Efficient Generation of Models of Fed-Batch Fermentations for Process Design and Control. Issue 7 (2016)

• Record Type: Journal Article
• Authors: Hebing, Lukas; Neymann, Tobias; Thüte, Tobias; Jockwer, Alexander; Engell, Sebastian
• Is Part Of: IFAC-PapersOnLine. Volume 49:Issue 7(2016)
• Page Start: 621
• Page End: 626
• Publication Date: 2016
• Subjects: Elementary Modes -- Macroscopic Modelling -- Dynamic Metabolic Flux Analysis -- Parameter Estimation; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2016.07.237
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: The paper presents a fast and efficient approach to the generation of process models for fed-batch fermentation processes. Elementary Modes (EM) with kinetic terms are used as macro-react ions in the model. A method based upon dynamic metabolic flux analysis (DMFA) is used to find an optimal selection of EM based on concentration measurements instead of reaction rates that must be calculated from the measurements. A multi-objective genetic algorithm is used to investigate the tradeoff between the number of EM and the difference of the estimated concentration profiles from the available measurements. Thus the number of macro-reactions for the model is kept small but the measured data are described sufficiently well by the selected EM. The influence of the process conditions on each macro-reaction rate are identified by a statistical analysis and kinetic expressions are fitted to the estimated evolutions by solving multiple simple optimization problems instead of a multi-dimensional large-scale nonlinear parameter optimization problem. The resulting process model provides a good fit of the measurement data. Provided that the data is generated by a systematic approach, e.g., using a design of experiments strategy, the modelling procedure provides models of adequate complexity that are valid within the so-called design space of the process.