Simulação estocástica na estimativa de assoreamento em reservatórios

Abstract

Reservoirs are built for various purposes such as power generation, water supply, etc. These structures are subject to some degree of sedimentation, since by altering the river's equilibrium, its capacity to transport sediments is altered. Such siltation, among other problems, may interfere with the use for which the reservoir was built. Thus, the height of deposited material must be estimated and when the accumulated sediments will begin to interfere with the functions of the reservoir. However, predicting the accumulation of sediments is difficult because the processes involved are complex, subject to temporal variability and uncertainties, which makes the study not only deterministic but also stochastic. Thus, the objective of this research was to develop a stochastic model and to evaluate its performance in the estimation of sedimentation in reservoirs. For the realization of this project, the hydrosedimentological and topographic data of the Salto do Paraopeba PCH was used, which was built in 1956, and the reservoir is intensely silted, which made it inoperative. In the CPH of the UFMG there is a reduced model of the reservoir of this PCH, and the result of silting observed in this model was used to validate the stochastic model. The discharge data and sediment concentrations in suspension of the PCH were obtained, which were converted to the reality of the reduced model according to the hydraulic similarity scales. From these data, thousands of synthetic series were stochastically generated, using statistical software R and model AR(1). The data generated were introduced in the HEC-RAS software to estimate the siltation in the reduced SHP model. For this, a computational code was developed that allows the automatic coupling of the stochastic model with the deterministic one. The result obtained by stochastic simulation was compared with the sedimentation measured in the physical model, observing that the actual siltation for the two periods analyzed (2008-2012 and 2013-2017) was always between the 1st and 3rd quartile of probability of the result stochastic, that is, the actual silting was always greater than 25% of the stochastically generated values and less than 75% of them. Thus, it is possible that the stochastic model can help in future projects of estimation of sedimentation in reservoirs, since it allows to obtain the probabilities of heights silted in sections of interest.