🎓 Lesson 8
D5
Trigger Thresholds for Mill Circuit Recalibration
Trigger thresholds for mill circuit recalibration are preset changes in key process measurements—like particle size or throughput—that tell the control system it’s time to automatically adjust grinding parameters.
🎯 Learning Objectives
- ✓ Calculate P80-based trigger thresholds using moving-window statistical deviation analysis
- ✓ Design adaptive threshold logic incorporating ore hardness variability (A×b index) and mill liner wear progression
- ✓ Analyze historical mill data to identify false-positive triggers caused by sensor lag or sampling bias
- ✓ Apply ISO 50001-aligned energy deviation thresholds to justify recalibration events in sustainability reporting
📖 Why This Matters
In modern integrated mines, a 2% deviation in P80 can reduce downstream flotation recovery by 3–5% and increase specific energy consumption by 8–12%. Without robust trigger thresholds, operators rely on manual intervention—delaying response by 15–45 minutes and risking off-spec product. This lesson bridges real-time analytics and closed-loop metallurgical control: mastering thresholds means turning raw sensor data into actionable, self-correcting intelligence.
📘 Core Principles
Trigger thresholds operate at the intersection of statistical process control (SPC), ore characterization, and dynamic circuit modeling. First, baseline performance is established during steady-state operation using 72-hour rolling windows of key variables (P80, Qfeed, kW·h/t, density). Second, thresholds are not static—they adapt using three layers: (1) short-term (15-min) standard deviation bands (±2σ) for noise filtering; (2) medium-term (4-hr) trend detection via linear regression slope (e.g., P80 drift >0.5 µm/hr); and (3) long-term (shift-based) ore-type compensation using real-time hardness proxies (e.g., SAG mill torque vs. feed F80). Finally, recalibration logic applies hierarchical priority: particle size error overrides throughput error, which overrides energy error—ensuring product quality remains primary.
📐 Adaptive P80 Trigger Threshold
This formula calculates the upper P80 trigger threshold based on statistical deviation and ore hardness compensation. It ensures recalibration activates only when change exceeds both random variation and expected hardness-driven shift.
Adaptive P80 Upper Threshold
P80ₜᵣᵢg = μₚ₈₀ + 2σₚ₈₀ + k × (A×bₜ − A×b₆ₐₛₑ)Calculates dynamic upper threshold for P80 to initiate mill circuit recalibration, accounting for statistical variation and ore hardness change.
Variables:
| Symbol | Name | Unit | Description |
|---|---|---|---|
| μₚ₈₀ | Rolling mean P80 | µm | 72-hour moving average of circuit product P80 |
| σₚ₈₀ | Standard deviation of P80 | µm | 72-hour rolling standard deviation of P80 measurements |
| k | Hardness sensitivity factor | µm per unit A×b | Empirically calibrated coefficient linking A×b index change to expected P80 shift |
| A×bₜ | Current A×b index | dimensionless | Real-time Bond work index proxy derived from SAG mill torque and feed F80 |
| A×b₆ₐₛₑ | Baseline A×b index | dimensionless | Reference A×b value established during commissioning or stable operation |
Typical Ranges:
Copper porphyry circuits: 0.45 – 0.75 µm per unit A×b
Iron ore SAG circuits: 0.20 – 0.35 µm per unit A×b
💡 Worked Example
Problem: Given: 72-hr rolling mean P80 = 142 µm, σ = 3.8 µm, current A×b index = 18.5 (vs. baseline 14.2), and hardness sensitivity factor k = 0.65 µm per unit A×b deviation.
1.
Step 1: Compute base threshold = mean + 2σ = 142 + (2 × 3.8) = 149.6 µm
2.
Step 2: Calculate hardness offset = k × (A×b_current − A×b_baseline) = 0.65 × (18.5 − 14.2) = 2.795 µm
3.
Step 3: Apply adaptive threshold = base threshold + hardness offset = 149.6 + 2.795 = 152.4 µm
4.
Step 4: Verify against safe limit: 152.4 µm < 155 µm (max allowable P80 for flotation feed) → valid trigger
Answer:
The adaptive upper trigger threshold is 152.4 µm. If real-time P80 exceeds this value for two consecutive 5-minute samples, the circuit initiates feed rate reduction and water addition recalibration.
🏗️ Real-World Application
At Newcrest’s Telfer SAG-Ball Circuit (Western Australia), implementation of adaptive P80 thresholds reduced off-spec product events by 68% over 12 months. During a transition from moderate- to high-hardness ore (A×b increased from 15.1 to 19.3), static ±5 µm thresholds triggered 14 unnecessary recalibrations in 48 hours—causing 12% overgrinding and 9% energy waste. Switching to the adaptive model (k = 0.58 µm/unit A×b, validated via 3-month test campaign) cut false triggers to zero while maintaining 99.2% on-spec production—even as ore hardness varied daily.