COUPLING OF DEM AND CFD

SIMULATION AND EXPERIMENT

25 March 2009

Christoph Kloss
Stefan Pirker

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling
Kopenhagen
 Outline

I. The Discrete-Element-Method (DEM)

II. DEM Simulation Examples

III. DEM CFD Coupling (Model Synthesis)

IV. Coupled DEM-CFD Simulation Examples

IV. Conclusions and Future Prospects

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 2
Kopenhagen
 I.

I: The Discrete-Element-Method

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 3
Kopenhagen
 Principles of DEM
In DEM, every single particle ist tracked in Lagrangian
Frame. The force balance is integreated explicitely.

DEM manages information about each individual particle

(mass, velocity,...) and the forces acting on it. It can also
take into account the particles shape, rather than assuming
that all particles are spherical.

Normally, soft-sphere contact models where particles are

allowed to slightly overlap (<0.5%) are used to resolve
every single contact

Normal Force:
Hertzian model: F~ 3/2 k
Cundall & Strack, 1979: F~ k

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 4
Kopenhagen
 Principle of DEM
Forces that can be included:
Contact forces (soft-sphere approach):
particle-particle, particle-wall
Gravity

Fluid drag

Electrostatic and magnetic forces

Contact force: Modelling of particles Cohesion

as spring-damper systems Chemical bonds

Characteristics of DEM:
DEM is limited by CPU resources and low time-steps

Boundary conditions like velocity inlet or pressure outlet are

not available and have to be programmed manually
Stiff materials require short time-steps

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 5
Kopenhagen
 DEM vs. Eulerian Granular Models

Eulerian Models
Calculation of conservation equations on a grid
Properties of indivual particles are smeared out (loss of
information
Closure relations are used (e.g. granular pressure and
temperature)
A yield criterion is used
Eulerian models are CPU-efficient

DEM
Closer to the physics by resolving the micro-scale
Contact laws are used
Efficient contact detection is important
CPU ressources limit applicability

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 6
Kopenhagen
 DEM Simulations: Practical Issues

Time-Step has to be a fraction (10%?) of the Rayleigh time:

Tr= r sqrt(/G)/(0.163 +0.8766)

(r: radius, density, G: shear modulus, poisson ratio)

Example (glass): r=4 mm, =2500 kg/m, G=26 GPa, =0.25 t=4.2e-7s

Simulation time varies with material properties

A maximum penetration of 0.5% of dp is allowed:

so that vmax t= 0.005 dp,min

Example: vmax=10 m/s, dp,min=1mm t=5e-7 s

Simulation time varies with dp,min and vmax

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 7
Kopenhagen
 DEM Simulations: Practical Issues

Simulation time increases when long-lasting particle

contacts come into play (contact search)

Practical numbers:

Cundall number (=particle time-steps / CPU second) of

~200.000 on desktop computers
Example: 100.000 particles , t=5e-6,
1 hour computation on one CPU=36 ms real-time

Particle limit of ~1 mio. on desktop computers

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 8
Kopenhagen
 II.

II: DEM Simulation Examples

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 9
Kopenhagen
 I. DEM Simulation Example
Hopper Discharge

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 10
Kopenhagen
 II. DEM Simulation Example
Particle Charging / Bed Formation

distribution density function

0.4
simulation - fine
0.35 simulation - coarse
experiment - coarse
experiment - fine
0.3

0.25

0.2

0.15

0.1

0.05

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

x in m

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 11
Kopenhagen
 III. DEM Simulation Example
Simulation of Bed Formation: API-Implementation

time-dependant boundaries

Particles are
kept in place
by controller

Partikel have
been removed

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 12
Kopenhagen
 III.

III: DEM-CFD Coupling

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 13
Kopenhagen
 Parallel DEM-CFD Coupling
Principle

init particle data

EDEM (C++)
FLUENT (C)
including fluid drag
1 time-step
100 time-steps
t ~ 1e-3s
t ~ 1e-5s

end
flow data

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 14
Kopenhagen
 Parallel DEM-CFD Coupling
Program FlowChart

EDEM calculates FLUENT calculates

start 100 time-steps, 1 time-step
and transfers particle
FLUENT fetches EDEM
data to FLUENT
particle data
FLUENT searches cell
each particle is in
FLUENT calculates
Parallel Processing is necessary to volume fraction and
deal with large-scale geometries! momentum coupling
FLUENT transfers flow
data to EDEM

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 15
Kopenhagen
 IV.

IV: Coupled DEM-CFD Simulation

Examples

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 16
Kopenhagen
 I. DEM-CFD Coupling Example
Proof of Concept - Acceleration of Single Particle
3

2.5

water flow, u = 3 m/s

2

v in m/s
1.5

glass particle
1
d = 4 mm
u0 = 0 m/s EDEM-FLUENT Coupled Simulation
Analytical Solution
0.5

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
t in s

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 17
Kopenhagen
 II. DEM-CFD Coupling Example
Trajectory Segregation

flow vectors
in the midplane
umax=1.7 m/s

detail
detail

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 18
Kopenhagen
 II. DEM-CFD Coupling Example
Trajectory Segregation

flow vectors
in the midplane
umax=1.7 m/s

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 19
Kopenhagen
 III. DEM-CFD Coupling Example
Hopper Discharge With Standpipe

Hopper geometry
(all lengths in mm)

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 20
Kopenhagen
 III. DEM-CFD Coupling Example
Hopper Discharge With Standpipe

particle velocity in m/s fluid velocity in m/s

Discrete Element Method Christian-Doppler Laboratory
25 March 2009 on Particulate Flow Modelling 21
Kopenhagen
 III. DEM-CFD Coupling Example
Hopper Discharge With Standpipe

fluid accelerates
particles

particles
accelerate fluid

z-momentum source on fluid in N/m solids fraction

Discrete Element Method Christian-Doppler Laboratory
25 March 2009 on Particulate Flow Modelling 22
Kopenhagen
 III. DEM-CFD Coupling Example
Hopper Discharge With Standpipe | Experimental Validation
Pressure of the fluid phase
0

-0.5

simulation
-1
measurement

-1.5
p in Pa

-2

-2.5

-3

-3.5

-4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
z in m

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 23
Kopenhagen
Z1

IV. DEM-CFD Coupling Example

Partial Fluidized Bed

side view

d in = 4 mm
bed 14 cm
7 cm

top view
3.4 cm
4 cm
Gas: 6 Nm/h = 0.02 kg/s, vIn=132m/s
30.5 cm
Discrete Element Method Christian-Doppler Laboratory
25 March 2009 on Particulate Flow Modelling 24
Kopenhagen
Slide 24

Z1 Bohrung leicht schief:

Aussen 8mm vom Rand entfernt innen 6mm
(auf 6cm Wandstrke)
ZID; 23-02-2009
 IV. DEM-CFD Coupling Example
Partial Fluidized Bed

Solids fraction in the injection plane

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 25
Kopenhagen
 V. DEM-CFD Coupling Example
Pneumatic Conveying

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 26
Kopenhagen
 V. DEM-CFD Coupling Example
Pneumatic Conveying

particles

gas

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 27
Kopenhagen
 V. DEM-CFD Coupling Example
Pneumatic Conveying

left middle right

particles

gas

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 28
Kopenhagen
 V. DEM-CFD Coupling Example
Pneumatic Conveying
particle volume fraction at pos. 1 from coupled DEM-CFD simulation
0.14
particle volume fraction at pos. 3 from coupled DEM-CFD simulation left
0.12 right
0.14 middle
0.1 no magnus force
left with magnus force
0.08
0.12 right

y in m
0.06
middle
0.04
0.1 no magnus force
0.02
with magnus force
0
0.08 0 0.5 1 1.5 2 2.5 3 3.5
y in m

volume fraction -3
x 10
particle volume fraction at pos. 2 from coupled DEM-CFD simulation
0.14
0.06 left
0.12 right
middle
0.1 no magnus force
0.04 with magnus force
0.08

y in m
0.02 0.06

0.04

0 0.02
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
0
volume fraction -3 0 0.5 1 1.5 2 2.5
x 10 volume fraction
x 10
-3

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 29
Kopenhagen
 V. DEM-CFD Coupling Example
Pneumatic Conveying
particle velocity at pos. 2 from coupled DEM-CFD simulation
particle velocity at pos. 3 from coupled DEM-CFD simulation 0.14
left
0.14 0.12 right
middle
left 0.1 no magnus force
with magnus force
0.12 right 0.08

y in m
middle 0.06

0.1 no magnus force 0.04

with magnus force 0.02

0.08
y in m

0
0 1 2 3 4 5 6 7 8
particle velocity
particle velocity at pos. 1 from coupled DEM-CFD simulation
0.06 0.14
left
0.12 right
middle
0.04 0.1 no magnus force
with magnus force
0.08

y in m
0.02 0.06

0.04

0 0.02
0 2 4 6 8 10
0
particle velocity 0 1 2 3 4
particle velocity
5 6 7

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 30
Kopenhagen
 V. DEM-CFD Coupling Example
Pneumatic Conveying
volume fraction at pos. 3 particle velocity at pos. 3

0.12 0.12

measurement
0.1 simulation

0.08
0.1 measurement

y in m
simulation 0.06

0.04

0.08
0.02

0
y in m

0 1 2 3 4 5 6 7 8 9 10
particle velocity in m/s
0.06 volume fraction at pos. 2
0.12
measurement
simulation
0.1

0.04
0.08

y in m
0.06

0.02
0.04

0.02

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
volume fraction -3 volume fraction -3
x 10 x 10

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 31
Kopenhagen
 VI.

VI: Conclusions and Prospects

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 32
Kopenhagen
 DEM-CFD Coupling
Conclusion

We showed that the the synthesis of DEM and CFD leads to a very
versatile tool. Many possibilities of application, applicable for
different kind of regimes:

High solids fraction (fluidized bed, hopper discharge)

Low solids fraction (trajectory segregation during charging,
pneumatic conveying).

One way coupling fluid flow induced by particle motion

(hopper discharge, trajectory segregation during charging)
One way coupling particle motion induced by fluid flow
(pneumatic conveying)
Two way coupling strong interaction between particle motion
and fluid flow (fluidized bed)

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 33
Kopenhagen
 DEM-CFD Coupling
Prospects Model Synthesis

Simulation snapshot Solids volume fraction (cap at 3%)

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 34
Kopenhagen
 DEM-CFD Coupling
Prospects Model Synthesis (DEM-DPM Coupling)

DPM Source
DEM Sink DEM-DPM
boundary

DEM Source DPM-DEM

DPM Sink boundary

DPM Injection
EDEM FLUENT
Discrete Element Method Christian-Doppler Laboratory
25 March 2009 on Particulate Flow Modelling 35
Kopenhagen
 DEM-CFD Coupling
Prospects Model Synthesis

time-dependant boundaries

Particles are
kept in place
by controller

Partikel have
been removed

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 36
Kopenhagen
 DEM-CFD Coupling
Prospects

As DEM is very CPU intense, it is desired to extend our coupling

software to incorporate further models for granular media for cases
where good results can be achieved with a less CPU demanding
method (model synthesis). These could include FLUENTs DPM,
FLUENTs EuEuGran model, and further particulate models.
As the applicability of such a coupled method strongly depends on
computational power, the parallelization is a key issue.

For these reason, our in-house DEM-CFD coupling code is

Fully parallel (on both EDEM and FLUENT side)

Extendable to incorporate further models

Further efforts will be taken in order to make reduce the costly DEM
method to regions where the key physical phenomena occur.

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 37
Kopenhagen
 Thank you for your attention

Discrete Element Method Christian-Doppler Laboratory

25 March 2009 on Particulate Flow Modelling 38
Kopenhagen