Progressive Pillar Degradation and Monitoring Driven Reinforcement Control in Hard Rock Room and Pillar Mining: Using Unki Mine, Zimbabwe as a Case Study

Room and pillar extraction remains one of the principal mining methods in hard-rock environments, yet maintaining the long-term integrity of pillars continues to pose significant challenges. In many operations, gradual spalling, geometric reduction of pillar dimensions, and redistribution of in-situ stresses progressively weaken load-bearing capacity, increasing both operational risk and the likelihood of production interruptions. At Unki Platinum Mine in Zimbabwe, deterioration was observed in the form of rib scaling, crack propagation, and measurable time-dependent loss of effective pillar width, suggesting a progressive damage mechanism rather than abrupt structural failure. This study proposes an integrated monitoring simulation machine learning framework for predictive reinforcement prioritization in hard rock room and pillar systems. Field observations indicate that instability is controlled by the interaction of decreasing width to height ratio, stress concentration, structural discontinuities, blasting effects, and ongoing surface degradation. FLAC3D simulations further clarify the redistribution of stresses within the mining panel, revealing elevated stress strength ratios at pillar edges and a steady decline in load bearing capacity as pillar width diminishes. Based on these findings, a combined rock bolt and shotcrete support system was applied underground, effectively limiting spalling and improving confinement of the damaged pillar surface. To supplement simulation-based analysis at the panel scale, a monitoring based predictive framework was developed. Using deformation records and key mechanical indicators, pillar conditions were classified into stability states. The Random Forest classifier achieved an overall accuracy of 88%, with unstable pillar recall exceeding 91%, demonstrating reliable early warning capability for reinforcement prioritization. The resulting monitoring prediction control approach provides a practical basis for proactive stability management in hard rock room and pillar operations.