Hello team, I have successfully run the rotating drum tutorial and have a basic understanding of read_kf and data_kf_0001.txt.
A question I have is what are the units of var1 in this file? How is the code understanding that the drum should be rotated? Thank you.
For this tutorial, the unit is rad/s (see pdf file in tutorial folder). In general, you can use any unit you want, and adjust how you use the keyframe data in usr1_des.f
Hello Dr. Dietiker, greetings. Thank you for pointing me in right direction.
A. I do not have coding background, are we specifying rotation axis as well in this tutorial? I could not find pdf file.
B. Does number of facets on STL surface play any role on solution accuracy? If cell width is dx = dy = dz, do we have to ensure any cell sizes on STL as well?
Thank you,
Jagan Mohan.
A. Yes the axis of rotation is specified in usr0.f . Once you open the rotating drum tutorial from the GUI, navigate to the project folder and you should see the pdf file in this folder.
B. As long as the STL file quality is good, it should not have any significant impact on the solution accuracy. By good quality, I mean no extremely stretched facets, no duplicates, no missing facets, and consistent normal vector orientation. I personally like to keep the STL facet sizes around the same size as the fluid cell size. Going much smaller than the cell size won’t provide more geometric accuracy because we use cut cells. It will slow down the simulation though with DEM since particles interact directly with the STL facets. If you have large planar sections, it is fine to use large triangles there.
Hello Dr. @jeff.dietiker, I did following steps, could you please look into them.
A. I created 2 concentric cylinders with a separation between them using primitive-STL feature and adjusted “domain box” in y-axis so that flat surfaces are trimmed-off.
B. STL normals are flipped so that computations are in separation only. [computational volume is annulus cylinder along y]. Cell width is 10 times particle diameter and TFM is physics. Bottom is gas only mass inlet, top is pressure outlet, outer and inner STLs are no-slip. BCID for inner is 1.
C. I copied following 8 files from drum tutorial
calc_collision_wall_mod.f, des_time_march.f, usr0.f, usr1_des.f, usr_init_namelist.f, usr_mod.f, usrnlst.inc. Set read_kf(1)=true and usr_def=true.
D. And only edited following in usr0.f
cyl_center1(list(1), 
= (/0.0, -0.05, 0.0/)
cyl_center2(list(1), = (/0.0, 0.05, 0.0/)
E. data_kf_0001.txt is
2, 1
linear
! time var1
0.0 75.0
80.0 75.0
Observations:
Initially, dt = 1E-3s but very soon landed into DT<DTMIN or negative void fraction error. Circumvented this by lowering dt=1E-6s. Any other ideas?
Gas flow is above Umf and even after running for 10 secs, no azimuthal flow. What could have gone wrong? I have exact same experimental setup where particles+air behave like fluid.
Can we export STL surface to check tangential velocity?
Mesh details are
NUMBER OF CELLS = 214775
NUMBER OF STANDARD CELLS = 74408 ( 34.64 % of Total)
NUMBER OF CUT CELLS = 25560 ( 11.90 % of Total)
NUMBER OF FLUID CELLS = 99968 ( 46.55 % of Total)
NUMBER OF BLOCKED CELLS = 114807 ( 53.45 % of Total)
I’m using 24 CPUs and progress is only ~2 secs in 6 hours. How can we speed this up?
Thank you.
The moving STL file capability as illustrated with the rotating drum is only available to DEM. This is not meant for TFM.
Hello Dr. @jeff.dietiker, thank you for your reply. I’m using mfix-20.4.3 which is the most latest version, I believe.
Do we have any other feature or tutorial where I can have a rotating/moving boundary in TFM framework?
If no, can we do this using any UDFs? This is important as CFD-DEM is not feasible due to large particle count. If you could point me in right direction, I’m ready to give time and explore writing a UDF.
Thank you.