Please use this identifier to cite or link to this item:
|Type:||Artigo de periódico|
|Title:||Genetic Algorithms (Binary and Real Codes) for the Optimisation of a Fermentation Process for Butanol Production|
de Angelis, DD
|Abstract:||In this work, the capability of genetic algorithms (GAs) to optimise an alternative fermentation process for the production of biobutanol was assessed. The process consists of three interconnected units, as follows: fermentor, cell retention system (tangential microfiltration) and vacuum flash vessel (responsible for the continuous recovery of butanol from the broth). The dynamic behaviour of the process is described by a non-linear mathematical model with kinetic parameters determined experimentally, whose non-linearity makes the solution of the optimisation problem difficult through conventional algorithms, thus justifying the use of an evolutionary method based on the GAs. The objective of the optimisation was the search of the process inputs that maximises the productivity of butanol for a desired substrate conversion. The potential of binary and real coded genetic algorithms to solve the optimisation problem was assessed. The GA parameters were evaluated making use of the statistical technique of the factorial design in order to identify the most significant ones to the GAs response and to determine the values of the parameters that improve the GAs performance. With both GA codes similar solutions to the optimisation problem were obtained. However, in relation to computational time, the binary code outperformed the real code. The optimised process ran on concentrated sugar solution (140.7 g/l), reaching a high final butanol concentration (27.1 g/l) and high butanol productivity (9.0 g/l.h). The use of mathematical optimisers in the butanol fermentation is a novel approach and sums up the efforts of recent researches in turning the biobutanol industry commercially viable.|
|Editor:||Berkeley Electronic Press|
|Appears in Collections:||Unicamp - Artigos e Outros Documentos|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.