uiuc, august 19, 2015 effect of nozzle port angle on mold...
TRANSCRIPT
CCC Annual ReportUIUC, August 19, 2015
UIUC: Seong-Mook Cho and Brian G. Thomas POSTECH: Hyoung-Jun Lee and Seon-Hyo KimPOSCO: Ji-Joon Kim, Hyun-Jin Cho, Jong-Wan Kim, and
Yong-Hwan Kim
Department of Mechanical Science & EngineeringUniversity of Illinois at Urbana-Champaign
Effect of Nozzle Port Angle on Mold Flow
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 2/35
Research ScopeObjectives:
- Optimize nozzle port angle to reduce flow-related surface defects in wide slab- Investigate mold flow patterns in wide slab caster with current casting conditions - Quantify effect of nozzle port angle on flow pattern and surface velocity, and surface level in a wide slab caster- Investigate similarity between water model experiments and plant measurements by comparing surface velocity
Methodologies: - 1/3 scale water model experiments to visualize mold flow patterns and quantify surface velocity and surface level fluctuations- Plant measurements to measure surface velocity, surface level, slag pool thickness using nail dipping tests and level sensor measurements- Computational modeling using Fluent on lab workstation or Blue Waters supercomputer to quantify nozzle flow and mold flow
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 3/35
Caster Dimensions and Process ConditionsReal STS caster (Case R)
Lab (1/3) scale Water model (Case W)
Casting speed 0.8 m/min 0.5 m/min
Volume Flow rate 256.0 LPM 16.4 LPM
Mold width 1600 mm 533 mm
Mold thickness 200 mm 67 mm
Aspect ratio between mold width and thickness
8.0 8.0
SEN depth 140 mm 46.7 mm
Nozzle port angle 15° (up) degree15° (up), 5°(up), -15° (down),
-30° (down) degree
Nozzle port size (width x height)
60 mm x 65 mm 20 mm x 21.7 mm
Nozzle bore(inner / outer)
60 ~ 65 mm (from bottom to top) / 110 mm 20.8 mm (average) / 36.6 mm
Area ratio between two ports and nozzle bore
2.54 2.54
Ar gas injection No gas 10 ml/min (0.06 % volume fraction)clear visualization of mold flows
Flow similarity between 1/3 scale water model (Case W) and real caster (Case R)
φ φ φ φ
Froude number (ratio of inertia force to gravitational force): ( ) ( )RW
gLu/gLu/ =
RWRc,Wc, /LLuu =Casting speed uc,W for 1/3 scale water model:
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 4/35
Physical Water Model Experiments
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 5/35
Schematic of 1/3 Scale Water Model
V2: surface bottom-up view
3 Video cameras:V1: NF end view
V3: WF front view
Stopper-rod
P P
Tundish
Dam
Weir
SENFlowmeter
Flow meter
PumpPump Reservoir
Valve
3-Ultrasonic displacementsensors
Electromagnetic-current sensor
Mold
30mm from NF
W/4
30 mmfrom SEN
30 mm from NF
Stopper-rod
Dam
Weir
Tundish
SEN
Electromagnetic-current sensor
3-ultrasonic displacement sensors
mold
533 mm1292
mm
380
mm
Nozzle port with+15 ° (up) angle
V1
V2 V3
Valve
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 6/35
Electromagnetic Current Sensor (for Surface Velocities) and Ultrasonic Displacement Sensor
(for Surface Level Variations) Measurements
8
W
Measuring positions of surface flow velocity and surface level
10mm
Front view / wide face
Measure transient surface velocity at
10mm below surface on W/4, W/8, 30
mm from NF, using electromagnetic
current sensor during 1000 sec.
Measure transient surface level on
30mm from SEN, W/4, 30 mm from NF
points, using ultrasonic displacement
sensors during 1000 sec.
4
W
4
W30mm 30mm 30mm
: Electromagnetic current sensor measurement position: Ultrasonic displacement sensor measurement position
Top view / surface
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 7/35
Videos to Visualize Mold Flow
<NF end view>
<WF front view>
<Surface bottom-up view>
Record 3 videos to show both nozzle and mold flow at same time Understand measured surface velocity and surface level with help of recorded videos
* Case W. +5 (up) angle nozzle port
Eddy-current sensor
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 8/35
Modeling and Results Analysis
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho • 9/35
Quarter Domain and Mesh: Standard k-ε Flow Model with Fluent on Lab Computer (LC)
Case W-LC. +15° (up) angle nozzle Case W-LC. -15° (down) angle nozzle
quarter domain: ~ 0.15 million hexahedral cells
quarter domain: ~ 0.15 million hexahedral cells
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •10/35
Full Domain and Mesh: LES with Fluent on Blue Waters (BW)
Full domain: ~ 2.4 million hexahedral cells
Case W-BW. +15° (up) angle nozzle Case W-BW. -15° (down) angle nozzle
Full domain:~ 2.6 million hexahedral cells
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •11/35
Governing Equations
Case W-LC: Reynolds Averaged Navier-Stokes (RANS) Model with standard k-εmodel
Case W-BW: Large Eddy Simulation (LES) with Wall-Adapting Local Eddy (WALE) subgrid-scale viscosity model
( ) 0uρx i
i
=∂∂ ( ) ( )
∂∂
+∂∂+
∂∂+
∂∂−=
∂∂
i
j
j
it
ji
*
jij x
u
x
uμμ
xx
puuρ
xMass conservation Momentum conservation
( ) ρεGx
k
σ
μμ
xuρk
x kjk
t
ji
i
−+
∂∂
+
∂∂=
∂∂
Turbulent kinetic energy
( )k
ερCG
k
εC
x
ε
σ
μμ
xuρε
x
2
2εk1εjk
t
ji
i
−+
∂∂
+
∂∂=
∂∂
Turbulent kinetic energy dissipation rate
( ) 0ρux i
i
=∂∂ ( ) ( ) ( )
∂∂
+∂∂+
∂∂+
∂∂−=
∂∂+
∂∂
i
j
j
it
ji
*
jij
i x
u
x
uμμ
xx
puρu
xρu
t
Mass conservation Momentum conservation
( ) ( )( ) ( )5/4d
ijdij
5/2
ijij
3/2dij
dij2
St
SSSS
SSLρμ
+=
Turbulent viscosity
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •12/35
Boundary Conditions
Case W-LC: Standard k-ε model Case W-BW: LES
Inlet (Tundish bottom region)
Constant velocity: 0.00573 m/secTurbulent kinetic energy: 10-5 m2/sec2
Turbulent kinetic energy dissipation rate: 10-5 m2/sec3
Constant velocity: 0.0573 m/sec
Outlet (Mold exit)
Pressure: 0 pascal gauge pressureTurbulent kinetic energy for backflow:
10-5 m2/sec2
Turbulent kinetic energy dissipation rate for back flow: 10-5 m2/sec3
Pressure: 0 pascal gauge pressure
Surface (interface between water and air)
Stationary wall with 0-shear stressStationary wall with 0-shear
stress
Wide face, Narrow face, Stopper-rod, and Nozzle walls (interface between
water and )
Stationary wall with no slip Stationary wall with no slip
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •13/35
Fluent Performance on BW: Speed-Up Test
Test problem: Case W-BW1/3 water model with 15o (up) nozzle portsMesh of nozzle & mold: ~6.94 million hexahedral cells
<Speed up of FLUENT calculation on Blue Waters>
With 96 cores (6 XE nodes), the simulation on Blue Waters runs ~ 41 times faster than on our Lab Computer (LC) (DELL Precision T7600 with Intel Xeon E5-2603 1.80 GHz CPU processor/node with 8 Cores): One iteration on Blue Waters using 96 cores (6 XE nodes) requires ~2.4 seconds of wall clock time. One the other hand, on the lab workstation (using 1 core), the same simulation requires ~ 98.4 seconds of wall-clock time per 1 iteration. Fluent computations on Blue Waters
show almost linear speed-up with increasing XE nodes.
0 16 32 48 64 80 960
1
2
3
4
5
6
7
8
9
10
11
12
13
14 Computing time per 1 iteration Speed-up ratio relative to lab workstation using 1core
Total cores used during computation (#)
Co
mp
uti
ng
tim
e p
er 1
iter
atio
n (
sec)
0
5
10
15
20
25
30
35
40
45
Sp
eed
-up
rat
io*1 BW_XE node usage per 16 cores
Speed-up ratio = Computing time per 1 iteration on LC / Computing time per 1 iteration on BW
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •14/35
0 5 10 15 20 25 30 35 40 45 500
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Computing time per 1 iteration Speed-up ratio relative to lab workstation using 1core
Number of XE node (#)
Co
mp
uti
ng
tim
e p
er 1
iter
atio
n (
sec)
0
5
10
15
20
25
30
35
40
45
Sp
eed
-up
rat
io
Fluent Performance on BW: Nodes-Cores Distribution Test
<Effect of distribution of nodes and cores on speed up of FLUENT calculation on Blue Waters>
3 nodes x 16 cores
6 nodes x 8 cores 24 nodes
x 2 cores12 nodes x 4 cores
48 nodes x 1 core
*Fixed number of total cores: 48 cores
For using same total cores (#48), speed-up of Fluent computation is more enhanced by increasing compute nodes.
Applying the distribution of 48 nodes x 1 cores (total 48 cores), shows more speed-up (~43 times vs ~ 41 times)than using total 96 cores (6 nodes x 16 cores): 48 HPC licenses of the Ansyslicense pool can be saved with more speed-up.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •15/35
0 5 10 15 20 25 30 35 40 45 500.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
BW
-co
re e
ffic
ien
cy
Number of XE nodes (#)0 16 32 48 64 80 96
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
BW
-co
re e
ffic
ien
cy
Total cores used during computation (#)
Fluent Performance on BW: BW-Core License Efficiency
Blue Waters (BW)-core license efficiency = Speed-up ratio / total assigned cores = Computing time per 1 iteration on LC / (Computing time per 1 iteration on BW x total assigned cores)
<Effect of distribution of nodes and cores on BW-core efficiency>
<Effect of number of cores on BW-core efficiency>
*Fixed number of total cores: 48 cores
3 nodes x 16 cores
6 nodes x 8 cores
24 nodes x 2 cores
12 nodes x 4 cores
48 nodes x 1 core
*1 BW_XE node usage per 16 cores
Use of full 16 cores per 1 node for Fluent calculation on BW, has low efficiency: only 0.4 ~0.5 of lab computer computation using 1 core. For using same total cores (#48), BW calculation per 1 core shows much higher efficiency by
increasing compute nodes.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •16/35
Time-averaged Jet Flow Patterns(Case W-LC)
Case W-LC. +15° (up) angle nozzle Case W-LC. -15° (down) angle nozzle
Back flow from mold to nozzle port, get more with 15 ° (up) nozzle port Jet flow in the mold get deeper with -15 ° (down) nozzle port
Nozzle port view
Nozzle port view
Center-middle plane view Center-middle plane view
Velocity magnitude (m/sec)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •17/35
Transient Swirl from Nozzle Port
Case W. +15° (up) angle nozzle portCase W-BW. +15° (up) angle nozzle port
Case W-BW. -15° (down) angle nozzle port Case W. -15° (down) angle nozzle port
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •18/35
Time-averaged Mold Flow Patterns (Case W-LC)
Case W-LC. +15° (up) angle nozzle Case W-LC. -15° (down) angle nozzle
Velocity magnitude (m/sec)
With the nozzle port having +15 ° (up) angle, jet flow goes down deep into wide mold cavity: downward flow is predominant, this is very harmful to produce internal defects.
Downward -15 ° degree nozzle port induces a classic double-roll pattern in wide mold.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •19/35
Vertical Velocity along NF (Case W-LC)
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20400
350
300
250
200
150
100
50
0
Dis
tan
ce b
elo
w t
he
surf
ace
(mm
)
Vertical velcotiy component along casting direction (m/sec)
15 (up) degree -15 (down) degree
~40 mm
~80 mm
Jet flow impingement point on narrow face, is moved deeper into the mold cavity, by decreasing nozzle port angle.
Upward angle of +15°causes strong downward flow along casting direction. °
°
*Error bars: expected velocity fluctuations = ( ) k3
2u
2' =
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •20/35
Turbulent Kinetic Energy in Mold(Case W-LC)
Case W-LC. +15° (up) angle nozzle Case W-LC. -15° (down) angle nozzle
Jet wobbling is more severe with +15 (up) degree nozzle port, resulting in higher surface flow instability
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •21/35
Transient Mold Flow Patterns
Measured: Case W. +15° (up) angle nozzle Measured: Case W. -15° (down) angle nozzle
Predicted: Case W-BW. +15° (up) angle nozzle Predicted: Case W-BW. -15° (down) angle nozzle
From both measurements and predictions, the case of 15 (up) angle nozzle shows more variations of jet flow: Jet flow from the nozzle port with 15 (up) angle, induces more severe jet wobbling by producing more swirl flow?
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •22/35
Water Model Measurements: Surface Velocity with +15°(Up) Angle Port
Velocity of flow from IR to OR (m/s)
Velocity of flow From NF to SEN (m/s)
Average Standarddeviation
Average Standarddeviation
30 mmfrom NF
-0.0175 0.0215 0.0535 0.0189
W/8 -0.0118 0.0173 0.0595 0.0310
W/4 0.0218 0.0335 0.0628 0.0292
<30 mm from NF> <W/8>
<W/4>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<30 mm from NF>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN <W/8 location>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN<W/4 location>
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •23/35
Water Model Measurements: Surface Velocity with +5° (Up) Angle Port
Velocity of flow from IR to OR (m/s)
Velocity of flow From NF to SEN (m/s)
Average Standarddeviation
Average Standarddeviation
30 mmfrom NF
-0.0125 0.0226 0.0561 0.0141
W/8 -0.00473 0.0215 0.0795 0.0282
W/4 -0.00492 0.0134 0.0985 0.0265
<30 mm from NF> <W/8>
<W/4>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<30mm from NF>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Ve
loc
ity
(m
/sec
)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN<W/4 location>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<W/8 location>
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •24/35
Water Model Measurements: Surface Velocity with -15° (Down) Angle Port
Velocity of flow from IR to OR (m/s)
Velocity of flow From NF to SEN (m/s)
Average Standarddeviation
Average Standarddeviation
30 mmfrom NF
-0.00381 0.0190 0.0768 0.0144
W/8 -0.00266 0.0204 0.118 0.0213
W/4 -0.00278 0.0126 0.124 0.0199
<30 mm from NF> <W/8>
<W/4>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<W/8 location>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Ve
loci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN<W/4 location>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<30mm from NF>
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •25/35
Water Model Measurements: Surface Velocity with -30° (Down) Angle Port
Velocity of flow from IR to OR (m/s)
Velocity of flow From NF to SEN (m/s)
Average Standarddeviation
Average Standarddeviation
30 mmfrom NF
-0.00158 0.0124 0.0651 0.00762
W/8 0.0000297 0.0121 0.112 0.0173
W/4 -0.00305 0.0155 0.105 0.0172
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oc
ity
(m
/se
c)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<30mm from NF>
<30 mm from NF> <W/8> (near NF)
<W/4>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oci
ty (
m/s
ec)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<W/8 location>
0 100 200 300 400 500 600 700 800 900 1000-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Vel
oc
ity
(m
/sec
)
Time (sec)
velocity of flow from IR to OR velocity of flow from NF to SEN
<W/4 location>
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •26/35
Model Validation (Results of Case W and Case W-LC) : Horizontal Surface Velocity
0 50 100 150 200 250-0.05
0.00
0.05
0.10
0.15
0.20
Ho
rizo
nta
l ve
loc
ity
co
mp
on
en
t to
wa
rds
SE
N (
m/s
ec
)
Distanc e from nozzle center (mm)
Predicted Measured15 (up) degree: -15 (down) degree:
Su
rfac
e-fl
ow
vel
oci
ty t
ow
ard
s S
EN
(m
/sec
)
*Error bars: expected velocity fluctuations = ( ) k3
2u
2' =
RANS model using standard k-ε model agrees with water model measurements of surface velocity and its fluctuation
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •27/35
Model Validation (Results of Case W and Case W-LC) : Surface Level
0 50 100 150 200 250-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
Su
rfa
ce
lev
el (
mm
)
Distance from SEN center (mm)
Predicted (Case W-LC) Measured (Case W)+15o (up) angle:
-15o (down) angle:
i AVGi
w
P -Ph =
ρ g
Average surface level profile (hi):
Surface level fluctuation ( ):
i
k∆h =
g
i∆h
RANS model using standard k-εmodel agrees with water model measurements of average surface level profile Discrepancy of level fluctuation:
Low resolution (0.5 mm) of level sensor is not sufficient to capture level fluctuations in water model mold.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •28/35
Effect of Port Angle on Average of Velocity Components
130 140 150 160 170 180 190 200 210 220 230 240-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
Vel
oci
ty (
m/s
ec)
Distance from mold center (mm)
15 (upward) 5 (upward) - 15 (downward) - 30 (downward)
<Comparison of average cross-flow velocity component towards OR>
<Comparison of average surface velocity component towards SEN>
130 140 150 160 170 180 190 200 210 220 230 2400.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
0.12
0.13
Vel
oci
ty (
m/s
ec)
Distance from mold center (mm)
15 (upward) 5 (upward) - 15 (downward) - 30 (downward)
The cases of downward port angle show higher surface velocity towards SEN, than upward angle cases -15° (down) angle port causes the highest
surface velocity component towards SEN.
Upward angle port produces more asymmetric flow between IR and OR than downward angle port.
w/4 point w/8 point 30 mm from NF
Su
rfac
e-fl
ow
vel
oci
ty t
ow
ard
s S
EN
(m
/sec
)
Cro
ss-f
low
vel
oci
ty t
ow
ard
s O
R (
m/s
ec)
Wy,u Wz,u
°°°°
°°°°
SENSEN
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •29/35
Effect of Port Angle on Fluctuations of Velocity Components
<Comparison of fluctuations of velocity component towards OR>
<Comparison of fluctuations of velocity component towards SEN>
130 140 150 160 170 180 190 200 210 220 230 2400.00
0.01
0.02
0.03
0.04
0.05
Vel
oc
ity
(m
/sec
)
Distance from mold center (mm)
15 (upward) 5 (upward) - 15 (downward) - 30 (downward)
130 140 150 160 170 180 190 200 210 220 230 2400.00
0.01
0.02
0.03
0.04
Vel
oci
ty (
m/s
ec)
Distance from mold center (mm)
15 (upward) 5 (upward) - 15 (downward) - 30 (downward)
Velocity fluctuations towards both SEN and OR, increase with increasing nozzle port angle
ST
DE
V O
f S
urf
ace-
flo
w v
elo
city
to
war
ds
SE
N (
m/s
ec)
ST
DE
V o
f C
ross
-flo
w v
elo
city
to
war
ds
OR
(m
/sec
)
( )2'Wy,u ( )2'
Wz,u
°°°°
°°°°
SEN SEN
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •30/35
Power Spectrum Analysis of Transient Surface Velocity towards SEN of Water Model measurements
1E-3 0.01 0.1 11E-15
1E-14
1E-13
1E-12
1E-11
1E-10
1E-9
1E-8
1E-7
1E-6
1E-5
1E-4
Po
wer
(m
2/se
c2)
Frequency (Hz)
15 (upward) 5 (upward) -15 (downward) -30 (downward)
<Power of velocity variation on W/8 location at the surface>
<Oscillation period plotted against ratios of submergence depthsand casting speeds[*]>
Upward port angle induces stronger surface velocity variations. For the nozzle port with +15° (up) degree angle, the frequency of ~0.11 Hz, for asymmetric flow
past the SEN predicted using Honeyands and Herbertson’s relation, is found in the power spectrum
*T.A. Honeyands and J. Herbertson: 127th ISIJ Meeting, Tokyo, Japan, March, 1994.
~5.6
~9.3
~0.11 Hz (~ 9.3 s oscillation)
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •31/35
Nail Dipping Tests
Total tests (for each IR and OR region)
5
interval of each test 1 minute
Dipping time 3 sec
Dipping test details
<Measurement locations in mold>
1600 mm
200
mm
W/8
200 mm<Case R>
SEN
600 mm
Measurement locations
50 mm50 mm
( ) ( )0.567
lump
0.696
lumps (mm)h(mm)φ0.624(m/s)u ⋅⋅= −
Empirical equation for surface velocity magnitude[*]:
* Liu et al., Proc. of TMS 2011,TMS, Warrendale, PA, USA
<Solidified lump from nail dipping test>
hlump
lumpφ
Surface velocity magnitude:
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •32/35
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0-0.25
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
Ho
rizo
nta
l vel
oci
ty c
om
po
nen
t to
war
ds
SE
N (
m/s
ec)
Distance ratio from nozzle center to NF
Computational model 1/3 scale water model Nail board measurement
Scale up method:
Su
rfac
e-fl
ow
vel
oci
ty t
ow
ard
s S
EN
(m
/sec
)
ys,u
×=
Win,
Rin,Ws,Rs, u
uuu
( )
( )
××
××
×=
2
W
WWWc,
2
R
RRRc,
Ws,
rπ
twu
rπ
twu
u
×=
Wc,
Rc,Ws, u
uu
W
RWs, L
Lu ×=
Ws,Rs, u1.73u ×= Limitation to understand surface flow in real wide-mold caster
from water model measurement and computational modeling of water model: Computational modeling for real caster case including steel shell and liquid mold flux layer is needed
Suggested by Singh, Thomas, and Vanka [*]
Froude number similarity
s: surface, in: nozzle inlet, r: nozzle inner radiusw: mold width, t: mold thickness, L: process length scale
* Sigh et al., MMTB, Vol. 44B, 2013, p.1201-1221
Comparison of Plant Measurements (Case R), 1/3 Scale Water Model Measurements (Case W), Computational Model (Case W-LC): All Scaled-Up of Surface Velocity for Real Caster
+15° (Up) degree nozzle
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •33/35
Summary: Modeling and Measurements
1/3 scale water model experiments, plant measurements, and computational modeling using both lab computer and Blue Waters supercomputer, were performed to investigate effect of nozzle port angle on nozzle and mold flow, for reducing surface defects in wide slab.
Reynolds-Averaged Navier-Stokes (RANS) model using standard k-εmodel on lab computer and Large Eddy Simulation (LES) on Blue Waters supercomputer, were used to quantify time-averaged and -dependent flow in 1/3 scale water model
RANS model shows a very good quantitative match with average surface velocity profile across 1/3 scale water model.
LES model can capture transient nozzle swirl and jet wobbling, which is important transient flow phenomena to cause surface instability, related to surface defect formation with nozzle port angle effect.
Water model surface velocities near narrow face exceed those in real wide-mold caster.
Transient flow modeling including steel shell and liquid mold flux layer, is likely needed to understand surface flow variations (especially, near NF) in wide mold of real caster
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •34/35
Summary: Effect of Nozzle Port Angle & Suggestion
Jet flow from +15° (up) nozzle port shows more severe wobbling than +5°(up), -15°(down), -30°(down) ports; this jet impinges first on top surface causing surface instability (the highest surface velocity variations even though it has the lowest surface velocity).
Higher surface velocity fluctuations not always caused by faster surface flow: surface instability depends on casting conditions
Maximum average surface velocity is produced by port angle of -15° (down). - Deeper port angle (-30° (down)) has slower surface velocity.- Shallower port angle (+5° (up) and +15° (up) degree) has slower surface velocity.
Maximum surface velocity when jet impinges on NF at upward angle: Surface velocity slower if jet first impinges on NF at downward angle or near top surface or corner
Up-angled nozzle with non-optimized SEN depth could be detrimental in causing both severe surface instability (surface defects) and abnormal downward flow (internal defects) deep into mold cavity.
Worst flow pattern (from +15° in this work) is unstable between single and double-roll: pressure sucks jet up to impinge top surface; or down to impinge on NF; wobbling between causes instability and defects)
Deeper submergence is suggested for up-angled nozzle in this caster system to enable jet flow-hit to impinge first on NF
High-frequency low-amplitude turbulence is optimal to get mixing, heat transfer to meniscus without surface instability: Avoid High-power lower-frequency oscillations with large spatial variations.
University of Illinois at Urbana-Champaign • Metals Processing Simulation Lab • Seong-Mook Cho •35/35
Acknowledgments
• Continuous Casting Consortium Members(ABB, AK Steel, ArcelorMittal, Baosteel, JFE Steel Corp., Magnesita Refractories, Nippon Steel and Sumitomo Metal Corp., Nucor Steel, Postech/Posco, SSAB, ANSYS/ Fluent)
• Blue Waters / National Center for Supercomputing Applications (NCSA) at UIUC
• POSCO (Grant No. 4.0011721.01)• Dr. Ahmed Taha, NCSA for help with the Fluent
Module Implementation into Blue Waters• Hyun-Na Bae, Dae-Woo Yoon, and Seung-Ho Shin
POSTECH for help with the 1/3 scale water model experiments