Efficient techniques for the optimization of complex chemical processes

Efficient and robust techniques for the optimization of complex chemical processes are presented. In particular, we address the solution of large, multistage optimal control problems for processes described by DAE models. Our boundary value problem approach (a simultaneous solution strategy) is based on a piecewise representation of the control functions and a multiple shooting discretization of the DAEs, combined with a specifically tailored SQP technique. The inherent problem structure is exploited on various levels, leading to an extremely efficient overall method. A reliable and easy-to-use implementation of this strategy is provided by the recently developed modular optimal control package MUSCOD-II. Two real-life applications of MUSCOD-II are discussed: 1. Optimal control of a fed-batch fermentation for the production of Nikkomycin. 2. Optimization of a hydrogen-air combustion process in a 1D steady-state tube reactor. The first application provides a realistic bioprocess optimization problem, and our method is shown to be so fast that it could be even used for the online optimization of such processes. The second problem demonstrates the suitability of our approach for the stiff differential equation models that are common in chemical kinetics applications. (orig.)SIGLEAvailable from TIB Hannover: RR 1606(96-40) / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekDEGerman