Module 9.7–9.8 — Communication and human error (SHELL, Reason, error types)

Scope and levels

9.7 Communication and 9.8 Human error are both level 2. Human error is the most heavily weighted Module 9 topic in the EASA AMC question allocation, together with 9.10 Dirty Dozen.

9.7 Communication

The communication process

Sender → message (encoded) → medium/channel → receiver (decoded) → feedback. Noise (literal or otherwise) can corrupt any stage.

• Verbal: spoken words, tone of voice. Fast, but easily forgotten or mis-heard; no record.
• Non-verbal: body language, gestures, facial expression, hand signals (marshalling).
• Written: work cards, tech log, manuals, handover sheets — provides a record; can be ambiguous if poorly written.
• Communication modes: one-way (instructions, notices) vs two-way (dialogue with feedback — more reliable).

Barriers: noise, language and jargon, assumptions, poor listening, distraction, status differences (junior engineer reluctant to question senior), time pressure, fatigue.

Key fact: Effective communication needs feedback — confirm understanding by read-back or by asking the receiver to repeat key points. Never assume a message has been understood.

Within and between teams

• Within a team: team briefings, clear task allocation, speaking up about problems.
• Between teams (shifts, departments, contractors, stores, planning, the operator): formal channels and written records are essential because the parties may never meet. 145.A.47(d) requires hazards from external working teams in the organisation's facility to be considered.

Work logging and recording

• Record work as it is done, not from memory at the end of the shift.
• Entries must be clear, accurate and complete: what was done, data reference, any incomplete steps, parts used.
• Incomplete stages of a task must be recorded so that the next person knows exactly where the work stopped.
• Falsifying records is a ground for licence suspension/revocation (66.B.500).

Shift handover

• 145.A.47(c): when tasks are handed over for a shift or personnel change, relevant information must be adequately communicated between outgoing and incoming personnel.
• AMC 145.A.47(c) lists three elements of effective handover:
  1. the outgoing person's ability to understand and communicate the important elements of the task;
  2. the incoming person's ability to understand and assimilate the information;
  3. a formalised process for exchanging information, with a planned shift overlap and a place for the exchange.
• Good practice: written handover log plus face-to-face briefing, ideally at the aircraft; cover task status, deviations, problems, parts on order, and safety hazards (systems de-activated, panels open).

Keeping up to date, currency

• Engineers must keep their knowledge current: recurrent training (145.A.35(d) — sufficient recurrent training in each 2-year period), type differences, new procedures.
• Licence holders must have 6 months of relevant experience in every 2 years (66.A.20(b)2 and 145.A.35(c)) to exercise privileges.
• Read airworthiness directives, service bulletins, company notices and amendments to maintenance data.

Dissemination of information

Organisations must get safety information to the people who need it: technical notices, quality notices, safety bulletins, briefings, intranet, read-and-sign systems. Feedback from occurrence reports (what changed as a result) maintains trust in the reporting system.

9.8 Human error

Error models and theories

SHELL model (Edwards 1972, adapted by Hawkins 1975; adopted by ICAO). The central Liveware (the person) interfaces with:

Interface	Example of mismatch in maintenance
L–S Software (procedures, manuals, checklists, computer programs)	Ambiguous task card; outdated manual
L–H Hardware (tools, equipment, aircraft design)	Poorly designed tool; connectors that can be cross-connected
L–E Environment (physical and organisational)	Poor lighting, noise, cold apron, commercial pressure
L–L Liveware (other people)	Poor handover, communication, teamwork, supervision

PEAR model (FAA): People who do the job, the Environment they work in, the Actions they perform, the Resources needed.

Reason's "Swiss cheese" model (James Reason): an organisation's defences (design, procedures, training, supervision, inspection) are like slices of cheese with holes. Holes are caused by:

• Active failures — unsafe acts by front-line people (e.g. not refitting a cover); effects are immediate.
• Latent conditions — created by decisions made higher up (e.g. under-staffing, poor procedures, inadequate training); can lie dormant for years.

An accident occurs when holes in several layers line up and allow a hazard to pass through all the defences.

Rasmussen's skill–rule–knowledge (SRK) levels of performance underpin error types: skill-based (automatic actions), rule-based (applying stored rules "if X then Y"), knowledge-based (working out new problems from first principles).

Types of error

Type	Definition	Maintenance example
Slip	Action not as intended — attention failure (skill-based)	Picking up the wrong tool; tightening the wrong nut
Lapse	Forgetting an intended action — memory failure	Forgetting to refit a cap after an interruption
Mistake — rule-based	Applying a wrong or inappropriate rule	Using the procedure for a similar but different variant
Mistake — knowledge-based	Wrong plan due to lack of knowledge	Misdiagnosing a fault in an unfamiliar system
Violation — routine	Habitual deviation, often tolerated ("norms")	Never using the torque wrench for "easy" fittings
Violation — situational	Deviation forced by circumstances	Skipping a step because the correct tool isn't available
Violation — optimising/exceptional	For personal gain or thrill, or a one-off in unusual circumstances	Taking a short-cut to finish early

Errors are unintentional; violations are deliberate deviations from rules (though usually not intended to cause harm).

Other classifications:

• Errors of omission (something not done) — the most common in maintenance (e.g. incomplete installation, item not refitted).
• Errors of commission (something done incorrectly or extra).
• Reversible vs irreversible errors; random, systematic and sporadic error patterns.
• Design-induced errors (a part that can be fitted wrongly — Murphy's law).

Key fact: In maintenance, omissions (incomplete installation, missing parts, items not refitted) are the most frequent type of error found in studies of maintenance occurrences.

Implications of errors

Maintenance errors can cause accidents, incidents, air turn-backs, delays and cancellations, injuries and damage to equipment. Because they are often latent, the person who made the error may never connect it with the outcome.

Organisational errors

Errors made by management or the organisation: inadequate staffing, unrealistic schedules, poor procedures, inadequate training, poor tool control, a culture that tolerates short-cuts. These are Reason's latent conditions — fixing them protects against many individual errors.

Avoiding and managing errors

• Error reduction: good design (error-proof connectors), clear procedures, training, adequate lighting and tools, good planning.
• Error capturing: checks after the task — independent inspection, functional tests, duplicate inspections (145.A.48).
• Error tolerance: the system copes with errors that are not caught — redundancy, multiple load paths, not having the same person work on both engines/systems in the same check.
• Personal strategies: follow procedures, use checklists, avoid working from memory, restart interrupted tasks from a known point, speak up.