Particles in ghost cells

Hi all.

I have been running into ‘Error 1100: Particles detected in a ghost cell’ while doing DEM of particles settling into a cylindrical hopper. The geometry is imported via a STL file and meshed. Initially I thought the error was due to particles leaving through the inlet, so I placed a non-permeable internal surface in the neck of the hopper. The aim is to let the bed settle before flowing a fluid through the bed.

From a previous thread, I saw that it can be due to particle softness or timestep being too large. From reviewing the results the particles start to settle when this error occurs. Decreasing the timestep and restarting from the last restart point does not remove the error.

Any recommendations to what can cause the problem?

A common error is to use an STL file with stretched triangles which are removed below a small angle tolerance. Check the standard output to see if you get any message such as:

STL file(s) successfully read.
 Total number of facets read =   208
 Number of valid facets      =   144
 Number of ignored facets    =    64

If there are ignored facets, either fix the STL file outside the GUI, or lower the small angle tolerance until there are no ignored facets.

Another common issue is to have the STL normals pointing in the wrong direction. Please see https://mfix.netl.doe.gov/doc/mfix/19.1.4/reference/faq.html#how-to-verify-flip-stl-file-facet-normals for more details.

If particles are too soft, reducing the fluid time step won’t help. You need to increase the spring stiffness for particle-particle and particle-wall collision.

If you get stuck too long, feel free to post your .mfx and stl files, and someone may be able to provide a better solution.

Good Morning Jeff

This is the output form my mesher


INFO get_stl_data.f:692
Reading geometry from geometry.stl …
<<<<<-----------------------------------------------------------------
STL file opened. Starting reading data…
Done reading file.
STL file(s) successfully read.
Total number of facets read = 512
Number of valid facets = 512
Number of ignored facets = 0
Geometry range from stl file:
x-range = -0.4485 to 0.4485
y-range = -0.4000 to 2.8000
z-range = -0.4500 to 0.4500

So the STL does not seem to be the issue. My STL normals are pointing towards my flow domain as I get about 0.3s and some decent settling before the error occurs.

Ill increase the stiffness to see if that will fix it.

Regards

Increasing the particle-wall stiffness eliminated the error.

Thanks

Hi jeff,
I met Error 1100 always. I saw that you mentioned the chekcing of STL file may be helpful. Where to see the output of STL file? I run the MFIX by source code. And I checked the RUNNAME.OUT file and did not find any information of STL listed in your reply. Thank you for your any help.
Regards,
Leina

Hi Jeff

by increasing the wall and particle spring stiffness the particle in ghost cell issue has been partially resolved. Every now and again I have a particle that “slips” through my wall and just floats in the ether outside my fluid domain, but inside where the initial background mesh was, no matter how high i make the stiffness. These “free floating particles” causes a particle in ghost cell error when it happens to reach the edge of the background mesh. These particle`s speed is a order magnitude larger than it should be.

Any Ideas to stop this from happening? I have decreases the time steps and increased the stiffness, but it keeps happening.

Regards

The summary of the STL file reading is displayed on the standard output. If you run MFiX outside the GUI, the standard output is displayed in the current terminal. You will need to scroll up because it is displayed at the beginning of the simulation. I usually pipe the output to a text file so I can keep the output:

./mfixsolver -f runmane.mfx |& tee run.log

The above will run mfix using runmane.mfx as the input file and will both display the standard output in the terminal and save it in run.log

Maybe decrease the “Max steps between neighbor search” in Solids>DEM Advanced pane

Thanks, Jeff. You mean I could find the information of STL file in run.log. I usually use this command for a DMP running: mpirun -np 4 ./mfixsolver > run.log &. So I can get the output file run.log. I will check it carefully. Thank you for your help again.

Morning Jeff.

Decreasing the neighbor_search_n, as suggested still causes the phantom particles to appear.

The goal is to develop a baseline for a spouting bed design. Currently I am trying to fill the geometry with 6million DEM particles and allow it to settle due to gravity, before the fluid momentum equations are solved.

From the baseline a more complex geometry will be investigated. IT will be beneficial to settle the bed once and use that to test different flow conditions.

Attached is the mfix case if someone can help tweak the parameters to allow the bed to settle.

spouting_bed.zip (413.0 KB)
Regards

Hi gbotha, I also meet this problem. Have you solved it? I have tried to increase kn, or decrease max steps between neighbour search but finally nothing changed. The phantom particles still exist. Can you give me some advice?