Title: Fitting Sovova's mass transfer model using an evolutionary algorithm and differential evolution

Authors: Dejan Hrncic, Marjan Mernik, Masa Knez Hrncic, Zeljko Knez

Addresses: Faculty of Electrical Engineering and Computer Science, University of Maribor, Smetanova ulica 17, 2000 Maribor, Slovenia. ' Faculty of Electrical Engineering and Computer Science, University of Maribor, Smetanova ulica 17, 2000 Maribor, Slovenia. ' Faculty of Chemistry and Chemical Engineering, University of Maribor, Smetanova ulica 17, 2000 Maribor, Slovenia. ' Faculty of Chemistry and Chemical Engineering, University of Maribor, Smetanova ulica 17, 2000 Maribor, Slovenia

Abstract: In chemical engineering, reliable models are necessary to reduce the cost of process design. An evolutionary algorithm with resizable population was used to estimate coefficients of Sovova|s mass transfer model and was compared with a global optimiser found in the literature and commonly used differential evolution algorithm. Comparison of the evolutionary algorithm to the global optimisation technique proved that the evolutionary algorithm is more robust, efficient, and significantly better than the global optimiser in regards to the deviation of the model from experimental data. It is also shown that the proposed evolutionary algorithm performed better than differential evolution algorithm.

Keywords: evolutionary algorithms; EAs; mass transfer; parameter estimation; Sovova model; differential evolution; chemical engineering; process design; modelling.

DOI: 10.1504/IJICA.2010.036811

International Journal of Innovative Computing and Applications, 2010 Vol.2 No.4, pp.237 - 243

Published online: 09 Nov 2010 *

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