When a run is recovered from stall / divergence, it leaves a haze

MFiX
#1

Hi,

I am using MFiX 22.3.1 for a vertical transport model with 2D TFM.

Sometimes when a run is automatically recovered from stall / divergence during the course of a simulation, it leaves a haze in the visualisation (please see the screen grab below). Worried that this may affect the accuracy of the results (for example when the data is time averaged)?

smudge_query
smudge_query717×924 355 KB

Would it be possible fetch the clean data from the end of the previous time step when a run is recovered from stalls / diverges (and not use the data that has been overwritten during the iterations of the current time step)?

Are there any other means to minimize this haze?

Thank you.

#2

How are you generating this figure (which file do you open, which software)? Does it look the same in the GUI? The output file is written after a successful iteration. Did you increase the tolerance or are you using default values?

#3

Hi Jeff,

I generated the figure with VisIt using the VTK output files. I ran the simulation in the CLI and hence have not seen it in the GUI. I have used default tolerances (I have attached the .mfx this time.)
urx29.mfx (10.3 KB)

Thank you.

#4

A couple of things:

  1. You have scaled the x direction quite a bit so it is distorting the image.
  2. When you look at solids velocity in TFM, it is better to filter out data where the void fraction is one or very near one, because the solids velocity has no meaning in these cells.
#5

Hi Jeff,

Thank you.

I have seen it in cases with higher solids loadings too; I will upload the results later.

In the meantime: It is seen not only in solids velocity, but also in gas velocity (please see below).

Smudge query
Smudge query1092×1001 224 KB

If we start an iteration with a different set of data (i.e. data that has been overwritten during the current time step) then the iteration may converge at a different set of maxima / minima ?

Thank you.

#6

What is the solids volume fraction in the region around y=4.25m? If it is close to the default value of dil_eps=1E-4, you may need to lower it.

#7

Hi Jeff,

This is a very dilute case (solids to gas mass ratio = 1) and the ep_s is close to the dilute threshold (1E-4) in the region as you say.

If I observe this at higher solids loadings I will upload the details.

Thank you.