Aplicação de modelos regionais e conceituais em estudos hidroenergéticos: uma abordagem bayesiana

• Autor(a) principal: Damazio, Jorge Machado
• Data de Publicação: 1980
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFRJ
• Texto Completo: http://hdl.handle.net/11422/2998
• topic: Monitoramento ambiental; Amostragem; Inferência Bayesiana; Reservatórios; CNPQ::ENGENHARIAS::ENGENHARIA CIVIL::ENGENHARIA HIDRAULICA::HIDROLOGIA

Aims to quantify the effect on reservoir design of regional and conceptual information as well as measures of sample variation. The subject is presented by means of the case study of Carolina hydro-electrical plant on the Tocantins River. The adopted method is general and may be developed for other cases. Bayesian theory is the framework from which the main results are derived. Background material of this theory for independent and normally distributed mean anual flows, are presented in concise form. Other stochastic models are not considered. The physiographic, precipitation and streamflow data for the Tocantins basin are obtained from files developed by the Projeto de Hidrologia e Climatologia da Amazônia (SUDAM) that used the "square grid method". A multivariate regression model has used these informations to estimate the mean and variance of the anual flows at Carolina. A conceptual model of the water balance is utilized to extend the streamflow data through the use of precipitation data. Then the extended serie is used to obtain new estimates. The generation of the hydro-electric plant is defined by the relation between annual power demand and the risk of falling below this target. Adopting the desired risk equal to 5%, the capacity of the plant is estimated for three cases: (i) Using only the sample information; (ii) Using the sample combined with the results yielding from the conceptual model results; (iii) Using the sample combined with the results yielding from the conceptual and regional models. The Monte-Carlo method is used to obtain confidence intervals for the risk.