Flow through a fibre-like structure

[anserko]

Hello,

I'm trying to model a longitudinal flow of a viscous fluid through a fibre-like structure. I need to extract velocity and pressure fields from exported vtk files for further permeability assessment

The obstacle is a set of cylindrical bodies like so:

Particles are introduced via an emmiter (velocity 1). The scene itself is constrained with a cylindrical surface.

I face a couple of problems I can't resolve yet. Can anyone help please?

1. When I run my solution, few particles "fly" away after initial contact with the obstacle, has anyone faced something similar?

2. At the emitter base there are regions where velocity gradient is "blinking/flashing" - rapidly changing values from almost zero to 4-6. It can be tackled by reducing particle size, but then overall solution takes way more time (as expected). Has anyone faced such problem? Why does it occur?

3. If I specify emmiter velocity too high (e.g. 5 instead of 1) simulation crashes after initial contact with the obstacle. Is there a way to avoid that (apart from lowering velocity)?

4. Is it possible to explicitly correlate exported vtk files with solution time? I do specify FPS in the json file, but I'd rather have that value explicitly as my timestep varies

5. I ran exactly the same analysis on linux and windows machines. Surprisingly, results vary a bit. On windows, particles are more prone to "flying away" at the flow front, while linux build provides smoother flow front. Has anyone had such experience?
   Here is an example of windows build simulation:

6. The last question is probably not that clever, but I couldn't find an answer. Is there a guideline for mapResolution parameter? What do these parameters mean exactly? My obj file dimensions don't exceed 4 in all directions and I specified mapResolution as [25,40,20] (took that from one of the examples). Changing this value greatly affects convergence, I'd like to understand how to assign that parameter properly.

7. I attach my scene file, any suggestions (solver, contact definition, fluid models etc) are very welcomed!

Thanks in advance!
scene.zip