Adaptive Optimal Control of Continuous Aqueous Two-Phase Flotation (ATPF)
Professor Dr. Moritz Diehl
Institut für Mikrosystemtechnik (IMTEK)
Prof. Dr.-Ing Hermann Nirschl
Karlsruher Institut für Technologie (KIT)
Institut für Mechanische Verfahrenstechnik und Mechanik
Arbeitsgruppe Verfahrenstechnische Maschinen
Active ingredients, like mRNA vaccines, are encapsulated within liposomes or liquid nanoparticles. Thereby, phospholipids form an outer shell around the active ingredient and ensure the transport into the human body. Specific modifications of the amphiphilic phospholipids by enzymes, for example the cleavage of fatty acid from the hydrophobic tail allow one to stabilize different active ingredients. There are various phospholipases, each of which hydrolyzes phospholipids at different sites and thus specifically alters their properties for an improved quality. Phospholipases are technical enzymes that play a crucial role as biocatalysts in many industrial applications. Their unique properties are used not only in the formulation of pharmaceutics but also in cosmetics and food production. This leads to a large demand for phospholipases, which can only be met by microbial fermentation and efficient downstream processing to recover the enzymes from the fermentation broth. The aim of this project is therefore to convert continuous aqueous two-phase flotation (ATPF) into an automatically controlled process for phospholipase recovery and purification that can be easily transferred to other biotechnological products from complex biosuspensions. The experimental focus is being handled by the Institute of Mechanical Process Engineering at the Karlsruhe Institute of Technology (Prof. Hermann Nirschl), including the establishment of suitable online measurement methods to determine the concentration and activity of the phospholipases. In close cooperation with the Institute of Microsystems Engineering of the University of Freiburg (Prof. Moritz Diehl) a new system model of the continuous ATPF process is created. The system model enables two-stage nonlinear model predictive control (NMPC), which is developed in Freiburg and combined with an estimation method capable of learning zero- and first-order model corrections online to ensure optimal steady-state operation under various process conditions. At the end of the first funding period, a closed-loop control system, that is able to adapt the ATPF model and to react to disturbances based on online measurements, shall be available on a laboratory scale. In the second funding period, the further separation of phospholipases from the concentrated enzyme solution by ultracentrifugation will be integrated into the controlled process chain.