Speaker
Description
Steel recycling in electric arc furnaces (EAFs) is crucial to achieve environmental as well as economic sustainability of the steel industry. Although the EAFs impose a smaller environmental imprint in comparison to basic oxygen furnaces (BOFs), they still represent an energy-intensive process, with numerous possibilities for improvements. Digitalization and informatization of the steelmaking processes opened numerous possibilities in the field of production optimization using large amounts of data, where EAFs are no exception. In the following paper, a fuzzy-model-based approach for estimation of the bath temperature in an EAF is presented. As known, in most EAFs, disposable measuring probes are used to determine bath temperature prior to tapping. Since the EAF needs to be switched off during the measurement procedure, each measurement increases the losses and imposes a certain operational delay. Considering that several temperature samples are taken, i.e. 3 – 6 on average, the losses and delays are no longer neglectable. Furthermore, as the bath temperature between the measurements can only be roughly estimated by the operators, the control of the EAF cannot be as precise as it would be if the temperature was known. Therefore, the proposed methodology combines computational intelligence in soft sensor design and the operational measurements of all influential EAF inputs, to continuously estimate the bath temperature during the refining stage of the recycling process. The goals of the proposed approach are twofold, both leading to higher plant efficiency, i.e. first, to reduce the number of necessary temperature measurements, and thus decrease the losses and delays; second, to provide more information to the operators, and thus allow more precise EAF control. The results have shown high prediction accuracy of the proposed approach; thus, the developed methodology will be implemented in an industrial environment, running in parallel with the actual EAF process.