Modélisation et Simulation d'une Hybridation Directe Pile à Combustible - Supercondensateur

Abstract

Today we face two major problems for our energy supply. One of them is the risk of depletion of both fossil fuels and nuclear fuels. The second problem is the environmental pollution caused by the use of these fuels. The burning of hydrocarbons contributes significantly to the greenhouse effect. Nuclear waste that may not be reprocessed must be stored for thousands of years and continues to be harmful.
It is an absolute necessity to find alternatives to fossil fuels, alternatives that present the same favorable properties with regard to transport, storage and energy density. Hydrogen is an ideal candidate, even if the gas is an energy carrier and not a primary raw material. Moreover, hydrogen can contribute to the expansion of electricity production through renewable sources. These sources are inherently intermittent and unpredictable, which makes them unsuitable as a primary source of electricity supply. However, they are not harmful to the environment and the resources they use (solar, wind, water ...) are free and inexhaustible. The electricity they generate can be used to synthesize hydrogen via electrolysis. The easy storage of hydrogen and its large energy density make it possible to build up a buffer for the intermittent operation of renewable resources and to meet the instantaneous demand for electricity. In this scenario, the fuel cell emerges as the missing link to convert the chemical energy with high efficiency into electrical energy.
Within this context, it is logical that the Laboratoire Laplace has been researching fuel cells for years. The aim of the research in this laboratory has always been to construct an electrical equivalent model of a fuel cell. Today, such a model for a unit cell exists in the laboratory, thanks to the PhDs of Guillaume Fontès and Olivier Rallières and the work of Christophe Turpin. This model is the first building block of this work.
Unfortunately, the fuel cell is still an expensive and fragile component that is very susceptible to premature aging. Exposing the fuel cell to steep changes in power consumption or to high frequency harmonics, caused by a static inverter, is a major cause of accelerated aging. However, such loads are difficult to avoid in real conditions of use. An important part of the research is therefore aimed at making the fuel more robust against these loads. A promising technique is the direct hybridization, the immediate parallel connection of a fuel cell with an ultracapacitor. Benoît Morin examines this architecture in his PhD. His work is the second building block on which this thesis is based.
This work has multiple purposes: First, reliable models must be created for the fuel cell and ultracapacitor. These tasks are the subject of chapter one. Then the interaction between these two components is studied in chapter two. The ultimate goal is the assembly of the models to an elementary system. In the third chapter, the simulations of this system are compared to the measurements B. Morin performs as part of his PhD.
The first objective is the parameterization of the unit cell model using the available measurements. We use two different types of measurements: impedance spectroscopy, and the static polarization curve of the fuel cell. We also adjust the model for use in a widespread program for simulation of electrical circuits. We subject the simulation model to a sinusoidal excitation that covers the entire working range of the real component. This test is performed in a frequency range from 1mHz to 200Hz. Comparison of the simulation results with the measurements shows that the model has a very strong performance at low (mHz) and average (Hz) frequencies. At frequencies from 10 Hzonward there is a clear rotation observed between the real and the simulated polarization curve. We show that this rotation occurs because it is impossible to represent non-integer impedance, a representation of the porosity of the electrodes of the fuel cell, as an electric circuit. A possible solution to augment the precision of the model is to distribute its discrete elements. Thanks to this first part we better understand the limits of the existing theoretical model. We understand the limitations arising from its adaptation to simulation software, and we know how the model can be modified if we desire to increase the simulation's precision.
However, this first chapter raises some questions. We realize that the parameter identification process is slow and not very robust. The result of the process is strongly dependent upon the number of spectroscopic measurements, measurement errors, and the experience of the researcher. Moreover, the impact of common phenomena, such as spiralisation or instability of the Nyquist path, is not known. In parallel with the work on the fuel cell, we propose an electrical equivalent model for an ultracapacitor. This model is also configured with the aid of the available measurement results.
In chapter two, we discuss the direct hybridization. We first give an overview of the standard hybridization techniques with emphasis on the advantages and disadvantages of each technique. Then we focus on the direct hybridization and its different architectures. In the second part of this chapter we study the interactions between the fuel cell and ultracapacitor in an elementary hybrid system. We propose a simplified impedance model so an analytical study becomes possible. We investigate the stability, damping, the frequency domain behavior and the transitory response of the basic system. It is shown that the interactions between the fuel cell and ultracapacitor are always stable and overdamped. The direct hybridization reduces the current variations for the fuel cell and filters higher order harmonic components. This slows down the aging of the fuel cell. Moreover, the hybridization stabilizes the voltage, a beneficial effect for the load. Comparison of the analytical results with simulations confirms these conclusions. This chapter also shows the indispensability of a reliable simulation model. Even for a simplified system that only takes the main physical phenomena into account, the calculations are too complex and often impossible to interpret.
Finally we compare simulations of a hybrid system with measurements of B. Morin in the third chapter. We observe that the simulation model closely approximates the reality system at low frequencies. At medium and high frequencies, however, there is always a difference of a few percent between the simulation results and the measurements. If we zoom in on the transition phenomenon we notice the same difference on a time scale of a few milliseconds. We try to explain these errors. Ultimately there are only two hypotheses. Either we have made an error in the identification of one or more dynamic elements, either the model is not able to replicate the behavior of the fuel cell on a small time scale, which might mean that the model ignores certain physical phenomena. Finally, we compare the analytical predictions from chapter two with the measurements. It appears that the established theory only allows qualitative predictions. The highly nonlinear nature of the components makes quantitative predictions impossible. Simulation shows itself again indispensable.

Date of Award	10 Sept 2012
Original language	French
Supervisor	Philippe Lataire (Promotor), Turpin Christophe (Promotor) & Guillaume Fontes (Co-promotor)