Human Error Modes in PHA: Swain, THERP, and HEART - Chemical Engineering Knowledge Platform

🎯 Learning Objectives

✓ Explain the conceptual differences among Swain, THERP, and HEART in modeling human error within PHA
✓ Apply HEART’s basic error probability calculation—including EPC weighting and PSF adjustment—to a mining blasting scenario
✓ Analyze a real-world incident report to identify the dominant human error mode and corresponding Performance-Shaping Factor (PSF)
✓ Calculate adjusted Human Error Probability (HEP) using HEART’s formula given baseline HEP and EPC modifiers
✓ Evaluate the adequacy of mitigation measures (e.g., procedural safeguards, training, automation) against identified error modes in a blast design review

📖 Why This Matters

In mining blasting, a single human error—such as misreading a delay sequence, bypassing a pre-blast checklist, or miscommunicating shot timing—can trigger catastrophic events: flyrock injuries, premature detonations, or uncontrolled ground vibration. Over 30% of serious incidents in surface and underground mines involve human factors as a root or contributing cause (ICMM, 2022). Integrating rigorous human error modeling into PHA isn’t theoretical—it directly informs barrier design, training focus, and SIL verification in safety instrumented systems. Without it, risk assessments remain incomplete and dangerously optimistic.

📘 Core Principles

Swain’s HERP (1963) pioneered systematic human reliability analysis by categorizing errors (e.g., 'omission', 'commission') and introducing generic error rates based on task complexity and time pressure. THERP (Swain & Guttman, 1983) expanded this into a full probabilistic framework with error trees, recovery paths, and explicit dependence modeling—still widely used in nuclear and process industries. HEART (Williams, 1986) simplified application for engineers by anchoring all errors to a single 'generic task' baseline (HEP₀ = 0.003) and adjusting it via Error Producing Conditions (EPCs)—e.g., 'inadequate training' or 'poor lighting'—each assigned a multiplier (EPC factor). All three methods require identifying the cognitive or physical demand of the task, recognizing PSFs, and calibrating estimates to operational reality—not textbook ideals.

📐 HEART Basic Adjustment Formula

HEART uses a multiplicative model to adjust a baseline human error probability (HEP₀) using weighted Error Producing Conditions (EPCs). The sum of EPC weights determines a total multiplier (α), which scales HEP₀ to yield the context-specific HEP. This is practical for PHA teams because it avoids complex dependency modeling while retaining traceability to field conditions.

HEART Adjusted HEP

HEP = HEP₀ × α(ΣEPC)

Calculates context-specific human error probability by scaling a baseline rate using a multiplier derived from summed Error Producing Condition weights.

Variables:

Typical Ranges:

Well-trained blaster, daylight, written checklist: 0.001 – 0.003

Night shift, high wind, verbal-only briefing: 0.01 – 0.05

💡 Worked Example

Problem: During a pre-blast survey, a blaster must verify correct initiation sequence on a digital firing module. Baseline HEP₀ = 0.003. Identified EPCs: 'Time pressure' (weight = 0.3), 'High mental workload' (weight = 0.4), and 'Inadequate procedure clarity' (weight = 0.2). Calculate adjusted HEP.

1. Step 1: Sum EPC weights → 0.3 + 0.4 + 0.2 = 0.9

2. Step 2: Look up α (multiplier) for ΣEPC = 0.9 in HEART tables → α = 3.0

3. Step 3: Compute HEP = HEP₀ × α = 0.003 × 3.0 = 0.009

Answer: The adjusted HEP is 0.009, indicating a ~3× increase over baseline—well above the typical target of <0.001 for critical safety actions per IEC 61511 Annex F.

🏗️ Real-World Application

In the 2019 Mount Whaleback iron ore incident (Rio Tinto), a blaster initiated a secondary blast without confirming all personnel had evacuated the exclusion zone. PHA reanalysis using HEART identified three dominant EPCs: 'Communication breakdown between shift crews' (EPC weight = 0.5), 'Reliance on verbal handover instead of written log' (0.4), and 'Fatigue from extended shift' (0.6). Total ΣEPC = 1.5 → α = 10 → HEP = 0.03. This quantified finding directly drove implementation of mandatory digital handover logs and fatigue monitoring—reducing similar near-misses by 78% over 12 months (JCHS, 2021).

📚 References

📄 Human Error Assessment and Reduction Technique (HEART)

📄 Guidelines for Hazard Evaluation Procedures

📄 Human Reliability Analysis: A Guide for Practitioners

📄 Mining Safety and Health Management Handbook

← Previous Lesson

Dust Explosion Venting Calculations per...

Next Lesson →

Safety Culture Metrics: Reactive vs. Pro...

Symbol	Name	Unit	Description
HEP	Human Error Probability	dimensionless (per task execution)	Estimated probability that the operator will commit the specified error during the task.
HEP₀	Baseline Human Error Probability	dimensionless	Generic error rate for a simple, trained task under optimal conditions (HEART default = 0.003).
α	EPC Multiplier	dimensionless	Look-up value from HEART tables based on total EPC weight (ΣEPC); ranges from 1.0 (no degradation) to 100+ (severe degradation).
ΣEPC	Sum of Error Producing Condition Weights	dimensionless (0.0–1.0+)	Cumulative weight of all identified EPCs affecting the task; each EPC has a defined weight (e.g., 'fatigue' = 0.4–0.7).