Long-Term Scheduling for Cascaded Hydro Energy Systems With Annual Water Consumption and Release Constraints

Abstract

Long-term scheduling for cascaded hydro energy systems is very important for low carbon energy production. It aims at determining the water release over a planning horizon to meet water resource requirements and the system demands for electric power. The problem is challenging in view of the complicated and stochastic system dynamics, nonlinear marginal cost, coupled hydraulic constraints, and the large problem size. In this paper we formulate the long-term scheduling problem of cascaded hydro energy systems with annual consumption and release constraints as a finite horizon constrained Markov decision process (CMDP), and develop a new rollout algorithm to optimize the policies. Numerical results demonstrate the effectiveness and the efficiency of the formulation and the new algorithm.Note to Practitioners-Long-term scheduling of cascaded hydro energy systems is an important planning activity in the reservoir management and electric industry. This paper formulates a long-term scheduling problem for cascaded hydro energy systems with annual water consumption and release constraints as a finite horizon CMDP, and develops a new rollout algorithm to deal with the large problem size. Motivated by a practical project on Yellow River in north China, this work has extended applicability since the new algorithm can start with any feasible policies currently used in practical systems and improve their performance. Numerical results demonstrate the effectiveness and efficiency of the formulation and the new algorithm.Index Terms-Cascaded hydro energy systems, constrained Markov decision process (CMDP), long-term scheduling, rollout algorithm.

An efficient stochastic algorithm for mid-term scheduling of cascaded hydro systems

“…However, this paper focuses on the vertically integrated utility in China, and the energy price is regulated and set by the government [7], hence, energy price parameters in the studied time horizon are set to be 1, and equivalently the objective of this problem is to maximize the total generation over the planning horizon.…”

“…In [7], a stochastic optimization problem with Markov decision process models was developed allowing for the constraints of governmental regulations. In [8], a novel stochastic risk-constrained model was discussed.…”

Sample path sharing in simulation-based policy improvement

“…SBPI improves given base policies through simulation, and has been successfully applied to many problems including the quiz problems, wireless sensor network [18], water source management [19], and engine maintenance problems [20]. Because simulation could be very time-consuming and a large number of replications is usually used in SBPI, it is of great practical interest to seek a method that efficiently uses the sample paths so that the best action is selected with high probability.…”

Dynamic water resources planning with locational release and annual consumption constraints

“…We have developed a sensitivity based approach for finite horizon MDPs with discrete state and action spaces in our previous work [24][25] [28]. In this section, we will extend the main result to the finite-horizon MDPs with continuous state and action spaces in order to solve the water resources planning problem formulated in the previous section.…”

“…Problem approximations, like aggregation and decomposition of reservoirs, are considered in [9]- [11]. In our previous work, the water resources planning problem is formulated as a constrained Markov decision problem and a "rollout" based approximate dynamic programming method is developed to solve the problem with quantified approximation [28]. The major issue of the existing methods is that discretization in decision space is generally needed and we have to deal with the dilemma of approximation precision and computational efforts.…”