Application of an extended Kalman filter to a binary distillation column model

Abstract One of the major obstacles for control of the distillation process is reliable online measurement of composition and liquid levels. This paper examines the feasibility of using an extended Kalman filter as a soft sensor for a binary distillation column model derived from first principles. The sensor estimates the composition, liquid level and pressure at various stages in the column via measurement of the column temperature profile, feed conditions and pressure at the top of the column.