Assessment of a Tailings Dam Breach by Experimental, Numerical, and Gene-Expression Programming Model

Empirical evidence from documented tailings dam failures highlights the severe consequences for economic systems, human lives, and ecological integrity. The spatial distribution and depositional configuration of tailings within the impoundment structure are regarded as the critical factors influencing the heterogeneous behavioral responses during failure events. This study uses experimental and numerical approaches to investigate the influence of a lateral slope of non-liquefied tailings on localized tailings dam breach mechanisms. The HEC-RAS 2D model was employed to simulate failure scenarios, with the numerical model calibrated against experimental data to evaluate flow characteristics and hydrograph profiles under conditions with and without a lateral slope. Gene-Expression Programming (GEP) was successfully applied to predict flood hydrographs at the failure location based on the simulated data. Results indicate that erosion in the direction perpendicular to the dam is more pronounced in the presence of a lateral tailings slope compared to the scenario without a lateral slope. While a 2% lateral slope exerts minimal influence on the outflow hydrograph, it reduces tailings erosion from the reservoir by approximately 1.3 times in localized failure scenarios. The GEP-derived formula demonstrated high accuracy in computing the flood hydrograph, offering a reliable approach for predicting tailings dam breach-induced flooding.