Title: Application of hybrid neural particle swarm optimisation algorithm to predict solubility of carbon dioxide in blended aqueous amine-based solvents

Authors: Reza Taherdangkoo; Mohammad Taherdangkoo

Addresses: Department of Petroleum Engineering, Omidiyeh Branch, Islamic Azad University, Omidiyeh, Iran ' Department of Artificial Intelligence, Tehran Business School, 2532 Valiasr Street, Tehran, Iran

Abstract: In this study, we proposed a new artificial neural network (ANN) trained with particle swarm optimisation (PSO) to predict solubility of CO2 in aqueous amine-based solvents over wide range of pressure, temperature, overall concentration, and MWa. The model was developed with mixtures that consisted of methyldiethanolamine (MDEA), monoethanolamine (MEA), diethanolamine (DEA), 2-amino-2-methyl-1-propanol (AMP), diisopropanolamine (DIPA), piperazine (PZ), triethanolamine (TEA), and tetramethylen sulfone (TMS). PSO is used to find best initial weights and biases of neural network. As input parameters, neural network considered the overall solute's concentration, temperature, CO2 pressure, and MWa. The PSO-ANN model was trained, and tested using 75%, and 25% of all experimental data points, respectively. The results show that the proposed model provides predictions in acceptable agreement with experimental data.

Keywords: blended amine solvents; carbon dioxide; CO2 solubility; artificial neural networks; ANNs; particle swarm optimisation; PSO; pressure; temperature; concentration; MWa.

DOI: 10.1504/IJSETA.2015.075637

International Journal of Software Engineering, Technology and Applications, 2015 Vol.1 No.2/3/4, pp.290 - 307

Received: 30 Mar 2015
Accepted: 30 Oct 2015

Published online: 30 Mar 2016 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article