Prediction of backfilling slurry viscosity based on high-lifting-velocity slump tests

The slump test is a well-established method for estimating the yield stress of filling slurries, and its ability to predict yield stress from slump height at low lifting velocities has been extensively studied. However, accurately determining the viscosity of slurries using the same test remains a significant challenge. Given that the influence of viscosity on slump height and spread diameter is negligible at low lifting velocities (e.g., 0.01 m·s–1), a modified slump test conducted at a higher lifting velocity (0.1 m·s–1) was proposed in this study. To precisely control the lifting velocity of the conical mold, a custom-designed slump apparatus was employed. For each slump test, the profile of the slumped slurry was measured using a laser rangefinder. To systematically investigate the coupled effects of Bingham yield stress and viscosity on slump height and spread diameter, numerical simulations were performed to analyze the slump behavior of fly-ash slurry under various combinations of yield stress and viscosity. The simulations specifically focused on conditions involving a high mold lifting velocity of 0.1 m·s–1, aiming to reveal the correlations between slump height/spread diameter and the Bingham parameters. Dynamic mesh techniques, together with the Volume of Fluid (VOF) method, were employed to simulate the collapsing process, enabling the acquisition of slump behavior under arbitrary yield stress and viscosity values. The numerical results were validated against experimental data obtained from slump tests conducted at a low lifting velocity of 0.01 m·s–1, showing good agreement. In parallel, lubrication theory was applied to predict the slumped slurry profiles at different solid mass fractions, and these theoretical predictions also coincided well with experimental measurements. The simulation results revealed several key trends. When the Bingham yield stress was held constant, the slump height decreased linearly with increasing Bingham viscosity. In contrast, the spread diameter also decreased with increasing viscosity, but the rate of reduction gradually diminished. When the Bingham viscosity was kept constant, the slump height decreased linearly with increasing yield stress. Meanwhile, the spread diameter exhibited an approximately linear decrease within the low-viscosity range, but a nonlinear decrease in the high-viscosity range. Under high lifting velocity conditions, the spread diameter showed more pronounced variations in response to changes in Bingham viscosity compared to slump height, making it a more sensitive indicator for characterizing viscosity changes. Consequently, spread diameter was selected as the primary parameter for constructing a predictive model of Bingham viscosity based on the slump test. Using four different regression methods, quadratic polynomial regression, random forest, XGBoost, and neural network, a predictive model for Bingham viscosity was developed. The model inputs included the known spread diameter and Bingham yield stress. Among the four methods, the random forest and neural network models exhibited superior predictive performance, yielding higher accuracy and better generalization capability. These findings validate the feasibility of employing high-lifting-velocity slump tests to predict the Bingham viscosity of mineral slurries. Given that the conventional slump test at low lifting velocity (0.01 m·s–1) has already been successfully applied to predict yield stress, the successful determination of Bingham viscosity using a modified slump test at a higher lifting velocity (0.1 m·s–1) implies that complete characterization of both Bingham yield stress and viscosity can be achieved through just two sets of slump tests—one at low velocity and one at high velocity. This dual-test approach offers a simple, rapid, and cost-effective alternative to conventional rheometers, which are often expensive and unsuitable for on-site or field applications.