🎓 Lesson 10 D5

Building 3D Geometallurgical Block Models

A 3D geometallurgical block model is a digital map of a mine that shows not just where rock is located, but also its metallurgical properties—like how easily it crushes or how much metal it contains—broken down into small 3D blocks.

🎯 Learning Objectives

  • Construct a 3D block model by integrating drill-hole assay, lithology, and metallurgical test data using geostatistical estimation
  • Analyze spatial continuity and uncertainty of metallurgical parameters using variogram modeling and cross-validation
  • Design optimal block support sizes and estimation strategies based on domain-specific anisotropy and data density
  • Apply conditional simulation to quantify risk in downstream process performance (e.g., flotation recovery variability)
  • Explain the impact of grade-metallurgical property correlations on mill feed planning and blending strategies

📖 Why This Matters

In modern integrated mining operations, knowing *where* the ore is no longer enough—you must know *how it will behave* in the plant. A poorly constructed geometallurgical model can lead to unexpected throughput bottlenecks, reagent overconsumption, or suboptimal recoveries—even when grade targets are met. At Newmont’s Boddington Mine, implementing a validated 3D geometallurgical model reduced grinding circuit energy use by 8% and improved gold recovery predictability by ±1.2% (vs. ±4.7% pre-model). This lesson equips you to build models that bridge geology and metallurgy—not just visually, but functionally.

📘 Core Principles

Geometallurgical modeling rests on three interlocking pillars: (1) *Metallurgical domain definition*, where lithological, structural, and alteration units are classified by their consistent processing behavior; (2) *Spatial characterization*, requiring rigorous geostatistical treatment—including non-stationary trend modeling and co-kriging of correlated variables (e.g., Cu grade vs. Bond Work Index); and (3) *Model validation*, which goes beyond cross-validation residuals to include metallurgical response testing (e.g., predicting SAG mill throughput from modeled Wi and hardness). Unlike resource models focused on grade, geometallurgical models prioritize *functional attributes*: variables must be measurable, process-relevant, and scalable from lab test to full-scale operation. Critical attention must be paid to support effects—lab tests on 2 kg samples cannot reliably represent bulk behavior without upscaling protocols grounded in fragmentation and liberation physics.

📐 Block Model Support Correction for Bond Work Index

Lab-scale Bond Work Index (Wi) values require support correction to represent full-blasthole or bench-scale material. The correction accounts for scale-dependent fracture density and grain boundary exposure. The standard empirical upscaling uses the 'size–scale factor' derived from field calibration datasets.

Wi Upscaling Correction

Wi_field = Wi_lab × (P80_lab / P80_field)^γ

Corrects laboratory-scale Bond Work Index to represent full-scale blast fragmentation conditions.

Variables:
SymbolNameUnitDescription
Wi_field Field-scale Bond Work Index kWh/tonne Estimated work index representing bench-scale material behavior
Wi_lab Laboratory-scale Bond Work Index kWh/tonne Measured Wi on prepared sub-sample (typically 2–5 kg)
P80_lab Laboratory P80 size mm 80% passing size of lab test feed
P80_field Field P80 size mm 80% passing size of actual fragmented muck pile (from image analysis or sieving)
γ Scaling exponent dimensionless Empirically calibrated exponent (0.18–0.25); 0.22 recommended for hard rock systems per AMIRA P970B
Typical Ranges:
Hard rock copper-gold porphyry: 7.0 – 12.5 kWh/t
Banded iron formation (BIF): 10.0 – 18.0 kWh/t

💡 Worked Example

Problem: Given: Lab Wi = 14.2 kWh/t (measured on 2 kg composite, 100% <6.35 mm); blasthole diameter = 250 mm; bench height = 15 m; average fragment size (P80) post-blast = 420 mm.
1. Step 1: Compute representative fragment size ratio: R = P80_lab / P80_field = 6.35 mm / 420 mm = 0.0151
2. Step 2: Apply scaling exponent (γ = 0.22, per AMIRA P970B calibration): Wi_field = Wi_lab × R^γ = 14.2 × (0.0151)^0.22
3. Step 3: Calculate: (0.0151)^0.22 ≈ 0.621 → Wi_field = 14.2 × 0.621 = 8.82 kWh/t
Answer: The upscaled Wi is 8.8 kWh/t, which falls within the typical range of 7–12 kWh/t for weathered porphyry ores at bench scale.

🏗️ Real-World Application

At Rio Tinto’s Pilbara Iron Ore operations, a 3D geometallurgical model was built integrating 2,400+ drill holes with 12,000+ laboratory metallurgical tests (crushing, grinding, magnetic separation response). Litho-geometallurgical domains were defined using hierarchical cluster analysis of combined geochemical (SiO₂, Al₂O₃, P) and mechanical (UCS, Wi) data. The resulting 10×10×5 m block model drove dynamic ROM pad blending—reducing magnetite concentrate silica variability from ±1.8% to ±0.4%, enabling consistent sinter feed specification without offline quality control delays. Model validation used blind 3-month mill feed predictions, achieving R² = 0.89 for grind size (P80) and 0.76 for Fe recovery.

📚 References