Intelligent Adaptive Systems for Optimal Energy Input in Steelmaking

40517 Neural Applications Corporation In the United States steel industry, the total annual electrical energy consumption by electric are furnaces (EAFs) is 16,000,000,000 kwh, at a cost of $600,000,000. Currently, the primary source of thermal energy in EAFs is the electric arc (65% of kwh), with additional energy input from oxy-fuel burners (5%), and other exothermic reactions (30%) that are supported by injecting oxygen into the furnace. Presently, energy input setpoint profiles are developed through trial and error, by simple linear algorithms, or are based on the experience of furnace operators. This project will develop an intelligent system, based on neural network technology, that optimizes the energy input from each major source, automatically adapting to changing operating conditions. In Phase I, an EAF will be outfitted with sensors to monitor furnace conditions, energy input, and furnace off-gas emissions so that more accurate neural-network models can be devised. In Phase II, a working prototype control system will be developed to optimally coordinate energy input from each major source. Anticipated Results/Potential Commercial Applications as described by the awardee: A 5% improvement in energy efficiency could save the US steel industry $50 million per year. Other benefits include greater flexibility to energy source availability and reduced emissions. Commercialization will be accelerated by targeting our existing customers for upgrades to their EAF installations.