Using operational data of a commercial plant and available simulator of chemical processes, the application of neural networks to the estimation of parameters of the model of the chemical process kinetics has been tested. It has been possible to quickly evaluate the usefulness of the commercial unit revamping without long lasting and expensive research program.