Opération périodique d'une pile à combustible microbienne par connexion intermittente à une résistance électrique

Amongst the XXIst century challenges, the renewable energy production and the drinkable water access represent two pressing problematics because of the world population growth and the average standard of living improvements. A recent technology, the Microbial Fuel Cell (MFC), may potentially bring major contributions, answering both problematics. The MFC is a system that generates electricity from the biological treatment of organic matter in wastewater. Its fundamental principle is based on bacterial activity: in the reactor, microorganisms named electricigens degrade organic matter dissolved in the wastewater, thus liberating electrons as part of their metabolism. These electrons can be directed through a circuit, enabling their use as an electrical energy source. Even though the system has proven to be stable and efficient in wastewater treatment, several obstacles prevent its industrial implementation. Amongst them is the lack of energy harvesting optimisation, especially given an environment where the quality of the wastewater is subject to frequent and random perturbations. Typically, a fixed external resistance is connected to the circuit, although the internal resistance is known to vary greatly depending on the organic matter concentration in the influent. Yet, the maximum power transfer theorem dictates that both internal and external impedances of a circuit must match in order to optimise the electrical power production. A loss in efficiency caused by the influent perturbation is thus inevitable. Also, the system is characterised by mechanisms of complex dynamics, some of which are yet to be studied. Thus the MFC is not adequately modeled and it is difficult to predict its behaviour when subjected to certain perturbations or modes of operation. Modeling efforts are thus required in order to track the evolution of internal electrical properties such as resistance and capacitance, and to assure an optimal energy harvesting. A possible solution is to operate the MFC under periodic operations, i.e., to intermittently connect and disconnect the external resistance to the system. A similar strategy is often used to optimise electrical motors. A series of laboratory experiments was conducted on an acetate-fed MFC. First, the periodic operations were characterised based on specific parameters, and different operating regions were identified. The Duty Cycle (DC), defined as the fraction of the time the system is in a closed circuit state within a cycle, was used in order to demonstrate the presence of an optimum in power generation. After, a control strategy based on voltage limits was studied and compared to current alternatives. In order to answer to the identification and modeling objectives, an equivalent circuit containing a resistance and a capacitance was suggested. This model was used to identify the electrical properties of the MFC, and was subsequently integrated to a model taking account for the bacterial population activities. Experiments have shown that periodic operation is optimal for power generation when the DC is close to 0.9 and the cycles are of the order of a the second. It is expected that the real optimum could be found at smaller cycles, but exploring this range was not possible due the equipment's precision. An observed operating point with a cycle time of 7 seconds, a DC of 0.9 and an external resistance of 20 $\Omega$ resulted in a power production of 2.84 ± 0.02 mW, a comparable performance to that of a continuously operated MFC giving 2.90 ± 0.28 mW. It was also observed that the power generation was usually greater when the external resistance was inferior to the system's internal resistance during periodic operations. On the other hand, the voltage limits based control was proven to offer a performance comparable to the actual alternatives, when subjected to perturbations in influent concentration. The identification of the system electrical properties through the use of an equivalent circuit was successful, especially when the MFC was subjected to varied excitations. The estimated and measured parameters were of the same order of magnitude. In order to further optimise the power production and to deepen the knowledge of the system, it is necessary to explore new ranges of periodic operations, with shorter cycle times. Also, developing an on-line identification strategy of the system's electrical properties should constitute the basis of a control algorithm adjusting the DC. Finally, simulations of the hybrid model should be performed to gain better insight about the impact of periodic operations on internal mechanisms of the MFC.