Application of global optimization methods to evaluation of chemical reaction terms in reactive solute transport analysis

Two global optimization methods based on Genetic Algorithms (GA) and Shuffled Complex Evolution (SCE-UA) method were examined to evaluate reaction terms in a reactive solute transport model involving cation exchange reactions. Although the both algorithms can minimize the objective function after succession of generations, SCE-UA method appears superior to GA in that its performance depends less on the selection of the parameter set. Moreover, GA case tends to fail to find best chemical reaction terms. In both algorithms, global optimum sought four times for each step, with narrowed search range for the efficient use of the algorithms.