Reactive transport parameter estimation: Genetic algorithm vs. Monte carlo approach

• Published in: AIChE journal Vol. 55; no. 8; pp. 1959 - 1968
• Main Authors: Majdalani, Samer, Fahs, Marwan, Carrayrou, Jérôme, Ackerer, Philippe
• Format: Journal Article
• Language: English
• Published: Hoboken Wiley Subscription Services, Inc., A Wiley Company 01.08.2009; Wiley; American Institute of Chemical Engineers
• Subjects: Applied sciences; Chemical engineering; chemical equilibrium constants; Exact sciences and technology; genetic algorithm; Genetic algorithms; Monte Carlo simulation; Monte-Carlo; Parameter estimation; reactive transport; Reactors; surface complexation modeling; Monte Carlo method; Equilibrium constant; Parameter estimation; chemical equilibrium constants; Porous medium; reactive transport; Surface complexation model; Monte-Carlo; Modeling; Optimization; Transport process; Chemical equilibrium; surface complexation modeling; Chemical reaction; Genetic algorithm; Cost function; Breakthrough curve
• ISSN: 0001-1541; 1547-5905
• DOI: 10.1002/aic.11796

This article concerns reactive transport in porous media with an emphasis on the optimization of the chemical parameters. The transport of Cadmium (Cd) and tributyltin (TBT) in column experiments were used as test cases. The reactive transport model is described by a set of chemical reactions with equilibrium constants as the main adjustable parameters. As such a problem is highly nonlinear and can have multiple minima, global parameter estimation methods are more suitable than local gradient‐based methods. This article focuses on the application of a genetic algorithm (GA) in estimating chemical equilibrium parameters of a reactive transport model. The GA is capable of minimizing the difference between the measured and modeled breakthrough curves for both Cd and TBT. A comparison between GA and Monte‐Carlo approaches shows that the GA performance is better than the Monte‐Carlo, especially for a small number of evaluations of the cost function. The results of this study show that the use of GA to estimate the parameters of reactive transport models is promising. © 2009 American Institute of Chemical Engineers AIChE J, 2009