Digital Twin-Enabled Decision Support Framework for Tailings Dam Operations and Safety

Abstract

Tailings storage facilities deal with urgent decisions amid unsure water and strength conditions. This study presents a digital twin–enabled decision support framework that connects live sensing, physics-based simulation, and clear decision rules to support safe and open operations. The goal is to design and test a closed-loop workflow that changes monitoring signals into suggestions for water release and deposition scheduling. The framework links a sensor-instrumented 1:60 physical model of a downstream-raised dam with a matched numerical model. State estimation applies a Kalman filter. Seepage and stability get solved with Richards flow and limit-equilibrium analysis. Scenario tests include routine operation, heavy rainfall, and a quick-load case. A weighted cost function balances safety standards, pond goals, and production changes to pick actions. Compared to a rule-based baseline, the framework reduced time to first action from 34 to 21 minutes, increased the minimum factor of safety from 1.28 to 1.36, and decreased peak pond level from 0.33 to 0.29 m. State estimation errors for pond level and pore pressure stayed small. The framework moves operations from reactive checks to proactive control, with water balance acting as the key tool. The method's new aspect is the closed loop that combines sensing, hydro-mechanical simulation, and direct action in one workflow. In practice, sites should start early pond release and shortterm deposition cuts as the factor of safety nears about 1.35.

Article history: Received (January 26, 2026); Revised (April 23, 2026); Accepted (May 5, 2026); Published online (June 26, 2026)