AISI/DOE Advanced Process Control Program Vol. 1 of 6: Optical Sensors and Controls for Improved Basic Oxygen Furnace Operations

The development of an optical sensor for basic oxygen furnace (BOF) off-gas composition and temperature in this Advanced Process Control project has been a laboratory spectroscopic method evolve into a pre-commercialization prototype sensor system. The sensor simultaneously detects an infrared tunable diode laser ITDL beam transmitted through the process off-gas directly above the furnace mouth, and the infrared greybody emission from the particulate-laden off-gas stream. Following developmental laboratory and field-testing, the sensor prototype was successfully tested in four long-term field trials at Bethlehem Steel's Sparrows Point plant in Baltimore, MD> The resulting optical data were analyzed and reveal correlations with four important process variables: (1) bath turndown temperature; (2) carbon monoxide post-combustion control; (2) bath carbon concentration; and (4) furnace slopping behavior. The optical sensor measurement of the off-gas temperature is modestly correlated with bath turndown temperature. A detailed regression analysis of over 200 heats suggests that a dynamic control level of +25 Degree F can be attained with a stand-alone laser-based optical sensor. The ability to track off-gas temperatures to control post-combustion lance practice is also demonstrated, and may be of great use in optimizing post-combustion efficiency in electric furnace steelmaking operations. In addition to the laser-based absorption spectroscopy data collected by this sensor, a concurrent signal generated by greybody emission from the particle-laden off-gas was collected and analyzed. A detailed regression analysis shows an excellent correlation of a single variable with final bath turndown carbon concentration. Extended field trials in 1998 and early 1999 show a response range from below 0.03% to a least 0.15% carbon concentration with a precision of +0.0007%. Finally, a strong correlation between prolonged drops in the off-gas emission signal and furnace slopping events was observed. A simple computer algorithm was written that successfully predicts furnace slopping for 90% of the heats observed; over 80% are predicted with at least a 30-second warning prior to the initial slopping events