Operational models for real time applications of nonlinear and moving boundary process distributed systems

Resumo

Thermal processes are widely employed in both Food and Biotechnology industries. Nevertheless, their operation and decision making is still ruled by rigid procedures which are difficult to adapt to sudden changes in the production conditions or to disturbances in the process caused by unforeseen events. Besides, it must be considered too, that in most of the cases, those operation policies have been not designed according to optimality principles, and therefore there exist opportunities to improve not only the process costs but also the final product quality. It is in this framework where the development and implementation of real-time oriented mathematical models, which would be employed as the core of computational tools and methods for optimal on-line control processes, constitutes an invaluable mechanism. Many are the advantages related to process modelling: from providing a virtual environment where new procedures and equipment can be evaluated to predicting the effects of failures or malfunctioning conditions on quality product. However the specific requirements of real-time tasks call for low-dimensional models aimed at reducing the computational times associated to its numerical solving. In response to this need, the concept of operational model emerges in the sense of a simplified while accurate representation of the system, based on first-principles, whose structure and dimensionality enable its role as the body of control and optimisation on-line schemes which would supply with the desired flexibility to the operation process. Thus, the objectives of this dissertation will be, in one hand, to provide with the insights of the methodology proposed, and in the other, to show the applicability and advantages of this integral modelling approach by means of examples consisting of real-time applications for different processes of interest in Food and/or Biotechnology industries.