Instrumentation
Mustafa R. Ozgu, Senior Research Consultant, Bethlehem Steel Corp.
20.1 Introduction
Instruments have been used on casters since the early days of continuous casting and, as is shown in Fig. 20.1, they can be found on every major component of a caster between the turret, or ladle car, and the run-out. The main functions of caster instruments are to: • Measure the parameters that are utilized for controlling the performance of mechanical and metallurgical functions of casting. • Assign a quality rating for each cast section. • Diagnose operating and machine problems.
ladle - weight - slag carry over tundish - weight - steel temperature - steel depth mold - copper temperature - steel level - oscillation - slab-mold friction - slag cover depth - in-gap slag thickness - wall deformation - steel flow velocity - water ∆T and flow rate
Fig. 21.1 Parameters measured on casters.
containment - strand surface temperature - strand shape and bulging run-out - final solidification point and segment/roll loads - length, width and weight - roll bending and temperature - hot surface quality - roll gap, allignment and rotation - cast speed and strand tracking
• Develop knowledge that correlates product quality and productivity to caster design and operation. The number and sophistication of instruments used on casters has been growing rapidly.1–3 The main reasons for the rapid growth are the ever-increasing demands for higher productivity and ascast product quality, and the availability of the modern on-line digital computer. This is particularly true for slab casters, where quality and productivity demands are the most stringent. Early on, emphasis was placed on mold instrumentation because mold practices and parameters have the most impact on product quality and productivity. However, lately, significant progress has been made in developing and applying instrumentation on the ladle, tundish, containment and runout. The text of this chapter is not meant to be an authoritative treatise on the history, theory and development, but rather a review of instruments used on casters. Sample measurement data and analyses results from several applications are reviewed as well.
Time (seconds)
Fig. 20.2 Typical processed accelerometer signal obtained during ladle teeming. From Ref. 8.
system incorporating slag alarm threshold points can be set up to activate an alarm and shut the flow off when the signal exceeds the threshold. Although the system is sensitive to environmental noise and hence is prone to inaccuracies, it can be used to reduce ladle slag carryover to the tundish. Because the accelerometer stays on the manipulator arm and can be used on the incoming ladle, the vibration analysis technique may appear attractive and preferable to other techniques that require sensor installation on each ladle. The electromagnetic method is the most widely used ladle slag flow detection method. It employs a transmitter and a receiver coil located around the exit nozzle in the ladle bottom, a pre-amplifier and a control unit. Two types of devices are used. In the first, the transmitter and receiver coils are opposite each other. In the second, the transmitter and receiver coils are concentrically oriented in the same housing. The sensor induces an electromagnetic field in the stream flowing through the exit nozzle and measures the resultant eddy currents. Because slag has a significantly lower electrical conductivity than liquid steel, the eddy currents induced in the slag are smaller than those induced in the steel. Hence, a transition from metal to slag and metal flow results in a rapid change in the output signal and the initiation of an alarm. Fig. 20.3 shows typical ladle weight and electromagnetic slag sensor output signals from a slab caster installation where the slide gate shutoff is manually activated upon the initiation of the alarm.10 As a result of the use of electromagnetic sensors, yield improvements ranging between 0.5 and 1.5%, and significant reduction in downgraded slabs attributable to ladle-to-tundish slag carryover have been reported by several steel plants. Other ladle sensors that are under development include an ultrasonic probe for the early detection of vortex formation13 and a microwave-driven slag thickness device.14
Fig. 20.5 Installation of electromagnetic level sensor in tundish bottom. From Refs. 19 and 21.
low tundish levels during transitions are increased yield during tundish exchanges and cast terminations, reduced mixed-grade-zone length during grade changes, and improved tundish life due to the ability to cast longer sequences with the same tundish. The electromagnetic sensors installed in the tundish bottom have a limited measurement range of 0–200 mm and thus cannot be used to measure and control tundish level during steady-state casting and ladle exchanges. Instead, the electromagnetic sensors can be installed on the sidewalls of the tundish, as depicted in Fig. 20.6.21,22 The transmitter and receiver coils can be installed on opposite walls, on adjacent wide and narrow walls, or side by side on the same wall. A tundish level measurement and control system incorporating electromagnetic sensors on the tundish bottom and vertical sidewalls may be preferable to the weight measurement system in some caster installations. Instruments such as displacement transducers, flow meters and pressure gauges are routinely used on the tundish to measure and control slidegate or stopper rod position, argon flows to the flow control system, and submerged tundish entry nozzle depth.
notebook computer
preamp level sensor loop level sensor loop
electronic unit and software main transformer
incoming power
flexible cable
instrument panel SEN control panel
Fig. 20.6 Electromagnetic level sensors in tundish vertical walls. From Ref. 20.
20.4 Mold
20.4.1 Introduction
The mold is the most complex and critical component of a continuous caster. The functions of the mold are to: • Act as a substrate for the formation of a thin, hot and crack-free solid shell. • Form the shape of the final product. • Extract heat from the strand at very high rates. • Facilitate the separation of nonmetallics from the solidifying shell. • Allow adequate production rates without breakouts. Since the early days of continuous casting, extensive studies have been conducted by numerous researchers to correlate the functions of the mold to its design and operation. As a result of these studies, both the design and operation of the conventional caster mold have been improved to the point where almost all steel grades can now be continuously cast with the desired surface quality and at high productivity rates. Significant improvements have also been made and continue to be made in the design and operation of the medium-thickness and thin-slab caster molds. The development and use of sensors have played a crucial role in the improvements made in mold design and operation. Because of the criticality of the mold and the relative ease with which sensors can be installed on it, a large number of mold parameters are now monitored for control and automation, on-line quality prediction and analysis in conventional as well as in medium-thickness and thin-slab casting. The following sections describe the mold parameters that are monitored and the various sensors used for monitoring.
path of heat flow (parallel to the hot face).30,34,35,38 The method is not suitable for routine installation of thermocouples because of the cost associated with the drilling of the thermocouple holes and the difficulties of handling thermocouple lead wires. However, it is the preferred method on molds that are specially instrumented for thermal analysis and evaluations because the disturbance caused by the thermocouples on heat flow—and thus absolute temperature measurement errors— are small. Thermocouples may be of the sheathed or intrinsic type. In sheathed thermocouples, at the measurement point a junction (“bead”) is formed between two wires of dissimilar metals, and the entire arrangement is encased in a protective sheath. With intrinsic thermocouples, the mold copper itself is used as one of the thermocouple metal components, while a second dissimilar metal, such as constantan, usually in the form of an electrically insulated rod, is welded or threaded to the desired measurement point in the mold copper. Sheathed thermocouples are more expensive and difficult to install than intrinsic thermocouples. However, they are more accurate than intrinsic thermocouples. Intrinsic thermocouples may be subject to electrical ground loop problems because of the possibility of variable electrical resistance paths between the mold coppers and the system ground potential of the measuring device. Response time of either type of thermocouple is determined mostly by the effective size of the “bead.” For general use, such as temperature, heat flux and sticker breakout prevention monitoring, response times of the order of one second are adequate. Two different instrumentation and analysis methods are used to determine the heat flux and the temperature variation along the heat path between the mold hot face and the cooling channels. In the first method, two thermocouples are used at different distances from the hot face along the path of heat flow.30,34 Fig. 20.7 shows the installation of dual thermocouples on the wide and narrow walls parallel to the hot faces of a slab caster mold.34 The heat flux and temperature distribution between the hot face and the water channels are then simply calculated from the gradient of the temperatures measured at the two points. In the second method, the inverse boundary solution method is used to calculate the local heat flux and temperature distribution from temperature measurements at a single point between the hot face and the cooling channel.38
615 485 102 152 229 635 615
457
900
711 bottom wide wall 21.6 water channel thermocouple holes 11.5 hot face section of top view dimensions: mm
narrow wall
259 narrow wall
Fig. 20.7 Dual termnocouple installation on wide and narrow walls parallel to hot face. From Ref. 34.
Thermocouple location from mold top (mm)
Fig. 20.9 Variation of copper temperatures with vertical distance from mold top in the funnel and parallel areas of a thinslab caster mold. From Ref. 41.
a funnel-shaped thin-slab caster mold during the casting of stainless steel. The two broad faces were instrumented with a total of 144 thermocouples. The specific objectives of the temperature measurements were to analyze early solidification, meniscus waviness, mold lubrication and shell sticking in the mold. Fig. 20.9 shows the copper temperature variation with vertical distance from the mold top in the funnel and parallel areas of the mold. The variation of copper temperatures in the width direction at 100 mm below the meniscus is shown in Fig. 20.10. As the figure illustrates, the temperatures on the two sides of the tundish nozzle are rather uniform. Such temperature measurements were helpful in improving mold powders and practices to achieve uniform mold temperatures and heat removal.
300
Temperature (˚C)
200
100
0
− 580
− 460
− 380
−280
−140
20
140
280
380
460
580
Thermocouple locations in horizontal plane 100 mm below the meniscus (mm)
Fig. 20.10 Variation of copper temperatures in width direction on either side of the tundish nozzle of a thin-slab caster mold. From Ref. 41.
Today, thermocouples are standard instrumentation on almost every modern caster mold. They are routinely used to predict mold sticker breakouts, which is explained in Chapter 19. Mold thermocouples are also used for process automation, such as automatic cast start-up and resumption through level sensing and confirmation. Attempts are also being made to use mold thermocouples for quality prediction and control during the casting of plate-grade slabs.26,31,32 Although the thermocouple data show good correlation between copper temperatures and longitudinal slab cracking,32 the correlation has not been strong enough for routine on-line quality prediction and as-cast slab disposition.
20.4.3 Steel Level
Next to copper temperatures, perhaps the most critical mold parameter is the level of the liquid steel pool in the mold.43–52 Maintaining a constant steel level in the mold is crucial for casting product with good surface quality, reducing the occurrence of breakouts, and process automation. The various methods that are commercially available or have been attempted for steel bath level measurement in the mold include: • • • Eddy current probe.1,43,50–52 Electromagnetic cassette.44 Radioactive source.44
Because the cassettes are located on top of the copper plate, they can be used for automatic startup; a “blip” in the sensor signal indicates the start of steel flow into the mold and thus the initiation of mold filling. As with the eddy current probe, the depth of the slag cover on top of the liquid steel has negligible effect on the measured electromagnetic signal. In the radioactive source method, a γ-ray emitter and a detector are installed in opposite mold walls. The accuracy of the method is better than ± 5 mm and is used on several casters throughout the world. However, it has the handling and disposal problems associated with radioactive materials. Another disadvantage of the radioactive source method is that the output signal is affected by the thickness of the mold powder on top of the steel pool. As-cast product quality requirements dictate mold level variation to be kept under tight control. A common measure is the standard deviation of the mold level, with a typical good value being at or below 2 mm in a 20-second window. The signal-to-distance relationship for the eddy current and electromagnetic sensors is highly nonlinear. Fig. 20.11 shows typical signal outputs of an eddy current probe and an electromagnetic sensor as a function of liquid steel depth below the top of the mold copper. When using measuring equipment that treats these sensors as providing a linear output, care must be exercised in the calibration and signal interpretation procedure. Otherwise, substantial errors and apparent nonrepeatable performance may result. The errors increase with increasing sensor nonlinearity and with departure from a targeted operating window. The effect of the mold level calibration procedure on the instantaneous level and level range variation will be described in the following example. Before the start of each casting sequence, a typical calibration procedure would adjust the electrical output of each sensor to two standard values at two fixed calibration points. Then, a straight line would be used between these two points to approximate the mold level. Fig. 20.12 shows such a calibration where the output at the two calibration points of 50 mm and 100 mm are fixed at 33.3 and 66.7% of the total signal output range. The desired range of operation is between the two calibration points. With this arrangement and the assumption that the signal response varies linearly as shown by the dotted line, it is clear that accurate representation occurs only at the two calibration points. When the actual mold level is between these two points, the signal response will be higher than that predicted by a linear relationship. Consequently, through the linear equation, the mold level will be interpreted as being lower in the mold than it actually is. The situation is reversed outside the range of the calibration points. In general, the situation may not be serious because the feature of importance for mold level control is not the actual instantaneous level, but rather the range of variation of level. For proper representation of this range, the local slope of the sensor curve at its instantaneous position is
4096
3072 Signal (counts)
2048
1024
electromagnetic sensor
eddy current sensor
0
0
25
50
75
100
125
150
Distance from mold top (mm)
Fig. 20.11 Input range of eddy current and electromagnetic sensors.
Instrumentation
100.0 idealized linear sensor 83.3 eddy current sensor electromagnetic sensor
66.7 Range (%)
50.0
33.3
16.7
0.0 0
25
50
75
100
125
150
Distance from mold top (mm)
Fig. 20.12 Calibration curves for eddy current and electromagnetic sensors.
important. Obviously, the local slope of each sensor curve is the closest to that of the assumed straight line midway between the two calibration points, and deviates as the level moves away from this central position. Therefore, when operating at a true mold level midway between the calibration points, the reported level will be greater than actual. However, the level variation about this point will have a small error because the slope of the actual level variation is very close to that of the linear curve. This discussion holds for electrical systems that make the assumption of linear signal response, and it is also true of older mold-level measuring equipment. However, the recent trend has been to use digital equipment, which facilitates the programming of a nonlinear response curve so that accurate interpretations of both level and level variation are simultaneously obtained. For level control and process automation, the measured level signal is fed into a control system that acts either on the tundish nozzle opening via a sliding gate or a stopper rod, or on the withdrawal speed. Fluctuations of the free surface of molten steel in the mold resulting from mold oscillation and recirculating flows can cause powder entrapment and thus defects at the product subsurface. A study was conducted to establish the relationship between mold oscillation and surface wave motion near the narrow wall in a slab caster mold.47 In the study, fiber optics were used to observe the meniscus through a quartz glass window mounted in a cutout in the mold top corner, as shown in Fig. 20.13. The casting conditions during the trials were: • • • • Cast speed = 0.53–1.60 m/min. Mold oscillation = ± 6 mm, 68–127 cpm, near-sinusoidal. Mold size = 250 x 920 mm. Grade = Low-carbon aluminum-killed.
Fig. 20.15 Relationships between mold displacement and meniscus displacement. From Ref. 47.
Absolute displacement of meniscus (mm)
(b) downstroke
5
castingspeed 0.53 m/min 1.60 m/min
• The magnitudes of the fluctuations are close to mold displacement and increase as the casting speed increases, as illustrated in Fig. 20.15. • The shape of the meniscus is always convex upward, but the radius of curvature of the meniscus varies with the displacement of the mold during an oscillation cycle. This is illustrated in Fig. 20.16.
time
upper dead point
narrow time wall
narrow wall
lower dead point
Fig. 20.16 Change in meniscus profile at casting speed of 0.53 m/min. From Ref. 47.
Fig. 20.17 Mold free surface fluctuations. From Ref. 48.
SEN sensor locations
wave motion
In a recent study, two suspended level sensors were used to study level fluctuations in a slab caster mold.48 As is shown in Fig. 20.17, one of the sensors was installed at 120 mm from the narrow wall. The other was located either at 50 mm away from the tundish nozzle on the same side as the first sensor, or on the other side of the tundish nozzle at 60 mm from the opposite narrow wall. Observations when casting at 1.3–1.4 m/min showed that rapid level fluctuations, which can have amplitudes of ± 20 mm, occur in small-sized slabs and that these fluctuations can cause surface defects. In larger sections, the level fluctuations have smaller amplitudes and occur at lower frequencies. Measurements also showed that the waves propagate from one side of the mold to the other in the width direction; the phenomenon is referred to as “pumping.” The effect of slab width on the principal frequency of the level fluctuations is shown in Fig. 20.18. The problem was partly solved by installing a band rejection filter on the control system. Another benefit of the dual sensor tests was the establishment of favored places for the single level measurement sensor.
1 0.9 Frequency (Hz) 0.8 0.7 0.6 0.5
Fig. 20.18 Effect of mold width on mold level fluctuation frequency. From Ref. 48.
The incidence and depth of subsurface inclusions resulting from mold level variation increase with casting speed.50 In several slab caster operations, the in-mold electromagnetic brake (EMBr) is used to suppress the mold level fluctuations when casting at high speeds.50,52–60 Plant data indicate that the EMBr decreases the surface wave fluctuations significantly and improves slab quality. However, it can adversely affect slab quality if the magnetic field strength is not properly adjusted for the various cast widths and speeds.
Instrumentation
6 4 Displacement (mm) 2 0 −2 −4 −6 0 0.4 0.8 1.2 Time (seconds) 1.6 2
Fig. 20.22 Mold vertical displacement versus time. From Ref. 73.
Figs. 20.19 through 20.23 show sample outputs of the triaxial accelerometer-based mold condition monitoring system.73 The data was collected on a straight slab caster mold during casting. The measurements showed that the actual frequency and stroke were 112.2 cpm and 8.77 mm, compared with the aim values of 112.0 cpm and 8.8 mm. Horizontal mold movements were 0.24 mm perpendicular to the broad face and 0.25 mm perpendicular to the narrow face. Phase was 0.0 degree, and rise/fall ratio was 0.977. The measured parameters in this trial were all within the acceptable ranges; thus, no corrective maintenance action was taken. Ideally, the acceleration-frequency spectrum presented in Fig. 20.20 should show only the primary oscillation frequency of 112 cpm. However, numerous secondary (extraneous) frequencies of higher values are observed, the most significant having a value of about 550 cpm. It might be possible to develop a correlation between the secondary oscillation frequencies and oscillator component repair, mold powder characteristics and product quality. However, the development of such a correlation would require designed powder trials, and long-term trending of oscillator repairs and extraneous oscillation frequencies. Casting practices, oscillator equipment, maintenance practices and quality standards vary from one installation to the next. For example, a survey of several major slab casting operations in North America and Europe revealed that the allowable mold movement in the horizontal directions varies between 0.2 and 0.5 mm to avoid corner and longitudinal cracks. Therefore, each caster operation should develop its own maintenance and quality criteria to be incorporated into an on- or off-line mold oscillation monitoring system.
0.3 0.2 Displacement (mm) 0.1 0 −0.1 −0.2 −0.3 0 0.4 0.8 1.2 1.6 2
Fig. 20.23 Mold horizontal displacement versus time. From Ref. 73.
20.4.5 Mold/Strand Friction
Interfacial friction between the strand and the mold affects surface quality and caster productivity. Depending on shell strength, if friction is not maintained below a critical level, shell sticking and tearing can occur. Also, if friction is not low enough, the desired casting speed and productivity levels might not be attained. Mold/strand friction is dependent on mold powder characteristics; steel grade; mold oscillation curve, e.g., sinusoidal vs. triangular; and oscillation parameters such as frequency, stroke, and duration of positive and negative strip times. In an effort to determine the proper combination of mold powders and practices that yield the best mold lubrication for different steel grades, several measurement and analysis methods have been applied or proposed to quantify mold friction. These include: • Strain gauges in oscillator arms.24,30,64,75 • Plurality of load cells under the mold support points.35,76,77 • Measuring the pressure variation between the inlet and outlet of the hydraulic cylinder in a hydraulic oscillator.65,67 • Measuring and analyzing the oscillator drive motor current.78,79 • Measuring the withdrawal roll current.80 • Accelerometer mounted on the side of the mold.81 Irrespective of the method used, friction is calculated from the difference between measurements made during casting and when oscillating with no casting. To illustrate the point, a recent study in which strain gauge bridges were used for slab/mold friction calculations will be cited. In this particular study,30 the strain gauges were installed in two pins connecting the mold table to the oscillator arms, as shown in Fig. 20.24. The intent was to evaluate and optimize mold powders for strip and plate grade slabs by comparing mold/slab interfacial friction and other mold parameters when casting with various mold powders. Mold/slab friction forces were obtained from pin forces and
LVDT
Instrumentation
cold 200 190 Pin force (kN) 180 170 5.0 Displacement (mm) 2.5 0 − 2.5 − 5.0 0 1 2 3 0 1 2 3
Fig. 20.25 Mold displacement and pin force under cold and casting conditions. From Ref. 30.
casting
Time (seconds)
mold displacements measured during casting and cold oscillation tests. In the cold tests, the mold was oscillated at frequencies ranging between 18 and 130 cpm as though a cast were made. Force and displacement measurements under cold and casting conditions are shown in Fig. 20.25. The force and displacement measurements were used to calculate the “work” expended by the pins to move the mold and mold table in a complete oscillation cycle from:
Pin work over a cycle = Ú (Pin force ¥ Mold displacement)
(Eq. 20.1)
Pin work over a cycle was then divided by the full mold stroke to determine a “work-averaged” pin force, and the difference between the work-averaged force during casting and cold oscillation test at the same oscillation frequency was defined as the average friction force. Friction per unit area was then determined by dividing the average friction force by the total slab surface area in the mold. That is:
Average friction force = Pin work during casting - Pin work under cold conditions Mold stroke
(Eq. 20.2)
Friction =
Average friction force Slab area in mold
(Eq. 20.3)
The method may best be understood by examining Fig. 20.26, which shows the pin work per cycle during casting and cold tests. Fig. 20.27 shows friction force per unit slab area in the mold versus casting speed for three different mold powders. As expected, friction decreases as cast speed increases. In the study, an attempt was made to calculate mold/slab friction from peak-to-peak strain gauge measurements. However, this yielded inconsistent and sometimes negative friction values. Another technique that can be used to calculate strand/mold friction from load cell signals is the application of Fast Fourier Transformation (FFT) on load cell signals measured during casting and when oscillating with no casting.24,75
1 In-gap slag thickness (mm) 0.8 0.6 0.4 0.2 0 film pool
12 Molten powder pool depth (mm) 10 8 6 4 2
Fig. 20.28 Relationship among in-gap slag thickness, molten powder pool depth and powder consumption. From Ref. 82.
0
0.2
0.4
0.6
0.8
1
Powder consumption (kg/m2)
• The accuracy is ± 0.1 mm for the film gauge and ± 2.0 mm for the molten powder depth gauge. • Changes in cast speed and powder viscosity cause variations in film thickness. • Variations in film thickness affect strand surface cracks. Sustained performance of the devices in the hostile caster environment is yet to be proven.
The plant measurements shown in Fig. 20.29 were matched with mathematically calculated deformations of the mold backplate.84 Dynamic changes of the cavity thickness at the two ends of the Burns Harbor straight mold during a seven-hour campaign are shown in Fig. 20.30. The measurements were made as part of a major undertaking to understand the causes of corner cracks on plate grade slabs. Slab dimensions were 254 x 1270 mm. As the figure shows, the cavity thickness opened by about 2.50 mm at both ends during steady-state casting. During tundish tube changes, the cavity thickness opening decreased significantly because of speed reduction and the resultant decrease in copper temperatures. After cast terminations, the walls moved back to their cold positions. The wall deformations measured in these studies were not large enough to restrain the movement of the narrow walls during in-cast width changes.
Displacement transducers fixed 0.00 mm 0.51 mm 0.96 mm 0.50 mm 0.05 mm
2.00 mm
1.91 mm
1.10 mm 0.38 mm 0.00 mm 0.38 mm 1.22 mm loose displacement transducers (a) six minutes after start-up; cast speed = 1.14 m/min displacement transducers
fixed 0.00 mm 0.36 mm 0.51 mm 0.30 mm 0.05 mm
1.27 mm
1.27 mm
0.96 mm 0.46 mm 0.25 mm 0.51 mm 1.10 mm loose displacement transducers (b) just prior to width change; cast speed = 0.6 m/min
Fig. 20.29 Measurement of in-cast mold deformation.
20.4.8 In-Mold Liquid Steel Flow Velocity
The surface and internal qualities of the cast product and the operability of the caster can be affected significantly by the flow of metal in the mold. Therefore, it is highly desirable to know the velocity distribution of liquid steel in the mold. Until recently, researchers relied on physical and mathematical models to study the flow field in the mold because instruments were not available to measure the flow velocities of liquid steel under production conditions. Although the researchers were mostly successful in correlating their model results to the quality of the cast product, they were not able to simulate critical in-cast events such as the plugging or wear of the tundish tube, and interactions between the liquid steel and the mold powder. Now, devices are available to measure liquid steel velocities in the mold during casting. Of the devices, the most notable is the electromagnetic sensor,85,86 which is schematically shown in Fig. 20.31. The device consists of a permanent magnet mounted on the water jacket, and a sensor mounted on the back of the copper plate. Neither the magnet nor the sensor contact the liquid steel and thus do not disturb the flow field at the measurement location. As the steel flows through the magnetic field in front of the device, it creates electrical currents in the liquid that can be determined from:
J = s (V ¥ B)
(Eq. 20.4)
where σ = electrical conductivity of liquid steel, → J = current density, → V = flow velocity of liquid steel, and → B = magnetic flux density. Thus, the velocity of the liquid steel can be determined by measuring the electromagnetic field of the induced currents in the vicinity of the sensor. The measurement represents the average
casting powder cooling water
MFC - sensor measured area
permanent magnet
liquid steel water jacket copper wall
Fig. 20.31 Installation of electromagnetic flow sensor on a caster mold. From Refs. 85 and 86.
137 mm
Fig. 20.32 Flow sensor locations on broadface of slab caster mold. From Refs. 85 and 86.
270 mm
electromagnetic flow sensor 2700 mm
horizontal (or vertical) component of flow velocity in a small volume of liquid steel in the vicinity of the sensor. Four such flow sensors were installed in a 200 x 2700-mm slab caster mold to correlate flow conditions in the meniscus region to slab internal quality.85,86 As Fig. 20.32 shows, the sensors were at two vertical locations on either side of the tundish nozzle. The flow velocities measured in front of the lower sensors and the casting speed during a time period of 1.5 hours are shown in Fig. 20.33. The velocities on the two sides of the tundish tube were about the same and varied similarly as the casting speed was changed. However, when all casting parameters were constant, the flow velocities on either side of the tundish tube varied randomly between about 2 and 40 cm/sec, indicating that even during steady-state casting the flow field in the mold is not steady. The flow pattern in a slab caster mold depends mostly upon the argon flow rate, tundish nozzle design, tundish nozzle submergence, casting speed and extent of clogging in the tundish nozzle. As illustrated in Fig. 20.34, the main flow patterns can be described as “single roll,” “double roll” or “meniscus roll.” Water and mathematical modeling indicated that, in the particular mold instrumented with the flow sensors, small tundish nozzle submergence would produce a “single roll” flow pattern, and large tundish nozzle submergence would produce a “double roll” flow pattern. To verify this, a trial was conducted in which the tundish nozzle submergence was varied over a 120mm range, but all other casting parameters were kept constant. The argon flow rate was 80% higher than in the water model. The flow velocities measured by all four sensors and the tundish nozzle position are shown in Fig. 20.35. It is clearly seen from the figure that, with all tundish nozzle positions, the flow was directed toward the narrow wall at all four sensor locations, indicating that the
50 right 25 Flow velocity (cm/sec) 0 −25 −50 1.6 1.2 Casting speed (m/min) 0.8 0.4 0 Time
Fig. 20.33 Variation of horizontal flow velocities measured by the lower sensor during casting. From Ref. 85.
double roll promoted by: low argon rate deep SEN immersion high casting speed
single roll promoted by: high argon rate shallow SEN immersion low casting speed
meniscus roll promoted by: clogged SEN
Fig. 20.34 Main flow patterns in slab caster mold. From Ref. 86.
flow pattern was always in the “single roll” mode. The velocities on the two sides of the tundish nozzle were about equal and showed negligible change with tundish nozzle submergence, except for the slight increase in flow velocities when the tundish nozzle was in its highest position. Further trials showed that tundish nozzle port angle had little effect on the flow pattern. Only when casting with a deeply immersed tundish nozzle and low argon flow rate was the “double roll” mode the dominant flow pattern. Cold mill data showed a correlation between coil surface defects and mold flow pattern. The coils produced with the “double roll” pattern had no defects, those
100 50 Flow velocity (cm/s) 0 -50 -100 50 25 right Flow velocity (cm/m) 0 left -25 -50 min 50 mm left lower sensors
upper sensors
right
SEN immersion depth (mm) 14 min max Time
Fig. 20.35 Variations of mold velocities with tundish nozzle submergence under steady-state casting conditions. From Ref. 85.
Fig. 20.36 Refractory probe and load cell arrangement for mold velocity measurement. From Ref. 87.
produced with the “single roll” pattern had slightly more defects than the plant average defect rate, and the coils produced with the “meniscus roll” pattern had the most defects. The “meniscus roll” type of flow pattern is promoted by the clogging of the tundish nozzle. These test results proved that the electromagnetic flow sensors are viable for on-line use to predict slab quality and to finetune mold parameters and practices. A simpler and more commonly used mold flow measurement technique involves the use of a circular ceramic probe as is schematically shown in Fig. 20.36. The circular ceramic probe is immersed in the liquid steel below the meniscus. The supporting rod at the upper end of the probe is equipped with a load cell or strain gauge bridge. Two approaches are used. In the first and more commonly used approach, the strain or load measurement is directly converted to the liquid steel velocity through the momentum imparted on the probe by the stream. Such an approach was used in several slab caster operations to see the effects of an electromagnetic brake (EMBr) on mold meniscus velocities.87–89 The measurement results from two different operations are shown in Table 20.1. In one operation, the measurement probe was kept at 155 mm from the narrow wall. In the other, it was held at 300, 250 and 180 mm from the narrow wall. Positive and negative velocity values in Table 20.1 imply flow toward and away from the narrow wall, respectively. Clearly, in both operations, the meniscus flow velocities were significantly lower when the EMBr was on compared with those when the EMBr was off. When the EMBr was off, the flow velocities varied between 4 and 110 cm/sec; and when the EMBr was on, they varied between 1 and 35 cm/sec. The second approach with the circular ceramic probe was tested in Wood’s metal and in liquid steel under laboratory conditions.90 In this approach, the linear relation between the velocity of the fluid flow approaching the ceramic probe and the shedding frequency of vortex streets generated behind the probe is used. The flow around the circular probe is governed by the Reynolds number, which is defined by:
Re = V iD u
If the Reynolds number is larger than 40, vortices are shed in very regular patterns from the cylinder, as shown in Fig. 20.37. These regular patterns are called Kármán vortex streets, and they result in oscillation of the refractory probe in a direction perpendicular to the flow direction. The oscillation frequency of the cylinder is the same as the shedding frequency of the vortex streets, which can be defined by:
f= St i V D
Casting Volume
D
Fig. 20.37 Schematic of Kármán’s vertex street. From Ref. 90.
flow
V
where f = oscillation frequency and St = Strouhal number. The Strouhal number remains nearly constant over the Reynolds number range of 300–200,000.90 Thus, once a setup is constructed and the Strouhal number is determined in off-line trials, it can be used to determine the flow velocities in the actual caster mold by measuring the probe oscillation frequency f. The concept was tested in Wood’s metal melted at 47°C in a cylindrical vessel, and in liquid steel melted at 1550°C in a cylindrical crucible using an induction furnace. The Kármán vortex probe was rotated in the melt by using the setup shown in Fig. 20.38. The speed of the probe was changed by means of a variable-speed motor on the rotation arm. The submergence of the refractory probe was 50 mm in both Wood’s metal and liquid steel. The test cylinder was made of sialon, and the supporting rod was made of stainless steel. The cross-section of the central part of the supporting rod was made rectangular to facilitate the attachment of strain gauge bridges. The erosion and deformation of the sialon test cylinder were negligible, even
30
bridge strain gauge supporting rod ∆t dynamic meter strain circular cylinder fast Fourier analyzer D furnace molten steel
50 mm
Fig. 20.38 Experimental apparatus for velocity measurement with Kármán vortex probe. From Ref. 90.
though the cylinder was submerged in the liquid steel without any preheating. The output of the strain gauge bridges was processed by means of a Fast Fourier Transformation analyzer (FFT) to determine the vortex shedding frequency f and subsequently the fluid flow velocity V. The test results with Wood’s metal and liquid steel are shown in Fig. 20.39. The Strouhal number with both liquids was 0.15. The measured velocities varied between about 20 and 70 cm/sec. The ceramic rod is simple and viable for experimental use in the laboratory or on the cast floor to fine-tune mold parameters. However, it is not practical for on-line continuous use in the caster mold to predict slab quality.
shedding frequency of Kormon vortex 100 Magnitude (mV) 7.5 Hz
0
0 Frequency (Hz)
100
supporting rod molten metal 15 ∆t Wood's metal molten steel D h 10 f− D/cms -1 h = 50 mm d = 5.6 mm Sialon cylinder k = 0.15 stainless steel material ∆t (mm) 2.0 1.5 2.0 1.5
5
0
20
40
60
80
Molten metal velocity, V/cms -1
Fig. 20.39 Velocity measurements in liquid steel and Wood’s metal. From Ref. 90.
20.4.9 Mold Cooling Water
The flow rate, pressure, and inlet and exit temperatures of mold cooling water are continuously measured and displayed in the caster control room. The difference between the water exit and inlet temperatures, and the water flow rate is used to calculate the heat removed from the mold. The measurements and the mold heat removal are then used to control the casting speed, issue alarms in the event of flow disturbances and call for emergency water in the extreme case of major coolant loss. They can also be used to detect mold problems. For example, if the measurements show low heat removal from the mold narrow side, this may indicate wrong taper and thus the need for mold maintenance. The sensors used to measure mold cooling water parameters are off-the-shelf industrial instruments and hence will not be elaborated here.
20.5 Containment
20.5.1 Introduction
After exiting the mold, the partially solidified strand enters the containment section of the caster. The containment is comprised of a large number of roll assemblies and is by far the largest and most mechanically complicated part of the casting machine. The main functions of the containment are to: • Extract the strand at a controlled speed. • Guide the strand with top and bottom rolls until the completion of solidification. • Remove heat from the strand. • Maintain a stable flow path to yield the desired strand profile and dimensions. • Perform the required mechanical functions such as the bending, squeezing and straightening of the strand. • Allow adequate production rates without yielding surface or internal defects. To assure the adequate performance of these demanding functions and to correlate the various functions to the design of the containment, numerous measurements have been conducted on the caster containment. A big challenge in these measurements has been the operational life of the sensors in the hostile containment environment. Although success has been limited compared with mold sensors, many sensors and application techniques have been developed. These developments enable the measurement and analysis of containment parameters that relate to slab quality, productivity and maintenance. The following sections summarize the instrumentation available for the measurement of critical containment parameters.
solidification occurs downstream of the unbending point. To avoid such internal defects, the secondary cooling practice and casting speed must be properly controlled to assure complete solidification before the unbending. Because of the strong correlation between product quality, productivity and secondary cooling practice, the achievement of proper spray practice for all casting conditions and grades has been one of the major goals of caster operators and researchers. The complex interactions among the oxidized and rough strand surface, rolls, sprays, surface water layers and containment environment make it difficult to mathematically or physically determine the proper spray practice for all casting conditions and grades. Hence, best results have been obtained by measuring critical parameters on actual casters and analyzing the data with the help of mathematical heat transfer models.35,91 The biggest challenge in establishing the heat transfer characteristics in the secondary cooling zone has been the accurate measurement of the strand surface temperature. Thermocouples and optical pyrometers have been used to measure strand surface temperatures. Thermocouples are either welded onto the strand surface with a welding gun or dragged and embedded onto the strand surface by feeding it between the strand and a roll upstream of the roll. Thermocouples are inexpensive and yield a continuous variation of the surface temperature of a particular strand section as it goes through the machine. This suits the approaches used in mathematical solidification models. However, thermocouples have two major disadvantages. First, they are subject to large errors because they act as a cooling fin on the strand surface, and they are affected by the water on the strand surface and the rolls that they come in contact with. Second, they are impractical for generating surface temperature data for a large number of casting practices, widths and grades. Optical pyrometers are preferable because, once compensated for the emissivity and energy absorption by water, they can be used for continuous temperature measurement at a particular spot on the caster. Several pyrometers might be needed along the metallurgical length to establish the cooling characteristics in the different spray zones. Laboratory tests have been useful in defining the effects of water flow rate, location in the spray, pyrometer offset distance, presence of steam, surface temperature, and ambient light on the accuracy of pyrometers.35 Compensation for energy absorption through water can also be accomplished in the laboratory tests. Table 20.2 summarizes the characteristics of various pyrometer types that were first tested under laboratory conditions and later used on two slab casters in a steel mill.35 In the mill, in order to eliminate spray water interference and obtain slab surface temperatures that can be compared to the model calculations, water-flows from metallurgical spray, cross sprays and fog sprays were turned off at the pyrometer location for short periods of time. The pyrometer read the surface temperature just before the strand entered the spray impact area. Fig. 20.40 compares the measured surface temperatures under actual casting conditions against those from a mathematical model before and after the spray impact area. The actual and calculated results are within 30°C. The mathematical solidification models developed from strand surface temperatures are mostly one-dimensional for practical reasons. Once developed and verified, they can be used to improve caster operation, productivity and product quality.
Table 20.2 Characteristics of Tested Pyrometers (from Ref. 35)
Minimum Measurable Temperature, °C 900 830 Effect of Water Spray on Measured Temperature Causes high readings Causes high readings Sensitive to ambient light Not workable in presence of water due to hardware problems Causes low readings Causes low readings Causes low readings
Manufacturer A B
Type 2-color 2-color
Sensor PM tube PM tube
Wave Length, µ 0.50 – 0.58 0.45 – 0.75
Special Features Flexible light guide Sights through tube
C
2-color
PbS cell
1.65 – 2.30
—
Sights through tube
D E F
1-color 1-color 1-color
Si cell Si cell PbS cell
0.4 – 1.1 0.9 1.6
700 700 330
Flexible light guide Sights through tube Flexible light guide
Instrumentation
1400 1300 1200 Surface temperature (˚C) 1100 ] 1000 900 800 700 600 500 0 5 10 15 20 25 30 35 40 before spray cast speed: 1.3 m/min. slab: 200 x 1168 mm
after spray measured calculated
Distance from meniscus (m)
Fig. 20.40 Strand surface temperature profile. From Ref. 35.
multiple points on the strand. Experimental and mathematical studies of bulging, and operating experiences with early casters, have led to the design of today’s modern slab casters equipped with split rolls of small diameter and pitch that can be operated at high casting speeds without the risk of excessive bulging and attendant internal cracking.
2388 mm bearing load cell 150 120 Bearing load (kN) 90 60 30 0 Roll length
Fig. 20.48 Measured roll bearing loads in bow. From Ref. 105.
slab
speed: 1.14 m/min. width: 1850 mm bottom roll 55 top roll 54
the load cells were installed in roll girders directly under the bearing blocks in the bow and in the unbending assemblies of the machine. The objective was to see if the forces sustained by the roll bearings during casting exceeded the dynamic load capacity of the bearings, thus contributing to the frequent bearing failures. Numerous trials were conducted with strip and plate grade slabs. Slab width varied between 960 and 1950 mm, and casting speed varied between 0.76 and 1.6 m/min. Roll loads were monitored during start-ups, cap-offs, steady-state casting and transients such as ladle changes, tundish tube changes, tundish flys and slowdowns. From the measured load cell data, the loads sustained by the bearings were then mathematically calculated. The results showed that the roll loads increased with slab width but were not significantly affected by steel grade and cast speed. As illustrated in Figs. 20.48 and 20.49, there was little or no difference between the loads sustained by the top and bottom roll bearings in the bow and in the straightener. The maximum measured bearing load in the straightener was 231 kN, which was only 43% of the rated dynamic load capacity of the bearing. In the bow, the maximum measured bearing load was 133 kN, which was 31% of the rated dynamic load capacity of the bearing. Another significant observation was that start-up and cap-off slabs, and speed transitions caused only small surges or drops in roll loads. This is verified in Fig. 20.50, which shows one load cell output from each of three test rolls in the bow and straightener over a 75-hour test period. Three start-ups, three cap-off’s, numerous width changes and speed transitions occurred during the trial period, yet no major surges or drops in load cell outputs were observed. These load cell measurements led to the conclusion that the cause of the frequent roll bearing failures was not the overloading of the bearings. Rather, further tests revealed that debris found in the roll bearings was the main cause of the bearing failures.
Instrumentation
away from strand be bp bs transient bend bic
into strand bp = roll permanent bend (manual turning of room) be = elastic bend bic = incast bend bs = stoppage bend
Fig. 20.51 Schematic illustration of roll bending. From Ref. 106.
front. Thus, long operational life for reduced cost and stability for improved product quality have provided the impetus for researchers to measure roll bending and temperature in the containment. Rolls bend because of mechanical and thermal interactions with the strand, misalignment with adjacent rolls and the condition of the roll itself. Fig. 20.51 schematically shows the types of bending that can occur on a roll.106 These are the permanent, elastic, in-cast and stoppage bends (bp, be, bic and bs). The permanent bend bp can be measured by turning the roll when cold, and it usually determines the need to change a roll. The elastic, in-cast and stoppage bends can be measured during casting by using linear displacement transducers contacting the roll surface. The transducer must be housed in a waterproof housing and attached to a water-cooled stable beam. Roll bending measurements were conducted on solid rolls in a slab caster and were found to be highly dependent on the type of roll cooling, which can be external, central-bore, peripheral or scrolled.106,107
2 roll deflection 1.5 Deflection at midspan (mm) Strand speed (m/min) strand speed 1.5
Time (seconds)
Fig. 20.53 Temperature distribution in solid roll body during speed transient.
Variation of total bending of solid rolls in the unbending zone with time and casting speed is shown in Fig. 20.52.108 Severe distortion due to nonuniform heating occurs in the shape of the roll during strand stoppages. When the strand is stopped, the nonuniformity of the temperature distribution in the roll causes it to distort and bend toward the strand. When the strand resumes motion, the distorted roll wobbles for a few revolutions until its temperature is more uniformly distributed. As Fig. 20.52 shows, at the mid-span the total roll bending was found to be about 1.5 mm during strand stoppage. The amount of roll bending that can be tolerated depends on the casting practices and
300
Maximum roll surface temperature (˚C)
250
200
150
100
50
0 0
0.25
0.5
0.75
1
1.25
1.5
1.75
Strand speed (m/min)
Fig. 20.54 Effect of speed on maximum roll surface temperature. From Ref. 108.
steel grades, and it should be determined for each caster installation. Roll bending has not been an issue on state-of-art slab casters, which are equipped with split rolls having multiple support points. Containment rolls are often replaced because of thermal cracking and spalling at the outer surface, or because of bearing failures under thermal loading. Thus, it is often desirable to measure roll body and bearing temperatures. This can be accomplished by using thermocouples. For bearing temperature measurements thermocouples can be inserted through grease lines or grooves machined into the roll axle, depending upon roll design. For body temperature monitoring special thermocouple plugs can be fabricated and installed at various locations on the roll body. Fig. 20.53 shows temperatures measured by such devices at various depths below the surface of a solid roll during steady state casting and strand stoppage in a slab caster.108 Fig. 20.54 shows the maximum roll surface temperature measured at various casting speeds in the same caster installation. As the figure shows, the roll temperature varied nonlinearly with casting speed because of the particular way the secondary water rate is controlled with casting speed. Roll temperature is determined by roll cooling, secondary cooling practices and roll design, which vary from caster to caster. Hence, roll temperatures should be measured at each caster installation or carefully extrapolated from one caster to the next.
Casting Volume
start-up 1 Misalignment (mm) 0 0.8 -1 0.4 -2 -3 0 0 1 2 Time (hours)
Fig. 20.55 Alignment change between bender and bow during casting. From Ref. 93.
tube change
tube change
tube change 1.0 Cast speed (m/min)
3
4
5
segment dynamics under thermal and mechanical loading.93,112 Hence, is often desirable to check the stability of the roll gaps and alignments during casting. Such in-cast checks enable the understanding of the causes of product defects and the control of gap tapering on casters that employ soft reduction for improved internal quality. The sensor that is commonly used to assess roll and segment stability during casting is the linear displacement transducer. Such sensors were used to measure hot-gap changes and frame movements under various casting conditions in a slab caster equipped with a straight mold, a bender and a bow.93 The intent was to determine if roll gap dynamics and roll frame movements at the upper end of the machine were the causes of crosssectional distortion and corner cracking on slabs of plate grades. Cold measurements were made before the start of cast to assure that gaps and alignments were within the machine tolerance of ± 0.5 mm. As is depicted in Fig. 20.55, the hot measurements showed that the bender and bow moved out of alignment by as much as 2.5 mm shortly after start-up. Slab inspections showed a definite correlation between the incidence of corner cracks and the magnitude of the bender-to-bow
Fig. 20.56 Linear variable displacement transducers (LVDTs) used to measure dynamic movements. From Ref. 112.
Casting Volume
2 before roll enhancement Gap opening (mm) 1.5 gap opening tolerance = ± 0.5 mm 1 after roll enhancement 0.5
0 stabilized roll
bow exit straightener inlet
Fig. 20.58 Roll gap changes at the bow-to-straightener transition before and after straightener roll enhancement From Ref. 112.
Today, on modern high-productivity slab casters, displacement transducers are routinely used to automatically change slab thickness between casting sequences.68 The objective is to avoid long outages, which would otherwise be taken to manually change roll gap settings. They are also used in conjunction with on-line solidification models to dynamically monitor and automatically change the tapering of roll gaps to assure that complete solidification occurs in the tapered section of the machine.68 This helps to improve the internal quality of slabs of critical grades.
20.5.7 Cast Speed and Strand Tracking
Cast speed is the most basic and fundamental parameter that is measured and controlled on a continuous caster. Cast speed is commonly measured with pulse encoders on drive roll motors. The measurement location varies from one caster to the next but is usually at the lower end of the machine where the strand is surely in contact with the drive roll. In some installations, a drive roll
1.0 machine enhancements and roll tapering
0.8 Open center index
0.6
Fig. 20.59 Decrease in the incidence of slab open centers. From Ref. 112.
0.4
0.2
0M A M J JA S O N D J F M A M J JA S O N D J F M A M J JA S O N 1991 1992 Month 1993
is designated for primary speed measurement, and one or more other drive rolls are used for comparison and backup measurements. Strand tracking is also carried out with pulse encoders, usually on the drive rolls that are designated for speed measurement and control.
20.6 Slab Processing Area (Runout)
20.6.1 Introduction
The caster runout begins at the end of the containment (or the last horizontal strand guide) and is equipped with bottom rolls only. In the runout, the completely solidified section is cut into ordered lengths, weighed, stamped with an identification number, inspected, stacked into piles and shipped for rolling. Other functions such as sample cutting for metallurgical testing (macro etching or sulfur printing), crop cutting and tail cutting are also performed in the runout. One of the major challenges for caster runout operators is to provide correct information on weight and dimensions for each cut section to enable direct application of product to intended orders. Sections with wrong weights or dimensions are inventoried for later applications, which is undesirable, particularly in mills that utilize direct or hot charging. Weight is routinely and reliably measured with scales before the cut sections are piled at the end of the runout. However, providing accurate information on dimensions is complex because of strand thermal shrinkage and creep under ferrostatic pressure below the mold. In most of the caster installations, dimensional information is provided through physical measurements on the hot sections in the runout. In others, dimensional information is provided through simple algorithms. Another big challenge for caster operators is to provide accurate quality information for each section. Quality information is provided mostly by computerized on-line models, which make quality predictions by monitoring and comparing actual caster parameters against those in lookup tables. Over the last 20 years, several devices have been developed for direct measurement of quality on hot sections in the runout. The following sections summarize the various devices that are available for measuring dimensions and quality on hot sections in the runout.
width in the runout. These are predictive algorithms, manual measurements with handheld calipers at the pilers, on-line physical measurements with contact devices on or around the torch machine, line-scan cameras, and lasers. Predictive algorithms are either experience-based or statistically developed. Both involve grade, secondary water practice and speed (or residence time in containment).115,116 Manual measurements with handheld calipers are conducted intermittently at the pilers. When the difference between the measured and ordered widths exceeds an allowable limit, the difference is keyboard-entered as a correction and remains in effect on all subsequent sections until the next correction. On-line contact devices are either built in-house or supplied by torch machine builders. In either case, a mechanical device from either side of the slab moves in and contacts the narrow side of the slab. Slab width is calculated from distance traveled from home position by each device. Accuracy can be within ± 2.5 mm and is highly dependent on buildup or wear on the contact points, the shape of the slab narrow sides and the condition of the encoder. Lasers are gaining popularity because they are accurate and require low maintenance.115 Predictive algorithms, manual measurements with handheld calipers and on-line contact devices provide single-point width information on each slab and are not conducive for identifying tapered slabs. Lasers and line-scan cameras provide continuous width measurement, thus enabling caster operators to identify tapered slabs. The major disadvantage of the camera systems is that they require light banks under the slab to highlight the dark (cold) edges of the slab. This can render camera systems maintenance intensive and impractical for use on casters.
20.6.3 Hot Surface Quality
In most modern slab caster installations, a computerized quality control system is used to assign a quality rating for each slab. The rating is based upon the comparison of actual process parameters, such as mold level and casting speed, which are monitored during casting against standards in predefined disposition tables. The concept is good and works adequately. However, it has several
=ak control equipment -A11 -A61 overall control system main control panel -A12 printer HC = CC continuous caster master control system = AW cooling water supply
= ME mearuring equipment
= DE descalling equipment
= CM crack marking
local control panel
= MC manipulator control system
Fig. 20.60 Eddy current-based surface inspection system. From Ref. 118.
drawbacks. First, it is as good as the information in the disposition tables, which are generally configured through statistical analysis of casting and quality data. The tables can be tuned for improved accuracy or new quality specifications, but this requires production trials that are costly and time-consuming, and it involves lag time in the actual implementation of practice changes. Second, the disposition tables are conservatively designed for improved quality assurance, which can lead to the rejection of acceptable product. Third, at times, defective slabs can flow through the system unnoticed. In an attempt to overcome these drawbacks and to achieve 100% quality assurance in an economical manner, several devices have been developed for on-line quality inspection of hot slabs in the runout. Based on the principle utilized, the devices can be broken into three predominant groups: electromagnetic, optical and ultrasonic.
6
Output signal
4
2
0
1.0
1.6
2.0
2.5
Crack depth (mm)
Fig. 20.61 Eddy current strength versus crack depth. From Ref. 188.
Fig. 20.62 Optical surface inspection system. From Ref. 126.
transfer table
CCD
receiving probes are held close to either side of the slab broad faces. The transmitting probes transmit an ultrasonic wave through the thickness of the slab, which interacts with a stationary magnetic field in the receiver and generates an induced current. The presence of internal defects is indicated by the attenuation of the induced current. Internal defects are associated with passline irregularities, bulging, machine operation and secondary water practices. Generally, they occur in plate-grade slabs, which constitute a small fraction of continuous casting, and are not caused by the more frequent mold perturbations. Hence, while being important, they are of a lesser concern in the industry than surface defects. As a result, ultrasonic and electromagnetic/ultrasonic devices for internal defect detection have received less attention than eddy current devices for surface defect detection.
20.7 Acknowledgments
The author expresses his gratitude to Dr. Donovan N. Rego of Bethlehem Steel’s Homer Research Laboratories for his numerous contributions. The help provided by Robert H. Klotz and Nancy A. Reszek is also greatly appreciated.
104. M. Izutani et al., Proc. of the 4th International Conf. on Continuous Casting, Brussels, Belgium (1988), 115–127. 105. M.R. Ozgu and J.L. Giazzon, Proc of 3rd European Conf. on Continuous Casting (1998), 75–84. 106. A. Perkins et al., Proc. of 2nd Continuous Casting Conf., London (1985), 67.1–67.11. 107. F. Nazzi, Steel Times Intl. (1988), 30–31. 108. K.M. Markarian et al., 74th ISS Steelmaking Conf. Proc. (1991), 585–598. 109. R.L. Hill, Proc of Conf. on Casting, Forging and Fabrication of Steel, Bangkok, Thailand (1987), 27/1–27/20. 110. G. Hyde and W.D.N. Pritchard, Condition Monitoring and Diagnostic Technology, 1:2 (1990), 50–55. 111. T. Soejima et al., McMaster Symposium on Iron and Steelmaking, 13, 226–245. 112. M.R. Ozgu et al., 77th ISS Steelmaking Conf. Proc. (1994), 297–304. 113. A. Diener and A. Drastik, Arch. Eisenhuttenwes, 53:1 (1982): 13–20. 114. K. Tsuboi et al., NKK Technical Report, Overseas, 36 (1982): 90–105. 115. B. Kocatulum et al., 80th ISS Steelmaking Conf. Proc. (1997), 209–213. 116. R. Mostert and J.P. Brockhoff, Iron and Steelmaker, 23:11 (1996): 35–41. 117. Anon., American Metal Market, 94 (1986): 12–16. 118. J-A. Holmstrom and H. Pettersson, Scanheating II Conf. Proc., Lulea, Sweden (1988), 135–153. 119. K. Berner, Metall. Plant and Technology Intl., 1 (1988): 76–85. 120. H. Smit et al,. 12th World Conference on Nondestructive Testing, Amsterdam (Amsterdam: Elseview Science Publishers, 1989), 896–902. 121. G. Backstrom and K.G. Bergstrand, Scanconditioning: 1st International Conference on Surface Conditioning and Detection of Surface Defects (1984), 23:1–23:13 . 122. T. Speninski and K.G. Bergstrand, Scanheating II Conf. Proc., Lulea, Sweden (1988), 124–125. 123. T. Hiroshima et al, Tetsu-to-Hagane 70 (1984): 1202–1209. 124. Y. Codur, Steel Technology Intl. (London: Sterling Publications Intl. Ltd., 1988), 250–252. 125. D. Ferriere et al., IRSID Report (1988), MCA-SG RE 88309. 126. Y. Nakai et al., Scanconditioning: 1st International Conference on Surface Conditioning and Detection of Surface Defects (1984). 127. M. Iwasaki et al, Tetsu-to-Hagane, 70 (1984): 1994- 1201. 128. M. Kitamura et al, Iron and Steel Engineer (1985), 55–60. 129. J.P. Birat et al, McMaster Symposium on Iron and Steelmaking, 15 (1985): 8–37. 130. K. Iwai and T. Ouchi, Tetsu-to-Hagane 70 (1984): 1181–1187. 131. K. Tsouboi et al, Rev. de Metall. (1982), 641–647. 132. M. Uesugi et al, Tetsu-to-Hagane 70 (1984): 1188–1193. 133. C. Thoma, Stahl u. Eisen 103 (1983): 217–223. 134. K. Iwai et al, 10th World Conference on Nondestructive Testing, Moscow (1982), 209–216.