Open Access Conference Proceedings

Estimating the elemental composition of scrap types used in Basic Oxygen Furnaces (BOF) and Electric Arc Furnaces (EAF) is fundamentally an ill-conditioned inverse problem. The composition of several scrap types must be recovered from a single scalar mass-balance measurement per heat, and the quality of any estimate depends critically on the condition number of the observation matrix formed by the scrap input masses. This paper systematically analyses how scrap-type usage frequency, charging diversity, and window length affect the condition number and explains why poor conditioning causes windowed least-squares methods to fail for rarely used scrap types or fixed charging recipes. We formulate scrap composition estimation as a linear Gaussian state-space model and show how the Kalman filter mitigates ill-conditioning and yields conservative and physically plausible estimates. We show on synthetic BOF data that the Kalman filter tracks time-varying composition more robustly than windowed non-negative least squares, particularly when the observation matrix is near-singular. The resulting scrap composition estimates can subsequently be used to support prediction of the elemental composition of the produced steel.

Keywords: EAF and BOF steelmaking, scrap composition, state-space model, Kalman filter

© This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.