Protective levees against flooding are large alignment structures that are critical for the safety of goods and people. In urban areas, it is common that these structures are used as mobility networks or are crossed by other technical networks.
Thus the failure of a protective structure affects not only the immediate safety of the protected area but also the risk of networks dysfunctions detrimental to the operation of the whole urban area. However the general lack of data to describe the behavior of dams during unwanted events led to estimate their resistance ability only on the basis of uncertain expert statements. This is especially true when we have to consider the presence of various networks inside the structures. In this context, our research aims to produce effective scientific methods and tools for the formalization and use of expert advice in an uncertain environment. In particular, we propose to develop an approach to reduce biases due to the subjectivity of expert evaluation and to assess the quality of information produced from calibration models. We propose to apply our approach on flooding protection structures in the territory of the city of Grenoble.