Mathematical models for the simulation of the thermal evolution of roller-compacted concrete (RCC) dams during construction constitute an important tool for preventing excessive temperature rise, which may lead to cracking and losses of functionality. Here, we present a framework for the simulation of the thermal process. We define the boundary conditions of the problem using a careful description that incorporates the main heat exchange mechanisms. We adopt both a non-adiabatic and an adiabatic heat generation model for the simulation of the cement hydration. Our numerical framework lets us study the effect of the adopted heat generation model on the thermal field. Moreover, we study the influence of the weather conditions on the evolution of the hydration, and on the starting date of construction. Our simulations have shown that the hydration model has an important influence over the temperature field during the construction and the heat generation rate. Moreover, the hydration process and the temperature evolution are driven by the weather conditions. Once the next lift is cast, its thermal insulation effect makes the hydration take place under quasi-adiabatic conditions. As expected, dams built in cold months are prone to dissipate more heat than those built in warm seasons.

Dam safety monitoring is of vital importance, due to the high number of fatalities and large economic damage that a failure might imply. This, along with the evolution of artificial intelligence, has led to machine learning techniques being increasingly applied in this field. Many researchers have successfully trained models to predict dam behavior, but errors vary depending on the method used, meaning that the optimal model is not always the same over time. The main goal of this paper is to improve model precision by combining different models. Our research focuses on the comparison of two successful integration strategies in other areas: Stacking and Blending. The methodology was applied to the prediction of radial movements of an arch-gravity dam and was divided into two parts. First, we compared the usual method of estimating model errors and their hyperparameters, i.e., Random Cross Validation and Blocked Cross Validation. This aspect is relevant not only for the importance of robust estimates, but also because it is the source of the data sets used to train meta-learners. The second and main research topic of this paper was the comparison of combination strategies, for which two different types of tests were performed. The results obtained suggest that Blocked CV outperforms the random approach in robustness and that Stacking provides better predictions than Blending. The generalized linear meta-learners trained by the Stacking strategy achieved higher accuracy than the individual models in most cases.

This paper presents the results of an extensive laboratory set of tests aimed to study the failure of the downstream shoulder of highly permeable rockfill subjected to overflow. The experimental research comprised testing 114 physical models by varying the following elements: (i) the median size of the uniform gravels (7 to 45 mm); (ii) the configuration of the dam, i.e., upstream and downstream shoulders and crest or just the downstream shoulder; (iii) the dam height (from 0.2 to 1 m), (iv) the crest length (from 0.4 to 2.5 m), (v) the downstream slope (from 1 to 3.5 H:V), (vi) the type of impervious element (i.e., central core, upstream face, and no impervious element). The tests allowed us to identify two failure mechanisms, slumping and particle dragging. In addition, the downstream slope was observed to be one of the most important variables in this parametric study, as it influenced the pore water pressures inside the dam, the failure discharge, and the occurrence of one or the other mechanism of failure.

In this article, a digital twin is a numerical model of a dam–foundation system with a physical basis. A dam–foundation system exhibits a complex behavior and, therefore, significant simplifications are necessary for a numerical model to be feasible, which reduces a model’s accuracy. Differences in the material characteristics of particular regions of the dam and the foundation are usually not considered. An important reason for that is the high computational cost of calibrating a model when many parameters must be optimized in such a way that calibration becomes unfeasible. A new, simple, accurate, and low time-consuming algorithm, the centroid method, is presented that allows the calibration of a dam–foundation system. The extreme efficiency of this algorithm opens the feasibility of calibrating a sectorized numerical model of an arch dam, so that a digital twin is obtained that takes into account the different characteristics of different regions or sectors of the dam and the foundation, more closely approaching the complex behavior of the dam–foundation system.

It is well established that the proper functioning of dams plays an important role in society. Therefore, as a fundamental part of their safety, the identification of anomalies by monitoring their deformations must be given special consideration. The ability to detect cracks based on monitoring device records depends largely on the number of devices included in the dam’s original design, their arrangement, and their accuracy. This paper aims to help determine the appropriate spacing and accuracy of the devices for detecting a given crack. For this purpose, numerical simulations of cracks that match the most likely cracks in arch dams are performed based on a real dam, ensuring that such cracks result in an opening wide enough to be of concern. Afterward, the study analyzes the impact of the length and depth of the crack and the accuracy and position of the monitoring device on the ability of the system to detect the crack.

Dam safety monitoring is critical due to its social, environmental, and economic implications. Although conventional statistical approaches have been used for surveillance, advancements in technology, particularly in Artificial Intelligence (AI) and Machine Learning (ML), offer promising avenues for enhancing predictive capabilities. We investigate the application of ML algorithms, including Boosted Regression Trees (BRT), Random Forest (RF), and Neural Networks (NN), focussing on their combination by Stacking to improve prediction accuracy on concrete dam deformation using radial displacement data from three dams. The methodology involves training first-level models (experts) using those algorithms, and a second-level meta-learner that combines their predictions using BRT, a Linear Model (LM) and the Greedy Weighted Algorithm (GWA). A comparative analysis demonstrates the superiority of Stacking over traditional methods. The GWA emerged as the most suitable meta-learner, enhancing the optimal expert in all cases, with improvement rates reaching up to 16.12% over the optimal expert. Our study addresses critical questions regarding the GWA’s expert weighting and its impact on prediction precision. The results indicate that the combination of accurate experts using the GWA improves model reliability by reducing error dispersion. However, variations in optimal weights over time necessitate robust error estimation using cross-validation by blocks. Furthermore, the assignment of weights to experts closely correlates with their precision: the more accurate a model is, the more weight that is assigned to it. The GWA improves on the optimal expert in most cases, including at extreme values of error, with improvement rates up to 41.74%. Our findings suggest that the proposed methodology significantly advances AI applications in infrastructure monitoring, with implications for dam safety.