Universitat Politècnica de Catalunya. Departament d'Enginyeria Minera, Industrial i TIC, Universitat Politècnica de Catalunya. RIIS - Grup de Recerca en Recursos i Indústries Intel·ligents i Sostenibles, Bonet Gil, Enrique, Yubero de Mateo, Maria Teresa, Sanmiquel Pera, Lluís, Bascompta Massanes, Marc, Universitat Politècnica de Catalunya. Departament d'Enginyeria Minera, Industrial i TIC, Universitat Politècnica de Catalunya. RIIS - Grup de Recerca en Recursos i Indústries Intel·ligents i Sostenibles, Bonet Gil, Enrique, Yubero de Mateo, Maria Teresa, Sanmiquel Pera, Lluís, and Bascompta Massanes, Marc
Historically, one of the most common causes of dam failure has been overtopping, primarily in earthfill dam, accounting for approximately 34% in the United States, according to the Association of State Dam Safety Officials. There have been other causes which has also been contributed to dam failures throughout history, with a significant issue in masonry dams being water infiltration through the dam body, leading to erosion of the mortar that binds the rocks forming the dam body. As a result, quantifying the flow rate from these cracks in the mortar is an important parameter to consider and measure in dam maintenance and operation. In this article, a tool is developed using Artificial Intelligence methodologies, specifically artificial neural networks, for predicting water leakages in a masonry dam. The tool learns from historical data collected from the Santa Fe del Montseny Dam (Barcelona) over the past 10 years (although not all dataset is completed during this period), taking into account factors such as temperature, precipitation, reservoir levels, water levels in observation wells within the dam, as well as water leakage measurements. The leakage flow prediction tool is developed in a MATLAB environment. The methodology used is an artificial neural network and different model options such as hold-out and k-folds were provided and tested. In this study, different layer sizes, different number of neurons, different k folds values are considered to minimize the leakage prediction error of the tool. The results indicate that the tool can predict infiltration flow with an accuracy close to 90% (using testing dataset), making it a valuable tool for decision-making in the masonry dam maintenance and operation tasks. In that sense, the leakage flow prediction is also a useful tool for dam monitoring to evaluate the dam's behavior., Postprint (published version)