Exploring Tailings Dam Stability Considering Uncertainties in the Critical State Parameters of the NorSand Model.

Stability analyses of tailings dams often utilize advanced constitutive models that require the calibration of many parameters. Yet, there remains limited research on the implications of parameter uncertainties within these constitutive models on the safety factor of tailings dams. This study successfully investigated the impact of uncertainty in critical state parameters of the NorSand model on the stability of a tailings dam. To achieve this goal, field and laboratory tests characterized the materials constituting the geotechnical profile of the dam, involving residual shale soils, alluvial soils, and granular-textured tailings. These tests were included in the finite element (FE) program PLAXIS 2D to estimate several parameters of the constitutive models that characterized the materials of the dam: Hardening Soil, Mohr–Coulomb, and NorSand. Monte Carlo simulations of FE analyses were conducted considering the construction period of the dam spanning 2500 days. Furthermore, the critical state line (CSL) parameters employed within the NorSand model were reconciled with void ratio data for the uncompacted tailings utilizing a Bayesian framework. The Bayesian framework allowed us to investigate: (i) the statistical distribution of the critical state parameters, (ii) confidence intervals for horizontal and vertical displacements, (iii) shear strains at various monitoring points across the tailings dam, (iv) and the marginal distribution of the factor of safety. The findings reveal a strong correlation between the three CSL parameters, which presented poorly defined statistical distributions. Nonetheless, the power law equation used to model the CSL closely matches laboratory data from triaxial tests (R² = 0.97). Uncertainties in the CSL, especially at lower stress levels, contribute to settlement prediction uncertainties. Monitoring points at the structure's base showed strains that could induce downstream slope instability and potential liquefaction failure. The histogram of the safety factors remained narrow, normally distributed, ranging from 1.39 to 1.47. This work is important because it explicitly addresses the uncertainty in the critical state parameters of the NorSand model within geotechnical analysis. [ABSTRACT FROM AUTHOR]