BCPGroup / Sibanye-Stillwater

Sibanye-Stillwater, one of the world's largest producers of platinum group metals and gold, operates energy-intensive underground mining facilities where artificial lift systems — pumps that remove water from deep workings — run continuously and consume substantial electrical power. Energy costs represent one of the largest operating expenses in deep-level mining, often accounting for 15–20% of total site operating costs. Without intelligent control, artificial lift equipment runs at fixed or manually adjusted setpoints, consuming peak power regardless of actual dewatering demand. At enterprise scale across multiple shafts and sites, this inefficiency compounds into millions of dollars in unnecessary energy spend annually.

BCPGroup, working with Rockwell Automation, implemented an AI-powered energy optimization system for Sibanye-Stillwater's artificial lift infrastructure. The solution applied machine learning to analyze real-time operational data from pumping systems — including flow rates, water ingress volumes, shaft conditions, and electricity tariff signals — to dynamically adjust pump scheduling and motor speeds. Rather than operating on static schedules, the AI continuously recalculates optimal lift configurations to match actual dewatering needs while shifting load away from peak tariff periods. Rockwell Automation's industrial automation and control platform provided the integration layer between the AI optimization engine and the physical pump assets, enabling deployment within the existing operational technology environment without requiring full infrastructure replacement.

The implementation delivered measurable reductions in energy expenditure across artificial lift operations at Sibanye-Stillwater's facilities. Key outcomes included:

• Significant energy cost reductions achieved through demand-responsive pump scheduling rather than fixed operational cycles
• Peak demand management enabled by AI load-shifting during high-tariff electricity periods
• Improved operational visibility giving site engineers real-time insight into energy consumption versus dewatering requirements
• Sustained optimization as the machine learning model continuously refined setpoints based on changing mine conditions

The project demonstrated that AI-driven energy management can be integrated into active mining operations without disrupting production continuity.

• Artificial lift is a high-leverage target — pump systems run 24/7 and respond well to dynamic AI scheduling, making them one of the best starting points for energy optimization in underground mining.
• Tariff-aware AI compounds savings — aligning pump cycles with off-peak electricity rates amplifies cost reductions beyond pure efficiency gains.
• Integration with existing OT infrastructure is critical; solutions that layer onto installed Rockwell or similar control systems reduce deployment risk and timeline.
• Start with data instrumentation — AI optimization depends on reliable real-time sensor data from pumps; gaps in telemetry limit model accuracy.
• Enterprise-scale rollout requires site-by-site calibration, as water ingress rates and shaft conditions vary significantly across a portfolio of mines.