In this thesis, a portable PEMFC stack, which can directly power or charge 3C products, will be developed. The stack is developed for portable applications, so the structure of the stack is simplified as possible as we can. The PEMFC stack is made with 32 carbon fiber bunches for current collectors and two 8-cell banded-type MEAs which are made with 8 sets of electrodes on a piece of membrane. The stack can develop a high voltage by serially connecting 8 cell or 16 cell outside of the reaction chamber. The resistance of each carbon bunch assembling with carbon cloth is measured before they are assembled into the stack. Under assembly pressure 3 bar, the total resistance is about 8.7m£[ or 11m£[¡Ecm2. The resistance is about one half of that graphite plate assembling with carbon cloth. Without being compressed greatly in diffusion layer, the fluid can easily flow through the gaps between carbon fiber and within diffusion layers, and then the reactive region will react more uniformly. In addition, the connecting wires are assembled to a wire collecting board, so that the stack is look more neat, and it easier assemble or dissemble. In this thesis, the volume of the developed 16-cell hydrogen fuel cell stack is about 9.6 cm*6.3 cm*2.2 cm. The total electrode area is 50 cm2 (16-cell¡Ñ3.15 cm2 per cell). When the stack is operating at room temperature and air-breathing, an 8-cell stack in series connection can generate 3.7V voltage. Its power at voltage 3.7V is about 3.6W. It can directly power PDAs, mobile phones or digital cameras. A 16-cell stack in series connection can generate 7.2V voltage. Its power at this voltage can offer 7W. The 16-cell stack can directly power digital single-lens reflex cameras. If two or more of this stack are connected in series, it will be able to power a notebook or other more power products.

Le développement de la technologie « piles à combustible » nécessite l'utilisation d'outils de diagnostic adéquats notamment pour le monitoring de l'état de santé des systèmes industriels (stacks) dans les conditions réelles de fonctionnement. L'utilisation des moyens traditionnels de diagnostic nécessite l'arrêt ou la perturbation du fonctionnement du système. Le travail de cette thèse vise le développement d'une approche innovante non intrusive pour le diagnostic des stacks PEM (Proton Exchange Membrane), basée sur la mesure des petites fluctuations électriques (bruits électrochimiques). Pour mesurer les bruits, un système d'acquisition des faibles signaux à haute fréquence a été utilisé sans filtrage analogique préalable. Ces mesures ont été dans le cadre du projet ANR « Propice » pour quatre campagnes de mesures avec la collaboration du FCLAB et du CEA LITEN. Les mesures des bruits électrochimiques, sur plusieurs semaines, ont permis de construire une base de données extrêmement riche. Pour traiter ces données, différents approches statistiques dans le domaine temporel, fréquentiel et tempo-fréquentiel ont été utilisés pour la génération de descripteurs fiables et robustes. Il a été démontré que la mesure des bruits permet d'obtenir une riche signature des stacks PEM dans un vaste domaine fréquentiel. Cette signature reflète les différents phénomènes physico-chimiques et est très sensible aux paramètres de fonctionnement du système. L'évolution de cette signature au court de temps peut être utilisée pour le diagnostic in-situ de d'état de santé des stacks commerciaux dans les conditions réelles de fonctionnement et pour le développement des moyens de pronostic. / Fuel cell technology development requires adequate diagnostic tools, in particular for monitoring the state of health of industrial systems (stacks) under operating conditions. Traditional diagnostic tools require to stop or disrupt the system operating. This thesis aims at the development of an innovative and non-intrusive approach for the diagnostic of PEM (Proton Exchange Membrane) fuel cell stacks. The methodology is based on the measurement of small electrical fluctuations (electrochemical noise). To measure this noise, a high frequency signal acquisition system was used without prior analog filter. These measurements were obtained within the ANR project « Propice » using four measurement campaigns with the collaboration of FCLAB and CEA LITEN. Electrochemical noise Measurements, over several weeks, made it possible to build a rich database. To process these data, different statistical approaches in time, frequency and tempo-frequency domains have been used for the generation of reliable and robust descriptors. It has been shown that the measurement of noise makes it possible to obtain a rich signature of the PEM stacks in a wide frequency range. This signature reflects the various physico-chemical phenomena and it is very sensitive to the operating parameters of the system. The evolution of this signature in short time analysis can be used for an in-situ diagnostic of the state of health of commercial stacks under real operating conditions and for the development of prognostic strategies.

An EU-funded project Fit-4-AMandA aims to establish a technological roadmap to scale-up from less than hundred stacks/year (manual assembly) to 50,000 stacks/year (automated assembly) in 2020 and beyond. Existing membrane-electrode assembly (MEA) and stack were redesigned/adapted for manufacturability and automation. The technology and machine system for the automated assembly of polymer-electrolyte-membrane fuel cell (PEMFC) stacks were developed, manufactured and are currently being tested. Fast in-line non-destructive quality-assurance methods for automated production of MEAs and stack assembly are being developed and implemented. For the final period of the project, a validation of the designs, hardware, tools and software for the automated production of MEAs and stack assembly as well as an integration of one of the prototype stacks manufactured by the automated processes into a light-commercial vehicle followed by a field-testing are scheduled. / Die breite Markteinführung von wasserstoffbasierten Antriebssystemen verlangt zunehmend nach einer kosteneffizienten und serientauglichen Produktion von Brennstoffzellenstacks. So sehen die Ziele der Europäischen Union vor, die Herstellung von aktuell unter 100 Stacks pro Jahr auf 50.000 Stacks pro Jahr bis zum Jahr 2020 zu erhöhen. Um dies zu erreichen, sollen im Rahmen des vom Fuel Cell and Hydrogen Joint Undertaking (FCH JU) der europäischen Union geförderten Projektes Fit-4-AMandA automatisierte Anlagen für solche Stückzahlen befähigt werden. Der Beitrag beschreibt, wie die bereits verfügbare Membran-Elektroden-Einheit (MEA) und der Stack bezüglich Herstellbarkeit und Automatisierung konstruktiv umgestaltet und angepasst wurden. Die neu entwickelte Technologie und das Maschinensystem für die automatisierte Montage von PEM-FC-Stacks sowie Verfahren der INLINE-Qualitätssicherung und der zerstörungsfreien Prüfung werden mit ihren Implementierungsmöglichkeiten in die automatisierte Fertigungsstrecke vorgestellt. Ein Ausblick gibt eine Übersicht über die weiteren Entwicklungsschritte wie die Validierung der Entwürfe, der Hard- und Software für die automatisierte Produktion der MEAs und Stacks. Eine vorgesehene spätere Integration der so gefertigten Stacks in ein Fahrzeug und die damit verbundenen Feldtests zur Untersuchung der Reproduzierbarkeit und Zuverlässigkeit der Stacks werden abschließend dargestellt. Die Förderung des Projektes erfolgt im Rahmen der Finanzhilfevereinbarung Nr. 735606 des FCH JU der EU.