Do we need subgrid drag models for Do we need subgrid drag models for

continuum and discrete simulations?

Sofiane Benyahia f y
US DOE, National Energy Technology Laboratory
MultiphaseFlowScienceWorkshop Multiphase Flow Science Workshop
May 4-6, 2010
Collaborations
- Research was conducted in collaboration with two groups:
Filtered drag models: Y. Igci and S. Sundaresan
1
(Princeton University)
EMMS drag model: W. Wang and J . Li
2
(Institute of Process
Engineering, Chinese Academy of Sciences) g g, y )
1
Igci, Y.; Andrews, A.T.; Sundaresan, S.; Pannala, S.; OBrien, T. AIChE J. 2008; 54; 1431-1448.
2
L B W W Li J Ch E S i 2009 64 3437 3447
2K-2571
2
Lu, B.; Wang, W.; Li, J . Chem. Eng. Sci. 2009; 64; 3437-3447.
2
Goals, Objectives, and Challenges
- Conduct fast and accurate gas-solids flow simulations
i id using coarse grids.
- Implement EMMS and filtered drag models in MFIX
- Conduct comparative study of these drag models with
experimental data
3
- Challenges:
Subgrid models are still being developed
Lack extensive validation Lack extensive validation
Consideration of other effects (e.g. polydispersity,
particle-wall interaction)
2K-2571
3
3
Benyahia, S. On the effect of subgrid drag closures. Ind. Eng. Chem. Res. DOI: 10.1021/ie900658k
Simulation Conditions
2
Twogassolids
outlets
0.35m
0.14m
Process temperature 297 K
Process pressure 1.01 x 10
5
Pa
Air density ~1.2 kg/m
3
0 09 m
Air viscosity 1.8 x 10
-5
Pas
Gas-phase turbulence length-scale 0.01 m
Solids density 930 kg/m
3
0.09m
60cells
10.5m
450cells
Particle diameter 54micron
Single-particle terminal velocity 0.074 m/s
Particle-particle restitution coefficient
0.9
P ti l ll tit ti ffi i t
2.78m
Particle-wall restitution coefficient
0.7
Specularity coefficient 0.0001
Particle-particle angle of internal
friction
pi/6
Gas inlet
Twosolidssideinlets
0.35m
0.14m
eps_mf =0.4
eps =0.3,Vs:computed
friction
Particle-wall angle of friction pi/16
Void fraction at maximum packing 0.4
Void fraction at minimum fluidization 0.6
2K-2571
4
Gasinlet
Vg=1.5m/s 2
Lu, B.; Wang, W.; Li, J . Chem. Eng. Sci. 2009; 64; 3437-3447.
Instantaneous values of friction coefficient
1.6
1.2
1.6
filtered
EMMS
WenYu
0 4
0.8
H
d
0
0.4
0.5 0.6 0.7 0.8 0.9 1 0.5 0.6 0.7 0.8 0.9 1
Voidage
2K-2571
5
Void fraction contour profiles
Wen-Yu drag
Filtered drag EMMS drag
Wen-Yu drag
Filtereddrag EMMS drag
2K-2571
6
g
Filtered drag EMMS drag
Instantaneous Time-averaged
Solids mass flux through outlets
400 50
Solids mass
flux (kg/m
2
s)
Solids mass
flux (kg/m
2
s)
EMMS
filtered
200
300
30
40
filtered
Wen-Yu
100
200
10
20
0 0
0 10 20 30 40
Time (sec)
Drag model Wen-Yu Filtered EMMS Experimental
Solids mass flux
(kg/m
2
s)
205 12.3 16.9 14.3
2K-2571
7
Standard deviation
(kg/m
2
s)
25.2 5.64 5.59 -
Vertical void fraction profiles
10
10
6
8
m
)
EMMS
Filtered
WenYu
experiment
6
8
m
)
EMMS
Filtered
WenYu
experiment
4
H
e
i
g
h
t

(
m
4
H
e
i
g
h
t

(
m
2
2
0
0.7 0.8 0.9 1
Voidfraction
0
0.7 0.8 0.9 1
Voidfraction
2K-2571
8
Numerical data based on
computed pressure gradient
Numerical data based on
computed void fraction
Horizontal profiles at 7 m above inlet
Solidsmassflux
(kg/m
2
s)
0.2
EMMS
Filtered
WenYu
0
400
(kg/m
2
s)
0.1
c
s
400
EMMS
Filtered
WenYu
0
0 0.03 0.06 0.09
Width(m)
800
0 0.03 0.06 0.09
Width(m)
2K-2571
9
Future work
- Why is the pressure gradient under-
di t d t th b tt ti f
8
10
HiFriction(phi=0.01)
predicted at the bottom section of
riser?
6
H
e
i
g
h
t

(
m
)
LoFriction(phi=0.0001)
experiment
Increase particle-wall friction in
J ohnson-J ackson BC
2
4
H
Effect of polydispersity is currently
investigated by considering three
particle sizes of 30, 60, and 120
0
0.7 0.8 0.9 1
Voidfraction
microns
2K-2571
10
Are subgrid models needed for discrete
i l i l i ? particle simulations?
- There is no reason why not if fluid flow is not properly
resolved (coarse grids)
- The same approach
3
used for continuum models can be
applied for discrete models, i.e. does the slip velocity
increase from that of homogeneous state as Eulerian
grid is refined?
2K-2571
11
3
Agrawal et al. The role of meso-scale structures in rapid gas-solids flows. J. Fluid Mech. 445 (2001) 151.
MP-PIC and continuum models MP PIC and continuum models
( )
( ) 0 = V +
c
c
g g g
g g
V
t

c
c
ct
( )
( ) ( ) ( ) g V x V P V V
t
V
g g
N
p
p p g
c
p
p g g g g g g
g g g
T

c
|
|
| t c c
c
+ V + V =
(
(

V +
c
c

=1
Continuum gas-phase
conservation equations
( )
( ) ( ) g V V P P V V
t
V
s s s g p s s g s s s s s
s s s

c | c c c
c
+ + V V =
(

V +
c
c
Continuum solids-phase
momentum equation

( ) ( )
s p
s
p p g
p
p
p
g p
P
V x V
P
g
t d
V d
c
|

V
+
V
=

Newtons law of motion for

discrete particles
( )
65 . 2
4
3

=
g
p p g g g
D p
d
V x V
C c
c
|

>
<
=
-
max
max
max
old
s s
s s
s
s
P
P
P
c c
c c
c c
c
|
2K-2571
12
4
pt
d

>
s s s
P c c
Simulation conditions
Fully periodic domain
Air density 1.3 kg/m
3
Air viscosity 1.8 x 10
-5
Pas
S lid d i 1500k /
3
0.4 m
Solids density 1500 kg/m
3
Particle diameter 75 micron
Void fraction at maximum packing 0.4
Eulerian grid density 8x32, 16x64, 32x128, 64x256
2K-2571
13
0.1 m
Snapshots of particle position-velocity (8x32)
2K-2571
14
1.0 sec 1.5 sec
2.0 sec 10 sec 20 sec
Snapshots of particle position-velocity (16x64)
2K-2571
15
1.0 sec
1.5 sec 2.0 sec 10 sec
20 sec
Snapshots of particle position-velocity (32x128)
2K-2571
16
1.0 sec 1.5 sec
2.0 sec
10 sec
20 sec
Snapshots of particle position-velocity (64x256)
2K-2571
17
1.0 sec
1.5 sec 2.0 sec
10 sec
20 sec
Transient Favre-averaged slip velocity
2K-2571
18
Slip-velocity for continuum & MP-PIC models
Grid size 8x32 16x64 32x128 64x256
Continuum slip
velocity & St. Dev.
(cm/s)
18.7
0.16
22.6
3.9
49.6
10.8
52.0
8.7
MP-PIC slip
velocity & St. Dev.
(cm/s)
18.5
0.005
18.5
0.005
50.7
15.1
55.4
13.8
2K-2571
19
Summary
- Sub-grid corrections are demonstrated to be
both needed and useful for both continuum
and discrete approaches to gas-solids flow
modeling modeling
- Our limited quantitative analysis indicates that
the same subgrid corrections developed for
continuum models may be used with DPM
2K-2571
20