Is it physically reasonable to expect a 0.16 m initial dense bed to expand to about 1 m in slugging flow?
Is my cross-sectional mesh too coarse for resolving bubble coalescence and slug formation?
3How should I choose the DMP domain decomposition in the x, y, and z directions for my 3D TFM case when using 32 or 64 cores? , so should I distribute the cores mainly in the vertical y direction, or use a more balanced decomposition such as nodesi = 2, nodesj = 8, nodesk = 2 for 32 cores and nodesi = 2, nodesj = 16, nodesk = 2 for 64 cores?
Does the uniform bottom inlet without a distributor/porous plate suppress or distort slug formation?
Which parameters are most important for forming slugging in a 0.06 m diameter gas-solid bed: gas velocity, initial bed height, wall boundary condition, frictional packing limit, or drag model?
Are there recommended TFM settings or benchmark cases for simulating slugging fluidization in a narrow 3D cylindrical bed?
. Thank you very much for any suggestions on whether this is mainly caused by insufficient solids inventory, coarse cross-sectional mesh, inlet distributor settings, or TFM model parameters.
Sorry I don’t have a good answer for you. Typically, if a CFD simulation doesn’t match experimental data, this could be due to a number of things, like the ones you listed.
First, I would recommend double checking all your input parameters (gas, solids definitions, Initial and boundary conditions) to make sure they are as close as possible as the experiment. What is the experiment solids inventory and pressure drop? How does the pressure drop across the bed compare with the experiment?
Once you are confident you have settings that represent the experiment, you can play with some grid, model, and numerical settings (one at a time) to see which one makes a difference. For example, what happens when you refine the mesh, change the drag law, max packing limit, solid inventory, or inlet velocity?
Thank you for your suggestions. I checked the pressure data and compared the simulation with the experiment at several axial locations.
The gauge pressure at the inlet/bottom is close between the simulation and experiment: about 3004 Pa in the simulation and 3054 Pa in the experiment. However, the pressure drops much faster in the simulation. At 0.2 m, the simulated gauge pressure is only about 303 Pa, while the experimental value is about 1004 Pa, which is less than one third of the experimental pressure. At higher locations, the difference is still obvious: at 0.46 m, 0.71 m, and 0.96 m, the simulated pressures are only about 11.5 Pa, 9.4 Pa, and 6.1 Pa, respectively, while the experimental values are about 523 Pa, 283 Pa, and 368 Pa.
Gauge pressure
0
0.2m
0.46m
0.71m
0.96m
simulation
3004.04Pa
303.48
11.52
9.38
6.13
experience
3054.42
1004.17
523.42
283.42
368.42
This suggests that in my simulation the gas pressure is lost mainly near the bottom region, and the bed is not maintaining enough solids holdup above about 0.4 m. In ParaView, the solids volume fraction also becomes very low above this height, so the upper part of the bed is almost dilute/freeboard rather than a dense slugging region.
Thank you for your suggestion. I checked the solids inventory and the pressure data from both the simulation and the experiment.
The solids inventory in my experiment is about 800 g. Based on the bed diameter of 0.06 m, the estimated bed pressure drop from the solids weight is about 2500–2800 Pa, which is close to the measured bottom gauge pressure of about 3054 Pa.
In the simulation, the bottom gauge pressure is also close to the experiment, about 3004 Pa. However, the pressure drops much faster along the bed height. At 0.2 m, the simulated gauge pressure is only about 303 Pa, while the experimental value is about 1004 Pa, which is less than one third of the experimental pressure. At higher locations, the simulated pressures are also very low: about 11.5 Pa at 0.46 m, 9.4 Pa at 0.71 m, and 6.1 Pa at 0.96 m, while the experimental values are still about 523 Pa, 283 Pa, and 368 Pa, respectively.
Gauge pressure
0
0.2m
0.46m
0.71m
0.96m
simulation
3004.04Pa
303.48
11.52
9.38
6.13
experience
3054.42
1004.17
523.42
283.42
368.42
This suggests that the simulation loses most of the pressure near the bottom region, and the upper part of the bed does not maintain enough solids holdup. In ParaView, the solids volume fraction above about 0.4 m is already very low, so the upper region behaves more like a dilute/freeboard region rather than a dense slugging region.
Therefore, although the total bottom pressure level is close to the expected solids weight pressure drop, the axial pressure distribution is very different from the experiment. I suspect that the gas may be channeling through the lower bed region, or that the current solids inventory, inlet/distributor treatment, drag/friction settings, or mesh resolution are not sufficient to maintain a dense slugging bed up to around 1 m.
It is the time-averaged gauge pressure. The pressures at these two locations are similar; there is an instrumental error, and the initial value is -50 Pa. The main issue is that during the experiment, the solid phase can be blown up to 1 m, whereas in the simulation it only reaches 0.4–0.5 m…
