Bauxite Residue: A Data-Driven Approach to Strength Characterisation

Cone penetration, pore pressure dissipation, and vane shear tests are commonly employed to gauge the state and potential response of saturated tailings. This paper illustrates cone penetration-derivedstate and strength characterization for bauxite residue deposited in a Tailings Storage Facilities. It presents the application of a model to predict normalized peak undrained shear strength based onthe net cone factor which was evaluated on mean absolute errorand correlation coefficient.Cone penetration, pore pressure dissipation and vane shear tests were conducted in near embankment residue.The state parameter was estimated based on a single set of presumed critical state parameters, as proposed by Plewes et al. (1992). The drainage boundary conditions were evaluated following the methodology proposed by Schneider et al. (2008). The general behavior of the residue was assessed in terms of the SBT indices proposed by Robertson (2016) and Jefferies and Been (2016).Using calibrated net cone factor, peak undrained strengthwas predicted from the empirical correlation between corrected cone resistance and total vertical stress whilenormalized peak undrained shear strength wasassessed in terms of state parameter. Peak undrained strength was measured by triaxial testing and estimated considering the theoretical formulation proposed by Wroth (1984).Characteristic strength was derived by statistical analysis of the state observed within the near embankment residue.

Keywords: Peak undrained shear strength ratio, Machine learning; data-driven, Red mud, Bauxite residue, Cone penetration test, Tailings behaviour, NorSand.