Search

By means of improved feedback control kLa measurements become possible at a precision and reproducibility that now allow a closer look at the influences of power input and aeration rate on the oxygen mass transfer. These measurements are performed online during running fermentations without a notable impact on the biochemical conversion processes. A closer inspection of the mass transfer during cultivations showed that at least the number of impellers influences mass transfer and mixing: On the laboratory scale, two hollow blade impellers clearly showed a larger kLa than the usually employed three impeller versions when operated at the same agitation power and aeration rate. Hollow blade impellers are preferable under most operational conditions because of their perfect gas handling capacity. Mixing time studies showed that these two impeller systems are also preferable with respect to mixing. Furthermore the widths of the baffle bars depict a significant influence on the kLa. All this clearly supports the fact that it is not only the integral power density that finally determines kLa.

Recombinant protein production processes are typically divided into two phases. In the first one, pure cell propagation takes place, while in the second one product formation is switched on within the cells by adding an inducer. In the initial biomass formation phase, the cell density is rather low and, hence, the measurement quantities that could be used to determine the process' state depict small values and are rather severely distorted by measurement noise. Because of these measurement problems, the fermentation cannot be reliably controlled by feedback control during this first production phase; instead, the process must be controlled in an open-loop fashion. The consequence, worked out in this paper, is to design substrate feed rate profiles for the growth phase in such a way that they are robust with respect to the main disturbances observed in practice. The robustness of the biomass formation is shown to be primarily dependent on the specific growth rate adjusted in the first hours. High batch-to-batch reproducibility can be obtained with exponential feeding profiles F(t) corresponding to specific growth rates micro(set) well below the maximal specific growth rate micro(max) of the organism. The reduction in the growth rate needed to obtain a robust process behavior depends on the inaccuracies in the initial biomass concentrations. Quantitative feed rate profiles were obtained by numerical simulation and these results were validated experimentally by means of a series of cultivation runs, where a recombinant pharmaceutical protein was produced. All experimental data confirmed the assumptions made in the robust process design study.

Heterologous protein production often causes a significant metabolic burden in Escherichia coli cells which manifests itself in a substantial decrease in their physiological characteristics such as the maximal specific growth rate on a given substrate, the maximal substrate uptake rate as well as the maximal specific oxygen uptake rate. In high-cell-density cultures, the substrate feed rate must be adapted to this changing capabilities of the cells in order to avoid overfeeding and thus the formation of by-products that inhibit the cell performance further. This requires the precise knowledge about the changes in these specific rates, particularly during the product formation phase. In order to precisely investigate the time profile of the critical specific substrate uptake rate σcrit of microorganisms, i.e. the maximal rate at which the cells can fully oxidize their substrate, a new online tracking technique is presented. The feed rate F is modulated in such a way that the specific substrate uptake rate σ is linearly raised towards its critical value σcrit. When this is reached the feed rate is automatically reduced and the procedure is repeated. In this way the method automatically follows the changing time profile of σcrit during the entire cultivation and avoids significant acetate formation rates. This procedure considerably increases the identifiability of σcrit. The high precision of the technique also results from replacing the pO2 measurements that seem to suggest themselves for monitoring maximal oxygen uptake rate, by measuring the total oxygen consumption rate tOUR, which is available at a much higher signal-to-noise ratio and is not as prone to distortions. An important advantage of measuring tOUR is that it allows keeping pO2 controlled at its optimal value. The applicability of the new tracking technology is demonstrated at E. coli cultures. The resulting σcrit(t) profile allows determining the substrate feed rate profile and other key variables that can be used for advanced feedback control in protein production processes. The technique can be considered a PAT tool and is well suited to industrial fermentations.

In practice, the performance of a biochemical conversion process, i.e. the bioreactor performance, is essentially determined by the benefit/cost ratio. The benefit is generally defined in terms of the amount of the desired product produced and its market price. Cost reduction is the major objective in biochemical engineering. There are two essential engineering approaches to minimizing the cost of creating a particular product in an existing plant. One is to find a control path or operational procedure that optimally uses the dynamics of the process and copes with the many constraints restricting production. The other is to remove or lower the constraints by constructive improvements of the equipment and/or the microorganisms. This paper focuses on the first approach, dealing with optimization of the operational procedure and the measures by which one can ensure that the process adheres to the predetermined path. In practice, feedforward control is the predominant control mode applied. However, as it is frequently inadequate for optimal performance, feedback control may also be employed. Relevant aspects of such performance optimization are discussed.

Some key aspects of obtaining hybrid process models which perform well and that can be used in process supervision, optimization and control are discussed from the point of view of the benefit/cost-ratio. The importance of starting with a clear definition of the problem and a corresponding quantitative objective function is shown. In order to enhance the benefit/cost-ratio above the threshold of acceptance, a series of procedures is proposed: in the beginning an exploratory process data analysis is suggested to classify the process variables according to their importance and to facilitate the development of black- and grey-box models. Efficient validation of the model is shown to be indispensable. Hybrid model approaches proved to have to significant advantages, since they allow the activation of a larger portion of the available a-priori knowledge. Applications of hybrid models with respect to process optimization require new techniques, since the classical approaches are too difficult to use and are restricted to well-performing models. Finally, powerful software tools are required to implement the different algorithms at the production plants and to allow the efficient conversion of the ideas to real benefits.

A comparative study of some random search optimization schemes that can be used to determine sub-optimal open-loop control profiles from given biotechnical process models is presented and compared with the classical optimization approach based on Pontryagin's maximum principle that has often been used in literature. The main objective of this study is to show that the much simpler to use random search procedures lead to essentially the same success as the classical procedure. Surprisingly little effort is required to use these random search techniques in order to determine feed rate, feed concentration, temperature or precursors profiles. Hence, using these techniques, bioengineers can focus their attention much more on creating and improving the underlying process models instead of tuning complicated optimization procedures. Three different techniques are described in detail and their application is illustrated with standard examples from literature.

An improved technique of hybrid modelling biochemical production processes is described, composed of a set of dynamical differential equations, an artificial neural network and a fuzzy expert system. The applicability of a hybrid model for state estimation, prediction, feed rate optimization, and process control is explained. Its performance is demonstrated by means of an the example of a fed-batch baker's yeast production process. The identification procedure is described in detail. Emphasis is placed on the most difficult part of it, the training of the neural network component of the hybrid model. New approaches to process control are developed based on neural networks.

The difficulty in using artificial neural networks (ANNs) to represent nonstationary processes such as biotechnical production processes has been coped with by process partitioning. By logically cutting down the whole process into segments corresponding to different process phases, or special process situations, and representing them by a corresponding modular set of neural networks it was possible to obtain well performing real-time state estimations and process predictions. At any given moment, more than one artificial neural net is employed to describe the process. A small fuzzy expert system, by which additional heuristic know-how about the process is exploited, is used to dispatch control to the individual ANNs appropriately, and to process the outputs of the individual nets. Knowledge conserved in the nets is continuously updated by means of a real-time learning procedure as they are activated. As compared with the process estimation and prediction by our previously developed distributed model which is based on the extended Kalman filter approach, the fuzzy-aided artificial neural network system turned out to be at least as accurate, and considerably faster to develop.

Cell cultures must tightly be kept under control in order to guarantee a sufficiently small variability in the protein product quality. A simple and efficient technique for CHO-cell cultures is presented that allows keeping the viable cell count X(v) and the specific growth rate μ of the cells on predefined trajectories. As X(v) and μ cannot directly be measured online, they are controlled indirectly via the total mass of oxygen consumed. Online values of the latter can precisely be estimated from off gas analysis, i.e. from the O₂ volume ratio measured in the vent line and air flow rate measurements. In glutamine-limited fed-batch cultivations, the glutamine feed rate can be manipulated in such a way that the viable cell density and the specific growth rate are kept on predefined profiles for nearly the entire cultivation time. The viability of the cells is not affected by the closed loop control actions. The technique was validated with CHO-cells cultured in a 2.5-L fully instrumented stirred tank bioreactor. It is shown that the controller is able to run the process exactly on predefined tracks with a high batch-to-batch reproducibility. By means of six fed-batch cultivations of CHO cells it was shown that a remarkable reproducibility of viable cell concentration could be achieved throughout 140 h cultivation time.

A new approach of dealing with complex mathematical models of biotechnological processes has been applied to real-time state estimation and the prediction of the time at which the beer main fermentation process must be stopped. The main idea behind the approach is to use some kind of distributed modelling, i.e., to break down comprehensive process models into pieces, corresponding to process phases or particular situations. In the case of beer fermentation the partial models are simpler and more accurate in describing a particular situation. Hence, the accuracy of the Extended Kalman Filter used as the estimation procedure could be improved. The additional effort of assigning the different models to a given situation is made by fuzzy reasoning, which allows to incorporate additional heuristic knowledge about the process. During several runs performed at the production scale of 220 m 3 working volume it could be shown that the method is superior to the conventional state estimation and prediction technique using an Extended Kalman Filter.

Batch-to-batch reproducibility of fermentation processes performed during the manufacturing processes of biologics can be increased by operating the cultures at feed rate profiles that are robust against typically arising disturbances. Remaining randomly appearing deviations from the desired path should be suppressed automatically by manipulating the feed rate. With respect to the cells' physiology it is best guiding the cultivations along an optimal profile of the specific biomass growth rate mu(t). However, there are two problems that speak for further investigations: Upon severe disturbances that may happen during the fermentation, the biomass concentration X may significantly deviate from its desired value, then a fixed mu-profile leads to a diminished batch-to-batch reproducibility. Second, the specific growth rate cannot easily be estimated online to a favourably high accuracy, hence it is difficult to determine the deviations in mu from the desired profile. The alternative discussed here solves both problems by keeping the process at the corresponding total cumulative carbon dioxide production-profile: it is robust against distortions in X and the controlled variable can accurately be measured online during cultivations of all relevant sizes. As compared to the fermentation practice currently used in industry, the experimental results, presented at the example of a recombinant protein production with Escherichia coli cells, show that CPR-based corrections lead to a considerably improved batch-to-batch reproducibility.