Application of Time Series Analysis for the Prediction of Liquid Metal Height in Basic Oxygen Furnace
6 Pages Posted: 15 Aug 2019
Date Written: August 15, 2019
In basic oxygen steelmaking (BOF), lance profile during the blow is very important as it impacts the oxygen impingement, which results in variation of the oxidation reactions with impurities like carbon (C), silicon (Si), manganese (Mn), and phosphorus (P). Improper measurements or estimations of initial bath height of the lance impairs the steel quality as well as the vessel refractory lining. Hence, estimation of the initial bath height is very crucial for efficient steel making operations. In Tata Steel, Jamshedpur, a 3D Laser profile measurement system is employed for evaluating the inner profile of the refractory lining of the BOF vessel. In this investigation, we developed a mathematical model which takes the inner profile data as input and estimates the bath height of the hot metal. Further, a 3D geometric model using CATIA software is designed to compare the findings of the above mentioned mathematical model. Results of the mathematical model are in agreement with the values of the 3D geometric model. Mathematical model estimations are validated with two different data sets. First data set is obtained from dip-stick measurements and the second data set is obtained from sublance (a tool used in BOF) data estimations. In the dip-stick measurements, a wire rod is tied to the lance and is immersed into the liquid pool. From the mark on the wire rod, the bath height is measured. sublance, a mechanical lance system, evaluates the bath height using the temperature and oxygen activity differences between slag and hot metal. The prediction from the mathematical model showed good agreement with the two validation data sets. Further, a time series model has been developed using a total of 150 batches belonging to one campaign. The aim of the time series model is to predict the bath heights for the upcoming batches in the absence of any measurement data. The predictions of the time series model are in excellent agreement with the plant measurements.
Keywords: Basic oxygen furnace, Bath height estimation, Time series modelling, Laser profile measurement, Sublance, Lance profile
Suggested Citation: Suggested Citation