The generation of synthetic inflows via bootstrap to increase the energy efficiency of long-term hydrothermal dispatches

Abstract

One of the most important techniques used to study long-term energy operation planning is the stochastic dual dynamic programme (SDDP). In large systems, hydraulic power plants are aggregated in so-called equivalent energy systems, where the inflows into hydro reservoirs are represented by the affluent natural energy (ANE) and the stored volumes are represented by the stored energy. The stochasticity of energy inflows is captured by the historical series ANE. Currently, ANEs are studied using the Box–Jenkins methodology to fit periodic autoregressive models (PAR(p)) and their order (p). A three-parameter log-normal distribution is applied to the residuals generated via PAR(p) modelling to generate synthetic hydrological series similar to the original historical series. However, the log-normal transformation incorporates non-linearities that affect the convergence in SDDP. This study incorporates the bootstrap statistical technique to determine the order p of the PAR(p) model to generate synthetic scenarios that will serve as a basis for SDDP application. The results indicate the adherence of the proposed method on the operational planning of hydrothermal systems. The proposed methodology in this article could successfully be applied in hydro-dominated systems such as Brazilian, Canadian and Nordic systems.