Hello, I am a pretty new user of MFIX. I have worked through the provided tutorials in the manual and I am attempting to recreate a simple DEM model as described in a paper as an additional learning exercise. The model is a simple thin, rectangular vessel with free-slip walls. 80,000 particles are initially distributed evenly throughout the system at a volume fraction of 0.2 and are allowed to compress under the force of gravity in a vacuum.
My problem is, the particle IC does not extend to the geometric bounds I’ve set in the Region. Here is my particle IC region in the GUI (colored in indigo):
And here is what my particle distribution looks like at t=0.0 s:
Here is what the Lagrangian version of that view looks like:
As you can see, my particles do not extend to the coordinates I would expect them to based on the definition of my region. Of course, I’m new so I fully expect this could be user error on my part. I have attached my zip file as well, so any help you could provide would be greatly appreciated.
Thanks,
Matt Black
2108_hong_dem_recreation_2023-09-15T113537.589540.zip (15.9 MB)
It is not an error on your part. The “Fit particle to region” is actually deprecated and should not be shown in the GUI. The volume fraction you specify as an input in relative to the region, it is not a local value. So a volume fraction of 0.2 means we want to fill 20% of the IC region with particles, not that the solids fraction will be uniformly equal to 0.2 everywhere. Since most applications deal with a bed of solids at the bottom of a reactor, and most users what to start the simulation with a fairly packed bed MFiX seeds particles by default at the bottom (y-direction) of the IC region. The option “Fit particle to region” used to spread particles but this is now deprecated. Instead we have more seeding options where users can control the lattice and spacing. Below is an option to spread particles a bit more in all directions (default value is 0.25):
Now you should always verify that the seeding was successful:
Phase 1: Number of particles to seed = 80362
|-------------------------------------------------------------------|
| IC Region: 3 |
|-------------------------------------------------------------------|
| Phase | Input | Input | Number of | EPs | Solid |
| ID | Specified | Value | Particles | | mass |
|-------|-----------|-----------|-----------|-----------|-----------|
| 1 | EP_S | 2.00E-01 | 80362 | 2.00E-01 | 8.42E-01 |
|-------|-----------|-----------|-----------|-----------|-----------|
Message from des/generate_particles_mod.f:693
Initial condition 3, Phase 1: Successful seeding
The above says that 20% of the IC region was filled with particles. Even if it looks like particles are only in the bottom half of the region, it represent 20% of the IC by volume.
Hi, thank you very much for the response. I understand the logic, although it does go against my intuition of how initial conditions are defined (my background in simulation is using Barracuda, so I’m probably carrying some assumptions based on how that software operates). So if I want to have this IC work the way I intended, is the best way to trial-and-error with this “particle spacing” parameter, or is there a more straightforward way? Thanks.
Yes, you can play with the seeding options. Please keep in mind that for thin domains when there are only a handful of particle layers in the z-direction, it may be best not to vary the spacing in the z-direction. A small change in z-spacing will results in an entire layer (XY plane) of particles being added or deleted, which makes it harder to adjust.
The settings below may be close to what you want:
1 Like
