Approximate particle size distribution (PSD) in TFM

MFiX
1

Hi folks,
by adding several phases in TFM it is possible to get a sort of “discretized” PSD - and by this a polydispersed model. By experience I know that added complexity or fidelity is not necessarily inherently good, and in this case meaning a higher number of bins or buckets, although mathematiclly it would represent the curve better. Is there a sweet spot here? Should one rather aim for 4 bins instead of 20, or does this not matter? It would be great with some insight here.

Adding this general image of this. Although it would be easier to see the curve-bin-approximation when not using a cumulative curve like here, but more like a normal distribution:

llgeneralfigure-3
llgeneralfigure-32040×1199 65.3 KB

2

The more bins you have the longer it takes… why not starting with picking say the D32 (Sauter mean) first? What system are you simulating? You want to make sure that if you discretise the distribution, your D32 and D43 remain essentially the same as the continuous psd (it should if you just coarsen the bins). https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018JB016544

3

I appreciate your suggestions, Eric, starting off with the D32 seems like a good idea - will read your paper in detail. The system is supposed to be a redox reaction of Geldart A particles, although I haven’t come to the reactions just yet.

1 Like