20
Content
Chapter 20
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
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
1
Casting Volume
• 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.
20.2 Ladle
In a continuous casting operation, the ladle is used to transfer liquid steel from the steelmaking shop to the caster with aim chemistry, temperature, cleanliness and slag cover. At the caster, steel from the ladle is teemed into the tundish through a nozzle in the bottom of the ladle. It is desirable to monitor three parameters on the ladle as a function of time to properly control the teeming operation. The first is the weight of steel in the ladle, which is routinely measured with load cells on the turret arms or the ladle car. The second is the temperature of steel flowing into the tundish. Direct measurement of steel temperature in the ladle with instrumentation on the ladle itself is not practical, because the ladle is a mobile unit that moves through harsh environments and thus is not conducive to temperature instrumentation. Instead, ladle temperatures are inferred from tundish temperatures, which are continuously or intermittently measured. The third is the slag carryover from the ladle to the tundish. As the end of the ladle drainage is approached, vortex formation above the ladle nozzle and slag carryover into the tundish can occur. The mixing of slag with steel can contaminate the steel in the tundish and cause an increase in the inclusion content of the cast product. It is possible to minimize the transfer of slag from the ladle to the tundish using one of the following slag detection methods: • Visual observation. • Tare weight monitoring. • Rate of teeming change.4,5 • Opto-electronic monitoring of stream surface.6 • Vibration analysis of the ladle shroud.7,8 • Electromagnetic methods.9–12 The ability to visually detect slag carryover into the tundish is highly dependent on operator experience. This can lead to the carryover of varying amounts of slag before detection and corrective action. Tare weight monitoring relies on correct ladle, ladle cover and slag weight estimates and is thus inherently inaccurate. The rate of teeming change monitors the variation in the rate of ladle weight change. A change from metal teeming to slag and metal teeming can be indicated by a rapid decrease in the teeming rate. The method relies on the accuracy of the load cells and can thus lead to errors at low teeming rates. It can also result in false alarms if the teeming rate is affected by the movements of the ladle flow control device or by the ladle. The opto-electronic method cannot observe the core of the stream through which large amounts of slag can be carried over. The vibration analysis method is based on vibration monitoring of the ladle shroud, using an accelerometer attached to the shroud manipulator arm. As is shown in Fig. 20.2, the amplitude of processed signal from the accelerometer increases with the onset of entrained slag flow through the shroud. A
2
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
slag visually detected by operator
Amplitude index
1.0
slag detected with accelerometer
0.5
0
25
50
75
100
125
150
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
20.3 Tundish
The tundish is an intermediate vessel between the ladle and the mold and is used mainly to perform the following functions: • Deliver liquid steel to the mold(s) at a controlled rate. • Maintain a steady supply of liquid steel to the mold(s) during a ladle change. • Remove nonmetallic inclusions from liquid steel before delivery to the mold(s). • Facilitate the control of steel superheat by means of plasma preheaters.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
3
Casting Volume
100 80 60 40 20 0 electromagnetic slag signal (%) Time
Fig. 20.3 Ladle slag carryover, weight and slidegate signals. From Ref. 10.
2 sec ladle slidegate position (%)
ladle weight (tons)
alarm
In order to properly perform the above functions and control the caster operation, it is essential to know the temperature and depth of liquid steel in the tundish. Temperature can be measured in two ways. The first is the manual immersion of disposable thermocouples. This method is reliable and is used in many caster installations. Its main drawback is that, if not exercised at sufficient frequency, rapid temperature changes can go unnoticed. Unexpected increases in steel temperature can result in breakouts, and rapid temperature losses can lead to freeze-off and cast terminations. Because of the push toward stable caster operation and reduced labor costs, continuous temperature measurement systems have been developed and are in use at several caster installations.12,15–17 The sensor used for continuous temperature measurement is comprised of a Pt-Pt/Rh thermocouple embedded in a refractory tube, and it is immersed into the steel bath through an opening in the tundish cover. Fig. 20.4 shows continuous temperature measurement from such a sensor, along with tundish weight and cast speed or throughput.16 The effect of tundish level (or weight) on tundish temperature profile can be clearly seen by examining the difference in profiles during tundish tube changes E1 and E2. When the tundish weight drops to 41,000 kg, about 25% of the tundish wall hot surface is exposed to the atmosphere. Consequently, significant heat loss and thus temperature drop occur from the tundish walls, which must be recovered upon refill. The extent of the temperature loss depends on the flow rate from the tundish, the length of time during which the tundish walls are exposed, and the wall area that is exposed to the atmosphere. Such temperature losses during ladle, mold width and tundish tube changes can be reduced through the continuous measurement of tundish temperature. Other potential benefits of continuous tundish temperature measurement are caster automation and accurate correlation of steel temperature to as-cast product quality. The depth of liquid steel in the tundish is either inferred from weight measurements with load cells on the tundish car or directly measured with electromagnetic sensors installed in the tundish bottom or the vertical walls. Knowledge of liquid steel depth from these measurements is used to: • Safely fill the tundish during the start of a cast. • Maintain a stable tundish level during steady-state casting. • Assure a minimum tundish level during ladle changes to avoid downgrades. • Safely drain the tundish to a very low level during grade changes, tundish changes (tundish flys) and cast terminations, without allowing any slag to flow into the mold.
4
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
55,000 Tundish weight (kg) (a) 1566 Temperatue (˚C) (a) (a) 1552 Temperature 1538 (e1) (e2)
Fig. 20.4 Tundish temperature, weight and cast speed variations. From Ref. 16.
(a) (h)
(a) 41,000
weight (h)
27,000
Cast speed (m/min)
1.18 (g) 0.64 Cast speed 2810 kg/min (6.2 klds/min) (f)
0 0 2 4 Time (hours) 6 8
(a) - ladle change (e) - tundish tube change (f ) - mold width change
(g) - machine slowdown for late ladle connection (h) - reduced tundish level prior to tundish tube change
Most of the caster operations use weight monitoring with load cells for bath height control because load cells are easy to install and maintain on the tundish car. Furthermore, load cells stay on the tundish car and can be used on the incoming tundish. The major disadvantage of weight monitoring is that it relies on correct estimates of tundish, tundish cover and slag weights, and is thus inherently inaccurate. The inaccuracies are particularly worrisome during draindowns, because they can result in slag drainage into the mold or excessive residual steel in the tundish. Slag in the mold can cause breakouts. An excessive amount of residual steel in the tundish compromises yield during a tundish fly or increases the length of the mixed-grade zone during a grade change. For improved draindown control, some caster operations use a bobber or a ceramic ball in the tundish in conjunction with the weight measurement system. The bobber or ceramic ball is positioned over the tundish nozzle. The bobber is used to indicate bath level. The ceramic ball is used to prevent vortexing and slag carryover into the mold during the draining of the tundish.18 Some caster operations use electromagnetic sensors in the bottom of the tundish for direct measurement of steel level.19,20 The sensor assembly consists of a primary and a secondary coil that are concentrically arranged in stainless steel housing. As Fig. 20.5 shows, the housing is installed between the refractory lining and the tundish bottom shell. An alternating current applied to the primary coil generates an electromagnetic field in the liquid steel. The electromagnetic field, in turn, generates eddy currents that attenuate the primary field in proportion to the steel level. By this means, the height of the slag/metal interface above the refractory bottom in the vicinity of the sensor can be measured very accurately in the range of 0–200 mm. The thickness of the refractory lining is compensated for. The electromagnetic sensor, in conjunction with an optimized tundish bottom design and an automatic drain control system, enables the operators to consistently drain the tundish to very low levels without allowing slag to flow into the mold. Other significant benefits of
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
5
Casting Volume
1. safety lining 2. spray lining 3. impact pad 4. electromagnetic level sensor
1 2 3 4
SEN
SEN
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.
6
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.4.2 Copper Temperatures
From the early days of continuous casting, numerous researchers have measured the copper temperatures on caster molds.23–41 The objectives of the measurements varied, but in general they were to: • Evaluate mold powders and practices. • Analyze mold heat transfer and strand solidification. • Evaluate mold design. • Predict mold/strand sticking. • Measure liquid steel pool level. In all cases, thermocouples were used for copper temperature measurements, because thermocouples are inexpensive and easy to install, and their signals are easy to interpret. The thermal behavior of all four walls between the top and bottom of a mold can be analyzed relatively easily by lacing the copper plates with thermocouples. There are two methods of installing thermocouples in mold coppers. In the first and most commonly used method, the thermocouples are introduced from the backside of the copper plates parallel to the path of heat flow (perpendicular to the hot face). The advantage of this method is the ease of thermocouple installation. The disadvantage is that, because the thermocouple holes are parallel to the path of heat flow, they affect the flow of heat and thus can result in erroneous temperature measurements. This method is suitable for routine relative temperature monitoring whereby the measurements are used for sticker breakout predictions, mold level measurement and other automation purposes. In the second method, the thermocouples are introduced from the top or bottom of the copper plates and perpendicular to the
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
7
Casting Volume
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.
8
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
400 low carbon strip speed: 1.27 − 1.62 m/min. width: 990.6 − 1435.1 mm 300 narrow wall
Mold wall hot face temperature (˚C)
200 loose wide wall
1000 300 medium carbon plate speed: 1.02 − 1.29 m/min. width: 1930.4 − 1955.8 mm
Fig. 20.8 Variation of mold hot-face temperatures with vertical distance from mold top. From Ref. 34.
200
loose wide wall
100 0 300
}
various trials 600 900
Distance from mold top (mm)
Fig. 20.8 shows the variations of hot-face temperatures with vertical distance below the mold top resultant from the two-thermocouple method shown in Fig. 20.7.34 The hot-face temperatures were calculated from wall temperature measurements made along the mid-plane of the loose and narrow walls when casting low-carbon strip and medium-carbon plate grades. As expected, the temperatures on both the wide and narrow walls decreased with increasing distance below the mold top. This is caused by three factors. First, as the distance below the meniscus increases the thickness of the solid shell and its resistance to heat transfer from the liquid core to the mold increases. Second, solid fraction of the slag layer between the slab and the mold increases with distance below the mold top, thus increasing the resistance to heat flow from the slab to the mold. Third, the fluid flow activity in the liquid core decreases with the vertical distance which, in turn, decreases the heat transfer from the liquid core to the solid shell. Because the liquid steel streams from the bifurcated tundish nozzle are directed toward the narrow walls, they affect higher heat transfer on and through the narrow side shell. As a result, copper temperatures on the narrow walls were higher than those measured on the wide walls. From Fig. 20.8 it can also be observed that, when casting low-carbon strip grade slabs, the mold hot-face temperatures exceeded 350°C, beyond which it is suggested that there is an increased potential for sticking.42 Thermocouples are also being used on medium-thickness and thin-slab caster molds to improve mold design, select mold powders, develop casting practices and control the operation of the caster.40,41 Figs. 20.9 and 20.10 show sample temperatures measured on the broad face coppers of
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
9
Casting Volume
300 funnel area
200 Temperature (˚C)
parallel area 100
0
0
100
200
300
400
500
600
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.
10
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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
• Thermocouples in mold copper.1,44,46 • Optical devices.1 • Link-arm type magnetic flux meter.44 • Telescopic magnetic flux meter.44 The eddy current probe, the electromagnetic cassette and the radioactive source are the most commonly used methods. The thermocouple, optical and flux meter methods have limited or no industrial acceptance. The eddy current probe is suspended over the mold and generates an electromagnetic field that is directed into the mold. The electromagnetic field induces eddy currents in a near-surface layer of the liquid metal in the mold via the primary coil. This eddy current produces a secondary field, which, in turn, induces voltages in the sensor’s secondary coil. The magnitude of the eddy current, and thus the induced voltage, is dependent on the actual distance between the sensor and the steel surface. The actual liquid steel level in the mold can be detected independently of any slag and powder layer on top of the liquid steel. Measurement accuracy with the suspended probe has been reported to be better than ± 2 mm at casting rates up to 7 tons/minute.51 The measurements are not affected by copper temperature and copper coating such as nickel. However, it is not suited for auto-start detection because the probe is brought over the mold after the initial filling of the mold is over and the mold level stabilizes. The electromagnetic cassette entails a transmitter and a receiver coil, which are mounted side by side on top of the fixed wide mold wall. This method differs from the suspended sensor method in that the transmitter’s electromagnetic field induces eddy currents in the exposed face of the copper wall. These eddy currents generate their own magnetic fields, which are detected by the receiver coil. Because the exposed surface area of the copper wall changes as the level of liquid steel in the mold changes, the strength of the signal at the receiver coil is dependent on the level of liquid steel in the mold. The sensor signal is also dependent on the temperature of the copper wall. The temperature effect becomes pronounced on molds that are coated with nickel at the top end. Although the coils in the cassettes can be oriented to offset the effect of nickel coating, experience shows that the nickel can cause a gradual drift in the mold level, which needs to be periodically readjusted.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
11
Casting Volume
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.
12
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
The trials showed that: • Contrary to the generally accepted belief, the meniscus does not slide freely along the mold wall and is not stationary when viewed from a stationary position. Rather, it fluctuates in phase with mold oscillation, as depicted in Fig. 20.14.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
13
Casting Volume
SEN displacemant transducer mold
quartz window optical fiber
data scanner
recorder
monitor
Fig. 20.13 Apparatus used for meniscus observation. From Ref. 47.
Mold displacement (mm)
casting speed: 1.60 m/min oscilation freq: 2.11 Hz oscilation amplitude: + 6 mm − 6
0
-6
10 mm from narrow wall Absolute displacement of meniscus 5 mm
3 mm from narrow wall
0
1.0 Time (seconds)
2.0
Fig. 20.14 Meniscus behavior. From Ref. 47.
14
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
(a) upstroke 5 casting speed 0.53 m/min 1.60 m/min
Absolute displacement of meniscus (mm)
0 −6
−3
0
3
6
displacement of mold (mm) 0 6 3 0 −3 −6
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.
(a) downstroke 5 mm
(b) upstroke
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
15
Casting Volume
50 mm 120 mm 60mm 120 mm
2
1
2
1
level sensor
level sensor
SEN
level sensor
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.
700
900
1100
1300
1500
1700
1900
Mold width (mm)
16
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.4.4 Mold Oscillation
Mold oscillation was introduced into continuous casting in 1949 to prevent shell sticking to the mold by providing sufficient lubrication between the strand and the mold.61,62 Today, mold oscillation is common practice on all conventional continuous casters. Numerous studies have shown that the surface quality of the product is highly dependent on the shape of the oscillation curve (e.g., sinusoidal, triangular), stroke and frequency.63 Until recently, conventional casters utilized mechanical oscillators and sinusoidal oscillation curves. However, most new and rebuilt casters are utilizing hydraulic oscillators because the profile, frequency and stroke of the oscillation curve can be easily changed and customized for each grade.59,64–68 The stroke and frequency of oscillation can vary from one installation to the next, and the oscillation practice can be different for various steel grades at a given caster installation. However, once the optimum oscillation practices are established, it is desirable that they remain constant. This requires that the quality of oscillation and the physical condition of the oscillator be checked and corrected periodically. Several methods are used to check the condition of the oscillator and the quality of mold oscillation. These include the manual pencil trace, human touch, periodic trials with displacement transducers and accelerometers,30,69,70 and on-line continuous monitoring with accelerometers.71–73 Displacement transducers have moving parts and are thus prone to failure under the harsh caster environment. On the other hand, accelerometers do not have moving parts. Furthermore, they have high sensitivity and accuracy and are thus the best suited for oscillation monitoring. A system comprised of four triaxial accelerometers mounted at each mold or mold table corner, a data acquisition and analysis subsystem, and a personal computer can yield all the parameters that are essential for complete assessment of the quality of oscillation and the condition of the oscillator.73 Each triaxial accelerometer sensor consists of three individual accelerometers oriented in the vertical and two horizontal directions perpendicular to the mold narrow and broad faces. The acquired acceleration signal is integrated to obtain time-based velocity and displacement data. The Fast Fourier Transformation (FFT) technique is utilized to convert the timebased mold acceleration data and the calculated mold velocity and displacement data into discrete frequency component.74 The procedure is repeated for acceleration signals collected in the three directions from each triaxial accelerometer sensor on the four mold or mold table corners. It has been demonstrated that the system can produce the following oscillation parameters: • Displacement, velocity and acceleration curves in three directions. • Peak-to-peak displacement, velocity and acceleration values in three directions. • Primary oscillation frequency. • Secondary (extraneous) oscillation frequencies in three directions. • Negative strip time. • Negative strip ratio. • Rise/fall ratio (i.e., peak-to-peak time up/peak-to-peak time down). • Phase (lead or lag of one mold table corner relative to the other in the vertical direction expressed in degrees).
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
17
Casting Volume
1200 Acceleration (min/sec2) 800 400 0 −400 −800 −1200 0 0.4 0.8 1.2 Time (seconds) 1.6 2
Fig. 20.19 Mold vertical acceleration versus time. From Ref. 73.
1500 Peak-to-peak acceleration (mm/sec2)
1000
500
0
0
400
800
1200
1600
2000
Fequency (cpm)
Fig. 20.20 Mold vertical acceleration versus frequency. From Ref. 73.
80 60 Velocity (mm/sec) 40 20 0 −20 − 40 − 60 − 80 0 0.4 0.8 1.2 1.6 2
Time (seconds)
Fig. 20.21 Mold vertical velocity versus time. From Ref. 73.
18
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
⊥ to broad face ⊥ to narrow face
Time (seconds)
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
19
Casting Volume
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
eccentric pin
instrumented pin mold table pin strain gauge bridges
thermocouple
Fig. 20.24 Schematic of strain gauge and displacement transducer installation on mold table oscillator. From Ref. 30.
20
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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
20.4.6 Mold Powder Film Thickness and Molten Pool Depth
Mold heat removal is strongly affected by the thickness of the powder film in the gap between the mold and the strand. Insufficient or uneven flow of mold powder into the strand/mold interfacial gap can cause surface cracks on the cast product. On the other hand, an excessively thick film of powder can result in low heat removal rates and thus low caster productivity. One of the major
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
21
Casting Volume
200 A1 190 A2 casting work cold work pin force (kN) 180
170
Fig. 20.26 Pin work under casting and cold conditions. From Ref. 30.
160
-5
-2.5 0 2.5 Displacement (mm) A2 − A1 [mold stroke] x [slab area in mold]
5
Friction (kN/m2) =
16
12
Mold/slab friction (kN/m2)
8
4 low carbon mold powders 0 0.7 0.9 1.1 1.3 1.5 1.7
Cast speed (m/min)
Fig. 20.27 Mold/slab friction versus cast speed. From Ref. 30.
factors that control the flow rate and uniformity of the film is the depth of the molten powder cover above the steel pool. It is thus desirable to measure the depth of the molten powder cover and correlate it to the thickness of the in-gap powder film. The depth of the molten powder cover is usually measured by wire burnoff. The thickness of the powder film is measured by recovering layers of powder from the mold hot face during major transitions when the mold level decreases or from the mold exit. However, these methods are manual and inherently inaccurate. An on-line measuring system comprised of an in-gap powder thickness gauge and a molten powder pool depth gauge was developed and applied on a slab caster.82 The film thickness gauge is thermal radiation-based and is installed below the mold. The molten powder cover gauge is an eddy current device, which is suspended above the mold and utilizes the difference in the electrical resistance of liquid steel and molten slag. Both gauges are of noncontacting type. Measurements made on a slab caster mold are shown in Fig. 20.28. The results showed that:
22
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.4.7 Mold Wall Deformation
The continuous caster mold is known to sustain large thermal stresses and undergo sizable deformation during casting.83,84 The deformation of the mold is critical because it distorts the shape of the strand and can thus impact surface quality. Furthermore, in slab casters, a gap is needed between the broad and narrow faces to facilitate width changes while casting. Excessive deformation of the wide walls, in combination with the thermal expansion of the narrow walls, can result in the damaging of the wide wall hot faces and the edges of the narrow walls. The deformation of the wide walls can sometimes be severe enough to restrain the movement of the narrow walls and thus prevent size changes while casting. It is thus desirable to know the extent of mold wall deformation under various casting conditions. Displacement transducers are commonly used to measure mold wall deformation during casting. Fig. 20.29 shows the installation of linear displacement transducers on a straight slab caster mold at Bethlehem Steel’s Burns Harbor plant to study wide wall and mold cavity deformation during the casting of 254 x 1933-mm slabs. Five transducers were installed on the backside of each wide side water jacket at the lower end of the mold, and two in the mold cavity at the mold ends. Measurements were made while casting at a steady-state speed of 1.14 m/min and just prior to automatic size change while casting at 0.6 m/min. The results are shown in Fig. 20.29 and indicate that: • The fixed and loose sides deformed differently during steady-state casting and just prior to size change; the deformed shapes were convex into the mold cavity. • The middle of the fixed side moved into the cavity. The middle of the loose side either stayed in the original position or moved back from the original position. • During steady-state casting, the mold cavity thickness decreased by 0.96 mm in the middle of the mold and increased by about 2.00 mm at the ends of the mold. During size changes, because of lower casting speed and lower copper temperatures, the walls sustained smaller deformation.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
23
Casting Volume
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.
3.5 3.0 2.5 Opening (mm) 2.0 1 1.5 1.0 0.5 0 0 1 2 3 capcast speed off 4 5 6 7 1.00 0.50 0 1.50 in-cast width change 2 1 2 Cast speed (m/min.)
Fig. 20.30 Variation of mold gap opening during a sevenhour cast.
Time (hours)
24
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
25
Casting Volume
720 mm 720 mm
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.
left
9 min
26
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
27
Casting Volume
load cell
SEN mold refractory rod
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
(Eq. 20.5)
where V = mean velocity of liquid approaching the cylinder, D = diameter of cylinder, and ν = kinematic viscosity of fluid.
28
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
Table 20.1 Typical Mold Velocity Measurements (from Ref. 87)
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
(Eq. 20.6)
29
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
31
Casting Volume
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.
20.5.2 Strand Surface Temperature
It is known that secondary cooling practices in the containment have significant effects on product surface, internal quality and also productivity. As the strand surface cools and reheats between rows of spray nozzles and rolls, thermal stresses are generated in the solid shell. If the stresses are high enough, cracks can form or existing cracks can worsen on the strand surface. At the same time, mechanical stresses arise in the shell due to inter-roll bulging under ferrostatic pressure. The resistance of the shell to ferrostatic pressure and bulging is determined by its temperature and hence by the spray-cooling practice. If the bulging stresses exceed a critical value, internal (refill) cracks can occur at the solidification front. Another potential problem is the occurrence of surface cracks during the unbending of the strand because of an improper combination of surface temperature and unbending stresses. Also, in single-point unbending machines, segregation and centerline laminations can originate in medium- and high-carbon grade slabs if complete
32
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.5.3 Strand Shape and Bulging
Devices have been developed and used to measure strand profile to understand and eliminate the causes of: • Excessive bulging or concavity on the narrow side just below the mold, which might cause breakouts, excessive wear on mold narrow side coppers, and surface cracks. • Deformed strand cross-section (e.g., trapezoidal), which might cause corner cracks.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
33
Casting Volume
34
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
• Bulging between containment rolls that might cause internal cracks at the solidification front. Because measurements are made in the spray chamber, where the environment is not conducive to instrumentation, most shape sensors have been custom-designed physical contact devices involving simple instrumentation. For example, to understand early solidification profile and concavity just below the mold foot rolls on the narrow side, a shape detector involving two contact rolls and a displacement transducer was built and used on a high-speed slab caster.92 Fig. 20.41 shows the shape detector and the measurement results. The mold taper was changed to make shell growth more uniform and reduce the narrow side depression to be within 2–4 mm. When trapezoidal cross-section and corner cracks became a problem on plate grade slabs on a caster equipped with a straight mold, a probe was built and used at the bender exit, as shown in Fig. 20.42, to assess slab shape coming out of the bender.93 The probe was equipped with two curved contacts that were pushed against the slab narrow side and two displacement transducers. When the probe measurements were compared with mathematical calculations, the results showed that the slabs were being excessively misshaped before exiting the bender. As one of the measures to prevent slab misshaping and corner cracks, the bender frames were strengthened and the bender rebuilding procedure was improved. Inter-roll bulging and attendant internal cracking at the solidification front has been a major concern for caster builders, operators and researchers. Sensors have been used to measure and correlate bulging to machine design and casting practices. These sensors include custom-designed physical contact devices,94,95 and more recently a laser bulge-meter.96 Figs. 20.43 and 20.44 show a contact bulge detector and a laser bulge-meter, respectively. Custom-designed bulge devices have the disadvantage that they have short operational life and generally can be installed in specific caster locations. On the other hand, noncontacting laser-based bulge-meters have long operational life and can be moved from one location to the next for continuous measurements at
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
35
Casting Volume
contact roll
measuring roll slab air cylinder displacement transducer
(de)
(dc)
molten steel solidified shell
depression (dd)
Ratio of retarded shell growth (de/dc)
1.0 X 0.8 0.6 0.4 0.2 0 -4.0 X X X X X cast speed 1.0 m/min. 1.2 X 1.4 1.6 20
0
4.0
8.0
12.0
Depression of narrow face (dd) (mm)
Fig. 20.41 Measurement of narrow face depression and retared shell growth. From Ref. 92.
36
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
curved contacts
bender rolls
Fig. 20.42 Schematic of narrow side shape probe. From Ref. 93.
LVDT
pusher rod slab
slab V ∅ 150 ∅ 80
split roll
linear transducer for roll deflection
air cylinder
linear transducer for bulging
motor for rotation of the device
drive motor
180˚
Fig. 20.43 Mechanical contact type bulge detector. From Ref. 94.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
37
Casting Volume
PC
laser blackbox
motor driver
water-cooled aluminum case bulgemeter mounting frame
optocater gauge probe
stepper motor
gearbox
support roll
segment frame strand surface
Fig. 20.44 Laser bulge-meter. From Ref. 96.
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.
20.5.4 Final Solidification Point (Liquid Core) and Segment/Roll Loads
It is often desirable to know the complete solidification point (or length of liquid core) under different casting conditions for quality and productivity considerations. For example, in casters where soft reduction is used to avoid the occurrence of centerline segregation, solidification must be controlled in the tapered zone of the machine. In machines equipped with single-point unbending, complete solidification must occur before the unbending point to avoid segregation and centerline laminations in medium- and high-carbon grade slabs. It is essential that solidification is completed within the containment zone under all casting conditions. Strain gauges97–100 and electromagnetic transducers101,102 have been successfully used to measure the end of the liquid core under different casting conditions. The results were then used to develop casting practices that would assure the completion of solidification at the desired location in the machine. Fig. 20.45 shows the installation of strain gauge bridges in the thickness setting pins on both sides of a caster segment.97 The sensors were installed at both the inlet and exit ends of the segment. The variation of the segment opening force with casting speed during about a 2.2-hour cast is shown in Fig. 20.46. In the same figure, the solid fraction calculated from a one-dimensional mathematical solidification model is also shown. The objective of Fig. 20.46 is to establish the relationship between the ferrostatic force
38
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
clamping force Fc
opening force Fo
Fig. 20.45 Installation of strain gauges in segment pins. From Ref. 97.
Fnet = Fc − Fo drive side thickness pins
upper s egm frame ent
free side thickness pins
lower s egm
ent fram
e
transmitted to the roll segment and the level of solidification. It is evident from the figure that sudden changes in segment force occur when the solid fraction is within 0.65–0.70. This implies that the liquid becomes isolated within the dendritic network, thus preventing transmission of ferrostatic force to the containment rolls. These findings confirm earlier studies that suggest that the application of soft reduction beyond the point where the solid fraction exceeds 0.65–0.70 might not be beneficial and can even have detrimental effects on quality.103,104 Segment load measurements with strain gauges showed that spray nozzle performance has a significant effect on the complete solidification point.98 The results showed that when nonoptimal spray nozzles were decreased from 58 to about 25%, the solidification constant increased by about 10%, which enabled about a 20% increase in caster productivity. Electromagnetic transducers have also been used to detect the liquid core in the containment of slab casters.101,102 They have the advantage that they can be more easily moved from one location to the next. Also, they more clearly indicate the tip of the liquid core. However, their longevity in the hostile containment environment has yet to be proven. Load cells and strain gauges are sometimes used to understand the causes of mechanical failures, particularly in the machine containment. For example, as part of a major undertaking to understand the causes of frequent roll bearing failures, load cells were used to measure the forces sustained by rolls in the bow and unbending section of a slab caster.105 As illustrated in Fig. 20.47,
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
39
Casting Volume
500
1.0
400 Segment opening force (kN)
0.8 Center solid fraction
300 solid friction 200 opening force 0.4 100 0.6
0 Cast speed (m/min) 1.3
0.2
0.9
0.5 0 2000 4000 Time (seconds)
Fig. 20.46 Variation of center solid fraction and segment opening force. From Ref. 97.
6000
8000
foot rolls bender
driven rolls numbered rolls instrumented with load cells and thermocouples
bow horizontal 1 horizontal 3
straightener 54
Fig. 20.47 Rolls instrumented with load cells and thermocouples. From Ref. 105.
horizontal 2
67 55 67 E
74 73
79 79 F G H
bearing load cell bearing thermocouple A B slab
C
D
40
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.5.5 Roll Bending and Temperature
A conventional slab caster can have 150–250 containment rolls, some of which sustain forces up to 100 tons while contacting a slab with a surface temperature of about 1000°C. An idle roll of split design costs in the range of US$8000 to $10,000, and repair costs are high. It has also been established that roll deflections in excess of 1 mm can cause intercolumnar cracking at the solidification
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
41
Casting Volume
2388 mm bearing load cell slab
250
Bearing load (kN)
200
speed: 1.0 m/min. width: 1930 mm bottom roll 67 top roll 67
150
100
50
0 Roll length
Fig. 20.49 Measured roll bearing loads in straightener. From Ref. 105.
Speed (m/min)
2
speed width
x - start-up o - cap-off
2.0 width (m)
1
1.5
0 x Load cell measurement (kN)
o
x o
x bottom roll 79 - load cell C
o
1.0
400
300 bottom roll 67 - load cell B 200 bottom roll 55 - load cell B 100
0 0
15
30 Time (hours)
45
60
75
Fig. 20.50 Bow and straightener load cell readings during a 75-hour period. From Ref. 105.
42
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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
1
1
0.5
0
0.5
-0.5 deflection into strand 1 0 100 200 300 400 500 600 700 800 900 0 1000
Time (seconds)
Fig. 20.52 Thermal distortion of solid roll during speed transient. From Ref. 108.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
43
Casting Volume
650
Depth, mm @ 2.54 @ 25.4 @ 6.35 @ 76.2
520
Temperature (˚C)
390
260
130 start slowdown 0 0 100 200 strand stopped 300 400 resume motion 500 600 700 800 900
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.
44
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.5.6 Roll Gap, Alignment and Rotation
Rotating rolls, correct roll gaps and well-aligned rolls are prime requirements for good as-cast product quality and high machine productivity. Stuck rolls result in gouging in the product surface, which can sometimes go unnoticed. Bad roll gaps and poorly aligned rolls can yield product defects such as: • Deformed cross-sections. • Corner cracks. • Refill (mid-face) cracks. • Segregation. • Open (laminated) centers. Stuck rolls, incorrect roll gaps and poorly aligned rolls also result in increased machine maintenance costs and reduced machine utilization. It is therefore essential that roll alignments, roll gaps and roll rotations be checked and maintained regularly. These functions can be performed manually by using handheld gap tools and templates, but they are time-consuming and require trained operators. The better approach is to use an automatic device that can be attached to the starter bar chain and that measures roll gap, roll alignment and roll rotation as it is moved up and down the machine. Such a device is referred to as the “gap tool.”109–111 The early versions of gap tools were provided by casting machine builders, and they measured data on roll gap, roll alignment and stuck rolls. Recent tools can provide additional information on roll eccentricity and spray-water pattern.110 Two versions of gap tools are available. The most commonly used version is attached to the end of the starter bar chain, replacing the starter bar head, and is drawn through the machine during scheduled maintenance downturns or turnarounds. The other version, referred to as “on-board gap tool,” is permanently installed on the starter bar chain in between links and is drawn through the machine during each start-up. The type of gap tool chosen should be determined by considering the downtime available for gap runs and maintenance functions, and by weighing gap tool purchase and maintenance costs against productivity and quality gains that can be achieved through its use. Gap and alignment tools provide data on the cold status of the machine. However, experience shows that roll gaps and alignments can significantly change during casting because of roll and
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
45
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.
46
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
2.0 1.5 1.0 0.5 0 bow exit -0.5 gap closing 0 0.5 Gap change (mm) 1.0 1.5 gap closing 0 0.5 1.0 1.5 2.0 2.5 0 2 4 6 8 10 12 14 16 18 gap opening straightener inlet gap opening
Fig. 20.57 Cast speed and roll gap changes at bow exit and straightener inlet. From Ref. 112.
Cast speed (m/min)
Time (hours)
hot misalignments. The bender and the bow were mechanically linked to keep them aligned during casting. This and other measures taken to stabilize roll gaps and roll frames minimized slab cross-sectional deformation and corner cracking. Displacement transducers were used to understand the causes of mid-face cracks and centerline laminations in slabs of high-carbon and plate grades.112 The transducers were installed at the end of the bow and on the straightener section of the slab caster to investigate roll gap and frame stability during casting. Fig. 20.56 shows the installation of the transducers in the bow exit and straightener inlet areas of the caster. The measured gap changes at the last bow roll pair and the first straightener roll pair are shown in Fig. 20.57 as a function of cast speed and time. The gap changes at both rolls exceeded the allowable limit of ± 0.5 mm significantly, and showed big swings during speed transients. Several mechanical improvements were implemented to minimize the large gap changes and other excessive frame movements measured with the displacement transducers. One of the improvements was the removal of the top driven roll at the inlet of the straightener from hydraulic cylinders and installation on rigid girders. Fig. 20.58 shows the improvement in gap change profile at the bow exit and straightener inlet area of the machine as a result of the mechanical improvements. The dramatic decrease in slab open centerline incidence as a result of the study and numerous mechanical improvements is shown in Fig. 20.59. Displacement transducers were used by others to investigate dynamic roll movements and roll gap distortions in the straightener area,113 and also stability of gap tapering in the soft reduction area114 of slab casters.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
47
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
48
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.6.2 Length,Width and Weight
Three methods are commonly used to measure the length of the as-cast product before the cutting operation begins. The first is by tracking the position of the strand and the position of the torch machine. The strand position is tracked by means of speed measurements on one or more drive rolls in the caster containment section. The position of the torch machine is tracked by means of an encoder that is tied to a rack-and-pinion mechanism connected to the torch machine. From the strand and torch machine positions, the distance between the head of the strand and the torches is calculated. When that distance equals the ordered length, the torch machine is clamped onto the strand and the cutting operation is started. The second method entails the use of low-level lasers. The third length measurement method involves the use of a wheel on the torch machine that contacts the narrow surface of the strand and directly measures the distance between the head of the strand and the torch machine. The measurement wheel is water-cooled, and the wheel rotation is converted to linear distance by means of an encoder. When properly maintained and regularly calibrated, all three methods yield length within the design tolerance, which is usually ± 1% of the aim length. For example, manual length measurements on numerous hot slabs showed that only 2% of the lengths measured with the wheels were outside the length tolerance of ± 76.2 mm.115 The width of the slab in the runout is different from that at mold exit mostly because of thermal shrinkage and creep under ferrostatic pressure.115 Other contributors to width variation include incorrect mold setup and roll gap profile. Width variance from one slab to the next is largely dependent on grade, secondary water practice and cast speed. Five methods are used to determine slab
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
49
Casting Volume
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.
50
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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. 20.60 shows one of the earliest hot-slab surface inspection methods that is based on the eddy current principle and was first installed in Domnarvet, Sweden117,118 It consists of a robust watercooled sensor that is moved back and forth on a beam, which is mounted transverse to the casting direction. The sensor consists of a coil that is mounted in a housing, about 5 mm above the slab. On detection of a crack, a paint mark is made on the slab. The sensor scans only the upper surface of the slab. Inspection of the slab surface is done above the Curie temperature of 768°C. High-pressure water is used to descale the slab surface. Several grades of steel, including medium-carbon, silicon, boron and HSLA steels, were tested on-line. The system can detect longitudinal, transverse and star cracks that are deeper than 2 mm. Reportedly, it also detects scabs, slag spots and doubleskin effects. Fig. 20.61 shows typical output signal as a function of average crack depth. In a test of 200 slabs, agreement between manual and on-line inspection was 95%. Some of the other eddy current devices developed for hot-surface inspection are described in Refs. 119–125. The features that are common to most of these eddy current systems are that the slab surface temperature must be above the Curie point, the surface must be descaled, and only the top surface can be inspected. The system described in Refs. 124 and 125 is used for the detection of transverse edge cracks on each side and is controlled by a robot. The slab surface temperature must be below 500°C. System reliability is better than 99% in detecting corner cracks that are deeper than 2 mm. Optical surface inspection systems employ industrial TV cameras (ITV), charge coupled devices (CCD), line-scan or TV cameras, laser scanning and photographic methods.126–132 Fig. 20.62 shows the system described in Ref. 126, which was installed on a slab caster to inspect both the top and bottom surfaces of slabs. Optical systems are suitable for the inspection of large surface areas but do not indicate crack depth and are not suitable for hairline or short cracks. Also, the surface must be properly cleaned, which is generally satisfied by high-pressure water systems that make it difficult to maintain uniform temperatures across slab width. Ultrasonic133 and electromagnetic-ultrasonic134 methods were developed and tested to detect internal defects, such as primary pipe, centerline lamination, refill cracks and segregation. In the ultrasonic method, water is employed as a coupling medium between a combination transmitter/ transducer head and the test slab. In the electromagnetic/ultrasonic method, transmitting and
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
51
Casting Volume
video signal
signal processor
data processor
process computer magnetic disk
VTR
hard copy unit CCD
floppy disk line printer CRT console slab
surface pre-conditioning machine
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.
References
1. A.T. Etienne and P.H. Dauby, 4th International Iron and Steel Congress (London: The Metals Society, 1982), 11.1–11.14. 2. W. Irving, 4th International Iron and Steel Congress (London: The Metals Society, 1982), 13.1–13.16. 3. M.R. Ozgu, Canadian Metall. Quarterly, 35:3 (1996): 199–223. 4. H-J. Ehrenberg et al., Int. Conf. Metall., Stahl u. Eisen (1987), 149–159. 5. R. Steffen, et al., Int. Conf. Metall., Stahl u. Eisen (1987), 97–118.
52
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52.
Sumitomo Metal Industries, Ltd, Trans. ISIJ, 26 (1986): 590. T. Itoh et al., Preprints for the 102nd ISIJ Meeting, Part III, 22:3 (1981): B90. D.J. Trotter et al., 74th ISS Steelmaking Conf. Proc. (1991), 743–746. D.J. Idstein et al., 77th ISS Steelmaking Conf. Proc. (1994), 545–549. P.H. Dauby et al., 9th ISS Process Technology Conf. Proc. (1990), 33–39. P. Andrzejewski and K-U. Kohler, 9th ISS Process Technology Conf. Proc. (1990), 41–42. M.C.M. Cornelissen et al., 9th ISS Process Technology Conf. Proc. (1990), 95–99. D.I. Walker et al., 9th ISS Process Technology Conf. Proc. (1990), 3–12. A. Zeewy et al., 9th ISS Process Technology Conf. Proc. (1990), 13–16. Y. Maruki et al., Trans. ISIJ, 28 (1988). T.J. Russo and R.M. Phillippi, 73rd ISS Steelmaking Conf. Proc. (1990), 237–246. M. Okabe et al., Nippon Steel Technical Report, 49 (1991): 23–28. G.T. Moulden et al., 77th ISS Steelmaking Conf. Proc. (1994), 265–270. D. Bolger and P. Krause, 80th ISS Steelmaking Conf. Proc. (1997), 297–305. D. Bolger and P. Krause, Proc. of 3rd European Conf. on Continuous Casting (1998), 687–697. P.W. Manos and R.T. McGuire, Steel Times (1989), 610–613. “Thin Section Casting Program—Final Report,” Vol. II of Report Submitted by U.S. Steel/Bethlehem Steel to the DOE under Contract No. FC07-84ID12545, 267–282. K. Kawakami et al., NKK Technical Report, Overseas, 36 (1982): 26–41. A. Delhalle et al., ISS Trans., 6 (1985): 69–80. J.K. Brimacombe et al., 6th International Iron and Steel Congress Proc., Nagoya, Japan (1990), 246–274. S.G. Thornton and N.S. Hunter, 73rd ISS Steelmaking Conf. Proc. (1990), 261–274. J-P. Birat et al., 74th ISS Steelmaking Conf. Proc. (1991), 39–50. J.K. Brimacombe and I.V. Samarasekera, 74th ISS Steelmaking Conf. Proc. (1991), 189–196. W.H. Emling and S. Dawson, 74th ISS Steelmaking Conf. Proc. (1991), 197–217. M.R. Ozgu et al., Proc. of 1st European Conf. on Continuous Casting (1991), 1.73–1.82. D.E. Humphreys et al., Proc. of 1st European Conf. on Continuous Casting (1991), 1.529–1.540. D. Steward et al., 79th ISS Steelmaking Conf. Proc. (1996), 207–214. R.B. Mahapatra et al., Proc. of 1st European Conf. on Continuous Casting (1991), 1.541–1.556. M.R. Ozgu and B. Kocatulum, 76th ISS Steelmaking Conf. Proc. (1993), 301–308. H.L. Gilles, 76th ISS Steelmaking Conf. Proc. (1993), 315–336. J.S. Jenkins et al., 77th ISS Steelmaking Conf. Proc. (1994), 337–345. S.L. Kang et al., 77th ISS Steelmaking Conf. Proc. (1994), 347–356. H.L. Gilles et al., 9th ISS Process Technology Conf. Proc. (1990), 123–138. M.R. Ozgu and M.B. Perrine, Proc. of 2nd European Continuous Casting Conf. (1994), 130–134. N.S. Hunter et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 289–300. A. Cristallini et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 327–343. T. Wada et al., 70th ISS Steelmaking Conf. Proc. (1987), 197–204. K. Sano et al., Nippon Kokan Technical Report, Overseas, 30 (1980): 24–33. H. Matsunaga et al., Nippon Steel Technical Report, 23 (1984): 55–67. H. Kato and M. Yamasita, IFAC Automation in Mining, Mineral and Metal Processing, Tokyo, Japan (1986), 253–258. A. Matsushita et al., Trans. ISIJ, 26 (1986): 313. A. Matsushita et al., Trans ISIJ, 28 (1988): 531–534. J.M. Ritter et al., Rev. Metall., Cah. Inf. Tech., 87:9 (1990): 761–769. J. Kubota et al., 6th Intl. Iron and Steel Congress Proc., Nagoya, Japan (1990), 356–363. J. Kubota et al., 74th ISS Steelmaking Conf. Proc. (1991), 233–241. G. Bocher et al., Metall. Plant and Technology Intl., 3 (1993): 50–55. H. Okita et al., 76th ISS Steelmaking Conf. Proc. (1993), 237–243.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
53
Casting Volume
53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103.
54
M. Washio et al., ATS Steelmaking Conf. Proc. (1992), 507–512. M. Zeze et al., 76th Steelmaking Conf. Proc. (1993), 267–272. G. Bocher et al., Proc. of 2nd European Continuous Casting Conf. (1994), 103–108. D.W. van der Plas et al., Proc. of 2nd European Continuous Casting Conf. (1994), 109–118. J. Fukuda et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 437–445. A.A. Kamperman et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 465–474. Y.K. Park et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 961–970. S. Yuhara et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 989–992. E. Herrmann, Handbook Les Stranggiessens, Alluminum, Verlag, Dusseldorf (1955). G.J. McManus, Iron Age, 224:4 (1951): MP7–9, MP11. M.M. Wolf, 74th ISS Steelmaking Conf. Proc. (1991), 51–71. J.P. Radot et al., Rev. Metall., Cah. Inf. Tech., 1 (1988): 71–79, 85. A. Shirayamo et al., NKK Technical Report, Overseas, 52 (1988): 1–7. M. Espenhahn et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 891–900. E. Becker et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 901–922. M. Jauhola et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 951–960. H.L. Gilles and B.R. Forman, 74th ISS Steelmaking Conf. Proc. (1991), 219–232. H.L. Gilles et al., 75th ISS Steelmaking Conf. Proc. (1992), 911–925. B. Mairy et al., 9th ISS Process Technology Conf. Proc. (1990), 73–81. B. Mairy et al., Proc. of 2nd Continuous Casting Conf., London (1985), 59.1–59.8. M. Kiss et al., 76th ISS Steelmaking Conf. Proc. (1993), 383–394. R.W. Ramirez, The FFT, Fundamentals and Concepts (Prentice-Hall, 1985). M. Nogues et al., Proc. of 2nd European Continuous Casting Conf. (1994), 119–125. D.A. Barnard et al., U.S. Patent No. 3,047,915 (1962). M. Komatsu et al., Trans. ISIJ, B-86, 23 (1983). F. Slamar, U.S. Patent No. 3,893,502 (1975). Y. Nakamori et al., Trans. ISIJ, 22 (1982). H.B. Osborn Jr., U.S. Patent No. 2,824,346 (1958). B. Mairy et al., Proc. 1st European Conf. Continuous Casting, Florence, Italy (1991), 1.225–1.234. Y. Nakamori et al., Nippon Steel Technical Report, 34 (1987): 53–61. B.G. Thomas, 74th ISS Steelmaking Conf. Proc. (1991), 105–118. B.G. Thomas et al., 80th ISS Steelmaking Conf. Proc. (1997), 183–201. K.-U. Kohler et.al., 78th ISS Steelmaking Conf. Proc. (1995). D. Gotthelf et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 825–833. Van der Plas et al., Proc. of 2nd European Conf. on Continuous Casting (1994), 109–118. K. Kariya et al., 77th ISS Steelmaking Conf. Proc. (1994), 53–58. K. Noro et al., 13th PTD Continuous Casting Conf. Proc. (1995). M. Iguchi et al., ISIJ Intl., 36 (1993): 190–193. S.G. Hibbins and J.K. Brimacombe, Continuous Casting, Vol. 2, (Warrendale, Pa.: Iron and Steel Society, 1984), 139–151. K. Tsuboi et al., Nippon Kokan Technical Report, Overseas, 36 (1982): 90–105. D.H. Kindt et al., 75th ISS Steelmaking Conf. Proc. (1992), 557–562. Y. Sugitani et al., 108th ISIJ Meeting (Oct. 1984), Lecture No. S898. K. Wunnenberg, Stahl u. Eisen, 98:6 (1978): 254–259. R.B. Mahapatra et al., 74th ISS Steelmaking Conf. Proc. (1991). S.R. Story et al., 76th ISS Steelmaking Conf. Proc. (1993), 405–413. J. Schade et al., 77th ISS Steelmaking Conf. Proc. (1994), 279–295. J. Gancarz et al., 74th ISS Steelmaking Conf. Proc. (1991), 15–22. K.D. Ives and R.V. Branion, 9th ISS Process Technology Conf. Proc. (1990), 61–68. P.W. Manos and C.Enstrom, 9th ISS Process Technology Conf. Proc. (1990), 139–150. R. Cornish and C. Baharis, Iron and Steelmaker (1991), 32–33. M. Yamada et al., 71st ISS Steelmaking Conf. Proc. (1988), 75–78.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
55
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
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
1
Casting Volume
• 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.
20.2 Ladle
In a continuous casting operation, the ladle is used to transfer liquid steel from the steelmaking shop to the caster with aim chemistry, temperature, cleanliness and slag cover. At the caster, steel from the ladle is teemed into the tundish through a nozzle in the bottom of the ladle. It is desirable to monitor three parameters on the ladle as a function of time to properly control the teeming operation. The first is the weight of steel in the ladle, which is routinely measured with load cells on the turret arms or the ladle car. The second is the temperature of steel flowing into the tundish. Direct measurement of steel temperature in the ladle with instrumentation on the ladle itself is not practical, because the ladle is a mobile unit that moves through harsh environments and thus is not conducive to temperature instrumentation. Instead, ladle temperatures are inferred from tundish temperatures, which are continuously or intermittently measured. The third is the slag carryover from the ladle to the tundish. As the end of the ladle drainage is approached, vortex formation above the ladle nozzle and slag carryover into the tundish can occur. The mixing of slag with steel can contaminate the steel in the tundish and cause an increase in the inclusion content of the cast product. It is possible to minimize the transfer of slag from the ladle to the tundish using one of the following slag detection methods: • Visual observation. • Tare weight monitoring. • Rate of teeming change.4,5 • Opto-electronic monitoring of stream surface.6 • Vibration analysis of the ladle shroud.7,8 • Electromagnetic methods.9–12 The ability to visually detect slag carryover into the tundish is highly dependent on operator experience. This can lead to the carryover of varying amounts of slag before detection and corrective action. Tare weight monitoring relies on correct ladle, ladle cover and slag weight estimates and is thus inherently inaccurate. The rate of teeming change monitors the variation in the rate of ladle weight change. A change from metal teeming to slag and metal teeming can be indicated by a rapid decrease in the teeming rate. The method relies on the accuracy of the load cells and can thus lead to errors at low teeming rates. It can also result in false alarms if the teeming rate is affected by the movements of the ladle flow control device or by the ladle. The opto-electronic method cannot observe the core of the stream through which large amounts of slag can be carried over. The vibration analysis method is based on vibration monitoring of the ladle shroud, using an accelerometer attached to the shroud manipulator arm. As is shown in Fig. 20.2, the amplitude of processed signal from the accelerometer increases with the onset of entrained slag flow through the shroud. A
2
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
slag visually detected by operator
Amplitude index
1.0
slag detected with accelerometer
0.5
0
25
50
75
100
125
150
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
20.3 Tundish
The tundish is an intermediate vessel between the ladle and the mold and is used mainly to perform the following functions: • Deliver liquid steel to the mold(s) at a controlled rate. • Maintain a steady supply of liquid steel to the mold(s) during a ladle change. • Remove nonmetallic inclusions from liquid steel before delivery to the mold(s). • Facilitate the control of steel superheat by means of plasma preheaters.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
3
Casting Volume
100 80 60 40 20 0 electromagnetic slag signal (%) Time
Fig. 20.3 Ladle slag carryover, weight and slidegate signals. From Ref. 10.
2 sec ladle slidegate position (%)
ladle weight (tons)
alarm
In order to properly perform the above functions and control the caster operation, it is essential to know the temperature and depth of liquid steel in the tundish. Temperature can be measured in two ways. The first is the manual immersion of disposable thermocouples. This method is reliable and is used in many caster installations. Its main drawback is that, if not exercised at sufficient frequency, rapid temperature changes can go unnoticed. Unexpected increases in steel temperature can result in breakouts, and rapid temperature losses can lead to freeze-off and cast terminations. Because of the push toward stable caster operation and reduced labor costs, continuous temperature measurement systems have been developed and are in use at several caster installations.12,15–17 The sensor used for continuous temperature measurement is comprised of a Pt-Pt/Rh thermocouple embedded in a refractory tube, and it is immersed into the steel bath through an opening in the tundish cover. Fig. 20.4 shows continuous temperature measurement from such a sensor, along with tundish weight and cast speed or throughput.16 The effect of tundish level (or weight) on tundish temperature profile can be clearly seen by examining the difference in profiles during tundish tube changes E1 and E2. When the tundish weight drops to 41,000 kg, about 25% of the tundish wall hot surface is exposed to the atmosphere. Consequently, significant heat loss and thus temperature drop occur from the tundish walls, which must be recovered upon refill. The extent of the temperature loss depends on the flow rate from the tundish, the length of time during which the tundish walls are exposed, and the wall area that is exposed to the atmosphere. Such temperature losses during ladle, mold width and tundish tube changes can be reduced through the continuous measurement of tundish temperature. Other potential benefits of continuous tundish temperature measurement are caster automation and accurate correlation of steel temperature to as-cast product quality. The depth of liquid steel in the tundish is either inferred from weight measurements with load cells on the tundish car or directly measured with electromagnetic sensors installed in the tundish bottom or the vertical walls. Knowledge of liquid steel depth from these measurements is used to: • Safely fill the tundish during the start of a cast. • Maintain a stable tundish level during steady-state casting. • Assure a minimum tundish level during ladle changes to avoid downgrades. • Safely drain the tundish to a very low level during grade changes, tundish changes (tundish flys) and cast terminations, without allowing any slag to flow into the mold.
4
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
55,000 Tundish weight (kg) (a) 1566 Temperatue (˚C) (a) (a) 1552 Temperature 1538 (e1) (e2)
Fig. 20.4 Tundish temperature, weight and cast speed variations. From Ref. 16.
(a) (h)
(a) 41,000
weight (h)
27,000
Cast speed (m/min)
1.18 (g) 0.64 Cast speed 2810 kg/min (6.2 klds/min) (f)
0 0 2 4 Time (hours) 6 8
(a) - ladle change (e) - tundish tube change (f ) - mold width change
(g) - machine slowdown for late ladle connection (h) - reduced tundish level prior to tundish tube change
Most of the caster operations use weight monitoring with load cells for bath height control because load cells are easy to install and maintain on the tundish car. Furthermore, load cells stay on the tundish car and can be used on the incoming tundish. The major disadvantage of weight monitoring is that it relies on correct estimates of tundish, tundish cover and slag weights, and is thus inherently inaccurate. The inaccuracies are particularly worrisome during draindowns, because they can result in slag drainage into the mold or excessive residual steel in the tundish. Slag in the mold can cause breakouts. An excessive amount of residual steel in the tundish compromises yield during a tundish fly or increases the length of the mixed-grade zone during a grade change. For improved draindown control, some caster operations use a bobber or a ceramic ball in the tundish in conjunction with the weight measurement system. The bobber or ceramic ball is positioned over the tundish nozzle. The bobber is used to indicate bath level. The ceramic ball is used to prevent vortexing and slag carryover into the mold during the draining of the tundish.18 Some caster operations use electromagnetic sensors in the bottom of the tundish for direct measurement of steel level.19,20 The sensor assembly consists of a primary and a secondary coil that are concentrically arranged in stainless steel housing. As Fig. 20.5 shows, the housing is installed between the refractory lining and the tundish bottom shell. An alternating current applied to the primary coil generates an electromagnetic field in the liquid steel. The electromagnetic field, in turn, generates eddy currents that attenuate the primary field in proportion to the steel level. By this means, the height of the slag/metal interface above the refractory bottom in the vicinity of the sensor can be measured very accurately in the range of 0–200 mm. The thickness of the refractory lining is compensated for. The electromagnetic sensor, in conjunction with an optimized tundish bottom design and an automatic drain control system, enables the operators to consistently drain the tundish to very low levels without allowing slag to flow into the mold. Other significant benefits of
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
5
Casting Volume
1. safety lining 2. spray lining 3. impact pad 4. electromagnetic level sensor
1 2 3 4
SEN
SEN
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.
6
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.4.2 Copper Temperatures
From the early days of continuous casting, numerous researchers have measured the copper temperatures on caster molds.23–41 The objectives of the measurements varied, but in general they were to: • Evaluate mold powders and practices. • Analyze mold heat transfer and strand solidification. • Evaluate mold design. • Predict mold/strand sticking. • Measure liquid steel pool level. In all cases, thermocouples were used for copper temperature measurements, because thermocouples are inexpensive and easy to install, and their signals are easy to interpret. The thermal behavior of all four walls between the top and bottom of a mold can be analyzed relatively easily by lacing the copper plates with thermocouples. There are two methods of installing thermocouples in mold coppers. In the first and most commonly used method, the thermocouples are introduced from the backside of the copper plates parallel to the path of heat flow (perpendicular to the hot face). The advantage of this method is the ease of thermocouple installation. The disadvantage is that, because the thermocouple holes are parallel to the path of heat flow, they affect the flow of heat and thus can result in erroneous temperature measurements. This method is suitable for routine relative temperature monitoring whereby the measurements are used for sticker breakout predictions, mold level measurement and other automation purposes. In the second method, the thermocouples are introduced from the top or bottom of the copper plates and perpendicular to the
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
7
Casting Volume
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.
8
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
400 low carbon strip speed: 1.27 − 1.62 m/min. width: 990.6 − 1435.1 mm 300 narrow wall
Mold wall hot face temperature (˚C)
200 loose wide wall
1000 300 medium carbon plate speed: 1.02 − 1.29 m/min. width: 1930.4 − 1955.8 mm
Fig. 20.8 Variation of mold hot-face temperatures with vertical distance from mold top. From Ref. 34.
200
loose wide wall
100 0 300
}
various trials 600 900
Distance from mold top (mm)
Fig. 20.8 shows the variations of hot-face temperatures with vertical distance below the mold top resultant from the two-thermocouple method shown in Fig. 20.7.34 The hot-face temperatures were calculated from wall temperature measurements made along the mid-plane of the loose and narrow walls when casting low-carbon strip and medium-carbon plate grades. As expected, the temperatures on both the wide and narrow walls decreased with increasing distance below the mold top. This is caused by three factors. First, as the distance below the meniscus increases the thickness of the solid shell and its resistance to heat transfer from the liquid core to the mold increases. Second, solid fraction of the slag layer between the slab and the mold increases with distance below the mold top, thus increasing the resistance to heat flow from the slab to the mold. Third, the fluid flow activity in the liquid core decreases with the vertical distance which, in turn, decreases the heat transfer from the liquid core to the solid shell. Because the liquid steel streams from the bifurcated tundish nozzle are directed toward the narrow walls, they affect higher heat transfer on and through the narrow side shell. As a result, copper temperatures on the narrow walls were higher than those measured on the wide walls. From Fig. 20.8 it can also be observed that, when casting low-carbon strip grade slabs, the mold hot-face temperatures exceeded 350°C, beyond which it is suggested that there is an increased potential for sticking.42 Thermocouples are also being used on medium-thickness and thin-slab caster molds to improve mold design, select mold powders, develop casting practices and control the operation of the caster.40,41 Figs. 20.9 and 20.10 show sample temperatures measured on the broad face coppers of
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
9
Casting Volume
300 funnel area
200 Temperature (˚C)
parallel area 100
0
0
100
200
300
400
500
600
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.
10
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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
• Thermocouples in mold copper.1,44,46 • Optical devices.1 • Link-arm type magnetic flux meter.44 • Telescopic magnetic flux meter.44 The eddy current probe, the electromagnetic cassette and the radioactive source are the most commonly used methods. The thermocouple, optical and flux meter methods have limited or no industrial acceptance. The eddy current probe is suspended over the mold and generates an electromagnetic field that is directed into the mold. The electromagnetic field induces eddy currents in a near-surface layer of the liquid metal in the mold via the primary coil. This eddy current produces a secondary field, which, in turn, induces voltages in the sensor’s secondary coil. The magnitude of the eddy current, and thus the induced voltage, is dependent on the actual distance between the sensor and the steel surface. The actual liquid steel level in the mold can be detected independently of any slag and powder layer on top of the liquid steel. Measurement accuracy with the suspended probe has been reported to be better than ± 2 mm at casting rates up to 7 tons/minute.51 The measurements are not affected by copper temperature and copper coating such as nickel. However, it is not suited for auto-start detection because the probe is brought over the mold after the initial filling of the mold is over and the mold level stabilizes. The electromagnetic cassette entails a transmitter and a receiver coil, which are mounted side by side on top of the fixed wide mold wall. This method differs from the suspended sensor method in that the transmitter’s electromagnetic field induces eddy currents in the exposed face of the copper wall. These eddy currents generate their own magnetic fields, which are detected by the receiver coil. Because the exposed surface area of the copper wall changes as the level of liquid steel in the mold changes, the strength of the signal at the receiver coil is dependent on the level of liquid steel in the mold. The sensor signal is also dependent on the temperature of the copper wall. The temperature effect becomes pronounced on molds that are coated with nickel at the top end. Although the coils in the cassettes can be oriented to offset the effect of nickel coating, experience shows that the nickel can cause a gradual drift in the mold level, which needs to be periodically readjusted.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
11
Casting Volume
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.
12
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
The trials showed that: • Contrary to the generally accepted belief, the meniscus does not slide freely along the mold wall and is not stationary when viewed from a stationary position. Rather, it fluctuates in phase with mold oscillation, as depicted in Fig. 20.14.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
13
Casting Volume
SEN displacemant transducer mold
quartz window optical fiber
data scanner
recorder
monitor
Fig. 20.13 Apparatus used for meniscus observation. From Ref. 47.
Mold displacement (mm)
casting speed: 1.60 m/min oscilation freq: 2.11 Hz oscilation amplitude: + 6 mm − 6
0
-6
10 mm from narrow wall Absolute displacement of meniscus 5 mm
3 mm from narrow wall
0
1.0 Time (seconds)
2.0
Fig. 20.14 Meniscus behavior. From Ref. 47.
14
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
(a) upstroke 5 casting speed 0.53 m/min 1.60 m/min
Absolute displacement of meniscus (mm)
0 −6
−3
0
3
6
displacement of mold (mm) 0 6 3 0 −3 −6
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.
(a) downstroke 5 mm
(b) upstroke
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
15
Casting Volume
50 mm 120 mm 60mm 120 mm
2
1
2
1
level sensor
level sensor
SEN
level sensor
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.
700
900
1100
1300
1500
1700
1900
Mold width (mm)
16
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.4.4 Mold Oscillation
Mold oscillation was introduced into continuous casting in 1949 to prevent shell sticking to the mold by providing sufficient lubrication between the strand and the mold.61,62 Today, mold oscillation is common practice on all conventional continuous casters. Numerous studies have shown that the surface quality of the product is highly dependent on the shape of the oscillation curve (e.g., sinusoidal, triangular), stroke and frequency.63 Until recently, conventional casters utilized mechanical oscillators and sinusoidal oscillation curves. However, most new and rebuilt casters are utilizing hydraulic oscillators because the profile, frequency and stroke of the oscillation curve can be easily changed and customized for each grade.59,64–68 The stroke and frequency of oscillation can vary from one installation to the next, and the oscillation practice can be different for various steel grades at a given caster installation. However, once the optimum oscillation practices are established, it is desirable that they remain constant. This requires that the quality of oscillation and the physical condition of the oscillator be checked and corrected periodically. Several methods are used to check the condition of the oscillator and the quality of mold oscillation. These include the manual pencil trace, human touch, periodic trials with displacement transducers and accelerometers,30,69,70 and on-line continuous monitoring with accelerometers.71–73 Displacement transducers have moving parts and are thus prone to failure under the harsh caster environment. On the other hand, accelerometers do not have moving parts. Furthermore, they have high sensitivity and accuracy and are thus the best suited for oscillation monitoring. A system comprised of four triaxial accelerometers mounted at each mold or mold table corner, a data acquisition and analysis subsystem, and a personal computer can yield all the parameters that are essential for complete assessment of the quality of oscillation and the condition of the oscillator.73 Each triaxial accelerometer sensor consists of three individual accelerometers oriented in the vertical and two horizontal directions perpendicular to the mold narrow and broad faces. The acquired acceleration signal is integrated to obtain time-based velocity and displacement data. The Fast Fourier Transformation (FFT) technique is utilized to convert the timebased mold acceleration data and the calculated mold velocity and displacement data into discrete frequency component.74 The procedure is repeated for acceleration signals collected in the three directions from each triaxial accelerometer sensor on the four mold or mold table corners. It has been demonstrated that the system can produce the following oscillation parameters: • Displacement, velocity and acceleration curves in three directions. • Peak-to-peak displacement, velocity and acceleration values in three directions. • Primary oscillation frequency. • Secondary (extraneous) oscillation frequencies in three directions. • Negative strip time. • Negative strip ratio. • Rise/fall ratio (i.e., peak-to-peak time up/peak-to-peak time down). • Phase (lead or lag of one mold table corner relative to the other in the vertical direction expressed in degrees).
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
17
Casting Volume
1200 Acceleration (min/sec2) 800 400 0 −400 −800 −1200 0 0.4 0.8 1.2 Time (seconds) 1.6 2
Fig. 20.19 Mold vertical acceleration versus time. From Ref. 73.
1500 Peak-to-peak acceleration (mm/sec2)
1000
500
0
0
400
800
1200
1600
2000
Fequency (cpm)
Fig. 20.20 Mold vertical acceleration versus frequency. From Ref. 73.
80 60 Velocity (mm/sec) 40 20 0 −20 − 40 − 60 − 80 0 0.4 0.8 1.2 1.6 2
Time (seconds)
Fig. 20.21 Mold vertical velocity versus time. From Ref. 73.
18
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
⊥ to broad face ⊥ to narrow face
Time (seconds)
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
19
Casting Volume
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
eccentric pin
instrumented pin mold table pin strain gauge bridges
thermocouple
Fig. 20.24 Schematic of strain gauge and displacement transducer installation on mold table oscillator. From Ref. 30.
20
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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
20.4.6 Mold Powder Film Thickness and Molten Pool Depth
Mold heat removal is strongly affected by the thickness of the powder film in the gap between the mold and the strand. Insufficient or uneven flow of mold powder into the strand/mold interfacial gap can cause surface cracks on the cast product. On the other hand, an excessively thick film of powder can result in low heat removal rates and thus low caster productivity. One of the major
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
21
Casting Volume
200 A1 190 A2 casting work cold work pin force (kN) 180
170
Fig. 20.26 Pin work under casting and cold conditions. From Ref. 30.
160
-5
-2.5 0 2.5 Displacement (mm) A2 − A1 [mold stroke] x [slab area in mold]
5
Friction (kN/m2) =
16
12
Mold/slab friction (kN/m2)
8
4 low carbon mold powders 0 0.7 0.9 1.1 1.3 1.5 1.7
Cast speed (m/min)
Fig. 20.27 Mold/slab friction versus cast speed. From Ref. 30.
factors that control the flow rate and uniformity of the film is the depth of the molten powder cover above the steel pool. It is thus desirable to measure the depth of the molten powder cover and correlate it to the thickness of the in-gap powder film. The depth of the molten powder cover is usually measured by wire burnoff. The thickness of the powder film is measured by recovering layers of powder from the mold hot face during major transitions when the mold level decreases or from the mold exit. However, these methods are manual and inherently inaccurate. An on-line measuring system comprised of an in-gap powder thickness gauge and a molten powder pool depth gauge was developed and applied on a slab caster.82 The film thickness gauge is thermal radiation-based and is installed below the mold. The molten powder cover gauge is an eddy current device, which is suspended above the mold and utilizes the difference in the electrical resistance of liquid steel and molten slag. Both gauges are of noncontacting type. Measurements made on a slab caster mold are shown in Fig. 20.28. The results showed that:
22
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.4.7 Mold Wall Deformation
The continuous caster mold is known to sustain large thermal stresses and undergo sizable deformation during casting.83,84 The deformation of the mold is critical because it distorts the shape of the strand and can thus impact surface quality. Furthermore, in slab casters, a gap is needed between the broad and narrow faces to facilitate width changes while casting. Excessive deformation of the wide walls, in combination with the thermal expansion of the narrow walls, can result in the damaging of the wide wall hot faces and the edges of the narrow walls. The deformation of the wide walls can sometimes be severe enough to restrain the movement of the narrow walls and thus prevent size changes while casting. It is thus desirable to know the extent of mold wall deformation under various casting conditions. Displacement transducers are commonly used to measure mold wall deformation during casting. Fig. 20.29 shows the installation of linear displacement transducers on a straight slab caster mold at Bethlehem Steel’s Burns Harbor plant to study wide wall and mold cavity deformation during the casting of 254 x 1933-mm slabs. Five transducers were installed on the backside of each wide side water jacket at the lower end of the mold, and two in the mold cavity at the mold ends. Measurements were made while casting at a steady-state speed of 1.14 m/min and just prior to automatic size change while casting at 0.6 m/min. The results are shown in Fig. 20.29 and indicate that: • The fixed and loose sides deformed differently during steady-state casting and just prior to size change; the deformed shapes were convex into the mold cavity. • The middle of the fixed side moved into the cavity. The middle of the loose side either stayed in the original position or moved back from the original position. • During steady-state casting, the mold cavity thickness decreased by 0.96 mm in the middle of the mold and increased by about 2.00 mm at the ends of the mold. During size changes, because of lower casting speed and lower copper temperatures, the walls sustained smaller deformation.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
23
Casting Volume
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.
3.5 3.0 2.5 Opening (mm) 2.0 1 1.5 1.0 0.5 0 0 1 2 3 capcast speed off 4 5 6 7 1.00 0.50 0 1.50 in-cast width change 2 1 2 Cast speed (m/min.)
Fig. 20.30 Variation of mold gap opening during a sevenhour cast.
Time (hours)
24
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
25
Casting Volume
720 mm 720 mm
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.
left
9 min
26
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
27
Casting Volume
load cell
SEN mold refractory rod
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
(Eq. 20.5)
where V = mean velocity of liquid approaching the cylinder, D = diameter of cylinder, and ν = kinematic viscosity of fluid.
28
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
Table 20.1 Typical Mold Velocity Measurements (from Ref. 87)
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
(Eq. 20.6)
29
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
31
Casting Volume
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.
20.5.2 Strand Surface Temperature
It is known that secondary cooling practices in the containment have significant effects on product surface, internal quality and also productivity. As the strand surface cools and reheats between rows of spray nozzles and rolls, thermal stresses are generated in the solid shell. If the stresses are high enough, cracks can form or existing cracks can worsen on the strand surface. At the same time, mechanical stresses arise in the shell due to inter-roll bulging under ferrostatic pressure. The resistance of the shell to ferrostatic pressure and bulging is determined by its temperature and hence by the spray-cooling practice. If the bulging stresses exceed a critical value, internal (refill) cracks can occur at the solidification front. Another potential problem is the occurrence of surface cracks during the unbending of the strand because of an improper combination of surface temperature and unbending stresses. Also, in single-point unbending machines, segregation and centerline laminations can originate in medium- and high-carbon grade slabs if complete
32
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.5.3 Strand Shape and Bulging
Devices have been developed and used to measure strand profile to understand and eliminate the causes of: • Excessive bulging or concavity on the narrow side just below the mold, which might cause breakouts, excessive wear on mold narrow side coppers, and surface cracks. • Deformed strand cross-section (e.g., trapezoidal), which might cause corner cracks.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
33
Casting Volume
34
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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.
• Bulging between containment rolls that might cause internal cracks at the solidification front. Because measurements are made in the spray chamber, where the environment is not conducive to instrumentation, most shape sensors have been custom-designed physical contact devices involving simple instrumentation. For example, to understand early solidification profile and concavity just below the mold foot rolls on the narrow side, a shape detector involving two contact rolls and a displacement transducer was built and used on a high-speed slab caster.92 Fig. 20.41 shows the shape detector and the measurement results. The mold taper was changed to make shell growth more uniform and reduce the narrow side depression to be within 2–4 mm. When trapezoidal cross-section and corner cracks became a problem on plate grade slabs on a caster equipped with a straight mold, a probe was built and used at the bender exit, as shown in Fig. 20.42, to assess slab shape coming out of the bender.93 The probe was equipped with two curved contacts that were pushed against the slab narrow side and two displacement transducers. When the probe measurements were compared with mathematical calculations, the results showed that the slabs were being excessively misshaped before exiting the bender. As one of the measures to prevent slab misshaping and corner cracks, the bender frames were strengthened and the bender rebuilding procedure was improved. Inter-roll bulging and attendant internal cracking at the solidification front has been a major concern for caster builders, operators and researchers. Sensors have been used to measure and correlate bulging to machine design and casting practices. These sensors include custom-designed physical contact devices,94,95 and more recently a laser bulge-meter.96 Figs. 20.43 and 20.44 show a contact bulge detector and a laser bulge-meter, respectively. Custom-designed bulge devices have the disadvantage that they have short operational life and generally can be installed in specific caster locations. On the other hand, noncontacting laser-based bulge-meters have long operational life and can be moved from one location to the next for continuous measurements at
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
35
Casting Volume
contact roll
measuring roll slab air cylinder displacement transducer
(de)
(dc)
molten steel solidified shell
depression (dd)
Ratio of retarded shell growth (de/dc)
1.0 X 0.8 0.6 0.4 0.2 0 -4.0 X X X X X cast speed 1.0 m/min. 1.2 X 1.4 1.6 20
0
4.0
8.0
12.0
Depression of narrow face (dd) (mm)
Fig. 20.41 Measurement of narrow face depression and retared shell growth. From Ref. 92.
36
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
curved contacts
bender rolls
Fig. 20.42 Schematic of narrow side shape probe. From Ref. 93.
LVDT
pusher rod slab
slab V ∅ 150 ∅ 80
split roll
linear transducer for roll deflection
air cylinder
linear transducer for bulging
motor for rotation of the device
drive motor
180˚
Fig. 20.43 Mechanical contact type bulge detector. From Ref. 94.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
37
Casting Volume
PC
laser blackbox
motor driver
water-cooled aluminum case bulgemeter mounting frame
optocater gauge probe
stepper motor
gearbox
support roll
segment frame strand surface
Fig. 20.44 Laser bulge-meter. From Ref. 96.
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.
20.5.4 Final Solidification Point (Liquid Core) and Segment/Roll Loads
It is often desirable to know the complete solidification point (or length of liquid core) under different casting conditions for quality and productivity considerations. For example, in casters where soft reduction is used to avoid the occurrence of centerline segregation, solidification must be controlled in the tapered zone of the machine. In machines equipped with single-point unbending, complete solidification must occur before the unbending point to avoid segregation and centerline laminations in medium- and high-carbon grade slabs. It is essential that solidification is completed within the containment zone under all casting conditions. Strain gauges97–100 and electromagnetic transducers101,102 have been successfully used to measure the end of the liquid core under different casting conditions. The results were then used to develop casting practices that would assure the completion of solidification at the desired location in the machine. Fig. 20.45 shows the installation of strain gauge bridges in the thickness setting pins on both sides of a caster segment.97 The sensors were installed at both the inlet and exit ends of the segment. The variation of the segment opening force with casting speed during about a 2.2-hour cast is shown in Fig. 20.46. In the same figure, the solid fraction calculated from a one-dimensional mathematical solidification model is also shown. The objective of Fig. 20.46 is to establish the relationship between the ferrostatic force
38
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
clamping force Fc
opening force Fo
Fig. 20.45 Installation of strain gauges in segment pins. From Ref. 97.
Fnet = Fc − Fo drive side thickness pins
upper s egm frame ent
free side thickness pins
lower s egm
ent fram
e
transmitted to the roll segment and the level of solidification. It is evident from the figure that sudden changes in segment force occur when the solid fraction is within 0.65–0.70. This implies that the liquid becomes isolated within the dendritic network, thus preventing transmission of ferrostatic force to the containment rolls. These findings confirm earlier studies that suggest that the application of soft reduction beyond the point where the solid fraction exceeds 0.65–0.70 might not be beneficial and can even have detrimental effects on quality.103,104 Segment load measurements with strain gauges showed that spray nozzle performance has a significant effect on the complete solidification point.98 The results showed that when nonoptimal spray nozzles were decreased from 58 to about 25%, the solidification constant increased by about 10%, which enabled about a 20% increase in caster productivity. Electromagnetic transducers have also been used to detect the liquid core in the containment of slab casters.101,102 They have the advantage that they can be more easily moved from one location to the next. Also, they more clearly indicate the tip of the liquid core. However, their longevity in the hostile containment environment has yet to be proven. Load cells and strain gauges are sometimes used to understand the causes of mechanical failures, particularly in the machine containment. For example, as part of a major undertaking to understand the causes of frequent roll bearing failures, load cells were used to measure the forces sustained by rolls in the bow and unbending section of a slab caster.105 As illustrated in Fig. 20.47,
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
39
Casting Volume
500
1.0
400 Segment opening force (kN)
0.8 Center solid fraction
300 solid friction 200 opening force 0.4 100 0.6
0 Cast speed (m/min) 1.3
0.2
0.9
0.5 0 2000 4000 Time (seconds)
Fig. 20.46 Variation of center solid fraction and segment opening force. From Ref. 97.
6000
8000
foot rolls bender
driven rolls numbered rolls instrumented with load cells and thermocouples
bow horizontal 1 horizontal 3
straightener 54
Fig. 20.47 Rolls instrumented with load cells and thermocouples. From Ref. 105.
horizontal 2
67 55 67 E
74 73
79 79 F G H
bearing load cell bearing thermocouple A B slab
C
D
40
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.5.5 Roll Bending and Temperature
A conventional slab caster can have 150–250 containment rolls, some of which sustain forces up to 100 tons while contacting a slab with a surface temperature of about 1000°C. An idle roll of split design costs in the range of US$8000 to $10,000, and repair costs are high. It has also been established that roll deflections in excess of 1 mm can cause intercolumnar cracking at the solidification
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
41
Casting Volume
2388 mm bearing load cell slab
250
Bearing load (kN)
200
speed: 1.0 m/min. width: 1930 mm bottom roll 67 top roll 67
150
100
50
0 Roll length
Fig. 20.49 Measured roll bearing loads in straightener. From Ref. 105.
Speed (m/min)
2
speed width
x - start-up o - cap-off
2.0 width (m)
1
1.5
0 x Load cell measurement (kN)
o
x o
x bottom roll 79 - load cell C
o
1.0
400
300 bottom roll 67 - load cell B 200 bottom roll 55 - load cell B 100
0 0
15
30 Time (hours)
45
60
75
Fig. 20.50 Bow and straightener load cell readings during a 75-hour period. From Ref. 105.
42
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
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
1
1
0.5
0
0.5
-0.5 deflection into strand 1 0 100 200 300 400 500 600 700 800 900 0 1000
Time (seconds)
Fig. 20.52 Thermal distortion of solid roll during speed transient. From Ref. 108.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
43
Casting Volume
650
Depth, mm @ 2.54 @ 25.4 @ 6.35 @ 76.2
520
Temperature (˚C)
390
260
130 start slowdown 0 0 100 200 strand stopped 300 400 resume motion 500 600 700 800 900
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.
44
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.5.6 Roll Gap, Alignment and Rotation
Rotating rolls, correct roll gaps and well-aligned rolls are prime requirements for good as-cast product quality and high machine productivity. Stuck rolls result in gouging in the product surface, which can sometimes go unnoticed. Bad roll gaps and poorly aligned rolls can yield product defects such as: • Deformed cross-sections. • Corner cracks. • Refill (mid-face) cracks. • Segregation. • Open (laminated) centers. Stuck rolls, incorrect roll gaps and poorly aligned rolls also result in increased machine maintenance costs and reduced machine utilization. It is therefore essential that roll alignments, roll gaps and roll rotations be checked and maintained regularly. These functions can be performed manually by using handheld gap tools and templates, but they are time-consuming and require trained operators. The better approach is to use an automatic device that can be attached to the starter bar chain and that measures roll gap, roll alignment and roll rotation as it is moved up and down the machine. Such a device is referred to as the “gap tool.”109–111 The early versions of gap tools were provided by casting machine builders, and they measured data on roll gap, roll alignment and stuck rolls. Recent tools can provide additional information on roll eccentricity and spray-water pattern.110 Two versions of gap tools are available. The most commonly used version is attached to the end of the starter bar chain, replacing the starter bar head, and is drawn through the machine during scheduled maintenance downturns or turnarounds. The other version, referred to as “on-board gap tool,” is permanently installed on the starter bar chain in between links and is drawn through the machine during each start-up. The type of gap tool chosen should be determined by considering the downtime available for gap runs and maintenance functions, and by weighing gap tool purchase and maintenance costs against productivity and quality gains that can be achieved through its use. Gap and alignment tools provide data on the cold status of the machine. However, experience shows that roll gaps and alignments can significantly change during casting because of roll and
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
45
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.
46
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
2.0 1.5 1.0 0.5 0 bow exit -0.5 gap closing 0 0.5 Gap change (mm) 1.0 1.5 gap closing 0 0.5 1.0 1.5 2.0 2.5 0 2 4 6 8 10 12 14 16 18 gap opening straightener inlet gap opening
Fig. 20.57 Cast speed and roll gap changes at bow exit and straightener inlet. From Ref. 112.
Cast speed (m/min)
Time (hours)
hot misalignments. The bender and the bow were mechanically linked to keep them aligned during casting. This and other measures taken to stabilize roll gaps and roll frames minimized slab cross-sectional deformation and corner cracking. Displacement transducers were used to understand the causes of mid-face cracks and centerline laminations in slabs of high-carbon and plate grades.112 The transducers were installed at the end of the bow and on the straightener section of the slab caster to investigate roll gap and frame stability during casting. Fig. 20.56 shows the installation of the transducers in the bow exit and straightener inlet areas of the caster. The measured gap changes at the last bow roll pair and the first straightener roll pair are shown in Fig. 20.57 as a function of cast speed and time. The gap changes at both rolls exceeded the allowable limit of ± 0.5 mm significantly, and showed big swings during speed transients. Several mechanical improvements were implemented to minimize the large gap changes and other excessive frame movements measured with the displacement transducers. One of the improvements was the removal of the top driven roll at the inlet of the straightener from hydraulic cylinders and installation on rigid girders. Fig. 20.58 shows the improvement in gap change profile at the bow exit and straightener inlet area of the machine as a result of the mechanical improvements. The dramatic decrease in slab open centerline incidence as a result of the study and numerous mechanical improvements is shown in Fig. 20.59. Displacement transducers were used by others to investigate dynamic roll movements and roll gap distortions in the straightener area,113 and also stability of gap tapering in the soft reduction area114 of slab casters.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
47
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
48
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
20.6.2 Length,Width and Weight
Three methods are commonly used to measure the length of the as-cast product before the cutting operation begins. The first is by tracking the position of the strand and the position of the torch machine. The strand position is tracked by means of speed measurements on one or more drive rolls in the caster containment section. The position of the torch machine is tracked by means of an encoder that is tied to a rack-and-pinion mechanism connected to the torch machine. From the strand and torch machine positions, the distance between the head of the strand and the torches is calculated. When that distance equals the ordered length, the torch machine is clamped onto the strand and the cutting operation is started. The second method entails the use of low-level lasers. The third length measurement method involves the use of a wheel on the torch machine that contacts the narrow surface of the strand and directly measures the distance between the head of the strand and the torch machine. The measurement wheel is water-cooled, and the wheel rotation is converted to linear distance by means of an encoder. When properly maintained and regularly calibrated, all three methods yield length within the design tolerance, which is usually ± 1% of the aim length. For example, manual length measurements on numerous hot slabs showed that only 2% of the lengths measured with the wheels were outside the length tolerance of ± 76.2 mm.115 The width of the slab in the runout is different from that at mold exit mostly because of thermal shrinkage and creep under ferrostatic pressure.115 Other contributors to width variation include incorrect mold setup and roll gap profile. Width variance from one slab to the next is largely dependent on grade, secondary water practice and cast speed. Five methods are used to determine slab
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
49
Casting Volume
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.
50
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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. 20.60 shows one of the earliest hot-slab surface inspection methods that is based on the eddy current principle and was first installed in Domnarvet, Sweden117,118 It consists of a robust watercooled sensor that is moved back and forth on a beam, which is mounted transverse to the casting direction. The sensor consists of a coil that is mounted in a housing, about 5 mm above the slab. On detection of a crack, a paint mark is made on the slab. The sensor scans only the upper surface of the slab. Inspection of the slab surface is done above the Curie temperature of 768°C. High-pressure water is used to descale the slab surface. Several grades of steel, including medium-carbon, silicon, boron and HSLA steels, were tested on-line. The system can detect longitudinal, transverse and star cracks that are deeper than 2 mm. Reportedly, it also detects scabs, slag spots and doubleskin effects. Fig. 20.61 shows typical output signal as a function of average crack depth. In a test of 200 slabs, agreement between manual and on-line inspection was 95%. Some of the other eddy current devices developed for hot-surface inspection are described in Refs. 119–125. The features that are common to most of these eddy current systems are that the slab surface temperature must be above the Curie point, the surface must be descaled, and only the top surface can be inspected. The system described in Refs. 124 and 125 is used for the detection of transverse edge cracks on each side and is controlled by a robot. The slab surface temperature must be below 500°C. System reliability is better than 99% in detecting corner cracks that are deeper than 2 mm. Optical surface inspection systems employ industrial TV cameras (ITV), charge coupled devices (CCD), line-scan or TV cameras, laser scanning and photographic methods.126–132 Fig. 20.62 shows the system described in Ref. 126, which was installed on a slab caster to inspect both the top and bottom surfaces of slabs. Optical systems are suitable for the inspection of large surface areas but do not indicate crack depth and are not suitable for hairline or short cracks. Also, the surface must be properly cleaned, which is generally satisfied by high-pressure water systems that make it difficult to maintain uniform temperatures across slab width. Ultrasonic133 and electromagnetic-ultrasonic134 methods were developed and tested to detect internal defects, such as primary pipe, centerline lamination, refill cracks and segregation. In the ultrasonic method, water is employed as a coupling medium between a combination transmitter/ transducer head and the test slab. In the electromagnetic/ultrasonic method, transmitting and
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
51
Casting Volume
video signal
signal processor
data processor
process computer magnetic disk
VTR
hard copy unit CCD
floppy disk line printer CRT console slab
surface pre-conditioning machine
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.
References
1. A.T. Etienne and P.H. Dauby, 4th International Iron and Steel Congress (London: The Metals Society, 1982), 11.1–11.14. 2. W. Irving, 4th International Iron and Steel Congress (London: The Metals Society, 1982), 13.1–13.16. 3. M.R. Ozgu, Canadian Metall. Quarterly, 35:3 (1996): 199–223. 4. H-J. Ehrenberg et al., Int. Conf. Metall., Stahl u. Eisen (1987), 149–159. 5. R. Steffen, et al., Int. Conf. Metall., Stahl u. Eisen (1987), 97–118.
52
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52.
Sumitomo Metal Industries, Ltd, Trans. ISIJ, 26 (1986): 590. T. Itoh et al., Preprints for the 102nd ISIJ Meeting, Part III, 22:3 (1981): B90. D.J. Trotter et al., 74th ISS Steelmaking Conf. Proc. (1991), 743–746. D.J. Idstein et al., 77th ISS Steelmaking Conf. Proc. (1994), 545–549. P.H. Dauby et al., 9th ISS Process Technology Conf. Proc. (1990), 33–39. P. Andrzejewski and K-U. Kohler, 9th ISS Process Technology Conf. Proc. (1990), 41–42. M.C.M. Cornelissen et al., 9th ISS Process Technology Conf. Proc. (1990), 95–99. D.I. Walker et al., 9th ISS Process Technology Conf. Proc. (1990), 3–12. A. Zeewy et al., 9th ISS Process Technology Conf. Proc. (1990), 13–16. Y. Maruki et al., Trans. ISIJ, 28 (1988). T.J. Russo and R.M. Phillippi, 73rd ISS Steelmaking Conf. Proc. (1990), 237–246. M. Okabe et al., Nippon Steel Technical Report, 49 (1991): 23–28. G.T. Moulden et al., 77th ISS Steelmaking Conf. Proc. (1994), 265–270. D. Bolger and P. Krause, 80th ISS Steelmaking Conf. Proc. (1997), 297–305. D. Bolger and P. Krause, Proc. of 3rd European Conf. on Continuous Casting (1998), 687–697. P.W. Manos and R.T. McGuire, Steel Times (1989), 610–613. “Thin Section Casting Program—Final Report,” Vol. II of Report Submitted by U.S. Steel/Bethlehem Steel to the DOE under Contract No. FC07-84ID12545, 267–282. K. Kawakami et al., NKK Technical Report, Overseas, 36 (1982): 26–41. A. Delhalle et al., ISS Trans., 6 (1985): 69–80. J.K. Brimacombe et al., 6th International Iron and Steel Congress Proc., Nagoya, Japan (1990), 246–274. S.G. Thornton and N.S. Hunter, 73rd ISS Steelmaking Conf. Proc. (1990), 261–274. J-P. Birat et al., 74th ISS Steelmaking Conf. Proc. (1991), 39–50. J.K. Brimacombe and I.V. Samarasekera, 74th ISS Steelmaking Conf. Proc. (1991), 189–196. W.H. Emling and S. Dawson, 74th ISS Steelmaking Conf. Proc. (1991), 197–217. M.R. Ozgu et al., Proc. of 1st European Conf. on Continuous Casting (1991), 1.73–1.82. D.E. Humphreys et al., Proc. of 1st European Conf. on Continuous Casting (1991), 1.529–1.540. D. Steward et al., 79th ISS Steelmaking Conf. Proc. (1996), 207–214. R.B. Mahapatra et al., Proc. of 1st European Conf. on Continuous Casting (1991), 1.541–1.556. M.R. Ozgu and B. Kocatulum, 76th ISS Steelmaking Conf. Proc. (1993), 301–308. H.L. Gilles, 76th ISS Steelmaking Conf. Proc. (1993), 315–336. J.S. Jenkins et al., 77th ISS Steelmaking Conf. Proc. (1994), 337–345. S.L. Kang et al., 77th ISS Steelmaking Conf. Proc. (1994), 347–356. H.L. Gilles et al., 9th ISS Process Technology Conf. Proc. (1990), 123–138. M.R. Ozgu and M.B. Perrine, Proc. of 2nd European Continuous Casting Conf. (1994), 130–134. N.S. Hunter et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 289–300. A. Cristallini et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 327–343. T. Wada et al., 70th ISS Steelmaking Conf. Proc. (1987), 197–204. K. Sano et al., Nippon Kokan Technical Report, Overseas, 30 (1980): 24–33. H. Matsunaga et al., Nippon Steel Technical Report, 23 (1984): 55–67. H. Kato and M. Yamasita, IFAC Automation in Mining, Mineral and Metal Processing, Tokyo, Japan (1986), 253–258. A. Matsushita et al., Trans. ISIJ, 26 (1986): 313. A. Matsushita et al., Trans ISIJ, 28 (1988): 531–534. J.M. Ritter et al., Rev. Metall., Cah. Inf. Tech., 87:9 (1990): 761–769. J. Kubota et al., 6th Intl. Iron and Steel Congress Proc., Nagoya, Japan (1990), 356–363. J. Kubota et al., 74th ISS Steelmaking Conf. Proc. (1991), 233–241. G. Bocher et al., Metall. Plant and Technology Intl., 3 (1993): 50–55. H. Okita et al., 76th ISS Steelmaking Conf. Proc. (1993), 237–243.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
53
Casting Volume
53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103.
54
M. Washio et al., ATS Steelmaking Conf. Proc. (1992), 507–512. M. Zeze et al., 76th Steelmaking Conf. Proc. (1993), 267–272. G. Bocher et al., Proc. of 2nd European Continuous Casting Conf. (1994), 103–108. D.W. van der Plas et al., Proc. of 2nd European Continuous Casting Conf. (1994), 109–118. J. Fukuda et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 437–445. A.A. Kamperman et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 465–474. Y.K. Park et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 961–970. S. Yuhara et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 989–992. E. Herrmann, Handbook Les Stranggiessens, Alluminum, Verlag, Dusseldorf (1955). G.J. McManus, Iron Age, 224:4 (1951): MP7–9, MP11. M.M. Wolf, 74th ISS Steelmaking Conf. Proc. (1991), 51–71. J.P. Radot et al., Rev. Metall., Cah. Inf. Tech., 1 (1988): 71–79, 85. A. Shirayamo et al., NKK Technical Report, Overseas, 52 (1988): 1–7. M. Espenhahn et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 891–900. E. Becker et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 901–922. M. Jauhola et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 951–960. H.L. Gilles and B.R. Forman, 74th ISS Steelmaking Conf. Proc. (1991), 219–232. H.L. Gilles et al., 75th ISS Steelmaking Conf. Proc. (1992), 911–925. B. Mairy et al., 9th ISS Process Technology Conf. Proc. (1990), 73–81. B. Mairy et al., Proc. of 2nd Continuous Casting Conf., London (1985), 59.1–59.8. M. Kiss et al., 76th ISS Steelmaking Conf. Proc. (1993), 383–394. R.W. Ramirez, The FFT, Fundamentals and Concepts (Prentice-Hall, 1985). M. Nogues et al., Proc. of 2nd European Continuous Casting Conf. (1994), 119–125. D.A. Barnard et al., U.S. Patent No. 3,047,915 (1962). M. Komatsu et al., Trans. ISIJ, B-86, 23 (1983). F. Slamar, U.S. Patent No. 3,893,502 (1975). Y. Nakamori et al., Trans. ISIJ, 22 (1982). H.B. Osborn Jr., U.S. Patent No. 2,824,346 (1958). B. Mairy et al., Proc. 1st European Conf. Continuous Casting, Florence, Italy (1991), 1.225–1.234. Y. Nakamori et al., Nippon Steel Technical Report, 34 (1987): 53–61. B.G. Thomas, 74th ISS Steelmaking Conf. Proc. (1991), 105–118. B.G. Thomas et al., 80th ISS Steelmaking Conf. Proc. (1997), 183–201. K.-U. Kohler et.al., 78th ISS Steelmaking Conf. Proc. (1995). D. Gotthelf et al., Proc. of 3rd European Conf. on Continuous Casting (1998), 825–833. Van der Plas et al., Proc. of 2nd European Conf. on Continuous Casting (1994), 109–118. K. Kariya et al., 77th ISS Steelmaking Conf. Proc. (1994), 53–58. K. Noro et al., 13th PTD Continuous Casting Conf. Proc. (1995). M. Iguchi et al., ISIJ Intl., 36 (1993): 190–193. S.G. Hibbins and J.K. Brimacombe, Continuous Casting, Vol. 2, (Warrendale, Pa.: Iron and Steel Society, 1984), 139–151. K. Tsuboi et al., Nippon Kokan Technical Report, Overseas, 36 (1982): 90–105. D.H. Kindt et al., 75th ISS Steelmaking Conf. Proc. (1992), 557–562. Y. Sugitani et al., 108th ISIJ Meeting (Oct. 1984), Lecture No. S898. K. Wunnenberg, Stahl u. Eisen, 98:6 (1978): 254–259. R.B. Mahapatra et al., 74th ISS Steelmaking Conf. Proc. (1991). S.R. Story et al., 76th ISS Steelmaking Conf. Proc. (1993), 405–413. J. Schade et al., 77th ISS Steelmaking Conf. Proc. (1994), 279–295. J. Gancarz et al., 74th ISS Steelmaking Conf. Proc. (1991), 15–22. K.D. Ives and R.V. Branion, 9th ISS Process Technology Conf. Proc. (1990), 61–68. P.W. Manos and C.Enstrom, 9th ISS Process Technology Conf. Proc. (1990), 139–150. R. Cornish and C. Baharis, Iron and Steelmaker (1991), 32–33. M. Yamada et al., 71st ISS Steelmaking Conf. Proc. (1988), 75–78.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Instrumentation
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.
Copyright © 2003, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
55
