Near-IR detectors show promise as a low-cost option for detecting the presence of slag in a stream of liquid steel poured from a basic oxygen furnace.
G. Raymond Peacock
A principal goal of most manufacturing facilities is to avoid waste. Such is the case with steel production. After processing in a basic oxygen furnace (BOF), molten material is poured into a transfer vessel or ladle. The goal is to transfer as much steel as possible, while minimizing the slag entering the ladle (see image sequence below). In the past, this has been difficult because an operator would usually control the pouring manually, basing judgements about slag on visual observation of the steel stream. Because the stream, with a temperature of about 1650°C (3000°F), is very bright, variations in brightness were often hard to discern, even with the use of filtered glasses.
In the 1980s, infrared (IR) thermal imaging was identified as an alternative to manual inspection of the melt stream. At the time, there were questions concerning reliability and ruggedness of such systems when used in a BOF shop operating 24 hours a day. There also were problems related to long sighting distances, a nearly constant shower of fine dust particles, smoke and dust in the sight path, and more.
In the last few years, the situation has changed radically with the development of focal-plane-array (FPA) cameras with better mean times between failure (MTBF) and the resulting enhanced reliability found in the imaging systems that incorporate them.1 Several companies now offer measurement systems that integrate thermal imaging with video archiving of liquid-steel pours and provide regular reports of slag caught in the ladle.
Determining which imaging system best fits a specific application can still be a challenge, though, especially when considering the trade-offs between measurement capability, functionality, and life-cycle costs. In many cases, the equipment typically used is still either a short- or long-wavelength passband device (3 to 5 µm and 8 to 12 or 14 µm, respectively). Since the processing temperatures of the melt are quite high, however, near-IR-wavelength (0.7-1.0 µm) equipment, which is restricted to operations with temperatures exceeding 600°C, also becomes a viable option.
Exploring equipment options
IR imaging works in this application because there is a significant difference between the apparent temperatures of slag and molten steel (see Fig. 1). A typical installed system will be required to operate continuously with at least two cameras and a display at each operator's pouring station (usually two in a BOF operation). The end user needs to be aware that the magnitude of relative temperature changes will be illustrated differently depending on the thermal imager used (see Fig. 2). The apparent differences in temperature between steel and slag, for example, appear larger with the long-wavelength systems. Even small changes in sight-path transmission will show up as large relative temperature changes.
Imaging devices with the shortest wavelength spread also show significant temperature differences between steel and slag, but these changes are relatively unaffected by sight-path changes. Ultimately, system selection will still come down to an evaluation of costs, especially upfront ones, which can exceed $200,000.
Typical detector elements for equipment operating in the short-wavelength passband include indium antimonide, lead selenide, and other special photovoltaic sensors. Early issues related to such equipment included cooling requirements and the fact that optical scanning devices used crossed scanning systems and complex electronics to create a two-dimensional (2-D) thermal image with one detector. Linear detector arrays helped reduce the complexity of the scanning devices. Two-dimensional FPA detectors have since improved the situation further. Typical detector arrays in platinum silicide have 256 x 256 elements, fewer moving parts, and reduced cooling requirements.
Cooling also was an issue with the long-wavelength passband devices based on mercury cadmium telluride point detectors and detector arrays. Equipment now includes microbolometers with arrays currently near 320 x 240 elements, with larger-resolution systems expected in the future. Although some cooling is still required with many of the 2-D FPAs now available, it is not as much of an issue as it was.
The good news is that both short- and long-wavelength passband systems now offer much higher reliability and MTBF ratings. There are still cost issues related to optics, though, which must be custom-made from IR-transmitting materials such as silicon and germanium. One reason for higher fabrication costs involves the care that must be taken to ensure good imaging properties because of the materials' relatively high indexes of refraction.
The cost of optics is one area that could give near-IR wavelength devices an edge over the competition in certain applications (see table).
In the mid-1990s, the Australian firm BHP Steel developed a practical near-IR thermal imaging system based on a silicon charge-coupled-device (CCD) FPA detector array. Applications for the prototype equipment included the detection of iron in a slag stream runner from a blast furnace. The device provided a good detection level because the blast-furnace slag had an emissivity some 50% higher than the liquid iron (very similar to the measuring situation for BOFs). Other potential applications included hot-rolling and continuous-casting operations.
While CCDs may have some limitations related to pixel-to-pixel uniformity and crosstalk between pixels, they are relatively low-cost compared to equipment required for devices operating at the other wavelengths. Another option, based on a General Electric Co. (Fairfield, CT) patent, was designed to overcome these limitations with the use of a silicon charge-injection-device (CID) array. The thermal imaging camera, which has been commercialized by Mikon Instrument Co. (Oakland, NJ), features uniform, long-life calibration; a MTBF of 10,000 to 100,000 hours, and a resolution of at least 640 x 480 elements. In essence, though, the choice of passband for a CID or CCD sensor array mainly depends on filter selection. This is because silicon FPA sensors are based on photojunctions in silicon materials, with the wavelength passband covering both the visible and the near-IR (0.4 to 1.1 µm).
Other near-IR equipment may loom on the horizon. Sensors Unlimited (Princeton, NJ), for example, has developed a 0.9- to 1.7-µm indium gallium arsenide 320 x 256-pixel camera, but has not yet packaged it for heavy-duty industrial use.
Advantages of near-IR imaging
Near-IR thermal imaging should be considered a viable option for slag detection in a BOF operation for more than just the relatively low cost of the equipment compared to short- and long-wavelength passband devices. Resolution imaging capability is high quality (640 x 480 elements). There is a high MTBF range. Systems use standard glass optical lenses, and no detector cooling is required. In addition, shop-hardened equipment is available, with more options expected in the future. Maintenance costs of equipment also will be relatively low compared to other options.
In addition, there also is less of a possibility that imaging-system manufacturers will have to deal with issues related to patent infringement. Bethlehem Steel Co. (Bethlehem, PA), for example, has been assigned a patent related to the use of 8- to 14-µm-wavelength imaging systems to detect slag. Claims such as this could add licensing costs to hardware solutions in the long and possibly the short thermal imaging passbands.
- G. R. Peacock, Proc. Thermosense XXII 4020, 50, SPIE, Bellingham, WA (April 2000).
G. RAYMOND PEACOCK is a senior staff engineer in the process-analysis group at LTV Steel Technology Center, LTV Steel Co., Independence, OH; e-mail: [email protected].