Simulación del proceso de separación de la mezcla acetonitrilo-agua mediante destilación extractiva en DWSIM y predicción de la concentración de acetonitrilo por RNA

Abstract

This research purpose dealt with simulating and validating a separation process of the acetonitrilewater mixture by extractive distillation in DWSIM, which serves as the basis for carrying out the design of the Artificial Neural Network (ANN) capable of predicting acetonitrile, ethylene glycol and water molar fractions. For the elaboration of the ANN, a database was generated from the simulation carried out in DWSIM; the database consisted of 100 pairs, with four entrances; feeding flow temperature, acetonitrile mole fraction at feeding, pressure of the preconcentration column (C1) and mole fraction of ethylene glycol. The ANN was designed in MATLAB software using nine hidden neurons and an algorithm Bayesian Regularization for testing, with a value of 3.0723e-05 Mean Square Error (MSE) and 0.99996 total regression coefficient. The network was validated through a statistical comparative analysis obtaining a 95% reliability. The simulation allowed to obtain 0.9830 acetonitrile in the distillate of the column two (extraction column), and in the distillate of column three (recovery) 0.9016 of water and 0.9737 of ethylene glycol are obtained at the bottom. Moreover, it should be mentioned that the recirculation of the products from the bottom of the column three towards column two was not performed; this proposal allowed to obtain results close to the reference research. It should be mentioned that the percentage errors are less than 10%, giving efficient results in the research. It is recommended to normalize the input values to ANN that are greater than one.