Tesis doctoral de Zivko Juznic Zonta
In engineering practice, when an explicit model of a system is available, numerical experiments can be performed in order to predict the future behavior of the system, explain or describe its hidden state, guide data collection, etc. Typically, the dynamics of the system are complex and difficult to observe with precision. Any approximation of the observed reality within an explicit model necessarily implies uncertainty, which should be characterized and quantified to build confidence over model results. Uncertainty associated with model-parameter and its implications for bio-process optimization are of main concern in this phd work. As a bio-process case study, the anaerobic digestion is considered for modeling. The production of biogas by controlled anaerobic digestion could be a profitable activity, apart of being a renewable energy source. However, the margins to improve this technology are wide. Anaerobic co-digestion with two or more input materials is a way to make low biogas yield biomass applicable at real scale. Among the possible co-substrates, lipids-rich wastes are attractive for their high energetic potential. The main limiting factor for this strategy is the inhibition of anaerobic digestion by long chain fatty acids. Modeling provides a useful approximation of the complex and delicate microbiology activity of this anaerobic digestion system. The underlying goal of the phd project is to to improve the wastewater treatment bio-processes with the aid of modeling and uncertainty analysis. With this goal in mind, a general purpose, user-friendly, simulation environment called ¿virtual plant¿ (vp) was build and it was applied to anaerobic co-digestion and activated sludge modeling. Within the vp tool, new core dynamics of the long chain fatty acids (lcfa) inhibition process were proposed and tested and different inferential procedures for the estimation of parameter-uncertainty were compared. Finally, a proposed multi-criteria analysis under uncertainty was applied to an industrial anaerobic co-digestion biogas plant. In conclusion, the developed vp toolkit was found reliable and user-friendly when modeling activated sludge and anaerobic digestion systems. The proposed lcfa-inhibition model was able to reproduce correctly the experimental data at hand and enable its interpretation. However, uncertainty estimation of parameters and falsification of the proposed model of lcfa-inhibition are still missing. The bayesian procedure was proved useful when addressing the estimation of parameter uncertainty of anaerobic digestion and activated sludge models. A considerable improvement in the operation efficiency and reliability of an industrial biogas plant was possible within the proposed multi-criteria analysis. However, future work is needed to improve the procedure of elicitation of the inputs for this multi-criteria analysis and decrease its computational burden.
Datos académicos de la tesis doctoral «Modeling biotechnological processes under uncertainty. anaerobic digestion as case study«
- Título de la tesis: Modeling biotechnological processes under uncertainty. anaerobic digestion as case study
- Autor: Zivko Juznic Zonta
- Universidad: Politécnica de catalunya
- Fecha de lectura de la tesis: 26/10/2012
Dirección y tribunal
- Director de la tesis
- Xavier Flotats Ripoll
- Tribunal
- Presidente del tribunal: joan Mata alvarez
- Jorge Rodríguez rodríguez (vocal)
- Jesús Colprim galcerán (vocal)
- joan De pablo ribas (vocal)
