Bilevel dynamic optimization of metabolic-genetic networks for bioprocess monitoring and control
Prof. Dr. Steffen Waldherr, KU Leuven/NL, CIT, Bio- & Chemical Systems Technology, Reactor Engineering and Safety Section
Monday, April 24th, 2017, 4:00 pm, Seminar Room 121, Building 58a
Invited by Jan Hasenauer
Models for metabolic networks are often formulated in terms of optimization problems where a biomass measure is optimized over the reaction fluxes under biophysical and stoichiometric constraints. In this talk, I first introduce a particular dynamic optimization based model that solves for the metabolic reaction fluxes, substrate concentrations, and biomass components as functions of time. Besides the standard biophysical constraints, it includes enzyme capacity constraints and biomass composition constraints. As will be seen from a simplified example network, this optimization is able to predict commonly observed metabolic adaptations, such as a diauxic switch with a preference ranking for different nutrients or re-utilization of waste products after depletion of the original substrate.
In the second step, I discuss the integration of such an optimization based models into common bioprocess engineering problems. When the engineering problem is also formulated as an optimization problem, typically with another objective and additional variables, one encounters a bilevel optimization problem. Specific applications to be discussed in this talk are in bioprocess design and state estimation.