Simone Brüning

• Complex bioprocesses require effective process control strategies. Especially model based control strategies show a high potential for the optimization of bioprocesses. The mathematical models in these algorithms must fulfill two major requirements. First, the model must be able to describe the mechanistic relationship between actuating variables like feeding rates and the main state variables, like biomass and product concentrations. Second, the parameterized model must be able to predict the time course of the state variables of a cultivation process. Model development appears to be a time consuming and laborious task. In order to reduce the time necessary for model development, a new structured compartment model was developed, which is easy to adapt to different cultivation processes, as will be shown in this contribution. The model developed is a new six-compartment model (6CM), which was used to describe the time course of batch and fed-batch cultivations of the bacteria Escherichia coli and Lactobacillus delbrueckii, the yeast Saccharomyces cerevisiae, the fungus Cyathus striatus and the cell line hybridoma. The model is divided into the compartments primary biomass Xpri, product forming biomass Xp, structural biomass Xs, inactive primary biomass Xi, inactive structural biomass Xsi and dead biomass Xd. The primary biomass represents all reactions related to cell growth. This compartment is reproduced autocatalytically by carbon and nitrogen source uptake. The product producing biomass is responsible for secondary metabolite or induced protein (e.g. GFP) production. The formation of this compartment is catalyzed by the primary biomass. The growth of the compartment of structural biomass is also catalyzed by the primary biomass. Xs consists of membranes, cell walls and complex polysaccharides. It is formed by utilization of a carbon sources. Primary biomass and structural biomass may be inactivated to form inactive primary and structural biomass and may be reactivated to represent the maintenance metabolism. Under certain process conditions the inactive biomass compartments will be converted into dead biomass. Major properties of the model are the following: 1. The substrate concentration influences its uptake rate and therefore the rates of all further reactions. 2. Substrates are used for biomass and for product formation as well as for energy generation via fermentation or energy generation via cellular respiration. 3. Uptake rates, inactivation and death rates as well as yield coefficients are functions of the state variables and are modulated by double sigmoidal functions. In order to adapt the model to a bioprocess a four-step modeling procedure was applied. First, a detailed process description was derived from experimental data and literature. Second, stoichiometric equations were formulated with respect to the metabolism under investigation. In the third step cause-effect relationship hypotheses, which were derived from process analysis, were used to identify the parameters of interest. Finally, these model parameters were estimated manually to adjust the model to the measured data. The model is able to describe the time courses of important state variables, such as the concentrations of carbon and nitrogen, intermediate products, and products of primary and secondary metabolism as well as induced products. It is capable of describing O2 uptake und CO2 production via exhaust gas composition (aerobic processes) or gas flow rates (anaerobic processes). The model was adapted to each cultivation process only by choosing different values for the model parameters. This reduces the modelling effort for a new process significantly.