Open Issues in Structural Health Monitoring and Local Failure Detection in Concrete Dams

Assessing the structural health of strategic infrastructures such as concrete dams is of paramount importance for clean energy production and flood control in the current context of climate change. Monitoring aimed at early detection of possible local failures is particularly important for long-standing, aging facilities located in the Alpine region. For this purpose, a sensor system is usually installed in the dam body and collects information on the structural response to external actions, mainly consisting of seasonal variations in temperature and water level. Additional measurements could be acquired using drone-mounted equipment. This large amount of data can be processed by various approaches, such as statistical models or machine learning tools trained to predict the dam behaviour and highlight anomalous trends. However, despite the enormous progress made in recent years by the available analysis tools, some limitations are difficult to overcome. In particular, information concerning damaged dams is scarce and hard to transfer from one situation to another. In fact, these structures are quite resilient, and almost all represent unique prototypes due to their particular geometry and environmental conditions. The scarcity of data can be partly overcome by developing a digital twin of the structure under study, with the aim of reproducing its behaviour under the conditions expected to be most critical. To return reliable predictions, the model should be calibrated and continuously updated based on monitoring data. However, measurable quantities may show limited sensitivity to key parameters that enable the identification of local faults. This document illustrates and discusses these aspects with reference to some specific examples.