Numerical and experimental studies of the natural mixing behavior between an uncemented paste backfill and dumped waste rock in stopes from laboratory toward field conditions. Part I: Calibration and validation of a numerical model
Yuyu Zhang, Li Li*, Serge Ouellet, Louis-Philippe Gélinas
Deep Resources Engineering. 2026, 3(1): 100232. 
doi.org10.1016j.deepre.2025.100232.pdf
Abstract: Underground mining operations generate large volumes of waste rock (WR). Transporting this material to the surface requires significant energy and incurs operational costs. As an alternative, WR can be directly dumped into stopes being filled with cemented paste backfill (CPB), reducing both costs and greenhouse gas emissions. However, inadequate dumping may lead to poor mixing between cohesionless WR and CPB, resulting in fill mass collapse during or after adjacent stope excavation. Understanding and quantifying the natural mixing between dumped WR and CPB is therefore critical; yet, such studies are scarce. A major challenge lies in replicating large-scale field behavior through limited laboratory-scale tests using scalped (truncated) WR samples. Numerical modeling becomes essential to capture size effects related to both stope and WR particle sizes. In this study, a discrete element method (DEM)-based numerical model was employed. It was first calibrated using repose angle tests on WR samples with varying maximum particle sizes (dmax), prepared using the scalping-down technique. All model parameters were determined through direct measurements, except for the rolling resistance coefficient (µr) between WR particles, which should be obtained through numerical calibration. Initially, it was assumed that the µr would vary with dmax, in line with the observed increase in repose angle with larger dmax. Surprisingly, calibration showed that µr was not very sensitive to changes in dmax, contradicting the experimental trend. Further investigation revealed that repose angle measurements are influenced by the quantity of material used; when sufficient WR mass is employed, the repose angle also becomes independent of dmax. This confirms that scalped samples can reliably represent in situ WR in repose angle tests. The scalping technique is thus validated for use in laboratory piles tests. The predictive capability of the calibrated model is further supported by strong agreement with additional experimental data. This calibrated and validated numerical DEM model can now be confidently applied to analyze the mechanical behavior of WR-based infrastructures across varying particle sizes and field conditions. Its application to simulate the natural mixing between dumped WR and uncemented paste backfill is presented in Part II of this companion study.
Highlights:
• Particle size effects on mechanical behavior of waste rock was taken into account using a DEM-based numerical model.
• Rolling resistance coefficient was initially assumed to vary with the maximum particle size, similar to measured repose angle.
• Numerical calibration showed a rolling resistance coefficient insensitive to the largest particle size (dmax) of scalped samples.
• Further investigation revealed that the measured repose angle is influenced by the quantity of material used in tests.
• Discrepancies with experimental data should prompt deeper investigation rather than arbitrary model parameter adjustment.
Keywords: Waste rocks; Paste backfill; Natural mixing behavior; Repose angle; Scalping method; Rolling resistance coefficient; Numerical modeling; Reliability
Cite: Zhang, Y.Y.; Li, L.; Ouellet, S.; Gélinas, L.-P., Numerical and experimental studies of the natural mixing behavior between an uncemented paste backfill and dumped waste rock in stopes from laboratory toward field conditions. Part I: Calibration and validation of a numerical model, Deep Resources Engineering 2026, 3 (1): 100232. https://doi.org/10.1016/j.deepre.2025.100232
Citations: ……
