Dynamic Modeling and Estimation of Carbon in a Basic Oxygen Furnace

Abstract

This paper describes an approach to dynamic modeling and on-line carbon estimation based on waste gas analysis in a basic oxygen furnace. As is well known, the basic oxygen process for steelmaking is highly stochastic in nature. There is, at present, no satisfactory model for predicting the decarbonization rate of the process. The first part of this paper considers approaches to on-line dynamic modeling. A two-stage dynamic model is developed using time-series analysis and empirical information on the decarbonization rate. This model is put in a state-vector form which is ideal for online estimation and control. The second part of the paper applies linear and nonlinear filtering to estimate the carbon and the decarbonization rate of the process during each heat. Since there is a time delay between the bath conditions and the measurements, the state of the system must be predicted. Two different methods for prediction are given. One is based on the present estimate of the carbon in the bath, and the other is based on the present estimate of the decarbonization rate of the bath. A combination of the two schemes is shown to provide better results than those achieved so far in practice. A total of 175 heats from a 220-ton BOF are processed and analyzed, and the results are discussed.