Promoteur : Prof. Alain Vande Wouwer

Résumé :

Anaerobic Digestion (AD) is an environmentally sustainable technology to treat waste (water) by biological means. During this process, several microorganisms break down organic material and convert it into biogas in the absence of oxygen. Biogas can in turn be combusted to generate electricity and heat, or can be processed into renewable natural gas and transportation fuels
Anaerobic digestion is used worldwide for the treatment of either industrial or municipal waste. Despite its increasing application and numerous advantages, the high complexity of the process and its dependence on many operational variables are some of the reasons that prevent its full potential use. Moreover, there are still unconsolidated matters and a large gap between the needs of the industry and the developments achieved in research works.
Anaerobic digestion systems are delicate to operate and suffer from instabilities (not in the strict mathematical sense but in an operational sense), including pH drop, large increase in the volatile fatty acid concentration or fluctuations in the biogas production rate.
For process optimization and control, mathematical models are particularly useful tools which allow the prediction of the process behavior under different scenarios, types of waste and/or operating conditions.
In this connection, the first objective of this thesis is to review the role of hydrogen in AD processes. In some cases, hydrogen appears as an interesting indicator of imminent instability and monitoring its evolution can therefore be a key component of an advanced control system.
The second objective is the development of a low-order dynamic model including hydrogen concentration as one of the state variables. This model is derived from an informative data set generated with the well-established ADM1 model. The derivation procedure is based on principal component analysis and nonlinear parameter estimation techniques.
The modelling approach is tested in two real-case studies, one based on winery wastewater experimental data, and the other on experimental data from a co-digestion process treating a mixture of three different residues.
The third part of the thesis is dedicated to the assessment of two hydrogen-based control strategies. The first one is heuristic whereas the second one is model-based and fully exploit the low-order model. The controllers offer satisfactory performance, even facing perturbations in pH or temperature levels.