Automated synthesis of multivariate QFT controller and pre-filter for a distillation column with multiple time delays

Abstract

In the refining and chemical industry, distillation is one of the prime processes and is a multivariable control problem, which involves interaction amongst several control loops and exhibit dead time in the dynamic behavior. The analysis and controller design of processes inheriting dead time is a challenging task, and the presence of plant uncertainties and disturbances makes it challenging to exert quality control over time. In this paper, a template and bounds free approach is explored for the simultaneous synthesis of quantitative feedback theory (QFT) based robust controller and pre-filter for Wood and Berry distillation column with multiple time delays. This presented method simplifies the QFT controller synthesis for MIMO systems and eliminates the need of ‘Schauder fixed point theory’ to express the m×m MIMO system into analogous m2 MISO plants before the QFT controller synthesis. Using the presented approach, an optimal MIMO QFT controller and pre-filter are synthesized in a single step, which else conventionally follows a long sequential design process. In the presented approach, the QFT controller design problem has been posed as an optimization problem and is solved using genetic algorithm. The work has been compared with the work already reported in literature. The designed multivariate QFT controller efficiently mitigate the effect of parametric uncertainties, satisfy the design requirements in a very tight envelope, minimize the interaction amongst the control loops and offers improved the disturbance rejection.