DFBI vs DEM

SiemenDiemon · December 1, 2020, 18:51

Hi everyone,

I'm trying to decide on a method for modelling particles in fluid flow in STAR-CCM+. I'm new to this but having read through the documentation it seems the choice is between DEM (discrete element method) or DFBI (dynamic fluid-body interaction). I am being advised that DFBI is the more accurate of the two (and hence also more computationally intensive) but am curious what everyone else thinks. I'm guessing there would be cases where one is preferable over the other. For example the particle is morphable which makes me err towards DFBI.

As an aside, it also seems like DEM makes uses of some fairly complicated empirically derived formulae for calculating forces on a particle (e.g. lift coefficients) whereas DFBI seems to not do this. Am I understanding correctly that DFBI then just calculates the forces on a particle via the stress tensor?

Cheers!

ping · December 3, 2020, 04:20

dbfi is not really meant for use as a method to create and model particles since you would generally only have one or free bodies, but i have seen it done with maybe a dozen balls bouncing off each other - painful to setup unless a macro is used. the flow around the bodies is captured much better assume your overset meshes are good. but the contact physics in dbfi is very basic compared to dem.
dem is designed to model many particles together with complex interactions including contact, breakup, thermal, chemistry etc but as the particle numbers get up to 1e6 it gets very slow unless you have a big cluster.

SiemenDiemon · January 24, 2021, 19:14

Thanks Ping.

Yep, that seems to be the case. It's a shame that DEM is more suited for modelling particles as I'm struggling to find a way in the DEM framework to integrate the stress tensor over the surface of each particle to find the force more exactly, rather than using the built-in DEM force models which make some assumptions about the geometry of the particles.

December 1, 2020, 18:51	DFBI vs DEM	#1
SiemenDiemon New Member Join Date: Sep 2020 Posts: 18 Rep Power: 6	Hi everyone, I'm trying to decide on a method for modelling particles in fluid flow in STAR-CCM+. I'm new to this but having read through the documentation it seems the choice is between DEM (discrete element method) or DFBI (dynamic fluid-body interaction). I am being advised that DFBI is the more accurate of the two (and hence also more computationally intensive) but am curious what everyone else thinks. I'm guessing there would be cases where one is preferable over the other. For example the particle is morphable which makes me err towards DFBI. As an aside, it also seems like DEM makes uses of some fairly complicated empirically derived formulae for calculating forces on a particle (e.g. lift coefficients) whereas DFBI seems to not do this. Am I understanding correctly that DFBI then just calculates the forces on a particle via the stress tensor? Cheers!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Modeling the gas flow through the packed bed of DEM particles	Legat	STAR-CCM+	6	January 17, 2024 09:58
Two-Way coupled DEM with heat transfer - Which is good?:+ LIGGHTS/OpenFOAM/CFDEM/<??>	yesaswi92	Main CFD Forum	1	November 27, 2023 05:49
Two-Way coupled DEM with heat transfer - Which is good? - disscus now	minhbaba	Main CFD Forum	0	December 1, 2018 00:50
Aligning a region with dfbi motion of body without dfbi superposed rotation	rockydongre	STAR-CCM+	2	July 21, 2014 02:11
Wind Turbine DFBI	BKaiser	STAR-CCM+	8	August 1, 2013 21:47

December 3, 2020, 04:20		#2
ping Senior Member Ping Join Date: Mar 2009 Posts: 556 Rep Power: 20	dbfi is not really meant for use as a method to create and model particles since you would generally only have one or free bodies, but i have seen it done with maybe a dozen balls bouncing off each other - painful to setup unless a macro is used. the flow around the bodies is captured much better assume your overset meshes are good. but the contact physics in dbfi is very basic compared to dem. dem is designed to model many particles together with complex interactions including contact, breakup, thermal, chemistry etc but as the particle numbers get up to 1e6 it gets very slow unless you have a big cluster.

January 24, 2021, 19:14		#3
SiemenDiemon New Member Join Date: Sep 2020 Posts: 18 Rep Power: 6	Thanks Ping. Yep, that seems to be the case. It's a shame that DEM is more suited for modelling particles as I'm struggling to find a way in the DEM framework to integrate the stress tensor over the surface of each particle to find the force more exactly, rather than using the built-in DEM force models which make some assumptions about the geometry of the particles.