Hello team, greetings. I’m new to MFiX and looking forward to using TFM. I’m fluidizing a bed with 500-micron size particles at its Umf. What is the appropriate cell size or cell volume? I want to resolve the system as accurately as possible.
The rule of thumb is to stay close to 10 particle diameters, i.e. around 5mm in you case. You should consider the overall volume of your domain, and see how many 5mm cells you need for the entire domain. Then based on your computational resources you can decide if this is doable or if you need to coarsen the mesh (and lose accuracy). Once you reach over a Million cells, you will probably need to run on a cluster, in parallel (DMP).
1 Like
Hello Dr. Dietiker, greetings. Thank you for your reply. I began setting up case as per your suggestions.
A. If we increase cell size, we lose accuracy. What happens if we decrease? (Let’s say experiment bed thickness is only 5 times, how to handle this numerically?)
B. Also could you advise me on deciding dt, dt_min, dt_max and dt_fac
Thank you,
Jagan Mohan.
I advice 'read the literature" to you. You need to find optimal grid size for you and your case. Very little cell size don’t give accurate results sometimes (depend of case).
A. Typically in CFD, as you refine the mesh you should reach grid convergence (results don’t get much better with refinement). The actual mesh size when this occur depends on the application and usually needs to be investigated. This is the sempiternal trade off between speed and accuracy.
B. I would always keep dt_fac=0.9, set dt_min between 1E-6 and 1E-7 s, start with dt=1E-3 s and set dt_max=1E-2 s. MFiX will try to adjust the time step so it runs as fast as possible. If you see the time step hits the dt_max and stays around dt_max, you can increase its value. If you are running chemically reacting flows, you may need to set a smaller dt_max (say 1E-4s). These are just guidelines though.
1 Like
The bed thickness/particle diameter is only 5, thus it is very tricky to use either TFM or DEM. For TFM or particle unresolved CFD-DEM fine grid simulation, the fluid grid size is usually in the
range of 2-3 times particle diameter in the bubbling fluidization regime and can be 10 times diameter in the fast fluidization regime. A researcher has identified a lower limit of cell size (1.63
particle diameters) to satisfy the assumptions (i.e., local mean fluid variables) of CFD-DEM
governing equations. So, you need to do a grid-independent study, use 5, 4,3, 2, 1 cells for the bed thickness.
3 Likes
Hello @gaoxi, thank you for your reply. Could you please point me to this article if the researcher has published it? It would be an interesting read to understand nuances, method of grid-independence to arrive at 1.63 particle diameters. Thank you.