Hi there, I am trying to simulate a laminar flow through a thin rectangular duct of length 20 inches, height 0.1 inches and width 5 in. I am using a simple mass flow inlet and an OUTFLOW boundary condition. I keep on getting reverse flow from the "outflow" boundary. Any idea as to why this might be happening ? ( i have seen this happening many times before when I was trying to simulate flow through thin and very long pipes !)

Thanks in advance .

(1). I always have flow separation at the exit when I run a commercial code,even now, I am getting a lot of reverse flow at the exit. (2). If I carefully adjust the time steps or relaxation factors, I can get a converged solution, but this requires the job to be subdivided into many sub-jobs. (3). So, you may want to try this tedious approach of (2).

In these cases something that helps is to extend the length of the model and place a porous jump at the end with a very low pressure drop. This normally gets rid of the reverse flow and leads to a fully converged solution within a short period of time. Obviously the location of the porous jump depends upon what is intended with your model. What is the purpose of it??

Dave

Thanks for all the help. ! I am maintly interested in the first few inches of the flow length, so yeah putting a porous jump with a small DP at the end wouldnt cause too much harm. will give it a try. thanks

Hi Sandeep

Why not use a converging outlet section (say a half sin wave), which will reaccelerate the outlet flow and give a better solution overall.

Regards oliver

(1). For converged solution, it is possible to do that. (2). For solutions during the iteration process, it's not going to work. (3). I am running a 3-D duct flow problem, and most of the time, there are flow separation at the outlet as well as the inlet. Sometimes, it is 100%. (4). I am using a different code, but the principle is the same. If your initial guess of the flow field variables of velocity , temperature and pressure is far away from the true solution (which you don't have yet), then reverse flow will appear at the outlet and the inlet in the iteration process. (5). The use of fine mesh, higher-order scheme tend to promote solution divergence. So, far, I have not found a method to solve this problem using commercial codes.