Progressive failure mechanisms and stability assessment of geotextile bag tailings dams using upper-bound analysis

Geotextile bag dam technology is widely adopted in tailings management owing to its material synergy and environmental sustainability. However, the stability evaluation of such dams remains constrained by the limitations of conventional homogenisation assumptions. To address this issue, this study investigates failure mechanisms under loading through reduced-scale model tests on unreinforced and reinforced dams with varying slope ratios. Based on the experimentally observed failure modes and the upper-bound theorem, a kinematically admissible failure velocity field is developed to propose a composite slip-surface limit analysis method, explicitly incorporating interface shearing. Results demonstrate that the inclusion of geotextile bags increases the ultimate bearing capacity by 62.6% compared with the unreinforced dam. Crucially, the failure mode is identified as a composite pattern of “circular shear within tailings coupled with interface sliding along geotextile bags,” with flexible confinement effects being more pronounced under gentler slopes. Analytically, predicted slip surfaces align closely with experimental observations. Furthermore, the calculated safety factors exhibit a relative error of only 5.3%, demonstrating significantly superior accuracy than that obtained using Bishop's method, which yields an error of 13.3%. Overall, this study elucidates interface-dominated failure mechanisms and establishes a high-precision framework for the stability assessment of geotextile bag dams. • Identified six-stage progressive failure in geotextile bag dams, revealing interface-dominated collapse mechanisms. • Novel upper-bound framework incorporates composite slip surfaces and interfacial shear, overcoming homogenisation limits. • Achieved superior accuracy (5.3% error) compared to conventional methods via dual validation. • Confirmed reinforcement effectiveness via lateral confinement and stress redistribution, enabling optimised slope design.