Risk Assessment Of Embankment Dams Event Tree And Bayesian Network Method

The total number of existing dams in China is approximately98000,of which more than90%are embankment dams. Most of these dams were built during the1950s-1970s,and are associated with poor design and construction due to limitation of knowledge and construction technology at that time. In addition, after operation of several decades, these dams will exist defects and risks of failure in different degrees, which seriously restrict normal functioning of the social and economic benefits of the reservoirs, and pose a significant threat to the downstream people’s lives and property, especially in densely populated areas. Therefore, there is pressing and practical sense to carry out risk assessment based on the safety of the structures of distressed dams, which can make sure of the comprehensive benefits of reservoirs, can ensure the public safety of downstream and improve the management level of dam safety. Combined with the Risk Mitigation and Strengthening of Endangered Reservoirs in Shandong Province project financed by Asian Development Bank loans, in this dissertation, an assessment method of potential risks of embankment dams during operation period is studied based on the in-depth understanding of the achievements of the risk analysis of dam failure at home and abroad, which is started from the identification of the potential risk factors and breach models. The main contents are as follows:Ⅰ. Summarize the current development of the basic theories on risk analysis of dam failure at home and abroad, and several commonly used risk assessment methods of dam failure and their application are introduced in detail.Ⅱ. Identify the potential risk factors and breach models of distressed embankment dams. A preliminary failure probability can be gotten using an ealier method based on history failure rates, and then identify the potential risk factors and breach models with this preliminary failure probability based on the Bayesian network backward reasoning process.Ⅲ. Taking the Qiangkuang reservoir in Shandong Province for example, establish the event tree models of embankment dam failure, analyze the relationship between the various events and dam failure, and calculate the probability of dam failure probability according to the basic principle of event tree.IV. Utilizing the Bayesian network which can solve the malfunction caused by the uncertainties of complex systems, a Bayesian network model of Qiangkuang reservoir is established according to the mechanism of the failure of embankment dams. Then the influence between various risk factors is analyzed and the failure probability of the dam is calculated.V. Compare the results calculated by Bayesian network method and event tree method, and study the process of transformation from the event tree to the Bayesian network, then come up with the Bayesian network method based on the event tree for the analysis of embankment dam failure.The Bayesian network method based on the event tree proposed in this dissertation is scientific and feasible, which can not only indicate the causal relationship between the dam breaks more intuitively, solve the problem of dam failure caused by common factors more easily, and the reliability analysis of dams can be carried out more effectively, which can be referenced for making rehabilitation scheme of distressed embankment dams.