Modelagem estocástica explícita da relação capacidade-regularização-confiabilidade-vulnerabilidade em reservatórios

Abstract

The study of the dynamics involving surface water reservoirs seeks to find ways to help design dams capable of withstanding the scarcity of rainfall, the hydrological uncertainties, and maintaining water availability for the population. This work proposes refinements to the explicit stochastic method of Fletcher-Ponnambalam (FP) for sizing reservoirs in order to build storage-yield-reliability-vulnerability (SYRV) relationships. One of the improvements to the original method is the incorporation of the monthly scale, instead of the originally used annual time period. Thus, based on pre-established demand and capacity, monthly reliabilities are estimated as well as probabilities of containment, deficit and spill together with statistical moments of storage. Another improvement is to include estimates for monthly vulnerabilities and for the moments of deficit and spill, absent in the old formulation. To validate the model, historical records of natural flow to the Sobradinho reservoir are used. After running the FP procedure, a monthly simulation of the operation is conducted for synthetic scenarios guided by the standard operating policy. The goodness of fit between model and simulation results is measured by the error sum of squares (SSE), the mean squared error (MSE) and the Nash-Sufcliffe efficiency index (NSE). In addition, a residual sample (model minus simulation) is created between all moments of each month. The sample is applied to parametric and nonparametric tests to verify the null hypothesis of zero mean. The results indicate that SSE and MSE are close to zero and that the NSE is close to maximum efficiency. Furthermore, both tests do not reject the null hypothesis, that is, the comparison between model and simulation is adequate. The study also includes the development of SYR curves and equations relating the variables involved. The FP-SYRV model demonstrates satisfactory results for the various conditions tested, reinforcing its potential for use in the evaluation and sizing of reservoir systems.