Développement d’un modèle de gestion des réservoirs pour la production hydroélectrique: Cas du bassin de la Sanaga au Cameroun
Files
FRITZ_BASSAMA_12192400_2026.pdf
Embargoed access from 2028-01-01 - Adobe PDF
- 12.73 MB
Details
- Supervisors
- Faculty
- Degree label
- Abstract
- The production of hydropower in large river basins relies on interconnected systems, where storage dams regulate the discharge that feeds power plants located hundreds of kilometres downstream. Operating such a system requires predicting where the water is, when it will arrive, and how much of it there will be. This task is particularly challenging when hydrometric data is scarce, as is often the case in many high-potential African basins. This thesis therefore presents a flow and storage simulation model designed for contexts with limited data and applies it to the Sanaga River basin in Cameroon. Due to the absence of bathymetric data, the geometry of the network, comprising over 1,000 km of rivers, is derived entirely from the global HydroSHEDS digital elevation model. In the absence of recent discharge records, the routing model, based on the variable-parameter Muskingum-Cunge method, is calibrated against reference HydroFlow1D hydraulic simulations solving the Saint-Venant equations. The model reproduces these with a Nash–Sutcliffe efficiency above 0.97 in 15 out of 18 configurations. The complete network is then represented as a directed graph connecting dams, confluences, and river reaches and is fed with synthetic seasonal hydrographs. The resulting simulator enables several release management rules to be compared during the low-flow season, ranging from open-loop control to predictive regulation. The results demonstrate that operating performance is more dependent on the operator's ability to distinguish natural inflows from artificial ones than on the refinement of the physical model. Thus, this work demonstrates that an operational management support tool can be built, calibrated and run in a context of data scarcity.