Development of monitoring and control systems for biotechnological processes

The field of biotechnology represents an important research area that has gained increasing success in recent times. Characterized by the involvement of biological organisms in manufacturing processes, its areas of application are broad and include the pharmaceuticals, agri-food, energy, and even waste treatment. The implication of living microorganisms represents the common element in all bioprocesses. Cell cultivations is undoubtedly the key step that requires maintaining environmental conditions in precise and defined ranges, having a significant impact on the process yield and thus on the desired product quality. The apparatus in which this process occurs is the bioreactor. Unfortunately, monitoring and controlling these processes can be a challenging task because of the complexity of the cell growth phenomenon and the limited number of variables can be monitored in real-time. The thesis presented here focuses on the monitoring and control of biotechnological processes, more specifically in the production of bioethanol by fermentation of sugars using yeasts. The study conducted addresses several issues related to the monitoring and control of the bioreactor, in which the fermentation takes place. First, the topic concerning the lack of proper sensors capable of providing online measurements of key variables (biomass, substrate, product) is investigated. For this purpose, nonlinear estimation techniques are analyzed to reconstruct unmeasurable states. In particular, the geometric observer approach is applied to select the best estimation structure and then a comparison with the extended Kalman filter is reported. Both estimators proposed demonstrate good estimation capabilities as input model parameters vary. Guaranteeing the achievement of the desired ethanol composition is the main goal of bioreactor control. To this end, different control strategies, evaluated for three different scenarios, are analzyed. The results show that the MIMO system, together with an estimator for ethanol composition, ensure the compliance with product quality. After analyzing these difficulties through numeric simulations, this research work shifts to testing a specific biotechnological process such as manufacturing bioethanol from brewery’s spent grain (BSG) as renewable waste biomass. Both acid pre-treatment, which is necessary to release sugars, and fermentation are optimized. Results show that a glucose yield of 18.12 per 100 g of dried biomass is obtained when the pre-treatment step is performed under optimized conditions (0.37 M H2SO4, 10% S-L ratio). Regarding the fermentation, T=25°C, pH=4.5, and inoculum volume equal to 12.25% v/v are selected as the best condition, at which an ethanol yield of 82.67% evaluated with respect to theoretical one is obtained. As a final step, the use of Raman spectroscopy combined with chemometric techniques such as Partial Least Square (PLS) analysis is evaluated to develop an online sensor for fermentation process monitoring. The results show that the biomass type involved significantly affects the acquired spectra, making them noisy and difficult to interpret. This represents a nontrivial limitation of the applied methodology, for which more experimental data and more robust statistical techniques could be helpful.