Determination of Kinetic Parameter in a Unified Kinetic Model for the Photodegradation of Phenol by Using Nonlinear Regression and the Genetic Algorithm

Abstract

Abstract This study reports the kinetic parameter estimation in the photocatalytic degradation of phenol over different TiO2 catalysts by using the Genetic Algorithm (GA) and nonlinear regression. Reaction networks are based on a previously reported unified kinetic model (UKM) of the Langmuir–Hinshelwood type. Nonlinear least-squares fitting and GA are used to find the values for the kinetic constants. The computed parameters were found to predict experimental data for phenol photodegradation at different levels of concentrations. It is shown that both methods render close values for the kinetic constants. This suggests that UKM approach gives the global minimum and as a result, this method provides good and objective parameter estimates with low to moderate cross-correlation among kinetic constants and acceptable 95% Confidence Intervals (CIs). Global optimization by using GA requires extensive computer times of up to 5 minutes. Least square fitting provides the same results with computer times of seconds only. It is then concluded that the UKM approach effectively avoids overparameterization by finding the global optimum when optimizing the kinetic constants.