Modeling of Separation of Ethylene Glycol-Water Mixtures

Abstract

The separation of ethylene glycol (EG)-water mixtures in a pervaporation system was performed in a wide range of percentages. These tests were performed at different temperatures from 60 to 80°C, using a dense cross-linked polyvinyl alcohol (PVA) polymer membrane. The authors in this research analyzed the amount of water- ethylene glycol separation in the pervaporation process by means of crosslinked PVA dense membrane with the help of Artificial Neural Network. The pervaporation process can be applied for separation of many liquids. Due to azeotropics with water, ethylene glycol has purity problems. In this research, the output information of the simulated model used to evaluate the experimental data and the results was analyzed. Modeling was performed by Multi Layers Perceptron (MLP) neural network feed forward. In this method of Propagation learning algorithm and Levenberg-Marquardt function include of 2 inputs and outputs were implemented. The graphs of the error percentage for the actual values of the flux outputs were compared to the achieved values from modeling through crosslinked PVA dense membrane for evaluating the efficiency of pervaporation process in separation of ethylene glycol from water. Finally, the graphs were drawn. The results obtained from comparing the experimental results of separation of ethylene glycol-water mixtures with the results obtained from the simulation data by the neural network model showed that the neural network models with the least errors were able to predict the experimental results.