Sudden drop in DT when injecting fluid to a vacuum

MFiX
#1

Hi,
I was injecting the fluid (oil) to the domain under TFM framework. The oil will becomes to fixed coal, gas and fluid. So I revised this case from the classic CO2 absorption case.

I found that when I inject the oil mixed with small amount of gas, the oil volume changes suddenly rather than smoothly. And at this rapid changing point, the FGS drag force changes suddenly, leading to a sudden drop in DT, the first drop happens at 1.6 s. Sometimes, negative void fraction happens, which reduce the computational efficiency. Moreover, i found the gas velocity above the fluid fluctuates with fluid injection. The injection speed is 0.005 m/s but the fluid velocity can reach 2.5 m/s. I am not sure whether it is right.

Could you please have a look at this problem?
sucess_plane3.zip (31.3 MB)

#2

Does the time step eventually go back up? It is OK to have the time step go down once in a while.

#3

Hi jeff,
It goes back a little bit for a while. However, with continuous injection, the DT keeps reduces at each time when the materials injects to the domain. However, when I output the F_gs, the value in X direction reaches 3000 at 0.6 s, then stabilizes. This value is nonphysical. I think it is the cause of a small time step. I do not know what’s the reason of this problem.

Kind regards
Yuanhe

#4

I am not able to run your case. It fails with unallocated F_gs in vtk_out.f and then it fails with array bound mismatch in usr1.f:

At line 144 of file sucess_plane3/usr1.f
Fortran runtime error: Array bound mismatch for dimension 1 of array 'x_eq' (9/14)

My suggestion is you should build the solver with debug flags and fix these errors first.

#5

Hi jeff,
Thanks for your test. But I have checked it before I uploaded it. I just download it and run it successfully again. By the way, I use mfix-21.4 in the Anaconda environment in windows system
image

#6

Can you please try with mfix-22.3 ?

#7

@jeff.dietiker how did you compile this? I tried with 22.3 and it didnt’ build for me either but I got a different error:

  445 |       IF(MODE==0) CALL UPDATE_AND_CLOSE_PVD_FILE
      |                                                1
Error: Missing actual argument for argument timestamp at (1)
/tmp/sucess_plane3/vtk_out.f:80:55:

   80 |          IF(DISCRETE_ELEMENT) CALL WRITE_VTP_FILE(LC,1)
      |                                                       1
Error: Missing actual argument for argument timestamp at (1)

Also note that the mfix build popup does not offer the “Build type” on Windows, so building with debug flags requires running CMake from the command line. We need to extend this functionality to Windows (in progress for 22.4)

#8

I built with 21.4 (I got the version from the .mfx file). It won’t build with 22.3 because vtk_out.f in the project directory is based off 21.4 and is no longer compatible with 22.3. There are many modified files so it could also be the case for other files.

#9

vtk_out.f (154.4 KB)
Hi, this is the vtk_out.f based on 22.3.

#10

I see, thanks Jeff. I generally try to test with the latest release to see if the problem is still reproducible.

#11

Hi cgw,
I have revised my case into MFIX 22.3 version.
The FGSX value remains very large and finally leads to a breakdown.

Could you help me figure out the underlying reason？
Attachment is the new case.
Ab-10.15.zip (4.3 MB)