|
[Sponsors] |
July 15, 2005, 12:46 |
Baffle_Convergence
|
#1 |
Guest
Posts: n/a
|
Hello,
I am doing a multiphase modelling (air+water in a rectangular tank). In this 5 meter high tank, there is a baffle, and water level is just 0.1 m above the top of the baffle. Also, the flow overflow to outside, so the outlet is also very shallow. I got convergence problem for this problem. The residual(RMS) never go blow 10^-4. Huge imbalance(80% to 120%) were observed for air, water, and the additional vaiable(o2). I adjusted timestep and refined mesh, but did not improve significantly. From cfx-post, I can see that the velocities at outlet and over the top of baffle are huge, compared with that in main flow. I know this is reasonable phenomena, but I am not sure if this is the reason for my convergence difficulty. How to solve this kind of problem with shallow flow passways? Thanks and have a nice weekend! Gab |
|
July 15, 2005, 13:27 |
Re: Baffle_Convergence
|
#2 |
Guest
Posts: n/a
|
Hi,
Are you performing steady-state or transient simulations. I'm simulating mold filling problems (transient simulations), and when the the flow front (interface liquid-air) gets close to the outlet B.C. it's hard to obtain convergence (a very small time step is necessary).The solution is to move the outlet B.C so that the interface never gets close to it. From my experience, it's enough if the outlet B.C. is 6 mesh elemts away form the interface. I think you see those huge velocity vectors because the solution is not properly converged at that zone. In CFX-Post, plot the residuals, and check what are their values at the interface/Outlet B.C. zone. Regards, Rui |
|
July 15, 2005, 16:34 |
Re: Baffle_Convergence
|
#3 |
Guest
Posts: n/a
|
Hello, Rui
Thank you for the reply. I am doing steady state simulation. I already move the outlet far very away from the actual location. In cfx-post, I can see that the highest residuals are before and after baffle top edge, and also before the outlet. What I can do based on this observation? Another thing is that my Y+ is higher than 1000(I am using k-e model). I was wondering if I need to refine the mesh further----my mesh number has already been very high--900,000. I tried to use 'inflation' in cfx-mesh, because it seems that if I define all the surfaces as inflation boundaries, it just shows inflation layers on one surface. Thanks again! Gab |
|
August 9, 2005, 10:57 |
Re: Baffle_Convergence
|
#4 |
Guest
Posts: n/a
|
||
|
|