LAUSR

Abstract Supercritical extraction is the utilization of a fluid operating above its critical temperature and pressure as a solvent to remove certain compounds from a mixture. It is an innovative technology for which there are relatively few models describing the interaction between the phases. In particular, one of the cases less studied is the counter-current extraction of compounds from liquids. Consequently, model-based control of such process is not usually considered in the literature. This work compares the performance of different control schemes to perform the supercritical extraction of methyl oleate. A dynamic model of the system, consisting of partial-differential equations, is implemented as a virtual plant. The main objective is to evaluate potential gains of using input shaping in addition to a feedback loop. Input shaping is a control technique that converts a step perturbation in the set point into a series of pulses at specific times and with different amplitudes, with the goal of avoiding oscillations. The calculations of this series of pulses is performed based on predictions of output response of a linearized plant model to changes in the set point. Controller performance has improved after the implementation of input shaping, with a reduction of the integral of the absolute error by 36% and of the settling time by 18%, despite the mismatch between the linearized model used by the input shaper and the actual non linear dynamics of the system.