Crew resource management

In short

Crew resource management (CRM) is the practised use of all available resources — people, equipment, procedures and information — by a crew to achieve a safe and efficient flight. It exists because most accidents are not caused by a failed engine or a lack of flying skill but by breakdowns in communication, leadership, monitoring and decision-making inside a technically normal aircraft. CRM is now a licensed and recurrent training requirement for Indian flight crew and cabin crew under DGCA Civil Aviation Requirements, with an approved CRM instructor/examiner system, and it extends to ATC, maintenance (MRM) and airport operations teams. In B.Sc Aviation programmes it is examined in human factors and safety papers, and in interviews at Air India, IndiGo, Akasa, SpiceJet, AAI, Air India SATS and ground-handling firms it is tested through situational questions ("the captain is descending below minima — what do you do?").

Explained

What CRM is, and how it evolved

• Definition (working): the effective use of all resources — human, hardware and information — to achieve safe, efficient flight operations; it manages the *liveware–liveware* interface of the SHELL model, i.e. people working with people.
• Trigger: NASA's 1979 workshop on the human factors in air-carrier accidents, prompted by crashes where the aircraft was serviceable — United 173 (Portland, 1978, fuel exhaustion while troubleshooting a gear light) and the Tenerife runway collision (1977).
• The generations of CRM (a standard exam question):
  1. 1st – Cockpit Resource Management: captain-focused management and assertiveness training, psychology-heavy, resented as "charm school".
  2. 2nd – Crew Resource Management: team concepts, briefings, workload distribution, situational awareness; cabin crew brought in.
  3. 3rd – integration with technical training: CRM taught in the aircraft/simulator context, LOFT becomes standard.
  4. 4th – procedural integration: CRM behaviours written into SOPs and checklists (e.g. mandatory call-outs, stabilised-approach gates).
  5. 5th – error management: error is accepted as normal and inevitable; the crew's job is to *avoid, trap and mitigate* it.
  6. 6th – Threat and Error Management (TEM)/organisational: crew performance seen as part of safety culture and the organisation's defences; links to the SMS.
• Related but distinct training: MCC (Multi-Crew Cooperation) teaches the mechanics of two-pilot operation for a licence; CRM teaches the behavioural skills; LOFT is the simulated line flight used to practise them; MRM/Maintenance Resource Management and Dispatch RM apply the same ideas on the ground.

The core CRM skills (the NOTECHS framework)

NOTECHS (Non-Technical Skills) is the European-origin framework used worldwide for *assessing* CRM; it has four categories, each with elements and observable behavioural markers.

• Co-operation: team building, supporting others, conflict solving, exchanging information.
• Leadership and managerial skills: use of authority, maintaining standards, planning and prioritising, workload management.
• Situation awareness: awareness of aircraft systems, of the external environment, and of time (anticipation) — Endsley's three levels: perception → comprehension → projection.
• Decision-making: problem definition and diagnosis, option generation, risk assessment and option choice, outcome review.
• Rating scale (typical): very poor / poor / acceptable / good / very good, with the rule that a rating must be justified by an observed behaviour, not an inference about personality.
• Many operators add Communication as a fifth category, and the ICAO/EASA competency framework (evidence-based training) now lists competencies such as application of procedures, communication, aeroplane flight path management (manual and automation), problem solving and decision making, leadership and teamwork, situation awareness and workload management.

Communication, briefing and the authority gradient

• Closed-loop (three-step) communication: sender transmits → receiver reads back/acknowledges → sender confirms. The loop is what makes a message a shared understanding rather than a noise.
• Standard phraseology and mandatory call-outs remove ambiguity: "1,000 feet, stable", "speed", "sink rate", "go around, flaps". Using standard words lets a non-standard word stand out as a warning.
• Briefings: pre-flight/departure, approach, cabin-crew joint briefing, and debriefing after the flight. A good briefing states the plan, the threats, the triggers for changing the plan, and invites input ("tell me if you see anything").
• Sterile cockpit discipline: below 10,000 ft (and during taxi, take-off, approach and landing) conversation is restricted to operational matters.
• Authority gradient (trans-cockpit authority gradient): the slope of perceived authority between captain and first officer.
  • *Too steep:* the FO will not speak up (the classic cause of monitored-approach accidents).
  • *Too flat:* no one is clearly in command, decisions drift and nobody resolves a disagreement.
  • *Correct:* clear command with open communication — "firm but not closed".
• Assertiveness tools for the junior crew member:
  • PACE: Probe → Alert → Challenge → Emergency (take over / declare).
  • Two-challenge rule: if two challenges get no satisfactory response and the flight path is unsafe, the monitoring pilot takes control.
  • "I'm concerned / I'm uncomfortable / this is unsafe" — graded, pre-agreed wording that the captain is trained to hear as escalation.
• Barriers: rank, age and seniority, language and accent, cultural deference (Hofstede's power distance — relevant to Indian and other high-power-distance cockpits), expectation bias, radio congestion, noise, task saturation.

Decision-making, workload and the handling/monitoring split

• PF/PM (pilot flying / pilot monitoring): the modern replacement for "PNF". Fly first, then navigate, then communicate (aviate–navigate–communicate). The PF never leaves the flight path to troubleshoot; the PM handles checklists, radio and the QRH.
• Structured decision models (quote at least two in an exam):
  • DODAR / TDODAR: (Time) – Diagnose – Options – Decide – Assign/Act – Review.
  • FORDEC: Facts – Options – Risks and benefits – Decision – Execution – Check.
  • GRADE: Gather, Review, Analyse, Decide, Evaluate.
  • DECIDE: Detect, Estimate, Choose, Identify, Do, Evaluate.
  • Naturalistic/recognition-primed decision-making (RPD): experts match the pattern to a known solution — fast, and the usual mode in time-critical events; vulnerable to expectation bias and plan continuation bias.
• Rule-, knowledge- and skill-based behaviour (Rasmussen) maps onto this: checklists are rule-based; a novel double failure is knowledge-based and needs an explicit model like FORDEC.
• Workload management: plan ahead, prioritise, delegate, use automation to buy time, offload non-essential tasks, and recognise task saturation (fixation, dropped radio calls, missed call-outs). Both extremes hurt: overload causes shedding of tasks, underload/complacency causes missed monitoring.
• Startle and surprise management: stabilise the flight path, breathe, then diagnose — the trained antidote to a freeze or an impulsive wrong action.

Threat and Error Management, monitoring and automation

• TEM model (University of Texas / ICAO Doc 9803):
  • Threats — events outside the crew's influence: weather, terrain, ATC pressure, MEL items, bird activity, congested apron, tailwind.
  • Errors — crew actions or inactions that deviate from intention or expectation: handling, procedural, communication.
  • Undesired Aircraft State (UAS) — an unsafe condition such as unstable approach, long landing, wrong altitude, runway incursion.
  • Outcome — recovery back to a safe state, or an incident/accident.
  • Countermeasures are the CRM behaviours themselves: planning, execution and review countermeasures.
• Error types: slips and lapses (execution), mistakes (planning), and violations (routine, situational, exceptional) — the last is a motivational/organisational problem, not a skill problem.
• LOSA (Line Operations Safety Audit): trained observers ride the jump seat on normal flights under a confidential, non-jeopardy agreement and code threats, errors and countermeasures. It gives an operator CRM data from *normal* operations, not just from accidents, and feeds the SMS.
• Monitoring is a primary duty, not a spare-time one. Accident studies repeatedly find ineffective monitoring; remedies are explicit PM duties, FMS entry cross-check, "both heads up by 1,000 ft", and automation call-outs ("mode?").
• Automation and CRM: mode confusion, automation surprise, over-reliance and the loss of manual skill are now CRM issues. Discipline: announce every mode change, verify it on the FMA, and know the levels of automation — if the automation is not doing what you want, disconnect and fly.
• Fatigue and FRMS interact with every CRM skill (first casualty: monitoring and decision quality); the regulatory limits sit in DGCA's flight duty time limitation CAR.

CRM training, assessment and Indian regulation

• ICAO basis: Annex 1 and Annex 6 require CRM/human-factors training; Doc 9683 (Human Factors Training Manual), Doc 9803 (LOSA), Annex 19 / Doc 9859 (Safety Management) place CRM inside the operator's SMS.
• India — the statute: the Bharatiya Vayuyan Adhiniyam, 2024 replaced the former Aircraft Act, 1934 and is in force (from 1 January 2025); the Aircraft Rules, 1937 and the DGCA's Civil Aviation Requirements (CARs) carry the operational detail. Accidents and serious incidents are investigated by AAIB-India under the Aircraft (Investigation of Accidents and Incidents) Rules, 2017; the investigation is no-blame and for prevention only, and its findings must not be used to apportion liability.
• India — CRM requirement: DGCA requires initial and annual recurrent CRM training for flight crew and cabin crew of scheduled operators, delivered by DGCA-approved CRM instructors/examiners (CRMI/CRMIE), with joint flight-crew/cabin-crew sessions and LOFT in the simulator. The requirement sits in the CARs on flight crew standards, training and licensing (Section 7) and the operator's approved Training Manual; the exact series/part and syllabus hours are revised periodically, so quote the current CAR revision rather than a remembered number.
• Typical CRM syllabus items: human error and reliability, error chain, error prevention and detection; company safety culture and SOPs; stress and stress management; fatigue and vigilance; information acquisition and processing; situational awareness; workload management; decision-making; communication and coordination inside and outside the cockpit; leadership and followership; automation; case studies; resilience and surprise.
• Reporting system: mandatory occurrence reporting plus a Voluntary Reporting Scheme that is confidential and non-punitive, and air safety reports/CVR-FDR analysis under the Flight Data Monitoring (FDM/FOQA) programme — all protected by a just culture policy that separates honest error from wilful violation.
• Assessment caution: CRM is *assessed*, and in India a pilot's proficiency check includes non-technical performance; but a single CRM course is not a pass/fail licence event in the way a skill test is — a crew member can be required to repeat training.

What CRM cannot do, and the Indian case studies

• Limitations: CRM cannot fix an unsafe roster, a commercial pressure to continue, poor SOPs, bad equipment or an organisation that punishes reports. It is also hard to measure — behavioural markers are subjective, and trainees can "say the right words" in a classroom.
• Cultural fit: material written for low-power-distance cultures must be adapted; blunt contradiction of a senior captain is unrealistic, so Indian operators train *graded, scripted* escalation (PACE, two-challenge) that is culturally usable.
• Air India Express IX-812, Mangalore, 22 May 2010 (158 fatalities): the Court of Inquiry (AAIB-India was set up only in 2012) found the captain, after prolonged sleep in cruise, continued a steep, unstable approach and ignored the first officer's repeated go-around calls — a textbook failure of monitoring, assertiveness and authority gradient.
• Air India Express IX-1344, Kozhikode, 7 August 2020 (21 fatalities): runway overrun after landing long on a wet tabletop runway in monsoon conditions; the inquiry highlighted non-adherence to SOPs, inadequate CRM, and the PM's failure to press a go-around call.
• Indian Airlines IC-605, Bangalore, 14 February 1990 (A320): confusion over autothrust/vertical modes and late recognition of a low-energy state — the early automation-and-CRM case taught in Indian syllabi.
• Charkhi Dadri mid-air collision, 12 November 1996: language/readback and ATC-crew communication breakdown; a direct driver of mandatory readback discipline and of airspace/ATC reforms in India.

Key points

• CRM – using all available resources (people, hardware, information) for safe, efficient flight; it is a non-technical skill set, trained and assessed.
• Why it exists – roughly 70–80% of accidents involve human-factors/teamwork failures rather than mechanical failure.
• Six generations – cockpit → crew → simulator-integrated → procedural → error management → TEM/organisational.
• NOTECHS – co-operation, leadership and management, situation awareness, decision-making (plus communication).
• Behavioural marker – an observable act that evidences a skill; CRM ratings must cite one.
• Closed-loop communication – transmit, read back, confirm; the only proof of shared understanding.
• Sterile cockpit – no non-operational talk below 10,000 ft, during taxi, take-off, approach and landing.
• Authority gradient – too steep silences the FO, too flat removes command; keep it firm and open.
• Two-challenge rule / PACE – graded escalation ending in the PM taking control.
• PF/PM – PF never abandons the flight path; aviate, navigate, communicate.
• TEM – threats → errors → undesired aircraft state → outcome; CRM behaviours are the countermeasures.
• Error vs violation – slips, lapses, mistakes are honest error; violations are intentional deviations and are handled differently under just culture.
• LOSA – confidential jump-seat observation of normal flights to measure threats, errors and countermeasures.
• LOFT – full-mission simulator scenario, debriefed from video, with no "right answer" scripted.
• Just culture – honest error is reported without punishment; recklessness and wilful violation are not excused.
• Automation CRM – announce mode changes, verify on the FMA, disconnect if the aircraft is not doing what you want.
• Indian framework – Bharatiya Vayuyan Adhiniyam, 2024 and the Aircraft Rules, 1937; DGCA CARs mandate initial and annual recurrent CRM with approved CRMI/CRMIE; AAIB-India investigates, no-blame.
• Stabilised approach gate – by 1,000 ft IMC / 500 ft VMC on speed, path and configuration; otherwise go around — the most common CRM-enforced hard rule.

Formulas and examples

Example 1 — cost of a delayed go-around call. A crew is at 140 kt groundspeed on short final. The PM sees the aircraft high and fast but hesitates for 3 s before calling.

• v = 140 × 0.5144 = 72.0 m/s.
• d = 72.0 × 3 = 216 m of runway and airspace lost before the call is even made; add a 2 s captain response and 1 s spool-up (3 s more) and the total is 432 m. On a 2,700 m runway after a 600 m long landing, that is the difference between a stop and an overrun — the arithmetic behind "call it the moment you see it".

Example 2 — time the sterile cockpit actually buys. An aircraft descends from 10,000 ft to touchdown at an average 1,500 ft/min.

• t = 10,000 ÷ 1,500 = 6.67 min ≈ 6 min 40 s. The entire approach-and-landing phase — where a large share of accidents occur — is under seven minutes of workload; this is why briefings are done *before* the descent and why non-operational talk is banned inside it.

Example 3 — risk index for a CRM hazard (ICAO 5×5). Hazard: "first officers on the fleet do not call go-around on unstable approaches."

• Worst credible outcome: runway overrun with fatalities → severity A = 5.
• FDM data show 4 unstable approaches continued in 2,000 flights → occasional, L = 3.
• R = 5 × 3 = 15 → in the red/unacceptable region (typically ≥ 12): operation to be restricted or mitigated immediately. Mitigations: mandatory PM go-around authority written into the SOP, simulator LOFT on assertiveness, no-jeopardy go-around policy.
• After mitigation, continued unstable approaches fall to 1 in 4,000 flights → L = 2, R = 5 × 2 = 10 (amber/tolerable with monitoring).

Example 4 — reading a LOSA report. 220 flights observed; 352 threats, 143 errors, of which 19 were mismanaged into an undesired aircraft state.

• Threat density = 352 ÷ 220 = 1.60 threats per flight.
• Error density = 143 ÷ 220 = 0.65 errors per flight.
• Error-management rate (mismanaged) = 19 ÷ 143 × 100 = 13.3 %; so the crews trapped or mitigated 86.7 % of their own errors.
• Interpretation: errors are normal and mostly caught — the training target is the 13.3 %, not the 0.65.

Example 5 — monitoring an altitude bust. Cleared FL090, the crew sets 9,000 ft but the PF leaves V/S at 1,800 ft/min with no level-off armed. The PM's altitude call-out is due at 1,000 ft to go.

• Time from the call-out to busting the level = 1,000 ÷ 1,800 = 0.556 min = 33 s of margin.
• If the PM is head-down on the load sheet and calls at 300 ft to go: 300 ÷ 1,800 = 0.167 min = 10 s. The call-out is not a courtesy — it is the only defence, and its value falls in direct proportion to the rate of descent.

Where it is used

• Airline flight operations: IndiGo, Air India, Air India Express, Akasa, SpiceJet, Alliance Air — annual recurrent CRM, LOFT in the simulator, FDM/FOQA review boards and stabilised-approach policies.
• Cabin crew and joint operations: joint flight-crew/cabin-crew CRM, NITS briefing (Nature, Intention, Time, Special instructions) before an emergency landing, evacuation command discipline.
• ATC and airport operations: TRM (Team Resource Management) at AAI centres; apron safety teams, ramp marshalling, push-back and de-icing crews use closed-loop communication and standard phraseology to prevent runway and apron incursions.
• Maintenance (MRM): shift handover documentation, duplicate inspection, task sign-off and the "challenge" of a senior AME — the engineering version of the same skill set, under CAR-145/CAR-66 human-factors training.
• Training organisations and recruitment: DGCA-approved FTOs teach MCC and CRM; airline pilot and cabin-crew selection uses group exercises, simulator assessments and situational interviews that score exactly these behaviours.
• Beyond aviation: operating-theatre and ICU crisis resource management in Indian hospitals, railway loco-pilot/guard crew protocol, offshore drilling and nuclear control-room teams all use CRM-derived training.

Interview questions

• Q: Define CRM in one sentence. A: The effective use of all available resources — people, equipment, procedures and information — by a crew to achieve the safest and most efficient conduct of a flight.
• Q: CRM is behavioural training, so how is it assessed objectively? A: Through behavioural markers under a framework such as NOTECHS or the ICAO competency model — the assessor must cite an observed action, not a judgement about the person's character.
• Q: What is the difference between MCC and CRM? A: MCC trains the mechanics of operating a multi-crew aircraft (task sharing, call-outs, SOPs); CRM trains the non-technical skills — communication, leadership, situation awareness and decision-making — that make that cooperation work.
• Q: You are the first officer; the captain is clearly below the profile and silent. What do you do? A: Use graded escalation: state the deviation factually ("500 feet, sink rate 1,200, we are one dot low"), then alert and challenge; if two challenges bring no correction and the flight path is unsafe, announce "I have control" and go around.
• Q: What is the two-challenge rule? A: If a crew member challenges an unsafe action twice without a satisfactory response, they are authorised and expected to take over control.
• Q: Explain the TEM model. A: Threats arise outside the crew's control, crew errors occur in handling them, unmanaged errors produce an undesired aircraft state, and the outcome is either recovery or an incident; CRM countermeasures act at every link.
• Q: What is an undesired aircraft state? Give an example. A: An aircraft position, speed, attitude or configuration that clearly reduces safety margins — for instance an unstable approach, a wrong altitude, or lining up on a taxiway.
• Q: What is the authority gradient and what is the ideal one? A: The perceived difference in authority between crew members; the ideal is firm, clear command combined with openness, so the junior pilot will speak and the senior will listen.
• Q: Does CRM dilute the captain's authority? A: No — the captain retains final authority and responsibility; CRM only ensures the decision is informed by all the information the crew holds.
• Q: What is LOSA, and why is it more useful than accident reports? A: A confidential, non-jeopardy observation of normal line flights that counts threats, errors and countermeasures; it measures how safety is actually produced every day instead of only what went wrong once.
• Q: What is just culture? A: A policy under which honest errors can be reported without punishment so the organisation learns, while gross negligence and wilful violations remain accountable.
• Q: Which Indian accident is the standard CRM case study, and what was the lesson? A: Air India Express at Mangalore in 2010 — the first officer's repeated go-around calls on an unstable, steep approach were not acted upon; the lessons are mandatory monitoring call-outs, PM go-around authority and a workable authority gradient.
• Q: Which law and regulator govern CRM training in India today? A: The Bharatiya Vayuyan Adhiniyam, 2024 with the Aircraft Rules, 1937, administered by the DGCA through its Civil Aviation Requirements; accidents are investigated separately by AAIB-India.

Common mistakes

• Treating CRM as "being polite and getting along" instead of a set of trained, assessable skills with defined markers.
• Writing that CRM transfers command authority to the crew — it informs the captain's decision; the authority and responsibility stay with the captain.
• Confusing MCC, CRM and LOFT, or listing LOSA as a training course rather than a data-collection audit.
• Naming only three or four CRM generations, or reversing the order — examiners mark the progression cockpit → crew → integrated → procedural → error management → TEM.
• Mixing up error and violation, and so recommending retraining for what is actually a motivational or organisational problem.
• Quoting a remembered DGCA FDTL figure or a repealed statute (the Aircraft Act, 1934 is the *former* Act); cite the current CAR revision or describe the provision without the number.
• In interviews, answering a scenario question with "I would inform the captain" and stopping — no escalation, no time awareness, no go-around decision.

Related topics

Human performance and limitations · Aviation safety and safety management systems · Air traffic control and aerodrome operations · Airline operations and flight dispatch · Aircraft systems and cockpit automation · Emergency procedures and cabin safety.