Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1
ANNUAL REPORT 2005Meeting date: June 1, 2005
Jun Aoki (M.S. Student) &Brian G. Thomas
Department of Mechanical & Industrial EngineeringUniversity of Illinois at Urbana-Champaign
Ladle Mixing andInclusion Removal by Bubbles
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 2
Acknowledgments
• The Continuous Casting Consortium• Dr. L. Zhang and Lab Members
• J. Peter and Dr. K. D. Peaslee, University of Missouri-Rolla
• Nucor Steel• Nucor Yamato Steel
• U.S. Department of Energy• National Science Foundation
• Fluent Inc., Lebanon, NH
2
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 3
BackgroundKey Phenomena in Ladle Refining
Slag entrainment
Re-oxidation
Inclusion collision & coarsening
Inclusion absorption into slag
Inclusion attachmentto bubble
Mixing
Alloy melting
Flow pattern
Slag-metal reactione.g.: Al+FeO Al2O3+Fe
Alloy addition
Gas injection
Slag layer
Inclusion
Bubble
Liq. steel
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 4
Outline
• Model 1: Multiphase ladle flow simulation
• Model 2: Alloy melting and mixing in ladle
• Model 3: Inclusion attachment to bubble
• Model 4: Inclusion removal by bubbles in ladle flow
• Model 5: Inclusion absorption at slag-metal interface
3
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 5
Model 1: Multiphase Ladle Flow Simulation
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 6
Model ApproachEulerian-Lagrangian Multiphase Model
Bubble
Liq. steel
Gas injection
Lagrangian:
Trajectories of Individual Bubbles
Eulerian:Molten steel flow pattern
Interaction:• Phase fraction• Momentum exchange• Source of turbulence
4
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 7
Model Configuration
Drag force
Buoyancy force
Virtual mass force
Pressure gradient force
Drag coefficient
Lift force
Bubble diameter
Bubble density
Source of turbulence from bubbles
Phase fraction
Momentum source
FLUENT UDF
Fluid velocityturbulence
phase fraction
Lagrangian bubble trajectory calculation
Eulerian fluid flow calculation
Local static pressure
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 8
Equations for Multiphase Turbulent Flowin Eulerian Frame of Reference
( ) ( ) 0=⋅∇+∂∂
lllll ut
ραρα
( ) ( ) ( )[ ] bltlllllllll Fupuuut
+∇+⋅∇+−∇=∇⋅+∂∂ µµαραρα
Continuity
Momentum conservation
Transport equation for turbulent kinetic energy
Transport equation for dissipation rate
Turbulent viscosity Generation of k
αl : Phase fraction of liquidul : Velocity of liquidρ l : Density of liquidp : Pressureµl : Viscosity of liquidµt : Turbulent viscosityC1, C2, Cµ, σk, σε :
Standard Empirical constants(=1.44, 1.92, 0.09, 1.0, 1.3)
Csk, Csε: Empirical constants(defined by matching withexperimental data0.1~0.7 suggested by (1))
ερµ µ
2kC lt =
kbsllt
llll Sk
Ck
CGk
Cut k
εεραεαεσµαεερα εε
+−+
∇⋅∇=
∇⋅+∂∂ 2
21
kbsklllk
tllll SCGkku
tk
k+−+
∇⋅∇=
∇⋅+∂∂ εραα
σµαρα
j
il
i
jl
j
iltk x
ux
uxu
G∂∂
∂
∂+
∂∂
= ,,,µ
Momentum source from bubbles
Turbulence source from bubbles
1. S. T. Johansen and F. Boysan, Met. Trans. B, Vol. 19B, 1988.
5
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 9
Equations of Motion of a Bubble(Lagrangian Approach)
( )iblibib
Dldrag uu
dCF ,2
,,4Re3
−=ρµ
( )y
ib
libbuoyancy egF
,
,
ρρρ −
=
( ) liblib
lLlift uuuCF ×∇×−= ,
,ρρ
llib
lgradientpressure uuF ∇⋅=
,_ ρ
ρ
gradientpressureliftmassvirtualbuoyancydragib FFFFF
dtud
__, ++++=
( )iblib
lVMmassvirtual uu
dtdCF ,
,_ −=
ρρ
Forces acting on a bubble
Drag force
Buoyancy force
Virtual mass force
Lift force
Pressure gradient force
ub,i : Velocity of ith bubbledb,i : Equiv diameter of ith bubbleρb,i : Density of ith bubbleCD : Drag coefficientCVM : Virtual mass coefficient (=0.5)CL : Lift coefficientxb,i : Position of ith bubbleζ :Normal Gaussian distributed
random number (σ=1)iR : Unit vector in random direction
Bubble position Turbulent effect
dtux ibib ∫= ,, Rll ikuu 32ς+=
78.000165.0 −= bLC α
1. S. W. Beyerlein et al., Int. J. Multiphase Flow, Vol.11, 1985.
(1)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 10
Drag Coefficient and Terminal Velocity of Bubble
8We38
We1.2065Re
3We
Re100Re
3.6
100Re49.0Re
68.20
49.0ReRe16
6.2
385.0
643.0
>=
>=
<=
<<=
<=
D
D
D
D
D
C
C
C
C
C
l
iblibl uudµ
ρ ,,Re−
=
lb
iblibl uudσ
ρ2
,,We−
=
Drag coefficient(1)
Weber number
Reynolds number
0.01
0.1
1
0.1 1 10 100Volumetric equivalent diameter db (mm)
Term
inal
vel
ocity
uT (m
/s)
Clift et al. "Contaminated Water"Air in waterN2 in Wood's metalAr in liquid steel
sphere spheroid
spherical cap
db=6mm db=42mm
1. J. T. Kuo and G. B. Wallis, Int. J. Multiphase Flow, Vol. 14, 1998.2. R. Clift et al., Bubbles, Drops, and Particles, Academic Press, New York 1978.
(2)
(2)
6
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 11
Source of Fluid Momentum and Turbulencedue to Bubble Motion
Source of momentum
[ ]∑ ∆+++∆
−=cellbN
i
sibgradientpressureliftmassvirtualdrag
cellb tQFFFF
VF
,
,__1
Source of turbulence
Nb,cell : Number of bubbles in the computational cell
Qsb,i : Flow rate of the bubble
stream∆t : Bubble timestep∆Vcell : Cell volume
[ ] ( ) sib
N
iiblgradientpressureliftmassvirtualdrag
cellkb QuuFFFF
VS
cellb
,,__
,1 ∑ −⋅+++∆
−=
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 12
Bubble Density and Diameter
( )TT
pyHgp il
bib0
0
00,
−+=
ρρρ
TT
pgHp l
bbottomb
0
0
00
ρρρ +=
( )bottomb
il
lbottomb
ib
bottom
ib dyHgp
gHpdd b 3
0
03
,, −+
+==
ρρ
ρρ
Bubble Density
Bubble density at inlet nozzle (ladle bottom)
Bubble diameter (volumetric equivalent diameter)
ρb,i : Density of ith bubbleρ0b : Density of the gas in STPp0 : Standard pressure (101,325Pa)T0 : Standard temperature (273K)T : Liquid temperature in the vesselH : Bath depthyi : Vertical location of the ith bubble
7
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 13
Bubble Size Distribution
( ) ( )[ ]( )[ ]
−−= 2
2
ln2lnlnexp
sddPP
mbb
m
0007.004.02.02
+
=
gQd bm
b
Bubble number distribution(assumed logarithmic normal function)(1)
Diameter with the maximumnumber distribution
1. Y. Xie et al., ISIJ Int., Vol. 32, 1992.
0.01
0.1
1
10
100
1 10 100
Volumetric equivalnet diameter db (mm)
Prob
abili
ty (%
)
Number distributionVolume distribution
N2 in Wood’s metalQb=204cm3/s (at nozzle)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 14
Model Validation
Wood’s metal experiment by Xie and Oeters
N2
Magnet probefor local velocity measurements
h=370mm
d=400mm
nozzle diameter = 3mm
0.0042
µPas
70940012.512.52550
M.P.ºC
ρkg/m3
Cd%
Sn%
Pb%
Bi%
Wood’s metal properties
Bath temperature = 100 ºCGas flow rate = 100~800 Ncm3/s
Y. Xie and F. Oeters, Steel Research, Vol. 63, 1992
8
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 15
Model Validation – Flow Field
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
-0.2 -0.1 0 0.1 0.2x (m)
Ver
tical
vel
ocity
uy (
m/s
)
y=0.1m calc.y=0.2my=0.3my=0.1m exp.y=0.2my=0.3m
Time averaged velocity (m/s) Comparison of vertical velocity with experiment
Experimental points : Y. Xie and F. Oeters, Steel Research, Vol. 63, 1992
Qb = 200 cm3/s (STP)→ 204(bottom)~273(top) cm3/s
(actual temperature and pressure)Csk=0.25, Csε=0.25
5.00e-01
4.50e-01
4.00e-01
3.50e-01
3.00e-01
2.50e-01
2.00e-01
1.50e-01
1.00e-01
5.00e-02
0.00e+00ZY
X
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 16
Model Validation – Gas Volume Fraction
0
0.1
0.2
0.3
-0.1 -0.05 0 0.05 0.1x (m)
Bubb
le v
olum
e fra
ctio
n
y=0.1m calc.y=0.2my=0.3my=0.1m exp.y=0.2my=0.3m
Time averaged gas volume fraction Comparison of gas volume fraction with experiment
Experimental data: Y. Xie and F. Oeters, Steel Research, Vol. 63, 1992
Qb = 200 cm3/s (STP)→204(bottom)~273(top)cm3/s
(actual temperature and pressure)Csk=0.25, Csε=0.25
9
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 17
Model ImplementationThe LMF Ladle of Nucor Yamato Steel
2832mm
305mm
254mm
406mm
2680mm
1143mm
3581mm
300mm
464mm
652mm800mm
Alloy addition directly above porous plug(randomly distributed in 400mm diameter circle region)
Alloy addition
Porous plug(φ = 64mm)
Sampling location
xy
xz
725mm
110 m-tonne ladle
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 18
Flow Field and Bubble Distribution(FLUENT 6.1 output)
Flow field (m/s) Gas volume fraction (%)Gas Flow rate = 0.113 Nm3/min→ 0.254 (bottom) ~ 0.775 (top) m3/min Csk=0.10, Csε=0.08
10
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 19
Flow Field in 3D(FLUENT6.1 output)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 20
Model 1 Summary
• An Eulerian-Lagrangian three-dimensional transient turbulent flow model is developed based on the commercial package FLUENT with extensive user defined subroutines.
• The calculated flow field and gas volume fraction match well with the experiments in Wood’s metal.
• Off-center Ar bubbling causes complex 3D swirling flow pattern in Nucor Yamato LMF ladle.
11
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 21
Model 2: Alloy Melting and Mixing in Ladle
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 22
Conditions of Alloy Addition in Nucor Yamato Steel(Experimental Heat 1)
Start alloy addition(t=0)
90secAlloy addition
37sec
Mixing 453sec
Plume
Operating Conditions
bal.0.070.0070.451.916.371.8
FeMoistureSPCSiMn
Chemical Composition of SiMn Alloy
Location of Alloy Addition
2m
0.4m
0 ~ 13mm : 5%13mm ~ 64mm : 90%~64mm : 5%
Size Distribution
Start bubbling Stop bubbling
12
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 23
Operating Conditions and Alloy Recoveriesin Nucor Yamato Steel
95.829735155215770.1708
94.724840153315770.1707
95.632734156415860.1136
95.133810156816050.1135
91.1231055152915660.1134
96.625415154115780.1133
93.030909155715900.1132
97.820380151615510.1701
FinalInitialFinalInitial
Mn recovery(%)
Oxygen (ppm)Temperature (ºC)
Ar flow rate
(Nm3/min)
Heat
Experiment: J. Peter and K. D. Peaslee, University of Missouri-Rolla
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 24
Concentration Histories duringHeat 1 Experiment
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
-100 0 100 200 300 400 500
Time since start of alloy addition (s)
Mn
& 4
*Si c
once
ntra
tion
(wt%
)
0.05
0.06
0.07
0.08
0.09C
con
cent
ratio
n (w
t%)
% Mn% Si % C
1551 ºC
slag
SiMnaddition
1516ºC
slag
Experiment: J. Peter and K. D. Peaslee, University of Missouri-Rolla
13
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 25
Concentration Histories duringHeat 2 Experiment
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
-100 0 100 200 300 400 500Time since start of alloy addition (s)
Mn
& 4
*Si c
once
ntra
tion
(wt%
)
0.05
0.06
0.07
0.08
C c
once
ntra
tion
(wt%
)
% Mn% Si % C
1590 ºC
slag
SiMnaddition
1557ºC
slag
Experiment: J. Peter and K. D. Peaslee, University of Missouri-Rolla
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 26
Estimated Time before Alloy Release(Shell Solidification and Melting)
0
5
10
15
20
25
30
0 20 40 60 80 100Steel superheat (°C)
She
ll ex
iste
nce
time
arou
nd S
iMn
allo
y (s
) Solid alloy particle
Steel shell growth
Alloy melting inside
Steel shell melting
Steel shell disappearLiquid alloy release
d=30mm
14
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 27
Equations for Alloy Melting and Mixing
( ) MMt
tMMM SC
ScDCvC
t=
∇
+−⋅∇+
∂∂ µρρρ
2. t ≥ t1 for each particle : Turbulent diffusion of alloy element CM : Mass fraction of element M
DM : Diffusion coefficient of element M in liquid steel
Sct : Turbulent Schmidt number(=1)
SM : Source or element M
sM
sAApA
TTTT
hdC
t−−
= 01 π
ρ
Steel shell growth and melting model(1) t1 : Shell existence timeCpA : Specific heat of alloyρA : Density of alloyh : Heat transfer coefficientTs : Solidification temperature of
steelT0 : Initial temperature of alloyTM : Liquid steel temperature 1. t < t1 for each particle :
Alloy particle transport with no diffusion
1. L. Zhang and F. Oeters, Steel Research, Vol. 70, 1999.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 28
Equations of Motion of Alloy Particle(Lagrangian Trajectory Calculation)
( )iplipip
Dldrag uu
dCF ,2
,,4Re3
−=ρµ
( )y
ip
lipbuoyancy egF
,
,
ρρρ −
= llip
lgradientpressure uuF ∇⋅=
,_ ρ
ρ
gradientpressuremassvirtualbuoyancydragip FFFF
dtud
__, +++=
( )iplip
lVMmassvirtual uu
dtdCF ,
,_ −=
ρρ
Forces acting on an alloy particle
Drag force
Buoyancy force
Virtual mass force
Pressure gradient force
up,i : Velocity of ith particledp,i : Equiv diameter of ith particleρp,i : Density of ith particleCD : Drag coefficient
(Spherical model in FLUENT)CVM : Virtual mass coefficient (=0.5)xp,i : Position of ith particleζ :Normal Gaussian distributed
random number (σ=1)iR : Unit vector in random direction
Particle position Turbulent effect
dtux ipip ∫= ,, Rll ikuu 32ς+=
15
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 29
Alloy Position at the Slag-Metal Interface
( )θθπ 33 coscos3224
+−= pupper dV
( )θθπ 33 coscos3224
−+= plower dV
slag
molten steel
alloy
Vupper
Vlower
2θ
Volumes of upper and lower parts
Force balance at the interface
yloweralloy
steelliqalloyupper
alloy
slagalloybuoyancy egVVF
−+
−= −
ρρρ
ρρρ
θ= 65.3 º for SiMn particleρalloy = 6120 kg/m3, ρslag = 2700 kg/m3
ρliq-steel = 7000kg/m3
Balance position
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 30
Alloy Particle Distribution
Alloy addition
SiMn particle distribution 20 second after alloy addition start Colored by particle diameter (including shell thickness)
Heat 1 (SH = 20 °C) Heat 2 (SH = 60 °C)
XY plane
16
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 31
Mixing Behavior: Mn Content(Heat 1, Qg=170Nm3/min, SH=60ºC)
XY plane YZ plane
15 second after alloy addition start
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 32
Mixing Behavior: Mn Content(Heat 2, Qg=113Nm3/min, SH=20ºC)
XY plane YZ plane
10 second after alloy addition start
17
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 33
Concentration Profiles at the Sampling Point
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0 100 200 300 400
Time after alloy addition start (s)
Nor
mal
ized
con
cent
ratio
n
Heat 1 (calculated)Heat 1 (measured)Heat 2 (calculated)Heat 2 (measured)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 34
Mixing Time at Various Monitoring Locations(Heat 1)
Z
Y
X
2
4
3
5
8
6
9
7
Porous plug
Plume
201
187
211
191
214
202
206
197
198
162
149
161
153
173
162
167
156
158
0 50 100 150 200 250
9
8
7
6
5
4
3
2
1
Mon
itorin
g lo
catio
n
Mixing time after alloy addition (sec)
±1% ±5%1: Sampling location
in industrial experiment
18
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 35
Model 2 Summary
• It takes 5~15 seconds for alloy particles to melt after they are added to molten steel.
• Alloy particles travel between slag and metal interface before they melt. Therefore, the place where alloy content starts to increase is different from where alloy particles are added.
• Using alloy particle transport and turbulent species diffusion model, the mixing behavior can be predicted.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 36
Model 3: Inclusion Attachment to Bubble
19
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 37
Inclusions Captured by Bubbles in Steel
1. L. Zhang, B. Rietow and B. G. Thomas, Investigation of Ingot Inclusions Using Microscope and SEM, Univ. of Illionis at Urbana-Champaign, IMF project report, May. 04, 2004.
2. L. Kiriha et al., CAMP-ISIJ, Vol. 13, 2000.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 38
Steps of Inclusion Attachment to Bubble
1.Collision-Particle trajectory
3.Slide- Sliding time ts- Film drainage time tf
2.Oscillation- Collision time tc
4.Attachment- Stability
ts=10-3~10-1sec
tc=10-7~10-5sec
tf=10-6~10-3sec
20
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 39
dC , SC
θC
ub
Bubble
Inclusion
Attachment Probability PA2
b
c
+==
pR
CA dd
dSSP
Inclusion Attachment Probability(Conventional Definition)
Assumption:Inclusion will be attached to bubble surface if sliding time ts is greater than film drainage time tf.The value of dc can be obtained by sliding time calculation.
L. Zhang and S. Taniguchi, Int. Mat. Reviews, Vol. 45, 2000.
SR
db dp
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 40
Example of Particle Trajectory Calculation(Conventional Method)
Particle Traces Colored by Particle Residence Time (ms)FLUENT 6.1 (axi, dp, segregated, ske)
May 01, 2004
6.95e+02
6.26e+02
5.56e+02
4.87e+02
4.17e+02
3.48e+02
2.78e+02
2.09e+02
1.39e+02
6.95e+01
0.00e+00
100µm silica inclusion trajectory toward 1mm Ar bubble
particle : SiO2 (spherical shape)
100db
yi
xy
attached if tf<ts
not attached
dc
Potential flow based trajectory calculation: C. M. Phan et al., Int. J. Miner. Process, Vol. 72, 2003.
• Assuming spherical bubble (db<1mm)• No consideration of turbulence
Particle trajectory calculation by computer simulation:L. Zhang, J. Aoki and Brian G. Thomas, MS&T 2004, New Orleans, LA. 2004.
OK for small bubbles and low turbulence (mold)Not applicable in steelmaking ladle condition
21
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 41
Bubble Morphology forSpheroidal or Spherical Cap Bubble
1
2
3
4
5
0.1 1 10 100
dp (mm)
e
0.7571/ 2 1 0.163e r r Eo= = +
Spherical cap
Sphere Spheroid
Spheroidal bubble(1)
( )σ
ρρ bbgdEo −=
2Eotvos number
db (mm)
Spherical cap bubble
r1r2
2θR
R = 9/8db (constant CD = 8/3)
θ=50.59°(volume matching)experiment: θ=46°~64°
1. R. M. Wellek et al., A. I. Ch. E. Journal, Vol. 12, 1966.
volumetric equivalent diameter
aspe
ct ra
tio
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 42
Inclusion Trajectories near Bubble in Various Global Turbulent Flow Fields
k = 10-6m2/s2 k = 10-4m2/s2
k = 10-2m2/s2 k = 10-1m2/s2
db = 100mm, dp = 100µm
22
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 43
Re-definition of Inclusion Attachment Probabilityfor Strongly Turbulent Flow Fields
( )4
2pbS
p
ap
A ddC
NP
+=
π
2
b
c
+=
pA dd
dP
Injecting surface (must include all inclusions which eventually attach)
bubble
Inclusion trajectory
Attachment probabilitynumber attached (#)
inclusion concentration on the injecting surface (#/m2)
reference surface area (m2)(conventional definition)
db
dp
Conventional method2
4 cSp
ap dCN π=
Number of inclusions attached is counted out of 40,000~400,000 injected inclusions per case
lCC Vp
Sp ∆=
CpV : Volumetric inclusion concentration (#/m3)
∆l : Bubble path length (arbitrary) (m)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 44
Attachment and Entrapment
( )4
2pbS
p
ap
A ddC
NP
+=
π
Attachment probability
( )4
2pbS
p
ep
E ddC
NP
+=
π
Entrapment probability
Total removal probability by a bubble
EATR PPP +=
Nep : number of entrapped inclusions in the
wake for more than bubble floatationtime (3sec)
k=10-2m2/s2
0
10
20
30
0 20 40 60 80 100 120
Volumetric equivalent diameter db (mm)
Pro
babi
lity
(%)
k=10-6m2/s2
0
10
20
30
0 20 40 60 80 100 120
Pro
babi
lity
(%)
Attached to the bubbleTrapped in the wakeTotal
23
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 45
Effect of Turbulence
0
5
10
15
20
25
1E-08 1E-06 1E-04 1E-02 1E+00
k (m2/s2)
Pro
babi
lity
(%)
attachmententrapmenttotalconventional method
db=100mm, dp=100µm
0
20
40
60
80
100
120
1E-08 1E-06 1E-04 1E-02 1E+00
k (m2/s2)
Pro
babi
lity
(%)
attachmentconventional method
db=1mm, dp=100µm
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 46
Inclusion Removal Probabilitywith Turbulent Motion
Turbulence greatly increases inclusion attachment probability (especially at large bubble diameter and high turbulence levels)
0.1
1
10
100
1000
1 10 100Volumetric equivalent diameter db (mm)
Atta
chm
ent p
roba
bilit
y P
A (%
)
Conventional methodk = 10-6m²/s²k = 10-4m²/s²k = 10-2m²/s²k = 10-1m²/s²
0.2 0.3 0.4 0.5 0.6 0.7Bubble rising velocity (m/s)
Particle diameter dp=100µm
Tota
l rem
oval
pro
babi
lity
PTR
(%)
24
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 47
Model 3 Summary
• The model for predicting inclusion attachment probability to a large distorted bubble in a turbulent flow field is developed for the first time.
• Inclusion entrapment in the recirculation zone behind large bubble at low turbulence level is also important.
• Probability of Inclusion removal by bubbles considering turbulence is 10 to 100 times higher than that of the conventional model.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 48
Model 4: Inclusion Removal by Bubblesin Ladle Flow
25
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 49
Modeled ProcessThe LMF Ladle of Nucor Yamato Steel
2832mm
305mm
254mm
406mm
2680mm
1143mm
3581mm
Porous plug(φ = 64mm)
xy
xz
725mm
110 m-tonne ladle
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 50
Flow Field and Bubble Distribution(FLUENT 6.1 output)
Flow field (m/s) Turbulent kinetic energy (m2/s2)Gas Flow rate = 0.170 Nm3/min→ 0.379 (bottom) ~ 1.166 (top) m3/min Csk=0.10, Csε=0.08
26
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 51
Equations for Inclusion Transport in Ladle Flow(Eulerian Approach)
( ) ( ) ( ) ( ) ( )p
Mp
tlMpll
z
MpllT
py
Mpll
xMpll SC
SczC
uyC
uuxC
uCt
−∇
∇=
∂∂
+∂
∂++
∂∂
+∂∂ µαραραρα
ρα
( ) ( )∑
−
+
∆=
cellbN
iibl
Mpll
pibRbpiTR
cellp uuC
ddSkddP
VS ,
2,
, 4,,1 ρα
π
2
2
,2
4b
pipb
R d
dNd
Sπ
ππ −
=
Inclusion transport in turbulent flow
Inclusion removal rate by attaching to bubbles
Surface area correction factorCp
M : Inclusion number concentration per unit liquid mass (#/kg)
ux, uy, uz: Liquid velocityup
T : Terminal velocity of inclusionSc : Turbulent Schmidt number (=1)PTR : Total inclusion removal probability (Model 4)
(function of db, dp, and k)dp = 100µm, k=0.1m2/s2 (assumed constant)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 52
Initial and Boundary Conditions
Boundary ConditionsCASE 1. Cp
M at top surface = 0(All inclusions reached to the interface will be absorbed into slag)
Bubble attachment(All inslusions attached to bubbles will be removed)
CASE 2. Bubble attachment only
CASE 3. Top surface removal only
Absorption into top slag
Attachment to bubble
Initial ConditionsInitial [O] = 50 ppm (75,156 inclusions/kg)Uniform distribution in the ladleinclusion diameter = 100 µm
27
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 53
Change of Oxygen Content
Case 1Bubble and Top surface
Case 2Bubble only
Case 3Top surface only
Snapshots at 60 seconds after bubbling start
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 54
Calculated Oxygen Content andDeoxidization Rate Constant
0
10
20
30
40
50
60
0 100 200 300 400 500 600Bubbling time (s)
Oxy
gen
Con
tent
(%)
Case 1 (bubble + top surface)Case 2 (bubble only)Case 3 (top surface only)
0
0.1
0.2
0.3
0.4
0.5
0.6
0 100 200 300 400 500 600Bubbling time (s)
Deo
xidi
zatio
n ra
te c
onst
ant K
o (1
/min
) Case 1 (bubble + top surface)Case 2 (bubble only)Case 3 (top surface only)
[ ] [ ]OOOK
dtd
−=Rate Constant
28
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 55
Comparison of KO with Various Processes
100 101 102 103 10410-3
10-2
10-1
100
KO (min - 1)
d[O]/dt=Ko [O][O]: ppmt: min
KO: min - 1
Ar gas bubbling ASEA-SKF (I) ASEA-SKF (II) VOD (NK-PERM) VOD (Convent.) RH (NK-PERM) RH (Convent.)
ε (Watt/ton)
bubble
L. Zhang and B. G. Thomas, 7th European Steelmaking Conference, Milano, 2002
bubble + top surface
top surface
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 56
Model 4 Summary
• Inclusion removal model in ladle flow considering absorption into top slag-metal surface and attachment to bubbles is developed.
• Assuming all inclusions will be absorbed into top slag, the rate constant is much higher than actual steelmaking processes. Thus, correct estimation of inclusion absorption at the slag-metal interface is very important.
• To predict the deoxidization rate more accurately, further modeling is necessary which includes:1. Inclusion absorption at the slag-metal interface.2. Re-oxidation from slag.3. Size distribution of inclusions.
29
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 57
Model 5: Inclusion Absorption at theSlag-Metal Interface
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 58
Inclusion (Alumina) Behavior at theSlag-Metal-Bubble Interface
contact angle in steel θ=130°
contact angle in slag θ=40°~60°
Alumina inclusion will detach from the bubble at the interface
molten steel
slag
bubble
inclusion
30
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 59
Inclusion Pile-up at the Slag-Metal Interface(Dissolution Rate < Transfer Rate)
Alumina dissolution into slag can be the rate-controlling process
molten steel
slag
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 60
Alumina Particle Dissolution in Slag
( )t
RCCMk
RR slag
OAlsat
OAlmt
0
)()(
0
32321ρ
−−=
Mass transfer (diffusion) control model
R : Particle radius (m)R0 : Particle initial radius (m)kmt : Mass transfer coefficient (m/s)M : Molecular weight of alumina
(=0.102kg/mol)C(sat)
Al2O3 : Saturation concentration of alumina (mol/m3)
C(slag)Al2O3 : Slag bulk concentration of
alumina (mol/m3)ρ : Density of alumina (kg/m3)t : Time (s)
W. D. Cho and P. Fan, ISIJ Int., Vol. 44, 2004.
Al
CaO
Si
31
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 61
Controlling Factors of Dissolution Rate
W. D. Cho and P. Fan, ISIJ Int., Vol. 44, 2004.
Temperature SiO2 content in slag Al2O3 content in slag
T SiO2 Al2O3
Time (s) Time (s)Time (s)
Par
ticle
Siz
e (µ
m)
Par
ticle
Siz
e (µ
m)
Par
ticle
Siz
e (µ
m)
Alumina dissolution rate increases with•Higher temperature•Lower SiO2 content in slag•Lower Al2O3 content in slag
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 62
Conversion from Dissolution Rate toAbsorption Rate per Unit Surface Area
( )0
)()(3232
RCCMk
dtdRR
slagOAl
satOAlmt
ρ−
−==&
dtAVd
dtA
Wd pOAl
A
−=
−=ρ
β 32
Rf
fNR
NR
AV
p
p
343
4
2
3
==π
π
Rf ppOAl
A&ρβ
34
32 −=
Dissolution rate (m/s)
Absorption rate (kg/m2s)
Top view of the slag-metal interface
R : Radius of particleW : Total mass of particlesA : Surface area occupied by particlesV : Total volume of particlesN : Number of particlesfp : Packing factor
(=0.907 for close packed spheres on a plane)
32
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 63
Deoxidization Rate Assuming Interface Control
(m/s)10128.1 6−×−=R&
1045451500
SiO2Al2O3CaO
Slag composition (wt%)Temp.(°C)
Experimental dissolution rate(1)
1. W. D. Cho and P. Fan, ISIJ Int., Vol. 44, 2004.
Absorption rate at the interface
Deoxidization rate in steel
s)(kg/m0048.0 332 =OAlAβ
[ ] (ppm/min)3.13 32
32
−=−=Steel
OAlA
OAl
O
WA
MM
dtOd β
0
10
20
30
40
50
60
0 100 200 300 400 500 600
Bubbling time (s)
Oxy
gen
Con
tent
(%)
Model 4: Case 1 Case 2 Case 3Interface control
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 64
Comparison of Deoxidization Rate Constantwith Various Processes
100 101 102 103 10410-3
10-2
10-1
100
KO (min - 1)
d[O]/dt=Ko [O][O]: ppmt: min
KO: min - 1
Ar gas bubbling ASEA-SKF (I) ASEA-SKF (II) VOD (NK-PERM) VOD (Convent.) RH (NK-PERM) RH (Convent.)
ε (Watt/ton)
Interface control
Model 4
33
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 65
Model 5 Summary
• Alumina inclusion particles carried by bubbles are considered to detach at the slag-metal interface.
• Alumina dissolution rate is strongly affected by temperature and slag composition.
• Interface absorption of inclusions may be one of the rate controlling process for deoxidization of steel ladle.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 66
Conclusion
• Ladle refining model is developed including multiphase turbulent flow simulation, alloy mixing, and inclusion removal.
• This model is ready to use to help optimize the operating conditions in actual steel plants, and to design the future continuous steelmaking process.
34
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • J Aoki & BG Thomas 67
Future Work
• Combine the interface rate control model (model 5) with the ladle deoxidization model (model 4).
• Consider the effect of re-oxidization.• Consider the effect of wall attachment.• Include the effect of inclusion size distribution using the
inclusion collision model by L. Zhang and B. G. Thomas.
• Develop surface reaction model (e.g. de-S) using the ladle flow model together with a chemical reaction model.