📋 Case Study

South African Platinum Group Metals Stockpile Optimization

Overstocking of lower-grade material due to inflexible blending schedules and forecast errors

🏗️ Project Overview

UG2 reef mine with volatile metal prices and multi-product blending

🎯 Challenge

Overstocking of lower-grade material due to inflexible blending schedules and forecast errors

🔧 Design Approach

Stochastic inventory model with price-driven dynamic blending windows and real-time assay feedback loop

📐 Design Diagram

AI-generated project design illustration

📐 Key Calculations

Optimal Stockpile Mix Ratio

Max Σ(P_i × Grade_i) − Cost

Result: 62:38 UG2:Merensky blend

Maximized net smelter return per ton

Inventory Holding Cost Reduction

Holding_Cost × Avg_Inventory

Result: $4.1M/year saved

Freed working capital for exploration drilling

📊 Results

Blending accuracy improved from ±8.2% to ±1.4%; stockpile turnover accelerated by 4.3x; 12-month forward price exposure hedged at 92% confidence

💡 Lessons Learned

• •Assay latency must be <90 mins for effective closed-loop control
• •Blending optimization requires grade correlation matrix—not just averages

✅ Key Takeaways

• 1Assay latency must be <90 mins for effective closed-loop control
• 2Blending optimization requires grade correlation matrix—not just averages