Avaliação das probabilidades de falhas em barragens, associadas a eventos de naturezas hidráulicas e hidrológicas: estudo de caso da PCH Cajuru

• Ano de defesa: 2014
• Autor(a) principal: Fernando Neves Lima
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Meio Ambiente e Recursos Hídricos; Saneamento; Método de Monte Carlo; Recursos hídricos Desenvolvimento; Avaliação de riscos ambientais; Engenharia sanitária
• Link de acesso: http://hdl.handle.net/1843/BUOS-9KYHX3

The concept of risk related to the potential breach of a dam is a current problem in engineering. Dams have an important role in society and the breakdown of such structures carries large disasters. This work aimed to summarize the quantitative methods, which most have been widely used, to evaluate the probability of a dam overtopping, caused by events of hydraulic and hydrological nature and apply them in case study. To this end, the fault tree analysis method (FTA), the technique of numerical simulation known as Monte Carlo simulation and event tree analysis method (ETA) have been used. In the first method, the goal was to describe an application of the Fault Tree Analysis to evaluate the probability of failure of a tainter gate of a dam spillway. Although FTA does not seem to be a well explored method in the technical literature, with just a few applications in the domain of water resources engineering, the focused method has proved to be useful and practical. In this analysis, the top event was defined as a failure at one of eight spillway gates that compose the surface of the small hydropower plant of Cajuru, located in the Pará river, in the brazilian state of Minas Gerais. All gate mechanisms and operating principles have been analyzed. The final result was considered satisfactory and compatible with those obtained in reported similar studies. In sequence, the Monte Carlo simulation was used with the objective to evaluate the probability of overtopping of Cajuru dam, due to a flood of a certain return period, considering the uncertainties in the local frequency analysis of maximum annual discharge. The Monte Carlo method, which account with the uncertainties of estimates of certain return period events, showed that can be useful in the complete evaluation of the probability of overtopping. In case study, the simulation showed that Cajuru dam has a not negligible degree of vulnerability to events with great scale. Finally, an application of event tree analysis was made, which revealed very useful for identification of sequences of events, if the floods with bigger return periods reached to Cajuru reservoir. Although be widely spread in the technical field the adoption of probabilities by judgement of experts, this research used a different approach, with assignment of probabilities in the ETA by fault trees and Monte Carlo simulation, showing compatibles results for the possible overtopping with values reported in literature.