This paper presents a method for the identification of kinetic models for a wide variety of bstch/semibatch proceases related to the production of fine and specialty chemicals. The developed methodology is particulary suitable for those industrial pmcesses where a detailed understanding of pnxxs~ fundamentals (reaction mechanisms, proasz kinetics) is not available. It is also readily applicable to those prccess es where it is difficult to obtain time dependent cooantmtion 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 pm. The physical/chemical insight gained from a set of two-level multifactor experiments in primary operating variables (e.g. temperature, catalyst concentration, initial amaotrationa of key reactants, promotors, etc) is used to select an approximate functional form (viz power-law kinetics, L.angrniur-Hinshelwood, etc) for various rate equations in the kinetic model. A priori process knowledge and proass understanding gained from statistical analysis of the data collected is used to aid in the model identification. The final values of naction orders and the remaining kinetic parameters are determined by minimizing a prespccifled model-fitting funaion. 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 aC4)unt for any observed disaepencies. 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 pmcess related to the production of falty acid epoxides which are used as stabilizer/plastidzers for PVC resins.