"NORMAL particle in a GHOST cell" in 24.1.1

MFiX How to
#1

I apologize for reopening my previous topic. A bug fix release was promised last week. I have downloaded release 24.1.1, and the simulation is now alive for 3 seconds, but after that, I get the same error message as before. (There exists a NORMAL particle in a GHOST cell. The particle may have gone OUT-OF-BOUNDS.)

test_1_2024-05-11T214716.080959.zip (7.3 MB)

Original topic:

NORMAL particle in a GHOST cell
#2

No problem. I locked the other thread, just to reduce confusion. I’m sorry the bug fix for 24.1.1 did not fix this completely. We will take a look and follow up ASAP.

#3

Are you sure you are running this in the 24.1.1 environment?

Your Test_1.env file has:

CONDA_PREFIX=C:\Users\86151\miniforge3\envs\mfix-24.1
MFIX_RUN_CMD=C:\Users\86151\miniforge3\envs\mfix-24.1\Lib\site-packages\mfix_solver\mfixsolver.bat -s -f D:\GuoqingLian\MFIX2024\Test\HeatTransfer\SingleParticle\Test_6\Test_6.mfx

which looks like you are running 24.1 not 24.1.1

#4

conveyor_2024-05-15T232647.600310 2.zip (2.3 MB)

Last time I ran my backup, which was attached to my previous post on version 24.1.1, I did not refresh the zip file because I thought it was the same. However, here is the refreshed one.

#5

Ok I will take a look on this case.

#6

@cgcglw
So I have checked and tested your conveyor case, belows are what I found:

  1. The boundary conditions are not set properly, you missed the vessel and the top belt location was not right.
  2. This case comes with a calc_collision_wall_mod.f as a udf. It is a little bit outdated.
  3. This case is kind of sensitive to particle shape, I tried with your glued sphere shape (needle-like), the simulation can run until some particles went through the gap between bottom belt and vessel. Then the “ghost in normal” error was triggered. I attached a snapshot for your reference:
    needle_shape
    needle_shape1652×1166 120 KB

4.After that, I changed the shape, decreased the height/radius ratio, kept the bounding diameter at the same level, the simulation can run to finish.. I attached the test folder for your reference.

new_shape
new_shape1652×1166 130 KB

Therefore, if you real need the origin shape you used, you might have to wait more time since it might not be a easy fixed, we have to carefully checkout the overall scheme for this particular needle shape as well as the particle-wall contact algorithm.

Thank you for testing the gsp model and detecting this issue, we will try to fix and improve the model ASAP.
conveyor_test.zip (1.7 MB)

#7

update here, so I used your particle shape, I can get simulation run without problem by increasing the contact stiffness to 10000 (before I only used 1000 to test). It was a little bit messy when particles contacted with bottom belt.

you can try with corrected BCs and play with contact stiffness.
But again we will still check out this gsp model for potential improvements.

tobottom
tobottom1652×1166 189 KB

result
result1652×1166 135 KB

#8

Thank you for your help. I tested the file you sent; it is okay. I will play with the contact stiffness and height/radius ratio this evening. The final goal is to simulate crop flow on a conveyor to investigate a wear phenomenon. I am trying to approximate a grain model as closely as possible. I am also interested in writing Fortran code if it is needed in the future to achieve my goal.