Pilot-Scale Investigation of Bauxite Tailings Dewatering by Decanter Centrifuge—Part 2: A Campaign-Calibrated Semi-Empirical Framework for Early Screening of Decanter Performance

Part 1 of this study identified the recovery of ultrafine particles as the main bottleneck in the decanter centrifuge dewatering of bauxite tailings. This study (Part 2) develops a semi-empirical framework that reconstructs the Whiten partition parameters (corrected cut size d50c, sharpness α, and hydraulic bypass Rf) and the cake solids content from bench-scale characterization, operating and equipment parameters, and coefficients calibrated on the pilot campaign. The d50c is estimated with Ambler’s Σ-theory, using an effective viscosity from a batch sedimentation test and a Σ-efficiency factor (ε = 0.0945) calibrated on the pilot cut sizes, which are reproduced with a mean absolute error of 0.33 µm (leave-one-condition-out errors from −4.5% to +7.6%). Hydraulic bypass and cake solids content are linear functions of centrifugal acceleration with offsets fixed by sedimentation test quantities, whereas sharpness is fully calibrated. The calibrated parameters reproduce the pilot fine-fraction (<20 µm) recovery within −0.3 to +2.1 percentage points, an internal consistency check rather than an independent validation. The recovery estimate is dominated by ε, whereas replacing the calibrated sharpness intercept by twice the feed PSD imperfection changes it by at most 0.5 percentage points. These results are combined into an early assessment workflow in which ε is evaluated through scenarios anchored on the campaign value, yielding a recovery range—transfer to other materials or equipment requires validation.