Hydrothermal scheduling with variable head hydroelectric plants: Proposed strategies using benders decomposition and outer approximation

Hydrothermal scheduling is a complex task that must deal with problems distinguished by a stochastic, multi-objective, dynamic, nonlinear and non-convex character. This paper proposes two decomposition methods to address the short term hydrothermal scheduling of systems that have hydroelectric plants with nonlinear or non-convex productivity functions. One is based on the Benders Decomposition principle and the other on the Outer Approximation algorithm. Both strategies allow one to deal with multi-stage mixed integer nonlinear problems, by decomposing the original problem into problems of known structure, smaller size and complexity level, which can be later approached using the powerful mathematical optimization solvers available today. By means of two case studies, it will be demonstrated that the proposed strategies are effective and useful.