Hi MFiX developers,
In my last post below, I tried to set an stl internal surface as movable to model the vibration of a distributor in CFD-DEM simulations. However, I found that such settings always give gas channeling and un-uniform gas distribution in the fluidized beds. This happens even if the movable internal surface does not move at all, as shown in the attached figure extracted from a run with the attached case files (the zip file). I have tried to change a series of parameters, including the mesh, interpolation scheme, and DEM time step, but none of them works. Is it a bug? How can we solve this problem? Thanks in advance for any insights.
https://mfix.netl.doe.gov/forum/t/how-to-set-a-dem-wall-for-a-layered-fluidized-bed/3127/14
movableIS.zip (17.6 KB)
To have a better demonstration of the problem, I have increased the particle size and mesh size. As such, the simulations can run faster to see the gas channeling within a half hour. Here are the case files.
demo.zip (16.7 KB)
The internal stl is not seen by the fluid phase. Are you saying that if you run with a stationary stl at the bottom and without an stl, you get different results? How much above minimum fluidization are you?
Hi Jeff @jeff.dietiker
Thanks for your reply. Yes, I indeed found that the results are very different between cases with and without a stationary stl at the bottom. I attached my cases and videos in the zip file here for your examination. The used superficial gas velocity is around 1.7-1.8Umf. I also tested other superficial gas velocities higher than this value (2-2.4Umf) and they all showed gas channeling.
I was wondering does it happen due to a lack of pressure drop “through” the stl internal surface? I found that for permeable and semi-permeable internal surfaces, we have a chance to define the gas resistance, but we cannot do the same for the movable internal surface because it is unseen for the fluid phase. With the pressure drop in mind, I tried to add a layer of stationary particles just below the stationary stl internal surface to increase the pressure drop “through” the internal surface and the results seem to be better, at least the gas channeling is not observed. The attached zip file also includes the case files and results for this trial. I am not sure whether this way is right or not. Is there any way to solve this?
caseANDresults.zip (56.4 MB)
Hi Jeff @jeff.dietiker
I also checked one more case with smaller particle size using the same method by adding a layer of stationary particles just below the stationary stl internal surface to work as a “virtual distributor”. I found that gas channeling can be avoided if the pressure drop through the “virtual distributor” is high enough. So, maybe this can be an effective method with using movable internal surface in CFD-DEM simulations
