L’Europe et, en particulier, la France ont engagé une politique de transition énergétique, consistant à diminuer de 20 % les émissions de gaz à effet de serre, réduire de 20 % la consommation d'énergie et à atteindre 20 % d'énergies renouvelables dans le bouquet énergétique d'ici à 2020. Afin d’atteindre ces objectifs, plusieurs pistes doivent être déployées afin de promouvoir les énergies renouvelables qui sont de nature intermittente. Dans le cadre de ce projet de thèse on propose trois pistes pour contribuer au développement de ces énergies : la première consiste à déterminer la combinaison efficace des sources intermittentes et de sources fiables telle que les combustibles, ainsi que le montant optimal à investir dans les technologies renouvelables intermittentes sachant leur productivité imprévisible et variable. La deuxième piste de recherche consiste à déterminer, à l’aide d’un modèle dynamique, dans quelles circonstances on stocke de l’électricité et dans quelles circonstances on la délivre. Dans ce cadre, on essayera de déterminer un seuil optimal de stockage. La troisième piste consiste à déterminer comment on organise le secteur de l’effacement électrique. on appliquera le même principe que pour les parties précédentes : on a le modèle de base qui est l’intermittence modélisée par la variable aléatoire à laquelle on rajoutera l’effacement électrique. Dans le cadre de la thèse on espère publier trois articles complémentaires, un article sur chaque piste. Le lien entre ces trois pistes sera l’analyse et à la modélisation des différents instruments politiques, qui sont en ce moment en discussion au niveau français. L’objectif de cette analyse sera de faire un classement en fonction de leur capacité à atteindre un état optimal (ou à s’en approcher) sur chaque piste. / L’Europe et, en particulier, la France ont engagé une politique de transition énergétique, consistant à diminuer de 20 % les émissions de gaz à effet de serre, réduire de 20 % la consommation d'énergie et à atteindre 20 % d'énergies renouvelables dans le bouquet énergétique d'ici à 2020. Afin d’atteindre ces objectifs, plusieurs pistes doivent être déployées afin de promouvoir les énergies renouvelables qui sont de nature intermittente. Dans le cadre de ce projet de thèse on propose trois pistes pour contribuer au développement de ces énergies : la première consiste à déterminer la combinaison efficace des sources intermittentes et de sources fiables telle que les combustibles, ainsi que le montant optimal à investir dans les technologies renouvelables intermittentes sachant leur productivité imprévisible et variable. La deuxième piste de recherche consiste à déterminer, à l’aide d’un modèle dynamique, dans quelles circonstances on stocke de l’électricité et dans quelles circonstances on la délivre. Dans ce cadre, on essayera de déterminer un seuil optimal de stockage. La troisième piste consiste à déterminer comment on organise le secteur de l’effacement électrique. on appliquera le même principe que pour les parties précédentes : on a le modèle de base qui est l’intermittence modélisée par la variable aléatoire à laquelle on rajoutera l’effacement électrique. Dans le cadre de la thèse on espère publier trois articles complémentaires, un article sur chaque piste. Le lien entre ces trois pistes sera l’analyse et à la modélisation des différents instruments politiques, qui sont en ce moment en discussion au niveau français. L’objectif de cette analyse sera de faire un classement en fonction de leur capacité à atteindre un état optimal (ou à s’en approcher) sur chaque piste.

The modern Anthropocene would not exist without electricity. It is the ticking clockwork that guarantees and dictates the rhythm of modernity in all of its beneficial and challenging extents. Electricity generation is a complex process with various inter-dependencies. It is thus important to survey, maintain and continuously adapt all contributors to guarantee a stable electricity generation. This is all the more valid as humanity became aware of the negative repercussions of its constant striving for growth and wealth that provoked various threats, ranging from the climate change over poverty to health issues. In this light, the Sustainable Development Goals were created that should help to assess and guide humanity to a more sustainable way of thriving. The Sustainable Development Goals address several targets, one of them being sustainable (electric) energy supply. The electricity generation’s dependencies are manifold and certainly dependent on the type of generation, but one constant prerequisite is (almost) inevitable for nearly every type of electric generation facility: water. There exists a vast energy-water context with various threads of dependency. It is safe to say that electricity generation without continuous water supply is not secure. And though water is also a special target within the Sustainable Development Goals, the analysis and evaluation of those inter-dependencies between energy and water are scant. Few are the tools that exist to survey, assess and remedy energy-water context challenges. What is more, electricity as pillar of our Anthropocene is already a globally implemented infrastructure that demands constant management action. Yet, the research of the sustainable and secure energy-water context management on i its various levels - ranging from macro-scale analysis and strategy development over meso measures of transference to the mindset behind the micro actions that maintain the electricity generation - is not too advanced, though the topic is of utmost importance. This dissertation investigates on various levels to develop and survey methodologies that reveal and remedy the energy-water context challenge. It does so with five studies. Three of them investigate the special issue of reservoir sedimentation as a prime example of threatened energy-water infrastructure, whose management needs to be surveyed. Out of the five articles, one is already published, one is in press and three are under review. The research will be presented in five chapters. Chapter 1 prepares the ground of the thesis as an introduction. The nature of energy-water dependency is demonstrated and the lack of energy-water sustainability research outlined. On various levels, central research questions for sustainable and secure energy-water management are developed for the thesis. As reservoir sedimentation as a special case and artificial neural networks as a research methodology are of key importance, their principles and backgrounds are illustrated. Chapter 2 surveys the possibilities to evaluate, analyze and assess the multifaceted nature of water-dependent electricity generation. It lights on an essential gap of holistic energy-water security assessment and fills this gap with a broad methodological approach for holistic energy-water security assessment. In Chapter 3 the transference of developed energy strategies to the level of application for an energy-water-(food)-context is investigated. A gap between public professionals and other stakeholder groups as major inhibitors is identified. Within the chapter, an approach to overcome this gap is developed and investigated in a case study in Ouarzazate, Morocco. Subsequently, the degree of security and sustainability thinking of the mindset behind applied energy-water management action is subject to investigation in Chapter 4. This is executed using the example of reservoir sedimentation in Japanese reservoirs. The optimism bias, an influential and non-sustainability mindset in infrastructure management, is used as a proxy to do so. Artificial neural networks serve as a prime tool to derive evidence. Management action is bound to have (expected or unexpected) effects. In the case of reservoir sedimentation in Japan, a mass data methodology based on artificial neural networks is developed in Chapter 5 to extract traces for such effect. It is based on a thorough data set of 1225 Japanese reservoirs with (among others) individual 18 year sedimentation and precipitation time series as well as continuous management action notations. The key element is a Gated Recurrent Unit (GRU) core of the neural networks that allows a memory function. The extensive research reveals evidence of concrete management action on a meso scale. The conditions of the energy-water context are globally quite different. Chapter 6 is another case to survey the effect of management action, investigating again via artificial neural networks on reservoir sedimentation. This time, the study is settled in the state of Ceará in Brazil and the focus object is a certain, ii presumably sustainable, management directive of the state governance. The results are discussed in Chapter 7, where a conclusion and outlook of the dissertation is given. The dissertation reveals the Gordian web of multi-leveled governance and management of the complex energy-water context. It is emphasized that the presented findings are not the only way to respond to the established research questions, since the results are by no means of panacea character. Rather, the outcomes of the dissertation are very worthwhile tools that bear the flexibility of being applied to the highly variable challenges of the energy-water context. The dissertation is thus a valuable contribution to establish a secure and sustainable utilization of (water) resources and (electricity) infrastructure within the modern Anthropocene.