Adjustable-structure design for ternary distillation columns

In this work, the problem of jointly designing the estimator structure and the algorithm for ternary distillation columns is addressed. In particular, the choice of the estimation structure (measurements location and innovated states set) and the design of a complete or reduced algorithm with the possibility of changing the estimation structure during the column operation are considered. The proposed estimators are tested using experimental data from a 32-stage pilot column, finding that: (i) the structural decisions play a key role in the estimator performance, regardless of the kind of estimation algorithm employed, (ii) the best estimator behavior is obtained by injecting the temperature information over a few column states, and (ii) the same functioning is obtained with geometric estimation and EKF.