Initial situation The aim of our work in the field of industrial biotechnology is to analyze and describe bioprocesses in as much detail as possible by means of modern measurement engineering and mathematical modeling. This facilitates the efficient optimiza- tion of processes and their transfer from laboratory to produc- tion scale as changes to processes can be calculated in ad- vance on the computer. This way only promising approaches will be validated experimentally. This approach saves devel- opment time and costs whilst increasing the understanding of the process. Additionally, software sensors provide further potential for cost savings. In this offline process parameters such as biomass and product concentration are estimated on the basis of online readings such as the pH value, exhaust gas data and consumption of acid and base, which considerably simplifies the monitoring of the process and reduces the ne- cessity of sampling. Approach The basic condition for the scale-up is a test system that comes as close as possible to the production plant in design and geometry. For this reason we mainly work with mini bio- reactors. Designed as multifermentation systems they per- mit the parallel screening of different strains and conditions. Through the selection of suitable systems for online meas- urement and control, it is possible to maintain critical param- eters constant and to generate a data base for a mathemati- cal description of the production process. In this regard we succeeded in establishing a real-time measurement of volatile components in a medium with an online mass spectrometer developed by the Fraunhofer ICT. We also possess offline pro- cess analytics to promptly determine substrate and product concentrations. Based on this data, a mathematical model de- scribing the behavior of different production strains can be designed. This covers the establishment of software sensors, hybrid models up to a first-principles description of the entire fermentation process. Screening of microorganisms The selection of a suitable production strain is a key compo- nent of the development of bioprocesses such as the genera- tion of ethanol from lignocellulose. It has been possible to successfully characterize different ethanol producers by using the multifermentation systems (Fig. 1). Important parameters in this were the ethanol productivity and yield, the product selectivity and the inhibitor tolerance of the individual micro- organisms. Process observations through online measurement We have already applied the online mass spectrometer to ob- serve enzyme catalysis in aqueous solution by measuring the concentration of several substances simultaneously. This way it has been possible to monitor reaction processes whose sensi- tivity is in the lower ppm region. Currently the mass spectrom- eter is being implemented and optimized as an analysis tool in a biotechnologically used pressure reactor in order to study biotechnological processes with gaseous substances and vola- tile products. Modeling and soft-sensing The Gibberellin Project, whose aim was the fermentative pro- duction of the plant hormone Gibberellin by the fungus Fusar- ium fujikuroi, is an example of software sensors and process modeling. A general problem regarding the fermentation of 9 4 A MODEL-BASED PROCESS DEVELOPMENT AND THE OPTIMIZATION OF BIOPROCESSES Dipl.-Ing. Tino Elter, Dipl.-Ing. Matthias Stier, Thilo Haitz M.Sc. CHEMISTRY 1 2 3