for some time. This is more puzzling since the symmetry should have been broken…

I shall check things again.

Thanks again.

Best, SK

This case is probally wrong codes to the bounce back schme at top and bottom. I use the period boundary and bounce-back scheme and body force, can gain good results.But, I use zou/he pressure boundary condition at inlet and outlet, and bounce-back scheme at top and bottom wall, can gain the the parabolic profile, but results were smaller than the analytical reuslts for the Poiseillue flow, I don’t know where were wrong to codes with c++. I wish your help.Tim Sir.
Thank you very much.

Thank you bent.
I had settle down that problem. but i found a strange performace that when applying zou/he pressure boundary, simulating velocity gradually increase from inlet and outlet, at the same time density between two plates shows parabolic curve. Did you meet over this case ?