Kun Xu - Continuous Casting Consortium - University of Illinois at ...

ANNUAL REPORT 2009 

UIUC, August 5, 2009 

Modeling heat transfer, precipitate 

formation, and grain growth during 

secondary spray cooling 

Kun Xu 

(Ph.D. Student) 

Department of Mechanical Science and Engineering 

University of Illinois at Urbana-Champaign 

University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Kun Xu • 1 

Objectives 

Cracks form during cooling due to: 

- tensile stress 

- low ductility 

To design temperature histories to 

avoid crack formation, need accurate 

predictive tools 

Models can now accurately predict 

temperature and stress histories 

Need tools to predict metallurgical 

behavior to estimate hot ductility, such 

as grain size, precipitate formation 

University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Kun Xu • 2

Specific application: 

transverse surface cracks 

casting direction 

transverse cracks 

Widespread crazing and fine 

transverse cracks at oscillation 

marks on the as-cast surface of 

a line pipe steel slab (top side) 

E. S. Szekeres,Proceedings of the 6th International Conference on Clean Steel, Hungary, 2002. 


mold 

gap 

Effect of depressions on grain size and cracks 

Steel shell 

Molten steel pool 

Larger grains and cracks beneath depressions & 

oscillation marks 

Due to: 

More heat flow resistance across gap 

Higher shell temperature 

Faster grain growth 

Causes: 

Tensile strain concentration area 

Make hot grains actually align (Secondary 

recrystallization) to form “blown grains” 

Embrittled with the large numbers of fine 

precipitates at weak grain boundaries — 

transverse cracks open up along boundaries 


Mechanism of surface crack formation 

with precipitate embrittlement 

STAGE I - Normal solidification on mold 

wall. Surface grains are small but highly 

oriented. 

γ 

STAGE III - Nitride precipitates 

begin to form along the blown 

grain boundaries. Microcracks 

initiate at weak boundaries. 

γ 

STAGE IV - Ferrite transformation 

begins and new precipitates form 

at boundaries. Existing microcracks 

grow & new ones form. 

STAGE II - Surface grains “blow” locally due to high 

temperature (>1350 o C) and strain, especially at the 

base of deeper oscillation marks. 

γ γ 

γ γ 


STAGE V - At the straightener, 

microcracks propagate and 

become larger cracks, primarily 

on top surface of the strand. 


Cracks open up at weak grain boundaries 

Casting direction 

~1mm 


Crazing around a transverse 

crack at base of an oscillation 

mark on the as-cast top surface 

of a 0.2%C steel slab 

Note larger grain size (typical 

size ~1mm) at the base of 

oscillation mark 


Hot ductility varies with temperature and 

grade due to precipitate formation 

B.G. Thomas et al, ISS Transactions, 1986, 7-20. 


Micrographs of precipitates in steels 

Complex precipitates with TiN core 

TiNb(C) precipitates at grain boundaries 

V. Ludlow et al, Precipitation of nitrides and carbides during solidification and cooling, unpublished work. 

M. Charleux et al, Metallurgical and Materials Transactions A, 2001, vol.32A, 1635-1647. 


Project Overview 

Heat Transfer 

Program: CON1D 

Ductility Prediction 

Phase fraction & 

temperature history 

Grain size 

Equilibrium 

Precipitation Model 

Equilibrium 

concentration 

Kinetic Model 

(precipitates) 

Grain Growth 

Model 

Volume fraction & 

size distribution 

Final goal Predict ductility and combine with other models to 

design casting practices to prevent cracks 


Equilibrium precipitation model 

Reaction xM + yX 

M M: metal, X: C, N, S or O 

x X y 

x y 

Solubility product: KM[ ] [ ] / 

x X = a 

y M ⋅a 

X a 

α (MxXy) : activity of precipitates 

( MxXy) [%i]: dissolved concentration of 

i j 

ai = fi[% i] 

where log fi = ei[% i] +∑ei 

[% j] 

element i 

Mutual Solubility: If two precipitates have the same crystal structures and similar 

lattice parameters (

Groups of precipitates 

The current model has 18 precipitates, 13 alloying elements and 4 groups of mutually 

soluble precipitates 

• TiN, TiC, NbN, NbC 0.87 , VN, V 4 C 3 (f.c.c structure) lattice mismatch ~7.92% 

– Validated by many experimental observations 

• Al 2 O 3 , Ti 2 O 3 (hexagonal structure) lattice mismatch ~8.06% 

• MgO, MnO (f.c.c structure) lattice mismatch ~5.54% 

• MnS, MgS (f.c.c structure) lattice mismatch ~0.38% 

• SiO 2 (trigonal structure) 

• TiS (trigonal structure) 

• Ti 4 C 2 S 2 (hexagonal structure) 

• AlN (hexagonal structure) 

• BN (hexagonal structure) 

• Cr 2 N (hexagonal structure) 

• A complete system of 35 nonlinear equations is solved with Newton-Raphson method 


Validation for mutually exclusive precipitates 

NbN after adding Al 

NbN after adding B 

AlN 

NbN 

BN 

NbN NbN before before adding adding Al B 

Three hypothetical steels 

1. 0.02%Nb and N 

2. 0.02%Nb and N, 0.02% Al 

3. 0.02%Nb and N, 0.01% B 

Second order equation and 

analytical solution exists 

Existing precipitate AlN or 

BN delays the precipitate 

NbN to form at the lower 

temperature and decreases the 

equilibrium amount of NbN at 

the same temperature 


Pseudo-ternary diagram for Al-Nb-N system 

Fix the sum of weight percent concentration (Al+Nb+N) as 0.05% 

0.75 

0.50 

0.25 

Nb 

0.00 

1.00 

none 

0.75 

0.50 

0.25 

Nb 

0.00 

1.00 

1.00 

1.00 

AlN 

0.00 

0.00 

Al 0.00 0.25 0.50 0.75 1.00 N 

Al 0.00 0.25 0.50 0.75 1.00 N 

0.75 

0.50 

0.25 

Nb 

0.00 

1.00 

0.75 

none 0.50 

NbN 

AlN 

0.25 

1.00 

0.00 

Al 0.00 0.25 0.50 0.75 1.00 N 


Ti 0.26 Nb 0.24 N 0.50 

T=1350 o C 

T=1150 o C 

0.75 

0.75 

0.50 

0.50 

0.25 

none 

0.75 

T=1300 o C 

0.25 

Nb 

0.00 

1.00 

none 

AlN 

0.75 

NbN 

NbN+AlN 

T=1125 o C 

0.50 

0.50 

0.25 

0.25 

1.00 

0.00 

Al 0.00 0.25 0.50 0.75 1.00 N 

Test problem for mutually solubility 

(Ti-Nb-N system) 

Ti 0.38 Nb 0.12 N 0.50 

Ti 0.48 Nb 0.02 N 0.50 


none 

Three hypothetical steels 

1. 0.01%Nb and N 

2. 0.01%Nb and N, 0.005% Ti 

At least fourth order equation 

and no analytical solution 

Mutual solubility decreases 

the activity of NbN, which 

allows more NbN to form 

Without consideration of 

mutual solubility, NbN 

precipitates would decrease 

with Ti additions

Molar fraction of precipitates 

1E-3 

1E-4 

1E-5 

Validation with stability of TiS and Ti 4 C 2 S 2 

Ti 2 O 3 

Ti 4 C 2 S 2 

TiC 

TiN 

Al 2 O 3 

TiS 

1E-6 

1000 1050 1100 1150 1200 1250 1300 1350 

Temperature (centigrade) 

1E-6 

1000 1050 1100 1150 1200 1250 1300 1350 


Steel 

Nb1 

Nb8 

Steel 

B 

C 0.0033 0.0040 0.081 0.0115 0.037 0.050 0.0022 0.0036 

Recorded types of observed precipitates in experiments (matches prediction) 

Measurements: X. Yang et al, ISIJ International, 1996, vol.36, 1286-1294. 

Steel 

B 

C 

C 

0.085 

0.081 

C 

0.0036 

1300 o C 

TiN 

TiN, TiS 

Si 

0.0050 

Mn 

0.081 

1250 o C 

TiN, TiS * 

TiN, TiS 

S 

0.0028 

1200 o C 

TiN, TiS 

TiN, TiS 

Molar fraction of precipitates 

1E-3 

1E-4 

1E-5 

Al 

0.045 

1150 o C 

TiN, TiS 

Ti 4 C 2 S 2 

0.095 

TiC 

Ti 

TiN, Ti 4C 2S 2 

TiN 

Al 2 O 3 

TiS 

Temperature (centigrade) 

Steel B Steel C 

N 

0.0019 

1100 o C 

TiN, Ti 4C 2S 2 

TiN, TiS * , Ti 4C 2S 2 


O 

0.0028 

Validation with equilibrium Nb(C,N) precipitation 

experiments (microalloy steel) 

Si 

0.28 

0.31 

Mn 

1.46 

1.44 

P 

0.008 

0.01 

0.002 

0.002 

Measurements: S. Zajac and B. Jansson, Metallurgical and Materials Transactions B, 1998, vol.29B, 163-176. 

Solution treated (1300 o C) 1 hour 

Followed by water quenching 

Then aged at different isothermal 

temperature for 24-48 hours 

Cool to ambient quickly 

Measure Nb amount in Nb(C,N) by 

the inductively coupled plasma 

(ICP) emission method 

Nb precipitated as Nb(C,N) (wt%) 

0.04 

0.03 

0.02 

0.01 

S 

Nb 

0.013 

0.033 

Al 

0.013 

0.017 

N 

0.005 

0.004 

V 

0.014 

0.011 

Prediction of Nb1 

Measurement of Nb1 

Prediction of Nb8 

Measurement of Nb8 

Ti 

0.003 

0.003 

0.00 

850 900 950 1000 1050 1100 1150 1200 1250 

Temperature (centigrade)

Temperature and steel phase fraction model 

CON1D Program: Solve the transient heat conduction in the mold and spray regions of 

continuous steel slab casters using finite difference method 

The program is calibrated to predict the temperature of the real casters. 

Example: thin slab casting of low-carbon 1006N2 steel 

Steel 

1006N2 

Phase fraction 

1.0 

0.8 

0.6 

0.4 

0.2 

C 

0.0472 

N 

0.0083 

S 

0.0013 

Liquid 

Delta ferrite 

Austenite 

Alpha ferrite 

Mn 

0.9737 

0.0 

650 800 950 1100 1250 1400 1550 

Temperature ( o C) 

Si 

0.2006 

0.0084 

0.0123 

0.0027 

0.0223 

0.0001 

0.0354 


Ti 

Nb 

V 

Slab size: 90mm thick, 1396mm wide 

Mold length: 950mm 

Pouring Temperature: 1553 o C 

Casting speeds: 2.8m/min and 4.6m/min 

Al 

Liquidus temperature: 1524.4 o C 

Solidus temperature: 1500.7 o C 

Equilibrium Precipitation in continuous casting of 

1006N2 Steel 

Mold length 0.95m, start of first spray zone 0.85m, end of last spray zone 11.25m 

Only precipitates predicted to form are: (Ti,Nb,V)(C,N), MnS and AlN 

7.5m: Ti 0.326 Nb 0.173 V 0.006 N 0.424 C 0.071 

0.85m: Ti 0.333 Nb 0.166 V 0.006 N 0.429 C 0.066 

V=2.8m/min 

7.5m: Ti 0.361 Nb 0.138 V 0.005 N 0.446 C 0.050 

15m: Ti 0.338 Nb 0.161 V 0.006 N 0.433 C 0.062 0.85m: Ti0.419 Nb 0.081 V 0.002 N 0.463 C 0.035 

V=4.6m/min 


B 

15m: Ti 0.428 Nb 0.072 V 0.002 N 0.466 C 0.032 

Higher casting speed causes higher temperature and less precipitates (same spray flows) 

TiN is most stable at high temperature because of its low solubility limit. For the higher 

casting speed, AlN does not form in the mold. 

Cr

Particle Collision 

Particle Diffusion 

Kinetic model for precipitate growth 

dni 

1 

= (1 + ) Φ nn − n (1 + ) Φ n ( i≥2) 

dt 2 

dn 

dt 

i 

i−1 

∑ δki , −kki , −kki−k nM 

i∑ δik 

, ik , k 

k= 1 k= 

1 

Generation of size i particle by 

collision of smaller particles k and i-k, 

only count once for reacting particles 

Loss of size i particle by 

collision with any particle j, 

double loss if both particle i 

=− βinn 1 i + βi−1nn 1 i− 1 − αiAn i i + αi+ 

1Ai+ 1ni+ 1 ( i≥2) 

Particle 

Diffusion 

(i→i+1) 

Particle 

Diffusion 

(i-1→i) 

Particle 

Dissolution 

(i→i-1) 

Particle 

Dissolution 

(i+1→i) 

Collision happens only in liquid, and diffusion happens in both liquid and solid 

ni : Number density of size i particle (m-3 ) 

Фij : Collision coefficient for collision between size i and size j particle (m3 /s) 

βi : Diffusion rate constant of size i particle (m3 /s) 

αi : Dissolution rate per unit area of size i particle (m-2s-1 i 4 1 i 

) 

Dr β π = 

α = βn exp( 2 σV 

/ RTr) / A 

i i 1, eq m i i 


Introduction of Particle Size Grouping (PSG) model 

The model always simulates from nuclei (~ 0.1nm) up to large coarsened particles 

(~100μm): particles could contain 1~10 18 molecules 

Serious computation and memory storage issues arise with such a large size range 

Solve with PSG method: Use N G groups (

Size group and particle radius for constant R V 


Validation of test problem for collision 

1/ 2 3 

Turbulent collision Φ = 1.3( ε / ν) 

( r + r ) with only single molecules initially 

Dimensionless form 

n 0 : Initial number density of single molecules; ε, ν, r 1 are input parameters 

Initial condition (t*=0 ): n 1 * =1 and ni *=0 for i≥2 (same for N i in PSG model)) 

Boundary condition: n M *=0, or N G *=0 at all t * 

Exact solution: n M =12000; PSG model: N G =16 (R V =2) or N G =11 (R V =3) 

Dimensionless numder density 

1 

0.1 

0.01 

1E-3 

1E-4 

1E-5 

N 1 

N 2 

N 3 

ij i j 

n = n / n 

* 

i i 

1E-6 

1E-3 0.01 0.1 1 

N 4 

N 5 

Dimensionless time 

0 

t = 1.3( ε / ν) 

r n t 

R V =2.0 

N6 N7 N8 N9 N10 * 1/2 3 

1 0 

N11 N12 Dimensionless numder density 

1E-6 

1E-3 0.01 0.1 1 



1 

0.1 

0.01 

1E-3 

1E-4 

1E-5 

N 1 RV =3.0 

N 2 

N 3 

N 4 

N 5 

N 6 

N7 N8 N 9

L 

n j 

Develop new PSG method for diffusion 

dN V V 

ceil( V ) floor( V ) 

dt V V V V 

U 

n j 

j 

= 1 U 

βjN1( Nj −nj ) − 1 

L 

αjAj( Nj − nj) + 

j− 1, j U U 

βj−1Nn 1 j− 1+ j, j+ 

1 L L L 

αj+ 

1Aj+ 1nj+ 1 

j j j j 

Diffusion 

inside group j 

Dissolution 

inside group j 

Diffusion 

group j-1→j 

floor( V ) ceil( V ) 

− Nn − An ( j≥2) 

V V 

j, j+ 1 U 

βj 1 

U 

j 

j−1, j L 

αj 

L 

j 

L 

j 

j j 

Diffusion 

group j→j+1 

Dissolution 

group j→j-1 

Dissolution 

group j+1→j 

: Number density of particles in group j, which jumps into group j-1 by dissolution after 

losing only one single molecule 

: Number density of particles in group j, which jumps into group j+1 by diffusion after 

getting only one single molecule 

Estimated from geometric progression of the number densities for two neighboring size groups 


Dimensionless number density 

10 

1 

0.1 

0.01 

1E-3 

1E-4 

1E-5 

1E-6 

Validation of test problem for diffusion 

Dimensionless form 

* 

ni = ni / n1, 

eq 

* 

t = 4π Drn 11 1, eqt 

Molecules produced by an isothermal first order reaction 

nM 

* * * * * 

n ( t ) = n ( t )/ n = i⋅ n = 9[1−exp( −0.1 

t )] 

s s 1, eq i 

i= 

1 

Initial condition (t*=0 ): n i *=0 for i≥1 (same for N i in PSG model) 

Boundary condition: n M *=0, or N G *=0 at all t* 

Exact solution: n M =16000; PSG model: N G =17 (R V =2) or N G =12 (R V =3) 

R V =2.0 

N 1 

N 2 

N 3 

N4 N5 N6 N7 N8 N9 ∑ 

N 10 

N 11 

N 12 

N 13 

1 10 100 1000 

N 14 

0 

0.01 0.1 1 10 100 




Dimensionless number density 

10 

1 

0.1 

0.01 

1E-3 

1E-4 

1E-5 

1E-6 

R V =3.0 

N 1 

N 2 

N 3 

N4 N5 N6 N7 

1 10 100 1000 

* 

N s 

10 

8 

6 

4 

2 

N 8 


N 9

Storage 

Computational 

time 

Compare computational cost for test problems 

Exact 

n M =12000 

~225 hours 

Collision 

PSG (R v =2) 

N G =16 

~0.8s 

PSG (R v =3) 

N G =11 

~7420s 


~0.4s 

Exact 

n M =16000 

Diffusion 

PSG (R v =2) 

N G =17 

~358s 

PSG (R v =3) 

For constant RV , the number of size groups in PSG method must satisfy 

V 

NG 

−1 

= R > n → N = Ceil(log n ) + 1+ 1 (last one for boundary group) 

NGV M 

G RVM N G =12 

~248s 

* Calculation is run with Matlab on Dell OPTIPLEX GX270 with P4 3.20GHz CPU and 2GB RAM 

The computational time is proportional to n M 2 or NG 2 for collision problem (double 

loop), and n M or N G for diffusion problem (single loop) 

Computation cost is dramatically reduced, especially for the problem with a large 

variety of precipitate sizes because of logarithm relation 

Sample results for AlN particles 

V M =1.33×10 -5 m 3 /mol, σ=0.2~0.75J/m 2 

DAl =3*10-3exp(-234500/RT) m2 /s, DN =9.1*10-5exp(-168500/RT) m2 /s 

6770 

Solubility product in austenite log K 

1.03 , Al0 =0.011wt%, N0 =0.009wt% 

AlN =− + 

T 

Choose RV =2 and 30 size groups, ∆t=1×10-3s, Implicit Euler with Gauss-Seidel 

iterative method 

Amount of dissolved elements and precipitate (wt%) 

0.0175 

0.0150 

0.0125 

0.0100 

0.0075 

0.0050 

0.0025 

0.0000 

700 750 800 850 900 950 1000 1050 1100 


Dissolved aluminum 

Dissolved nitrogen 

AlN precipitate 

1E-15 

700 750 800 850 900 950 1000 



D(m 2 /s) 

1E-11 

1E-12 

1E-13 

1E-14 

N 

Al

Choose σ=0.5J/m 2 

Calculated size distribution 

Growth (10~100s) is quick, 

coarsening (100-500s) is slow 

Lower temperature causes higher 

nucleation rate, thus large number of 

fine precipitated particles are observed 


Π= N1/ N1, 

eq 

Temperature 

900 o C 

850 o C 

800 o C 

750 o C 

Generate PTT diagram 

Nt / N1, 

eq −Π 

f = 

N / N −1 

5% precipitated 

178.1s 

60.2s 

40.3s 

38.5s 

t 1, eq 

Precipitated (500s) 

94.4% 

97.0% 

97.3% 

97.6% 

Experimental PTT curves of 

AlN precipitated in austenite of 

0.051%Al-0.0073%N steel 

Nose temperature (~800 o C) 

J. P. Michel et al, Acta Metallurgica, 

1981, 513-526. 

Quicker precipitation and higher 

nose temperature: interfacial energy 

value are between 0.5-0.75J/m 2 


Grain growth model 

Grain growth in austenite under the presence of precipitates 

(1/ n−1) 

dD * ⎛ Qapp 

⎞⎡1 1 f ⎤ 

= M 0 exp⎜− 

⎟⎢ 

− 

dt RT ⎣Dkr⎥ ⎝ ⎠ ⎦ 

M0 *: Kinetic constant that represents grain boundary mobility (m2 /s) 

Qapp : Apparent activation energy for grain growth (J/mol) 

n: Exponent to measure resistance to grain boundary motion 

f and r: the volume fraction and radius of precipitates 

K: Zener coefficient related to pinning efficiency of the precipitates. 

The maximum grain diameter in the presence of precipitates is defines as D = kr/ f 

Calculation begins from the temperature of totally austenite structure, the initial austenite 

grain size is assumed to be with the order of the primary dendrite arm spacing (PDAS) 

( ) 

m n 

1 K CR( C0) 

λ = K=278.75, m=–0.20628, n=–0.3162+2.0325C 0 for 0≤C 0 ≤0.15 (wt pct) 

I. Andersen and Ø Grong, Acta Metallurgica Materialia, 1995. 

M. E. Bealy and B. G. Thomas, Metallurgical and Materials Transactions B, 1996. 


Grain growth of 1006N2 steel in continuous casting 

Neglect the influence of precipitates 

dD * ⎛ Qapp 

⎞⎡1 1 f ⎤ 

= M 0 exp⎜− 

⎟⎢ 

− 

dt RT ⎣Dkr⎥ ⎝ ⎠ ⎦ 

Fully austenite temperature 1384 o C 

Cooling rate estimated by CON1D 

Without precipitates: the grains large 

enough to cause cracks 

Rough estimation: At surface, volume 

fraction of TiN 1.45*10 -4 at mold exit 

by choosing ρ steel =7500kg/m 3 and 

ρ TiN =5420kg/m 3 , typical average TiN 

particle size 10~100nm and k=4/3. the 

maximum grain size is calculated 

within the range of 90~900µm 

(1/ n−1) 

Grain size (mm) 

1.6 

1.2 

0.8 

0.4 

Lim 

M 0 * =4×10 -3 m 2 /s, n=0.5, 

Q app =167686+40562(wt%C p ) 

wt%C p =wt%C-0.14wt%Si+0.04wt%Mn 

J. Reiter, C. Bernhard and H. Presslinger, MS&T 2006, 

Cincinnati, USA. 

surface (V=2.8) surface (V=4.6) 

0.5mm (V=2.8) 0.5mm (V=4.6) 

0.0 

10mm (V=2.8) 10mm (V=4.6) 

0 3 6 9 12 15 

Distance from meniscus (m) 


mold 

exit

Conclusions (equilibrium precipitation 

model) 

1) An equilibrium precipitation model is established, which includes 

consideration of solubility limit with Wagner interaction effect, mutual 

solubility and complete mass conservation of alloying elements. 

2) The influence of mutual solubility on precipitation behaviors is carefully 

considered. Pseudo-ternary precipitation diagram is produced. 

3) The model is validated with analytical solutions of simple cases and 

experimental measurements of literatures. Good agreements are found. 

4) The model is applied for practical 100N2 steel and continuous casting 

with two casting speed. TiNbV(CN), Mns and AlN are only possible 

precipitates that could form. TiN is the mainly precipitates at high 

temperature especially in the mold because of its lowest solubility limit. 

With higher speed, AlN does not form in the mold. 


Conclusions (kinetic model) 

1) Comparing with some empirical or semi-empirical kinetic precipitation models, 

the suggested precipitate growth mechanisms by collision and diffusion are 

fundamentally based and all parameters are physically significant. 

2) The PSG method is suggested and validated for particle agglomeration due to 

collision and diffusion. Within a wide range of volume ratios, R V , mass balance 

is satisfied and good matches are found. Because the accuracy of the PSG 

method increases with decreasing R V , an R V of 2.0 seems optimal. 

3) PSG method needs only a small number of size groups to simulate particle 

agglomeration, and huge saving of memory storage and computational time is 

found by comparison with solving initial problem without losing mass balance 

and sacrificing the desired accuracy. 

4) In addition to generate volume fraction and size distribution of precipitated 

particles, the PSG method is used to generate the PTT diagram. 


Conclusions (grain growth) 

1) In continuous casting, the grain growth model predicts grains approaching 50% 

of their final size by mold exit. The grains under 0.5mm oscillation marks are at 

least 10% larger than those on the slab surface due to local high temperature. 

Without precipitates, the grains are large enough to cause ductility problems 

2) It is expected that the larger grains and lack of fine precipitates are likely the 

controlling factors to cause coarse grains and susceptibility to surface cracks 

beneath the oscillation marks. A program for coupling all the above models is 

necessary and important in future work 


Acknowledgements 

• Continuous Casting Consortium Members 

(ABB, Arcelor-Mittal, Baosteel, Corus, Delavan/Goodrich, 

LWB Refractories, Nucor, Nippon Steel, Postech, Steel 

Dynamics, ANSYS-Fluent) 

• Prof. B. G. Thomas 

• Other graduate students in CCC groups

Kun Xu - Continuous Casting Consortium - University of Illinois at ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?