Estimation of the stability of tailings dams based on genetic algorithm-supported calibration results and monitoring data

In consideration of the potential consequences, ensuring the safety of tailings dams is of paramount importance. A probable failure mechanism is the loss of global stability, which, in certain cases, may involve shearing within the subsoil. A significant issue is the inability to directly verify stability on site, in contrast to the routine collection of other monitoring data, such as displacement or groundwater level measurements. This study investigates, using numerically generated (synthetic) data, the feasibility of accurately estimating dam stability. To this end, a genetic algorithm was used to calibrate material parameters, aiming to achieve results that closely match the numerically generated monitoring data. The primary focus of this study is to assess whether successfully reconstructing the history of displacements and piezometric heads is sufficient for accurate estimation of the factor of safety (FOS), a principal indicator of stability. The findings indicate that an accurate estimation of the FOS strongly depends on the proper reconstruction of piezometric heads and the correct determination of the angle of internal friction. Consequently, this research has led to the development of a novel method for FOS estimation using genetic algorithm calibration applied to synthetic data, which holds considerable promise for improving the accuracy of tailings dam safety assessments.