Análise da influência dos períodos críticos na operação otimizada do sistema de usinas hidrelétricas da bacia do rio São Francisco utilizando programação não linear

Abstract

The evolution of water resource management, especially reservoir operation, is constantly changing over the decades. Currently, one of the greatest challenges is to operate hydrothermal networks subject to the randomness of hydrological events, atmospheric fluctuations and climatic change and, as a consequence, to the possible reduction of reservoir inflows. Conflicts among users are already registered and there are direct impacts on the generation and commercialization of electricity. In addition, the definition of the critical period directly interferes with the energy sharing of Brazilian hydroelectric plants. Northeastem Brazil has historically faced periods of water deficit, making the São Francisco River Basin (SFRB) important for the region’s development, mainly with the implementation of the São Francisco River and Northern Northeast River Basin Integration Project. In view of the above, the present study aims to conduct a study regarding the statistical behavior of time series related to the management of the SFRB hydrosystem, as well as to measure the performance of the hydroelectric reservoir system plants in supplying the multiple users: electrical, consumptive and environmental uses, given the different critical periods. Historical records of natural inflows, natural inflow energy and stored energy were employed. As the basic assumptions of randomness, independence, homogeneity and stationarity are considered in hydrological assessments, the following assessments were conducted: regarding the presence of trends, the Mann-Kendall, Spearman and Pettitt tests were applied; as for randomness, the single-sample runs test was applied; as for independence, the Wald and Wolfowitz test was applied and, for homogeneity, the Mann-Whitney test was applied. Two methodologies for defining the critical period were compared: according to the natural inflows and according to the stored energy. Subsequently, the commitment of a deterministic operation optimization model in assessing different uses was assessed, using nonlinear programming, for the different critical periods. It was observed that the series of natural inflows, natural inflow energy and stored energy are predominantly non-stationary, with a change in trend in the 1990s, directly interfering in the critical period of the basin, assessed between 2013 – 2019. When comparing the performance of operating models in nonlinear programming (NLP) with the standard operating policy model, there was a significant decrease in the vulnerability of the system.