Avaliação do comportamento de escoamentos provenientes de rupturas hipotéticas de pilhas de rejeito por meio de métodos empíricos e hidrodinâmicos

Abstract

This study aimed to analyze the tailings runout behavior resulting from hypothetical dry stack failures. The research emphasized two-dimensional hydrodynamic analyses and compared the obtained results with empirical methods for predicting landslides and tailings flow. Driven by the search for greater safety in mining structures, there is an expansion in the production of filtered tailings and the use of dry stacks as alternatives for disposal. However, gaps remain concerning studies of hypothetical failures of these structures under different topographic configurations, rheological formulations, and downstream elements. Thus, there is an opportunity to deepen discussions regarding the sensitivity of required input variables, as well as to compare results from different existing methods.Two hypothetical dry stack structures were proposed. Stack 1 was set up in a valley and tied to topographic constraints, while Stack 2, with a heaped configuration, was built on flat terrain. The tailings in both structures were assumed to be susceptible to liquefaction. Dynamic simulations were performed using the RiverFlow2D and MADflow models. The concept of apparent friction angle was employed to quantify the pore fluid pressure. The longitudinal extents provided by both modeling packages were similar, indicating similarity in the treatment of rheological terms for characterizing flow resistance. In MADflow, significant lateral spreading for the tailings runout due to Stack 1 failure was observed for the lowest apparent friction angle. The hypothesis that this spreading was due to the lateral earth pressure coefficients, which mediate active and passive stress states, was validated. Incorporating centripetal acceleration added stability to the model and solved the numerical discrepancy. The variation of the two parameters – centripetal acceleration and lateral earth pressure – suggests the high sensitivity of purely frictional dynamic models. In both programs, the frictional formulation provided higher average velocities and abrupt deceleration at the trailing portion of the flow. When a turbulence component was added, gradual decelerations and credible average velocities were observed. Thus, when there is susceptibility to liquefaction as a mechanism of failure propagation, the incorporation of turbulence is expected to yield more realistic results. When comparing the flow behavior on the two topographies, a difference in the lateral spreading of the tailings for the structure sitting on flat topography was observed, even with similar outflow volumes. This distinction may be helpful in preliminary locational studies for heaped tailings stacks. The interaction of the tailings outflow from the failure of Stack 1 in the presence of watercourses, as well as the susceptibility of a downstream sediment containment dam breach, was also analyzed. The entry of the filtered tailings into the dam reservoir led to its overtopping, from which the cascading failure of the two structures was considered. It was indicated that traditionally used techniques for the hydrologic design of dams may be insufficient when located downstream of dry stacks. Finally, empirical equations for predicting the reach angle and the planimetric inundated area were calculated and compared with the numerical solutions. In this comparison, discrepancies were mainly observed when the frictional model was solely used, and the lowest apparent friction angle was analyzed. It was highlighted that the use of empirical techniques should proceed with caution, especially in mechanisms of failure propagation associated with liquefaction.