Parametric study and control of a pressure swing adsorption process to separate the water-ethanol mixture under disturbances

Abstract

Dehydration is a necessary step in the conversion chain of fuel grade ethanol, whether biochemical or thermochemical methods are used for synthesizing bioethanol from biomass feedstocks. This paper aims to propose a controlled configuration for a Pressure Swing adsorption process to separate a preconcentrated ethanol-water mixture. The closed-loop system is co-simulated in the Aspen Adsorption-Matlab Simulink environment and is composed of the following elements: A virtual plant predicts the transient cyclic behavior of the PSA process using a nonlinear rigorous mathematical model with distributed parameters. A novel nonlinear reduced control-oriented model with a Hammerstein-Wiener structure improves the fit of the plant dynamics compared to classical linear and Hammerstein structures and allows straightforward controller designing. A fundamental element is a (Fuzzy PD+I and Optimal Model-based Predictive) control law calculated with the linear dynamic part of the reduced nonlinear model, stated in the standard state space form. As a result, the process performance is observed without control and using the controllers (Optimal MPC and Fuzzy PD+I), evaluated by perturbation analysis, it was observed that the controllers are able to attenuate and maintain purity 99% (by weight) ethanol within the range allowed by international standards to be used as fuel, likewise, the time (time optimization) was reduced to reach the new steady state cyclic without having to adjust the nominal starting parameters again. Another result obtained when using the controllers was that, in the presence of combined disturbances, the recovery was maintained at a value of 73%.