Incremental Identification of Reaction and Transport Models

The lack of predictive mathematical models still impedes the routine application of model-based techniques to product and process design on the one and control and operations on the other hand. Furthermore, mechanistic models are indispensible for a better understanding of observed phenomena in engineered systems across the various length and time scales. Because of a lack of ab initio prediction models, the identification of models from informative experiments is mandatory.

This lecture will present a unifying framework for the identification of models covering both, the discovery of and the discrimination between alternative model structures as well as the estimation of valid parameters. The suggeted incremental identification strategy builds on a decomposition of the identification problem into transparent steps which derives from the structure inherent to all kind of chemical process models. Incremental identification is shown to facilitate proper control on the identification process, to clearly reveal deficiencies in the information content of the experimental data and to reduce the engineering and computational effort.

The identification strategy will be illustrated by a number of applications of increasing complexity including reaction kinetics in homogenous reaction systems, the intrinsic reaction kinetics in multi-phase enzymatic reactions, multi-component diffusion in liquids and transport phenomena in falling film flow systems.

The examples reveal the potential of the suggested strategy to properly link experimentation and modeling to obtain valid mechanistic models at minimum engineering and computational effort.