Abstract

This paper presents a method for the identification of kinetic models for a wide variety of batch/semibatch processes related to the production of fine and specialty chemicals. The developed methodology is particulary suitable for those industrial processes where a detailed understanding of process fundamentals (reaction mechanisms, process kinetics) is not available. It is also readily applicable to those processes where it is difficult to obtain time dependent concentration data.The modeling effort is aimed at developing a low-order, nonlinear “Tendency Model” which is descriptive of the qualitative and approximate quantitative behavior of the overall process. the physical/chemical insight gained from a set of two-level multifactor experiments in primary operating variables (e.g. temperature, catalyst concentration, initial concentrations of key reactants, promotors, etc) is used to select an approximate functional form (viz. power-law kinetics, Langmiur-Hinshelwood, etc.) for various rate equations in the kinetic model. A priori process knowledge and process understanding gained from statistical analysis of the data collected is used to aid in the model identification. The final values of reaction orders and the remaining kinetic parameters are determined by minimizing a prespecified model-fitting function.A factorial analysis of the model's prediction for process responses is used to determine the accuracy of the model. An inaccurate model is modified to account for any observed discrepencies. The developed “Tendency Model” is used for the optimization of the process in an evolutionary manner. As the process operates and additional data becomes available, the model parameters are updated using the information on the sensitivity of the plant-model mismatch with respect to various reaction orders. The developed methodology is illustrated by means of experimental data for an example process related to the production of fatty acid epoxides which are used as stabilizer/plasticizers for PVC resins.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call