ISIJ International, Vol. 51 (2011), No. 10, pp.

1674–1681

Molten Steel Level Measurement in Tundish with Heat Transfer

Analysis

Zhenwei HU, Ying CI and Zhi XIE

College of Information Science and Engineering, P.O. Box, No.321, Northeastern University, Shenyang 110819, Liaoning, P.R.
China.
(Received on April 5, 2011; accepted on June 8, 2011)

Temperature variation rate is different between “molten steel – tundish cover flux(TCF) – air” layers in
tundish because heat transfer between these three materials with different thermal conductivities. Numer-
ical calculation proves that temperature gradient extremums exist at the interfaces of “molten steel – TCF
– air” layers in tundish. At the same time the change of initial temperature, thermal conductivity, convec-
tion coefficient have no effect on the interfaces locating. Thus temperature gradient extremums can be
used for measuring molten steel level in tundish. In the process of measurement, a CCD camera is used
to capture the thermal image of measuring sensor which is inserted into tundish to get the temperature
distribution. By calculating temperature gradients of the measuring sensor in the thermal images, the
interfaces of “molten steel – TCF – air” can be located, finally the molten steel level is obtained. The influ-
ence of measuring sensor adhering by TCF is also solved by gray image projection algorithm. Steel met-
allurgical field experiment shows that the method of molten steel level measurement is authentic and
measuring error is less than 5 mm.

KEY WORDS: molten steel level; measurement; heat transfer; temperature distribution; interface location.

top layer. For great different thermal conductivities of these

1. Introduction
three materials (steel, TCF and air), the temperature gradients
In metallurgy industry field it is significant to get molten at the interfaces between them are obvious characteristics.
steel level in continuous casting tundish for billet quality In order to obtain the differences of the temperature gra-
and safety production.1,2) Harsh measuring circumstance dient between the three mediums, we use the method as
caused by high-temperature of molten metal especially the follow which is shown in Fig. 1(a): insert a measuring
TCF at top layer makes it difficult to measure molten steel sensor made of refractory material (called “measuring
level accurately. sensor” below) into tundish through all the mediums includ-
As the different densities between molten steel and TCF, ing the air, TCF and molten steel layer to perceive the tem-
the traditional weighting method can’t give the accurate perature. When the measuring sensor reaches thermal bal-
molten steel level information.3) Many specialists and schol- ance, raise it and use a charge coupled device(CCD) camera
ars have done remarkable researches on molten steel level to acquire the thermal radiation. When the measuring sensor
measurement techniques.4–9) The isotope method4,5) has radi- has been raised, the raising parts drill through the TCF layer
ation pollution which is quite dangerous. Katakin and and it is adhered by the TCF. Thus the surface of raising
Pokrovskiy using eddy-current transducer to detect steel parts has the same emissivity, so the distribution of the tem-
level,6) the theoretical maximum measuring range is about perature can be measured due to the distribution of radiant
300 mm, but the molten steel level is over 700 mm in most energy. By analyzing the thermal image of the measuring
tundishes. Ultrasonic waves,7) laser,8) microwave9) methods sensor, detect the temperature gradient between the layers
are also studied to measure molten steel level, but because by means of image processing and recognition, the molten
the uneven TCF layer spatial distribution and the time-vary- steel level is measured finally. The temperature distribution
ing thickness, obtaining accurate molten metal level and the of measuring sensor is shown in Fig. 1(b).
thickness of TCF layer is still a difficult problem. The measuring process in this paper is shown in Fig. 2.
In order to overcome the measurement disadvantages of LMB is the length of the measuring sensor, HD is the vertical
TCF layer, a new method of molten steel level measurement distance between the top of the measuring sensor and optical
based on temperature gradient analysis is proposed. center of camera lens, TTCF is the thickness of TCF layer.
The interface of air layer and TCF layer is expressed as
TCF-level, which is designated as HTCF The interface of
2. Measuring Principle
molten steel layer and TCF layer is called molten steel level,
2.1. Measuring Method and Process which is designated as HSTEEL.
There are three layers in tundish: molten steel at the bot- The molten steel measuring process is separated into two
tom layer, TCF layer at the middle layer and air layer at the steps as follow:

© 2011 ISIJ 1674

 ISIJ International, Vol. 51 (2011), No. 10

Fig. 1. Schematic of measuring principle.

are recognized. The distance between P*AS, P*SS and the top
of measuring sensor is designated as H *ASL and H *SSL. Then
the thickness of the slag layer is:
TTCF = H *SSL – H *ASL ......................... (2)
Finally the molten steel level is:
HSTEEL = HTCF – TTCF = LMB – HASL + HDEEP – (H *SSL – H *ASL)
.......................................... (3)

2.2. Physical Parameters of Measuring Sensor

Physical parameters of measuring sensor should meet the
following conditions: the time when measuring sensor reach-
es thermal balance between TCF layer and molten steel layer
should be as short as possible. Meanwhile, the temperature
distribution of measuring sensor should be maintained as
long as possible when being elevated from the TCF and mol-
ten steel layer. Therefore the material of measuring sensor
needs thermal conductivity and specific heat of the selection,
Fig. 2. The measuring process of molten steel level measuring
alumina-carbon is a suit one, its thermal conductivity is 8.7
system (a) measure TCF level HTCF (b) measure molten W/(m·K) and specific heat is 1 000 J/(kg·K).
steel level HSTEEL. The measuring sensor is made in rod shape; at the same
time it is hollow-core for the purpose of measuring internal
temperature in order to validate ANSYS calculation results
Step 1: measure TCF-level HTCF. in the following part of this paper. The parameter of mea-
Usually, the measuring system is in the state of TCF-level suring sensor is shown as follow: the length is 960 mm, the
measurement, as shown in Fig. 2(a). During this time, imag- diameter is 85 mm and the hollow-core diameter is 25 mm.
es of the measuring sensor inserted into molten metal is
acquired by the camera; obtain the air-TCF layer interface
3. Temperature Distribution Analysis of Measuring
PAS by pattern recognition; then calculate HASL (the vertical
Sensor
distance between measuring sensor apex and PAS). HDEEP is
calculated according to the encoder, the length of measuring Finite element analysis is used to prove the validity and
sensor and the depth of tundish. Finally, the TCF-level HTCF reliability of molten steel level measuring method not only
is expressed as: in the standard state but also when physical parameter
changes. According to the physical parameters of measuring
HTCF = LMB – HASL + HDEEP .................... (1)
sensor, initial conditions and boundary conditions, the heat
Step 2: measure molten steel level HSTEEL. transfer differential equation between measuring sensor and
The mechanism takes the camera and the measuring three layers, the temperature distributions of the measuring
sensor move upwards when the measuring system is in sensor is calculated by the ANSYS software.
TCF-level measurement state and measuring sensor has
reached thermal balance. The measuring sensor is lift to a 3.1. Heat Transfer Differential Equation
position of a setting height which is recorded by the encoder The heat transfer differential equation can be described
and shown in Fig. 2(b). In the TCF thickness measurement according to the Fourier equation:10)
state, images of measuring sensor are acquired by the
camera, and then the temperature distribution at surface of ∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂T
⎜ kx ⎟ + ⎜ ky ⎟ + ⎜ kz ⎟ + qsh = ρ c
the measuring sensor is extract. The character of tempera- ∂x ⎝ ∂x ⎠ ∂y ⎝ ∂y ⎠ ∂z ⎝ ∂z ⎠ ∂t
ture extremums of TCF-steel layer interface P*AS and P*SS .......................................... (4)

1675 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 10

For isotropic thermal conductivity of materials, kx = ky = km: thermal conductivity of measuring sensor which is 8.7
kz = k, formula (4) can be transferred into: W/(m·K), TTCF: temperature of TCF layer which is 1 200°C,
Ts: temperature of molten steel layer which is 1 530°C, Tev:
⎛ ∂ 2T ∂ 2T ∂ 2T ⎞ ∂T ............ (5) temperature of air layer which is 30°C, Tm: initial tempera-
k ⎜ 2 + 2 + 2 ⎟ + qsh = ρ c
⎝ ∂x ∂y ∂z ⎠ ∂t ture of measuring sensor which is 30°C, hss: heat transfer
coefficient between molten steel layer and TCF layer which
The temperature distribution analysis of the measuring
is 200 W/m2·K, hsa: heat transfer coefficient between TCF
sensor is three-dimensional unsteady thermal analysis, fur-
layer and air layer which is 0.4 W/m2·K, hs: heat transfer
ther more there is no heat source inside the sensor, formula
coefficient between measuring sensor and molten steel layer
(5) can be simplified to:
which is 3 000 W/m2·K, hTCF: heat transfer coefficient bet-
∂ 2T ∂ 2T ∂ 2T ρ c ∂T ween measuring sensor and TCF layer which is 40 W/m2·K,
+ + = ................... (6)
∂x 2 ∂y 2 ∂z 2 k ∂t hf : heat transfer coefficient between measuring sensor and
air layer which is 5 W/m2·K.
Formula (6) is the heat transfer differential equation of
measuring sensor.
3.3. Temperature Distribution of Measuring Sensor
Using finite element method to solve temperature distri-
3.2. Physical Parameters and Boundary Conditions
bution model based on the heat transfer process and definite
Physical parameters of three mediums11,12) in tundish is
conditions, appending loads of temperature, convection,
shown in Table 1 and the thickness of TCF layer is 30 mm.
heat conduction and radiation. After the calculation by
The initial temperature of measuring sensor is described
ANSYS, nephogram of the temperature distribution of the
as formula 7 when heat transfer begins:
measuring sensor is shown in Fig. 3.
T t =0 = ϕ ( x, y) .............................. (7) Because of the thermal conductivity difference between
air, TCF layer and molten steel, temperature variation rate
ϕ (x, y) is the function of initial temperature and in this
is different between three materials. Calculating the temper-
paper measuring sensor is set to 30°C which is supposed to
ature gradients in the region of [A, B] near the TCF layer,
an isothermal body.
the results are shown in Fig. 4.
Boundary conditions include as follows:13)
There are two obvious inflexions in the temperature gra-
(a) boundary between molten steel layer and TCF layer:
dient curve. According to the geometrical relation the first
∂TTCF inflexion at position “a” is the interface of molten steel layer
− ks = hss (Ts − TTCF ) .................... (8)
∂n and TCF layer, the second inflexion at position “b” is the
interface of TCF layer and air layer. So the position of
(b) free surface of TCF layer:
inflexion can be used as criterion for locating the interface.
∂TTCF In order to demonstrate the universality of the conclu-
− ks = hsa (TTCF − Tev ) ................... (9)
∂n sions, change the initial temperature of measuring senor,
(c) ektexine of measuring sensor where contacting with
molten steel layer
∂Tm
− km = hs (Ts − Tm ) ..................... (10)
∂n
(d) ektexine of measuring sensor where contacting with
TCF layer
∂Tm
− km = hTCF (TTCF − Tm ) ................. (11)
∂n
(e) ektexine of measuring sensor where contacting with
air layer

∂Tm
− km = h f (Tm − Tev ) .................... (12)
∂n
In formulas (8)–(12), Fig. 3. Nephogram of calculated temperature distribution of the
ks: thermal conductivity of TCF layer which is 0.6 W/(m·K), measuring sensor.

Table 1. Physical parameters of three mediums in tundish.

Thermal conductivity
Density (kg/m3) Specific heat (J/(kg·K))
(W/(m·K))
Air 1.29 1.004 0.023
TCF 1 210 1 172.3 0.6
Molten steel (1 530°C) 7 000 880 48

© 2011 ISIJ 1676

 ISIJ International, Vol. 51 (2011), No. 10

Fig. 5. Influence of different initial temperature (a) temperature distribution (b) temperature gradient distribution.

Fig. 4. Surface temperature and gradient distribution of measuring

sensor along its vertical direction.
Fig. 6. Influence of different thermal conductivity on distribution
of temperature gradient.

convection thermal conductivity of TCF layer and coeffi-

cient between molten steel and TCF layer for calculating.
In metallurgical field, refractory usually preheats before
using. So initial temperature of measuring senor is set as
400°C, the results is shown in Fig. 5.
When initial temperature increases, the thermal equilibri-
um time decrease and the surface temperature of measuring
sensor is different, but position change of two inflexions in
the temperature gradient curve is less than 1 mm.
In order to verify the heat conduction between the mea-
suring sensor and different kinds of tundish cover flux, ther-
mal conductivity of TCF layer is changed from 0.8 W/(m·K)
to 1.6 W/(m·K). The calculating result is shown in Fig. 6,
it can be made out that the change of thermal conductivity
has small affect on the temperature gradient, the interface
Fig. 7. Influence of different convection coefficient on distribution
location doesn’t change. of temperature gradient.
For verifying the heat convection between measuring sen-
sor and different steel grades (including the effect of molten
steel flow speed), convection coefficient between molten the interface of “molten steel layer - TCF layer” and “TCF
steel and measuring sensor is changed from 2 000 W/(m2·K) layer - air layer” can be identified according to the position
to 8 000 W/(m2·K). The calculating result is shown in Fig. of extremums in temperature gradients distribution.
7, when the convection coefficient increases the difference
of temperature gradients increase, but it has no effect on 3.4. Verification of the ANSYS Calculation
locating the interface of “molten steel layer - TCF layer” In order to validate the correctness of the model men-
and “TCF layer - air layer”. tioned above, measuring temperature of measuring sensor at
Therefore the two position of extremums which are cal- metallurgy field is compared with the result of ANSYS. The
culated from temperature gradients of ektexine of measuring method is described as follows: the measuring sensor is
sensor, have no effect on the initial temperature, thermal hollow-core, when being inserted into the molten steel; the
conductivity and convection coefficient, the only change is hollow-core becomes a blackbody cavity.14) Then using a
the value of inflection point of temperature gradients. Thus radiation thermometer pointed to the hollow-core at the top

1677 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 10

of the measuring sensor, the temperature of measuring sen-

sor can be measured by the blackbody radiation from the
hollow-core. Field temperature measuring device is shown
in Fig. 8.
Depend on this method the temperature of measuring
sensor is measured at metallurgy field during the first 400
second after the measuring sensor being inserted into
tundish, Fig. 9 shows the comparison with the measuring
result and calculating result by ANSYS.
The comparison shows, the maximum difference temper-
ature between the measuring and modeling result is 9°C,
relative error is less than 1%, so the model satisfy the appli-
cation requirements.
The model proves that temperature gradient has extre- Fig. 8. Temperature measuring device of molten steel level mea-
mums at the interfaces between molten steel layer, TCF suring sensor.
layer and air layer. At the same time it provide evidence for
interface orientation in the image of measuring sensor
inserted into the molten steel layer.

4. Result and Discussion

4.1. Interface Locating in Typical Image
Figure 10 shows that the system has been applied to mea-
suring molten steel level of continuous casting tundish in
steel metallurgy industry field. During the continuous cast-
ing process the CCD camera captures the images of measur-
ing sensor for interfaces locating.
A typical image when the measuring sensor inserts into
molten steel is shown in Fig. 11(a), then the system mea-
sures the level of TCF layer continuously. Figure 11(b) is an
image of the measuring sensor when it was raised up. The Fig. 9. Comparison of calculating data and measuring data.
temperature distribution of the measuring sensor is shown in
Fig. 11(c). The green dots in the figure represent reliable
temperature distribution points on the surface of the measur-
ing sensor extracted by system; the white curse represents
axial gray distribution of the measuring sensor; and the two
vertical white lines acquired by discrimination according to
local extremum which PAS represents the air-TCF layer
interface, PSS represents the TCF-steel layer interface. Final-
ly the molten steel level can be calculated by formula 3.

4.2. Influence of TCF Flow

During the measuring process, the surface of the measur-
ing sensor adheres TCF when it rises through the TCF layer,
it seriously affects getting reliable temperature distribution.
The TCF has good flowability, it flows downward
through the surface of the measuring sensor in the rising Fig. 10. Measuring system working on tundish.
process. The interface of TCF layer and molten steel layer
are influenced, furthermore, temperature distribution can’t
be extracted or formed pseudo interface at the surface. TCF
flow images are shown in Fig. 12.
For this case, gray projection method is used to recognize
the flow of TCF on the surface of measuring sensor. TCF
flows about 10 mm per second. The thickness of TCF layer
in tundish is 40–120 mm for different casting conditions. It
may cost about 4 seconds that the TCF flow through the part
which the measuring sensor is raised up from thinnest TCF
layer. So the time interval of two TCF flow feature images Fig. 11. Interface locating (a) measuring sensor inserted into the
can be selected as any one between 1 and 4 seconds. For molten steel layer (b) measuring sensor raised for TCF
reliability and resolution, the TCF flow distance is calcu- measurement (c) interfaces location by temperature distri-
lated between t second image and t+1 second image in this bution.

© 2011 ISIJ 1678

 ISIJ International, Vol. 51 (2011), No. 10

paper. By comparing with the interface location distance, image pixel, y is vertical coordinates of image pixel, H is
pseudo interface is identified and removed. The concrete the height of image, W is the width of image, g1(xi, y) is
steps are shown as follows: pixel-gray value of t second image in coordinate (xi, y),
Step 1: get the TCF flow image of t second and t+1 g2(xi, y), is pixel-gray value of t+1 second image in coordi-
second. nate (xi, y), p(xi) is horizontal gray projection of t second
Step 2: calculate the gray projection of the images: image in column i, q(xi) is horizontal projection of t+1 sec-
Horizontal gray projection can be calculated as follow: ond image in column i, ε is noise reduction threshold. The
H
relationship of coordinate system is shown in Fig. 13.
p( xi ) = ∑ g1 ( xi , y) g1 ( x , y ) >ε (i = 0 ⋅⋅⋅ W ) ........ (13) the integration of p(xi) and q(xi) can be expressed:
y =0 xi

H
S p ( xi ) = ∑ p( x ) ......................... (15)
q( xi ) = ∑ g2 ( xi , y) g2 ( x , y ) >ε (i = 0 ⋅⋅⋅ W ) ........ (14) x = x0
y =0 xi
Which xi is the i position in horizontal coordinates of Sq ( xi ) = ∑ q( x ) ......................... (16)
x = x0

Which x0 is integral lower limit, xi is integral upper limit

Step 3:calculate the distance of TCF flow:
Set xpi is the abscissa of TCF edge in t second image,
then the integral value can be expressed as Sp(xpi), we can
found a point xqi in the abscissa of t+1 second image that
S p ( x pi ) − Sq ( xqi ) reaches minimum. Finally the flow distance
of TCF at the surface of measuring sensor is:
l = xqi – xpi ............................... (17)
Step 4: Calculate the gradient of temperature distribution
of measuring sensor in t and t+1 second images, search the
interface according to the extremum.
Step 5: The validity of interface can be compared the
Fig. 12. Typical images of TCF flow (a) TCF drops flow (b) TCF
measuring interface distance with the calculating result
complete coverage flow (c) TCF block flow (d) thick TCF according to formula 18. If the extremum position moves
flow. between successive images match the calculating result it
can be confirmed that the moving extremum position is a
pseudo interface.
According the successive TCF flow images in Figs. 14(a),
14(b), the integration of horizontal gray projection is calculated
which is shown in Fig. 14(c). Using formula S p ( x pi ) − Sq ( xqi ) ,
xqi is calculated on the curve of horizontal gray projection in
t+1 second image. Comparing the position of xpi and xqi, the
flow distance of TCF at the surface of measuring sensor is
9 mm.
At the same time the interface is located which is shown
in Fig. 15, taking the distance between real interface and
pseudo interface as the research object which is “ab” in Fig.
Fig. 13. The relationship of image coordinate system. 15(a) and “cd” in Fig. 15(b). “ab” is 8 mm, after TCF flow

Fig. 14. (a) TCF flow image at t second (b) TCF flow image at t+1 second (c)gray projection of image (a) and (b).

1679 © 2011 ISIJ

 ISIJ International, Vol. 51 (2011), No. 10

Fig. 15. Identify pseudo interface (a) real interface and pseudo interface of image at t second (b) real interface and pseudo
interface of image at t+1 second.

Table 2. Comparison of measuring and manual result.

Automatic Manual
Error
Measuring condition measuring Measuring
(mm)
result (mm) result (mm)
784 788 –4
Casting process 1: 787 783 +4
temperature of molten steel:
1 475°C–1 490°C; 747 751 –4
thickness of TCF layer: 719 724 –5
40 mm–60 mm.
(around the ladle change period) 762 758 +4
Fig. 16. (a) schematic of manual measuring method (b) field appli-
cation. 784 789 –5
674 670 +4

Casting process 2: 682 685 –3

“cd” is changed to 16 mm. On the other hand the flow dis-
temperature of molten steel: 672 674 –2
tance of TCF is about 8 mm, that meets the result of the gray 1 511°C–1 523°C;
projection calculating results. Thus the pseudo interface is thickness of TCF layer: 677 682 –5
identified. By removing the pseudo interface the molten 60 mm–80 mm. 683 685 –2
level in tundish is calculated according to interface a and b 689 691 –2
using formula 3. 676 681 –5
Casting process 3: 626 730 –4
4.3. Validation of Measuring Results temperature of molten steel:
1 535°C–1 548°C; 573 568 +5
In order to validate the accuracy of molten steel level
thickness of TCF layer: 465 462 +3
measuring system, a manual measurement method is used.15) 100 mm–120 mm.
A steel bar is insert into the tundish which has the same steel (at the end of casting period) 421 417 +4
grade as the molten steel casting in tundish. The part of the 385 380 –5
steel bar is melted which being inserted into the molten steel
layer, on the other hand the part in the TCF layer is still solid
state, its surface is adhered by TCF, the adhere length is the
thickness of TCF layer. Recording the length of steel bar
which is from melting position to the top of tundish which
expresses as HM, the depth of tundish is known as HT. So
the molten steel level is H *Steel = HT – HM . Figure 16 shows
the schematic of manual measuring method and its applica-
tion.
The automatic measuring result and the manual measur-
ing result at a steel corporation is shown in Table 2, for
three casting processes which the thickness of TCF layer is
cover from 40 mm–120 mm, temperature of molten steel is
cover from 1 475°C to 1 548°C. Deviation between auto-
matic measuring method and manual method is lower than
5 mm. Fig. 17. The history data of molten steel level measuring system and
In casting process 1 the measuring data during two ladle the manual validation.
change period is shown in Fig. 17, in the first ladle change
period which is 178 seconds long the maximum steel level the increasing thickness of TCF has no effect to the measur-
is 792 mm and the minimum is 711, it is intuitional for ope- ing result and during the end of casting period the measuring
rator better than tundish weight. In casting process 2 and 3, range fits the change of molten steel level. In metallurgical

© 2011 ISIJ 1680

 ISIJ International, Vol. 51 (2011), No. 10

field operator believes that exact steel level value is suit for Acknowledgements
continuous casting process and steel quality control, so this The authors would like to thank the field research support
method can meet the requirement of the filed well. by Nanjing Iron & Steel Corporation of China.

REFERENCES
5. Conclusion
1) S. Yokoya and S. Takagi: ISIJ Int., 38 (1998), 1086.
(1) Temperature gradient analysis method is used to 2) A. Mehdi, E. Hossein and R. Ahmad: ISIJ Int., 48 (2008), 28.
3) R. Fash and G. Moulden: Iron Steel Tech., 4 (2007), 76.
measuring molten steel level in tundish. In measuring pro- 4) F. Y. Gao: J. Test Meas. Tech., 22 (2007), 505.
cess a measuring sensor is inserted into molten steel, after 5) Y. G. Tang: J. Trans. Tech., 20 (2001), No. 3, 39.
heat balance temperature gradients exist at the “ air layer – 6) R. A. Katakin and A. D. Pokrovskiy: Russ. J. Nondest. Test., 45
(2009), 542.
TCF layer ” interface and “TCF layer – molten steel layer” 7) Nippon Steel Corp.: JP Pat, 2904467-B2.
interface. 8) P. Bohlaender and H. P. Hippler: DE Pat, 2945251-A1.
9) Y. Kawata, T. Kusaca and C.Manabe: Ind. Metrol., 1 (1990), 19.
(2) The initial temperature, thermal conductivity and 10) W. Q. Tao: Heat transfer, Northwestern Polytechnical Univ. Press,
convection coefficient of the measuring sensor have no Xian, (2006), 24.
effect on locating the “air layer – TCF layer” interface and 11) L. G. Zhu, S. M. Wang and A. Z. Wan: Res. Iron Steel, 107 (1999),
22.
“TCF layer – molten steel layer” interface. 12) L. G. Zhu, Y. W. Zhang and S. H. Zhang: Steelmaking, 15 (1999), 29.
(3) A CCD camera is used to detect these two inter- 13) S. H. Seyedein and M. Hasan: Can. Metall. Q., 37 (1998), 213.
14) Z. Xie, Y. Ci and H. J. Meng: Chin. J. Sci. Instrum., 26 (2005), 446.
faces, and then molten steel level can be calculated. At the 15) J. Kadota and K. Kohama: JP Pat, 8168860-A.
same time TCF flow is studied and solve by gray projection
method.
(4) The measuring error is less than 5 mm comparing to
the manual measurement at the steel metallurgical field.

1681 © 2011 ISIJ