01 Edem Fluent

DEM Solutions Training: EDEM-CFD Coupling Module for FLUENT Introduces the EDEM-CFD Coupling Module for FLUENT, how it works, and some of the module’s features

Revision 2.1/1

EDEM-CFD Coupling Module for FLUENT

DEM Solutions Training: EDEM-CFD Coupling Module for FLUENT

1

Overview  EDEM  DEM software used for integrating particle, fluid, and machine dynamics

 FLUENT  Configurable single-phase and multi-phase algorithms  Provides a framework for coupled software development

 EDEM-CFD Coupling Module for FLUENT  EDEM couples to Fluent with Eulerian or Lagrangian coupling  EDEM replaces approximation of solid phase in Fluent with explicit calculation of particle dynamic

2

Page 2 of 15


EDEM-Fluent Process Flow DEM timestep(s) started at end of fluid simulation timestep Calls EDEM

Fluid iterated to convergence

Forces on fluid from particles are introduced into fluid through a series of momentum sinks

Drag forces on particles calculated using data extracted from fluid mesh cells

Particle positions input into Fluent

Particle positions updated

3

Scheme Panel

4

Page 3 of 15


Coupling Methodologies

5

Euler-Lagrangian Coupling Method  EDEM’s Lagrangian method similar to Fluent’s DPM (Discrete Phase Model)  Only momentum is exchanged between the two phases  Lagrangian coupling method best used when the solid fraction is low (less than 10%)  Lagrangian coupling faster to calculate relative to Eulerian coupling

6

Page 4 of 15


Euler-Euler Coupling Method  Eulerian coupling method better than Lagrangian for flows with a higher solid fraction  Energy not transferred during calculation of the coupling; only mass and momentum need to be conserved  EDEM prevents particles from moving during the fluid phase and removes all other phasic interaction  Particle forces and positions updated in DEM phase

7

Momentum Exchange Momentum is exchanged between solid and fluid phases for both Eulerian and Lagrangian coupling methods: 

CFD iterated to convergence for a timestep



EDEM takes control of the simulation and performs one or several iterations. EDEM particle positions are updated due to contact forces, gravity and additional forces applied by the fluid



Control passed back to Fluent. A momentum sink is added to each of the mesh cells to represent the effect of energy transfer from the DEM particles

8

Page 5 of 15


Volume Fraction Exchange  Particle volume fraction transfered to Fluent for Eulerian coupling only  Ideally EDEM particles are smaller than Fluent mesh cells  A single particle in EDEM can be made from multiple spheres. The volume of a multi-sphere particle is passed to Fluent (the drag model assumes a particle is spherical)

9

Relaxation Factors  Momentum transfer and volume fraction transfer can be under-relaxed to provide greater stability on the CFD side: Pnew = xPcalculated + (1 – x)Pold Where P is the momentum, x is the relaxation factor

 Typically relaxation factors vary from 0.1 (very slow dense phase simulations) to 1.0 (fast flowing dilute simulations)

10

Page 6 of 15


Sample Points 





Sample points allow large particles to transfer volume fraction to mesh cells up to 2x smaller than the particles Increasing the sample points increases the accuracy and stability of Eulerian simulations A sample point of 1 is enough where particles are more than 60% smaller than the mesh cells

Small mesh cells unlikely to contain particle volume

m

α DEM =

∑SV i =1

i

particle

N sampleV fluidcell 11

Timesteps  CFD typically uses a timestep of orders of 10 – 100 greater than DEM  Performing a single CFD timestep for every DEM timestep is inefficient, so a ratio is chosen:  Must be small enough to assume the fluid flow pattern doesn’t change significantly during the DEM iteration  Each particle should be in a fluent mesh cell for a minimum of three iterations  Ratio must be such that the CFD can successfully iterate to convergence on return from the DEM  Smallest possible edem:fluent timestep ratio is 1:1

 EDEM and Fluent timesteps synchronized automatically  EDEM time data stored in each Fluent case file 12

Page 7 of 15


Drag Models

13

Drag Models The EDEM-CFD Coupling Module for FLUENT has several drag model options:  Freestream Drag Model – free stream drag model modified to calculate forces on particles  Ergun and Wen & Yu – modified freestream drag  Di Felice – adds porosity correction term  User-Defined – write and use your own plug-in drag models

14

Page 8 of 15


Drag Calculation   

Bounding sphere used to calculate the drag force for non-spherical particles Limited particle size since fluid data is taken from the mesh cell containing the center of the particle The drag coefficient, CD, is dependent on the Reynolds number, Re:

Re =

ρvl µ

Where ρ is the fluid density, μ is the viscosity, l is the diameter of the particles bounding sphere, and v is the relative velocity between the fluid and the particle



Buoyancy must also be taken into account:

FB = − ρVg 15

Drag Model Theory  Freestream Drag Model Free steam drag for a sphere is calculated from:

F = 0.5CD ρ Av |v | Where A is the projected area of the sphere



Ergun and Wen & Yu Drag Model According to the work of Twente:

F=

βV v 1− e

Where V is the volume of the sphere, e is the voidage of the Fluent cell, and:

16

Page 9 of 15


Drag Model Theory  Di Felice Drag Model Adds porosity correction term to the freestream drag model to take into account the effects on drag of neighboring particles. Calculated from:

Where e is the voidage/porosity and x is given by:

17

Other Models

18

Page 10 of 15


Lift Models The EDEM-CFD Coupling Module for FLUENT has several lift model options:  Saffman Lift – lift due to velocity gradient in fluid flow  Magnus Lift – lift due to particle rotation  Fluid-induced Torque – handles particle rotation due to fluid shear

19

Heat Transfer Models The EDEM-CFD Coupling Module for FLUENT includes both convective and radiative heat transfer models as optional licensable features: Convective Heat Transfer  Ranz & Marshall – suitable for a particle Reynolds number of up to 200  Gunn – more suitable for granular flows  Li & Mason – set the exponential constant to fine-tune your model Radiative Heat Transfer  Set the surface emissivity of particles 20

Page 11 of 15


User-Defined Functions EDEM

UDF

FLUENT

EDEM and Fluent are linked using User-Defined Functions      

Export the meshed geometry from Fluent case files directly into EDEM Periodically update volume fraction and particle drag forces Calculate simulation time and end time for EDEM Call EDEM to update particle positions and forces Transfer momentum source terms to Fluent Transfer mass source terms to Fluent (for Eulerian coupling)

21

Examples

22

Page 12 of 15


Examples Fluidized bed

Sedimentation

Entrainment 23

Examples Aerating Chemical Slurry

Pneumatic Conveying

24

Page 13 of 15


Examples - Combustion  Particles input at the top of the model at a set mass flow rate  Fluid velocity high enough to mix the particles and prevent them from leaving the base of the domain  Objectives: Find fluid flow velocity where particles fall into the gas inlet Gas inlet

Gas inlet

V=50 m/s

V=60 m/s

25

Examples – Particle Strings  Modeling transportation of strings of flexible bonded particles using a fluid drag model  Used to investigate required particle bond strengths and fluid flow rates  Result: Improved flow efficiency

26

Page 14 of 15


Summary         

Coupling through Lagrangian or Eulerian model In-built or user-defined drag models In-built lift and heat transfer models Eulerian stabilized for particles up to twice grid cell size Automated link between EDEM and FLUENT Automatic model set-up Time-step matching Joint post-processing using EnSight Pause-and-continue

27

Page 15 of 15

01 Edem Fluent

Recommend Documents