Article Number: DRJEIT11164535
Vol. 8, pp. 11-20, 2021
Copyright © 2021
Author(s) retain the copyright of this article
Original Research Article
Prediction and optimization of vapor-liquid equilibrium (VLE) data for equimolar ethanol/water mixture using adaptive neuro-fuzzy inference system
This research work explores the calculation of vapor phase fraction (VPF) data for an equimolar mixture of ethanol/water system through applications of Artificial Neuro-Fuzzy Inference System (ANFIS). The calculation of VPF data by conventional thermodynamic methods is tedious and requires the determination of “constants” which is arbitrary in many ways hence the need to adopt ANFIS with its associative property and its ability to learn and recognize highly nonlinear and complex relationships. An empirical model was first developed to attempt predicting the highly nonlinear system. A Sugeno type ANFIS model having 5 layers with 9 hidden neurons representing the If then, the fuzzy logic rule was developed to predict the vapor phase fraction of the system. The inputs to the model were the temperature and pressure values simulated on Honeywell UNISIM® design software and the model was trained to calculate the mole fraction and vapor phase fraction of the system taken as outputs. The triangular membership function was used for the inputs and the constant membership function for the outputs. The ANFIS model returned R-square values of 0.9915 and 0.9981 for the mole fraction and molar flow rate respectively as against 0.7715 and 0.7733 respectively for the empirical model. The ANFIS model was further optimized using data-based particle swarm optimization to give a prediction of 89 mole% and flow rate of 2.11Kgmole/h at a temperature of 62.13oC and pressure of 55.92KPa for the distillate. It can therefore be concluded that the ANFIS model gives a superior predictive capability than conventional thermodynamic models for system evaluation and design. This work is a basis for the design of flash columns to substitute conventional distillation columns in ethanol recovery.
Keywords: Optimization, prediction, ethanol, mixture, artificial, fuzzy-inference