Dynamic water resources planning with locational release and annual consumption constraints

Water resources planning is very important for water resource management and electric energy production. The problem is challenging in view of the stochastic system dynamics, nonlinear rewards, coupled hydraulic constraints, and large problem size. Existing methods based on discretization are facing the dilemma of solution accuracy and computational efforts. This paper formulates the dynamic water resources planning problem with locational release and annual consumption constraints as a finite-horizon Markov decision process (MDP) with continuous variables, and develops a sensitivity-based approach to optimize the policies. Numerical results based on a practical system on Yellow River in north China demonstrate the effectiveness of the formulation and the efficiency of the new algorithm.