🎓 Lesson 9
D5
Alarm Rationalization Using ISA-18.2 Methodology
Alarm rationalization is the process of reviewing every alarm in a control system to decide which ones are truly needed, how they should behave, and why—so operators aren’t overwhelmed during critical events.
🎯 Learning Objectives
- ✓ Explain the purpose and scope of ISA-18.2 alarm rationalization in ROC environments
- ✓ Analyze an existing alarm database to identify nuisance, missing, or improperly prioritized alarms
- ✓ Design alarm attributes—including priority, deadband, shelving rules, and response actions—for a given mine equipment subsystem (e.g., conveyor, ventilation fan, or blast initiation system)
- ✓ Apply the Alarm Flood Index (AFI) metric to assess and improve alarm system health
📖 Why This Matters
In a Mine Remote Operations Center (ROC), operators monitor dozens of assets across hundreds of kilometers—often with only seconds to respond to safety-critical events. Unmanaged alarms can drown out real threats: one Australian iron ore ROC recorded over 1,200 alarms per hour during commissioning, leading to delayed blast evacuation responses and near-miss incidents. Alarm rationalization isn’t about silencing alerts—it’s about engineering trust: ensuring every alarm demands attention *only when action is required*, directly supporting human factors, safety integrity, and regulatory compliance.
📘 Core Principles
ISA-18.2 establishes a lifecycle model for alarm management: philosophy development → identification → rationalization → detailed design → implementation → operation → maintenance → management of change. Rationalization sits at the heart—requiring each alarm to pass four key tests: (1) It must be necessary for safe, reliable, or efficient operation; (2) Its cause must be detectable and actionable; (3) It must have a defined, trained operator response; and (4) Its priority must reflect risk severity, urgency, and consequence. In mining ROCs, this integrates with functional safety (IEC 61511), human-machine interface (HMI) design standards (ISA-101), and mine-specific hazard registers (e.g., from JSA or HAZOP studies).
📐 Alarm Flood Index (AFI)
The Alarm Flood Index quantifies the likelihood of alarm overload during a 10-minute window. Per ISA-18.2 Annex F, AFI > 1.0 indicates unacceptable flood risk and triggers redesign. It combines alarm frequency, duration, and concurrency into a single normalized metric.
Alarm Flood Index (AFI)
AFI = (R × D) / CQuantifies alarm system overload risk over a 10-minute window; values >1.0 indicate unacceptable flood potential.
Variables:
| Symbol | Name | Unit | Description |
|---|---|---|---|
| R | Average alarm rate | alarms/minute | Total alarms in 10-minute window divided by 10 |
| D | Average alarm duration | minutes | Mean time from alarm activation to return-to-normal (RTN) |
| C | Peak concurrent alarms | count | Maximum number of active alarms observed simultaneously |
Typical Ranges:
Well-rationalized ROC: 0.2 – 0.5
Pre-rationalization ROC: 2.0 – 8.0+
💡 Worked Example
Problem: A primary crusher subsystem generates 42 alarms in 10 minutes; average alarm duration is 92 seconds; peak concurrent alarms observed is 7.
1.
Step 1: Calculate average alarm rate per minute = 42 ÷ 10 = 4.2 alarms/min
2.
Step 2: Compute AFI = (alarm rate × avg duration in minutes) ÷ peak concurrent alarms = (4.2 × 1.533) ÷ 7
3.
Step 3: AFI = 6.439 ÷ 7 ≈ 0.92 — falls below 1.0 threshold, but warrants review due to high concurrency (7) and proximity to limit
Answer:
The result is 0.92, which falls within the acceptable range (<1.0), though operational review is recommended given peak concurrency and duration.
🏗️ Real-World Application
At Newmont’s Boddington Gold Mine ROC (Western Australia), rationalization reduced total alarms by 68% post-ISA-18.2 implementation. Critical blast-related alarms—such as 'Blast Initiation System Fault' and 'Seismic Threshold Exceeded'—were retained with Priority P1 (immediate action <30 sec), while redundant 'Motor Vibration High' alarms were suppressed unless exceeding 2× baseline RMS for >15 sec. This cut average operator alarm response time from 47 sec to 19 sec and eliminated false positives during simultaneous dewatering and blasting sequences.