With the rapid development of industry and projects for ensuring the amounts of water that those industries require, the capacities and scales of hydraulic structures including reservoirs, dams, and weirs have increased sharply. Numerous dams experience some seepage, when water is held behind them; thus, dam leakage detection has been and continues to be a major concern. The basic task involves finding the path of the leakage, especially the concentrated leakage path. Using tracer techniques, we treat dam leaks as multiple attribute evaluation problems, which is one innovation presented in this paper. Entropy weight-set pair analysis, a new method of identifying leaks, is introduced. The water samples and the information flows including conductivity, pH, Cl<Superscript>−</Superscript>, D, and <Superscript>18</Superscript>O are set as alternatives and indexes, respectively. Then, the connection degree is calculated according to set pair analysis, followed by the weight of each index using an information entropy weight method based on the principle of maximum entropy. Finally, the integrated degree of connection of all the samples is calculated, to provide a basis for classifying water samples and judging the adequacy of the water resources. To verify the validity and feasibility of this proposed model, the Ling’ao reservoir, which is situated in Guangdong province of China’s southeastern monsoon zone and notable for its particularly heavy seepage is taken as a case study. The results indicate that the model is suitable for objectively analyzing dam seepage problems, especially given that the results match those obtained for the horizontal velocity as measured with drillings and the temperature distribution across the dam body. Copyright Springer Science+Business Media Dordrecht 2015
