| Most nonlinear observer designs to date are either too difficult to apply to high-order distillation models, or are applied to simpler equimolar overflow models that neglect the energy balance. In this study, two computationally efficient nonlinear observers are developed that use rigorous multicomponent distillation models that include both dynamic mass and energy balances.; The first estimator, called a vapor-bypass observer (VBO), functions by adjusting the apparent vapor rate through the stages proportionally to a measurement error. An adjustable parameter is used to regulate the size of the observer gains, and tuning guidelines are presented to achieve robust performance. The second estimator, called a McCabe-Thiele observer (MITO), is based on an unsteady state McCabe-Thiele diagram and functions by dynamically adjusting the position of an operating line. Unlike the VBO, this observer does not require tuning.; The observers are stable over a broad range of operating conditions. This was verified by initializing the observers at various different states on a steady state manifold while holding the plant column at steady state. A more rigorous stability test using plant columns initialized at various dynamic states also indicated stable operation.; The primary measurements used in both observers are compositions, as either mole fractions or a single stage temperature. Level measurements in the reboiler and reflux drum can also be included to further improve performance. Both observers consistently provide better performance than open- loop observers. The VBO can outperform the MTO when tuned aggressively. However, when tuned for robustness, both the VBO and MTO will have equivalent performance. |
