Stope Dewatering at Deep-Level Gold Mine in South Africa

Case Study 1: Stope Dewatering at Deep-Level Gold Mine in South Africa

Scenario A 3,200-m-deep underground gold mine in the Witwatersrand Basin experienced unexpected inflow into a newly excavated stope after encountering a fractured dolomitic aquifer. The stope (12 m × 8.5 m cross-section) required rapid dewatering to resume stoping operations within 72 hours. Constraints included limited shaft space for pump installation, high ambient temperature (>42°C), abrasive groundwater containing suspended silt (up to 120 mg/L), and strict energy efficiency targets due to grid instability.

Given Data

• Cross-sectional area: 10.2 m² (12 m × 0.85 m effective flow path height — accounting for ore pile obstruction)
• Flow velocity: 0.85 m/s (selected to minimize erosion while ensuring turbulent flow to suspend solids)
• Static head: 315 m (vertical distance from stope floor to surface discharge via existing shaft infrastructure)
• Velocity head: 0.037 m (calculated as v²/(2g) = 0.85² / (2 × 9.81) ≈ 0.037; rounded per tool input constraint to 0.04 m — but tool accepts 0.037 → clamped to min 0.01 → entered as 0.04)
• Friction head: 3.8 m (based on 210 m of DN150 HDPE pipe with fittings, Hazen–Williams C = 140, flow rate ~0.86 m³/s — back-calculated and validated)

Calculation Using the Dewatering Pump Capacity Estimator:

• Pump capacity = cross_sectional_area × flow_velocity = 10.2 m² × 0.85 m/s = 8.67 m³/s
• Total Dynamic Head (TDH) = static_head + velocity_head + friction_head = 315 m + 0.04 m + 3.8 m = 318.84 m → rounded to 318.84 m (precision 2 → 318.84 m)

Result and Decision The estimator indicated a minimum capacity of 8.67 m³/s at 318.84 m TDH. No single-stage centrifugal pump met both requirements. Engineers selected a two-stage, submersible multistage pump (Grundfos SP 500 series, 10-stage configuration) rated at 8.7 m³/s @ 320 m TDH, with hardened tungsten-carbide wear parts and integrated VSD. Installation occurred within 68 hours using modular shaft conveyance; real-time SCADA integration enabled adaptive speed control during variable inflow events.

Lesson Always validate estimated friction head with actual pipe routing, material roughness, and solids content — in this case, initial friction head estimates were 22% low due to unaccounted elbow losses and silt-induced wall roughness, necessitating field recalibration before final pump selection.