Probing Control. Analysis and Design with Application to Fed-Batch Bioreactors

Abstract: In most control problems the objective is to control the output at a desired value in spite of disturbances. In some cases, the best setpoint is not known a priori and it should be found online to optimize the process performance. This thesis examines a probing strategy that can be applied for this class of problems. The focus is on the application of the technique to the control of feed supply in fed-batch fermentations of the bacterium Escherichia coli. The thesis is divided into three parts. In the first part, the convergence properties of the probing algorithm are examined. The analysis is limited to processes modeled by a linear time-invariant dynamic in series with a static nonlinearity. Stability and performance analysis taking into account the process dynamic is performed. Tuning guidelines that help the user for the design are also derived. The second part presents a novel cultivation technique based on the probing approach. The fermentation technique combines the advantages of probing control and temperature-limited fed-batch technique. The feeding strategy is well adapted for prolonged operation at the maximum oxygen transfer capacity of the reactor. The efficiency of the method is demonstrated by simulations and experimental results. The strategy leads to a high biomass and it limits the degradation of the recombinant protein activity in the late production phase. In the third part, the probing feeding strategy is evaluated in industrial-scale bioreactors. Based on experimental results the influence of scale and complex medium is discussed. It is shown that the flexibility and robustness of the technique makes it a useful tool for process development.