The objective of this research is to accurately predict and control temperature in real time during the continuous casting of large, semi-finished steel shapes. The approach is to create a fast, accurate transient computer model of heat transfer during the solidification process that serves as a "software sensor", calibrated in real time through online temperature measurements to provide feedback to a control system, based on algorithms which will be designed specifically for this class of problem. The new software system will continuously read in operating conditions and mold temperatures and continuously adjust the spray-water flow rates in the secondary cooling zone of the caster, in order to maintain the desired temperature profile throughout the steel. This profile will be set by steel plant engineers, in order to minimize the formation of cracks and other defects. The system will be calibrated using thermocouple and optical temperature sensors, tested and implemented at an operating U.S. thin slab caster.
This project is important because 96% of the 100 million tons of steel produced in the U.S. each year is continuously cast, and the fraction produced by the new high-speed thin-slab casting process grows every year. This process experiences many defects caused by undesired temperature variations during spray cooling, which are unavoidable using current control systems. Conventional feedback control cannot be used because temperature sensors are too inaccurate and expensive. The model-based predictive control system proposed here must overcome many challenges, including the high speed of the process and increased relative importance of mold solidification. This project will directly benefit the steel industry, allowing it to become more competitive by increasing steel quality and lowering rejects. It will augment the research efforts of the Continuous Casting Consortium at the University of Illinois, which also helps to disseminate new knowledge to its member companies in this industry. In addition, improved understanding of the modeling and control issues will broadly benefit other manufacturing processes. The research will educate students who will take these new technologies into industry.
