In the present paper a comparison between different parameter estimation procedures commonly used for the kinetic modeling of chemical reaction is performed, based on experimental measurements of the cyclohexane dehydrogenation to benzene. The obtained results show that, when the Arrhenius equation parameters are estimated from estimates of the rate constant taken at different temperatures, larger parameter uncertainties and correlations are obtained, particularly when the variances of the experimental measurements are not considered during the estimation process. It is also observed that an apparent kinetic compensation effect occurs when the experimental data are separated according to the inlet partial pressure and catalyst mass in the reactor, mainly due to the existing and unavoidable experimental uncertainties and parameter correlations. Additionally, it is shown that larger uncertainties and correlations are obtained when the parameter estimates are computed through the differential method, which can also lead to poorer model predictions of the experimental data. Finally, it is shown that the simultaneous one-step estimation of all model parameters through the integral method and considering the available experimental uncertainties can provide the most accurate parameter estimates, making use of mathematical expressions that describe how variances of the experimental measurements depend on the experimental conditions.
Read full abstract