Heat and Mass Transfer Research Journal (HMTRJ)

Research Article

Studies on Artificial Neural Network Based Prediction of Wall Temperature Profiles for Methanol-Water System in a Natural Circulation Thermosiphon Reboiler

M. A. Hakeem¹, M. Kamil²

¹Department of Chemical Engineering, AMU Aligarh -202 002, U.P., India

²Department of Petroleum Studies, AMU Aligarh -202 002, U.P., India

Submitted: September 11, 2017; Revised: November 29, 2017; Accepted: December 13, 2017

Abstract

The present study was undertaken to predict the temperature profiles for a binary system (methanol-water) at various operating conditions using artificial neural network in a natural circulation vertical tube thermosiphon reboiler. The heat flux values ranged from about 4.1 to 43.0 kW/ m2. The liquid submergence levels were maintained around 100, 75, 50 and 30%. Two main operating parameters namely heat flux and liquid submergence affecting the wall temperature profiles were considered as inputs, while the output parameter was temperature profiles. The network was then trained to predict the wall temperature profiles as outputs. A feed-forward back-propagation network was developed and trained using experimental data from the literature. It was observed that the predicted values are in very good agreement with the measured ones indicating that the developed model is fairly accurate and has the great ability for predicting the temperature profiles. If more exhaustive input data are fed: heat flux, submergence and mass percent then the capability of the network to predict the temperature profile would had been better. The predicted temperature profiles yielded the relative error of the order of 0.1% in majority of the cases.

Keywords

Artificial neural network; Temperature profiles; Natural circulation loop; Thermosiphon reboiler; Methanol-Water

Downlaod Full Text >>