📋 Case Study

Gold Mine Real-Time Grade Control at Development Drift Face

Manual chip sampling caused 24–48 hr delay in stope boundary decisions, leading to 14% dilution

🏗️ Project Overview

Underground high-grade gold vein system, Western Australia

🎯 Challenge

Manual chip sampling caused 24–48 hr delay in stope boundary decisions, leading to 14% dilution

🔧 Design Approach

Edge-AI rig-mounted hyperspectral imager + XRF fusion with LSTM-based grade regression; decision logic embedded in MineSuite workflow

📐 Design Diagram

Edge-AI Rig-MountedHyperspectral + XRFFusionLSTM GradeRegressionMineSuiteWorkflowDecision latency: 92 secChallenge−14% dilutionDilution avoidance: −3.6 kt/stopeGold Mine Real-Time Grade Controlat Development Drift Face

AI-generated project design illustration

📐 Key Calculations

Decision Latency

Time from scan to boundary update
Result: 92 sec
Enables shift-level stope optimization

Dilution Avoidance

Dilution rate × waste density × volume
Result: −3.6 kt/stope
Direct recovery uplift

📊 Results

Dilution reduced from 14% to 5.8%, gold recovery increased by 9.3%, 22% reduction in development rework

💡 Lessons Learned

  • Spectral calibration drift required weekly field reference standards
  • LSTM memory depth must match structural repetition interval
  • Operators required simplified UI overlay on face camera feed

Key Takeaways

  • 1Spectral calibration drift required weekly field reference standards
  • 2LSTM memory depth must match structural repetition interval
  • 3Operators required simplified UI overlay on face camera feed