This paper aims at improving control systems for hydro-power production, by combining model predictive control techniques with decomposition-coordination methods for a global optimization over a whole hydro-power valley. It first recalls the model predictive control formulation for a centralized solution presented as the reference for comparison, and emphasizes the possible use of explicit solutions in the considered problem, making easier its practical use. It then highlights the further interest of such solutions in a decomposition-coordination approach, allowing to reduce the computational cost even more with a purpose of realtime implementation, and at the same time to take advantage of the distributed nature of the considered system. The results are illustrated on the basis of a real-data-based case-study provided by EDF group.