Thèse Co-Design Optimal Robuste du Navire-Hôpital « Bongo » Doté de Systèmes de Production et de Stockage d'Hydrogène H et d'Oxygène O H/F Doctorat.Gouv.Fr

Établissement : Institut National Polytechnique de Toulouse École doctorale : GEETS - Génie Electrique Electronique,Télécommunications et Santé : du système au nanosystème Laboratoire de recherche : LAPLACE - Laboratoire PLAsma et Conversion d'Énergie Direction de la thèse : Xavier ROBOAM ORCID 0000000233206668 Début de la thèse : 2027-09-01 Date limite de candidature : 2026-11-23T23:59:59 Ce proposition de recherche porte sur le ''co-design'' (intégrant dimensionnement et gestion énergétique) robuste et optimal d'un microréseau d'énergie intelligent isolé. Ce microréseau est destiné au navire-hôpital ''Bongo'', un navire de la Marine colombienne opérant dans la région amazonienne. Le système proposé intègre une production d'énergie solaire photovoltaïque (PV) à une électrolyse de l'eau, permettant ainsi la production simultanée sur site d'hydrogène vert - destiné à la régénération d'électricité via une pile à combustible - et d'oxygène de qualité médicale, répondant ainsi à deux besoins opérationnels essentiels au sein d'une architecture unique et unifiée. Un système de stockage d'énergie par batterie (BESS) est envisagé comme tampon complémentaire afin d'améliorer la qualité de l'énergie et la dynamique du système d'hydrogène. L'approche de ''co-design'' associe le dimensionnement des composants du micro-réseau à l'optimisation du système de gestion de l'énergie (EMS), minimisant à la fois le coût total (NPC: Net Present Cost) sur un cycle de vie de 20 ans et le potentiel de réchauffement global équivalent (GWP: Global Warming Potential, en kg COeq), tout en respectant les contraintes de fiabilité de l'approvisionnement en électricité et de production d'oxygène. La quantification des incertitudes inhérentes à ce microréseau est intégrée tout au long du processus de conception afin de tenir compte de la variabilité des profils d'irradiance solaire, de la demande de charge et des paramètres de modélisation critiques - en particulier la dégradation des dispositifs de stockage. La dimensionnalité de ce problème d'optimisation stochastique atteint plusieurs centaines de milliers de variables de décision, ce qui nécessite des avancées méthodologiques fortes, basées sur la programmation linéaire et la conception robuste par scénarios.
Une validation expérimentale de type « Power-Hardware-In-the-Loop » (PHIL) est prévue lors de la phase finale du projet, en s'appuyant sur la plateforme d'essais de micro-réseaux du LAPLACE à Toulouse.
Ce projet est mené en co-tutelle entre le LAPLACE (Université de Toulouse, France) et la Pontificia Universidad Javeriana (Bogotá, Colombie). Javeriana (Bogotá, Colombie). Integrated resource planning frameworks have evolved progressively from early Water-Energy (WE) approaches toward more complex configurations: the Water-Energy-Food (WEF) nexus, the Water-Energy-Carbon (WEC) nexus, and the Water-Energy-Hydrogen (WEH) nexus. Each extension responded to the need to incorporate new dimensions of sustainability, emissions, and energy security within integrated planning frameworks. Besides the WEH nexus, oxygen generated during water electrolysis has been treated consistently as a residual by-product without strategic value, excluded from optimisation models and from energy value chains. Most existing models represent only partially the water-energy-hydrogen interactions, excluding oxygen as an endogenous decision variable in planning processes. There is a limited use of multi-objective formulations capable of simultaneously capturing trade-offs between energy performance, environmental sustainability, and economic profitability. Furthermore, applications in the hydroelectric, hospital, isolated zones, and industrial sectors are separately analyzed, without integrated frameworks that enable the evaluation of multi-sectoral synergies within circular energy systems. The global oxygen market has grown to a scale that gives this co-product significant economic relevance within future electrolysis-based energy chains. In response to these limitations, the Water-Energy-Hydrogen-Oxygen (WEHO) nexus is proposed as a circular and integrated evolution of existing paradigms, one that explicitly incorporates oxygen valorization within long-term energy planning and operation frameworks. The WEHO framework transforms the linear hydrogen production chain into a circular, multi-product architecture: while hydrogen is distributed toward energy, industry, and transport sectors, oxygen is redirected toward environmental, industrial, and health applications, generating additional economic and environmental benefits that conventional models do not capture. Several sectors concentrate the greatest opportunity of electrolytic Oxygen: environmental restoration in hydroelectric systems due to deep-water deoxygenation, supply security in hospital infrastructure, process efficiency in industrial sectors such as steelmaking, combustion, and chemical synthesis, which constitute high consumers of oxygen. Waste water treatment through aeration processes are also developed. Oxygenation of aquaculture systems and the restoration of degraded water bodies also represent emerging applications with high environmental and socioeconomic value. Finally, space exploration presents the ultimate version of the energy-oxygen autonomy problem. More information about that context and state of the art is provided in Appendix 3 (see joint Appendix document.pdf).
Beside this WEHO nexus, integration of renewable energy production, especially with solar photovoltaic generators, is a key lever in terms of decarbonisation. However, renewable generators provide intermittent power, which must be managed in smart grids. In that context, smart microgrids constitute key local entities able to couple renewable generators (solar array or wind turbines) with storage capabilities to satisfy local load demand. Standalone microgrids decoupled from public grids are essential for remote areas or for electromobility (automobile, railway, aircraft and naval applications) as in the case of an electric ship.
The Colombian Navy thus operates a hospital ship known as the Bongo Hospital Ship, stationed in Puerto Leguízamo (Putumayo), which carries out significant social and humanitarian work along the binational riverbanks of Colombia-Ecuador and Colombia-Peru, providing medical care and development support to some of the most remote communities in the Amazon region. However, the Bongo Hospital Ship currently depends on energy supplied by an auxiliary motor ship with subsequent fuel burn, as it does not have its own installed capacity for electricity generation. In addition, there is a significant logistical challenge associated with the supply of medical oxygen, which is essential for healthcare services and difficult to provide in hard-to-access areas such as the Colombian Amazon.
'The Bongo Hospital Ship represents a particularly compelling implementation of the WEHO nexus because it integrates, within a single isolated platform, the four resource dimensions that define the framework: water as the feedstock for electrolysis, renewable electricity as the conversion driver, hydrogen as a long-term energy carrier for energy storage and regeneration, and oxygen as a critical output for healthcare services. Unlike most stationary applications, where oxygen can be supplied through established logistics networks, a hospital vessel operating in the Colombian Amazon faces significant geographical and operational constraints that make on-site oxygen production highly desirable and strategically important. Under these conditions, oxygen cannot be regarded merely as a secondary by-product of hydrogen generation, as is commonly assumed in the literature. Instead, it becomes a strategic resource whose availability directly affects the continuity and quality of medical services.
This perspective fundamentally redefines the role of oxygen within the WEHO nexus. The design challenge is no longer limited to maximizing energy autonomy or minimizing operational costs, but must also guarantee the continuous production of medical-grade oxygen under variable operating conditions. Although Proton Exchange Membrane (PEM) electrolysis can produce high-purity oxygen suitable for healthcare applications, maintaining the quality, reliability, and safety requirements associated with medical use introduces additional constraints related to system operation, water management, storage, and monitoring. Consequently, oxygen quality and availability must be considered from the earliest stages of system planning and optimization rather than being treated as secondary operational considerations.
The Bongo Hospital Ship therefore serves as more than a representative application of the WEHO nexus. It provides a highly demanding testbed in which energy security, water utilization, hydrogen storage, and healthcare continuity are intrinsically interconnected. The coexistence of these objectives within a geographically isolated and resource-constrained environment creates a unique scenario for evaluating integrated resource management strategies and demonstrates the broader potential of the WEHO framework for critical infrastructure applications.'
In order to increase operational autonomy, reduce logistical dependence, and strengthen environmental sustainability, this project proposes equipping the Bongo Hospital Ship with an isolated microgrid based on solar PV and Hydrokinetic production hybridized with a battery bank, which helps stabilizing this standalone system. This supply part is coupled with hydrogen technologies, capable of supplying the electrical energy required for its operation while producing medical oxygen on site. Illustrations of the Bongo Hospital Ship and on the corresponding microgrid topology are provided in Appendix 2 (see joint Appendix document.pdf).
This system will reduce CO emissions, eliminating the use of fossil fuels, and optimize operational costs, while strengthening medical response capacity in remote areas. It must be pointed out that, beyond the numerous technologic and scientific challenges addressed in this project, coupling hydrogen and oxygen production is original in itself while oxygen production from electrolysis is usually wasted! Benefits of the proposed project are both applicative and scientific especially with the development of methodological aspects:
- The applicative objective deals with the design and the optimal control of an isolated smart microgrid coupling both solar PV and Hydrokinetic generators with an electrolyzer. A battery storage may be added to support the hydrogen system. The electrolyser splits water molecules into hydrogen and oxygen: the oxygen will be used for medical purpose, while the hydrogen will be stored and later used in a fuel cell to generate electricity hybridized with the battery device. In this way, two critical needs-energy and medical oxygen-are addressed simultaneously, expanding the operational scope and healthcare capacity of the Bongo Hospital Ship.

The scientific objectives are double:
- First, the optimal design of the microgrid also called energy planning will be achieved. We propose to focus on an optimal co-design which integrates microgrid sizing (PV array surface and hydrokinetic turbine size? Electrolyzer and fuel cell stack power? H2-O2 storage capacity? Battery capacity?) with Energy Management System (EMS) dealing with management of storage trajectories. Objectives of this optimal co-design are twofold: minimizing the Net present Value (NPV) of the microgrid over its life cycle (typically over 20 years) and its carbon impact by minimizing the Global Warming Potential (GWP in kg CO2eq) while ensuring constraints in terms of electricity and oxygen production. In terms of economical KPIs, the levelized cost of energy (LCOE), levelized cost of hydrogen (LCOH) will be assessed besides the NPV. The critical innovation of WEHO planning lies in incorporating oxygen as an endogenous decision variable alongside hydrogen and energy flows.
A particular attention must be paid on the aging of certain devices (electrolyzer, fuel cell, battery) which must be considered as it affects the system cost depending on the number of device replacements during the life cycle. Thus, the complexity of this optimal co-design process is huge due to its dimensionality -number of decision variables is classically in the range of several hundred thousand decisions variables-. Furthermore, as we look for a Robust optimal co-design under uncertainties, the complexity is all the more important as a subsequent number of scenarios must be played to cope with the uncertain character of the system. Uncertainties are due to variability of environment profiles (renewable energy production, local load demand in terms of electricity and oxygen production) and to the model itself as modeling parameters are often clearly uncertain: as example, the storage device aging is typically an uncertain phenomenon, which has to be integrated in its model.
- Second, once the microgrid has been sized, its EMS must be optimized at real time by developing robust optimal control. Indeed, the quality of any planning solution is ultimately determined by not only the optimality of infrastructure sizing in isolation, but also by how well that infrastructure performs under real operating conditions, which is precisely where the operational dimension becomes essential. Operation governs short-term decisions spanning minutes to weeks: how much hydrogen is produced at each moment, how oxygen is allocated across competing sectoral demands, how storage is cycled, and how the system responds to renewable variability, demand fluctuations, and contingency events. Within the WEHO framework, this operational complexity is compounded by the need to simultaneously manage energy balances, oxygen supply continuity across sectors with heterogeneous criticality levels, and environmental constraints such as minimum dissolved oxygen thresholds in reservoirs.
A receding-horizon Model Predictive Control (MPC) framework is particularly well-suited to this layer: by continuously re-solving the optimization problem over a rolling time window using updated measurements and short-term forecasts of renewable generation, oxygen demand, and load profiles, MPC ensures that dispatch decisions remain dynamically consistent with both physical system states and the strategic priorities established during planning. Critically, this operational layer must embed non-negotiable priority hierarchies that cannot be traded against economic objectives under any circumstances, in hospital applications, for example, oxygen supply continuity and minimum safety inventory levels must remain guaranteed regardless of energy prices or renewable availability. These hard constraints, if not encoded at the operational level, render planning solutions infeasible in practice, and this is precisely the fundamental reason why planning and operation cannot be designed in isolation.
At the end of the project, experiments on real time optimal control of the whole microgrid may be developed thanks to LAPLACE facilities by implementing a PHIL (Power Hardware In the Loop) experimental process. Addressing the WEHO Nexus gap requires a structured research framework built on a fundamental premise: planning and operation are not sequential tasks but inherently interdependent layers of a unified optimization problem. Conventional approaches treat long-term infrastructure decisions and short-term operational control as decoupled stages solved in succession, first size the system, then operate it. This separation is inadequate for WEHO systems, where the full value of oxygen can only be captured when infrastructure design and operational strategy are developed jointly, as each directly constrains and informs the other.
The complexity of the robust optimal co-design under uncertainties induces actually methodological breakthroughs. As mentioned before, the dimensionality (number of decision variables, number of constraints) is so huge that the tractability of the design process is questioned. Optimization based on linear programming theories is often the efficient and quasi-unique solution in systemic context. Several studies are developed on that aspect, some of them proposed by LAPLACE [Antunes 2025] [Bergougnoux 2026], [Boennec 2025], [Radet 2022]. But if the EMS is not the same in the co-design process as in the real time optimal operation of the control, the question of the adequation between those two phases must be addressed. [Radet 2022] has shown that the level of optimality of the control used both in the co-design and in the real time operation must be close. The case study related to the Bongo electric Ship is perfect to address these methodologic issues.
A set of models must be built to address all objectives previously described and to be in accordance with the optimization methods addressed in each project step. A particular attention must be paid on the development of aging models: existing studies will guide this modeling effort [Antunes 2025], [Boennec 2025]. Models for the robust optimal co-design have to be linearized if linear programming methods are used in that process. Depending on the real time and operation control methodologies applied in the second step, complementary models should be developed. On the Colombian partner side, complementary expertise on energy management systems for hydrogen-based microgrids in the regional context [González-Madrid 2025] and on robust energy management under forecast uncertainty [Valencia 2015] will support the operational layer, while the strategic relevance of hydrogen deployment across Latin America is documented in [Gómez 2025].