27 th International Aircraft Cabin Safety Symposium Orlando, Florida, USA 27-29 April 2010 Recent Major Accidents & the Capacity of Training To Avoid Them.

27th International Aircraft Cabin Safety SymposiumOrlando, Florida, USA 27-29 April 2010

Recent Major Accidents & the Capacity of Training To Avoid Them

World Aviation Training ConferenceOrlando, Florida

April 2010

Robert Matthews, Ph.D.Senior Safety Analyst

Office of Accident Investigation, FAA


2

Training is a good thing; it reduces the risk of accidents.

Start with the premise that training is part of a 3-legged stool:Establish good standard operating procedures (SOPs);Train repeatedly to those good procedures; andEnforce those good procedures.

This presentation reviews 19 recent hull-losses involving 3 common accident categories to illustrate training’s capacity, as part of this 3-legged stool, to reduce risk.

Bottom Line: training can reduce the likelihood of many serious accidents, though its capacity to do so has practical limits.


3

Six Take-Off AccidentsFrom 2006 through 2008


4

27 August 2006; Comair CRJ2 at Lexington (49 fatal)

Early morning flight in night VMC. Flight crew planned to take off on

22 but lined up on a much shorter runway 26 & began takeoff roll

Ran off the runway end and struck the airport perimeter fence, trees, and terrain

Causes, Factors & Issues: Non-pertinent conversation on taxi & loss of positional awareness. Failed to cross-check & verify they were on the correct runway. Crew failed to use available cues & aids to identify location during taxi. Add airport configuration (short, single taxiway to 2 near-by runway ends) &

possible fatigue for Captain. Crew issues are basic in training programs, including private pilot training. Score for More Training: Low. Limited to the capacity to instill discipline &

adherence to SOPs on every flight.


25 Jan 2007, Regional Air, Fokker 100, at Pau-Uzein Airport (No injuries to 54 onboard; 1 ground fatal)

•T/O after 50-minutes on ground. WX: snow, broken 1,100, Temp/dew 0/-1C.•Rotated “abruptly” at 128 knots when the crew saw flocks of birds.•A/C immediately banked 35 degrees left, 67 right, then 59 left.•Reached 107 feet, sank, right gear struck runway & A/C bounced. •PIC (PF) aborted at 160 knots. •Touched down right of runway end, crossed a road & hit a truck, killing its driver, then rolled out into a field.

Causes: Limited awareness by crew & company of effects of ice on the ground; •No tactile verification of wing condition;•Wing design’s sensitivity to icing.•Rapid rotation & lack of “crew vigilance.”

Photo: BEA

Training Score: Moderate; high if good SOPs were in place; they were not.


25 May 2008, Kalitta Air B742 at Brussels (Minor injuries to 5 onboard)

•BRU uses 25 for T/O but 20 on Sunday for noise 9,573 feet (1,950 shorter).•Began T/O at taxiway 800 feet down runway; had computed T/O parameters for full runway.•Bird strike & compressor stall at V-1 & ATC noted flames on right side. •Captain continued several seconds, then decided he could not T/O safely & aborted 12 knots above V-1. •All 4 engines were brought to idle and did not deploy thrust reversers.•Overran 1,000 feet down embankment & came to rest with nose overhanging second embankment next to busy rail line.

Causes: RTO 12 knots past V-1; Miscalculated takeoff parameters.Failed to use thrust reversers.Situational awareness.

Training Score: Moderate - - Event complicated by bird strike & compressor stall exactly at V-1. Training still may have helped to emphasize SOPs, including check of computed T/O parameters & use of thrust reversers

Aviation safety.net


7/708, Kalitta Air B742, Bogota (3 ground fatal; 2 of 8 onbd serious)

El Tiempo & Reuters

•Night T/O (02:50); ceiling 1,700; calm.•Engine #4 failed on rotation.•Climbed out & began circular engine-out return in Bogota Bowl.•#1 failed 55 seconds after rotation 8.5 miles out at ¼ through circling return.•Seconds later, #2 engine entered compressor stalls.•Made emergency landing in dark field 6 miles NW of airport & struck a farm house; 3 fatal on the ground.

Cause: No report yet.

Broader Issues still not clear: investigation has become costly JT9D science project.

Training Score: Zero. Crew performed well under conditions; 3 engines failed in a mountainous black hole


8 Aug 2008, Spanair MD-82 at Madrid (154 fatal & 18 serious)

Reuters

•Awaiting T/O, A/C returned to gate for faulty Ram Air Temperature probe.•Maintenance pulled 2 CBs for ground control relay switch but tagged just 1.• Flight crew later reset tagged CB but not the other CB.•Crew ran all check- lists, yet failed to set flaps (FO simply replied “check” for flaps, but did not set them).•Result: no-flap T/O & configuration warning system was disabled.•Rotated at 154 knots, reached 40 feet AGL, then sank & rolled.

Issues: No final report yet but issues appear to include maintenance SOPs (Tagging) & checklist discipline (flaps) & risk from small disruptions.Bottom line: Moderate. Like LEX, must assume capacity to ensure adherence to SOPs every time, but a better chance here: if either flight crew or maintenance had followed procedures, no accident.


20 Dec. 2008, Continental B737-500 at Denver (5 serious, 115 onboard)

Rocky Mountain News

•B737 with winglets began T/O on 34R; wind 290 at 24, gusting 32.•A/C weather-vaned & drifted left as engines were nearly spooled up.•Ran off left 40 seconds into T/O roll.•A/C continued accelerating several seconds, then speed brake handle was deployed & A/C began decelerating.•Rolled across turf & down slope, with engines starting to burn.•Successful evacuation

Under investigation but likely to include: Demonstrated crosswind with winglets.Crew performance & response to drift.

Training Score: Low. Training for crosswinds & use of tiller may have had some effect, but, crew was experienced in make-model. On net, more targeted pilot training probably had only limited hope of avoiding this accident.


10

3 cases illustrate the value of training to & following good SOPs (BRU, Madrid & PAU-Uzein), while Pau-Uzein also illustrates the need to have appropriate SOPs & guidance in place to start.

Madrid also illustrates the same lessons in maintenance.

But, training had limited hope of directly influencing or averting LEX, DEN or Bogota.

Bottom line for training is mixed:•3 low scores & “only’ 3 moderate scores, but somewhat higher with good SOPs in place.

Findings for 6 recent Take-off Accidents


11

Four Runway Overruns on LandingFrom 2007 & 2008


12

18 Feb 2007: Shuttle America as Continental ConnectionERJ-170 at Cleveland (No Injuries to 4 Crew & 70 Pax)

•FO flying in first pairing of this crew. Contradicted SOP but PIC was fatigued.•Cleared to runway 28 (6,017 ft).•RVR reported at 6,000 & braking action fair, but glideslope unusable due to deep snow.•Passed FAF & ATC reported RVR at 2,000.•PIC (PNF) reported approach lights in sight & runway in sight at 50 feet AGL. •FO then turned off the A/P to land. •At 30 feet AGL, PIC briefly lost sight of the runway, then regained it, & continued.• In strong gusty winds, high sink rate developed in flare & “likely stalled.” Landed long & hard; gear trunnion fractured.•Reverse thrust peaked at 70% for 2 seconds, then slowly to reverse idle. •Applied 20% max brakes for 8 seconds, then 75%, then 90% when PIC applied his brakes. •Ran off into snow-Ccovered grass & penetrated a fence 150 feet past runway end.

Causes, Factors & Issues: •Lost visual cues & failed to go around. •Descent to ILS DH instead of localizer MDA (glideslope out).•Landed long on contaminated runway•No max reverse thrust or max brakes.•Captain’s fatigue & fear of reprisal.

Photo: NTSB

Training: Moderate on net. Winter ops training & go-around gates score high, but offset by fatigue & violation of several SOPs.


17 July 2007, TAM A320-200 at Congonhas, Sao Paolo (All 181 pax & 6 crew fatal; 12 Fatal & 11 severe on ground)

•6,365-foot runway was repaved but not yet grooved; no overrun area.•T/O Porto Alegre with #2 thrust reverser deactivated before flight. •ATC advised crew of light rain & wet runway (35L); wind 330 at 8.•Landed at normal spot but software requires both reversers at or near idle for spoilers to deploy.•#1 engine went to reverse, #2 to “climb.” •Overran left at high speed, over adjacent, low highway; into TAM Express building & gas station.

Causes: Runway characteristics; construction; inadequate crew training; Airbus’ lack of warning for braking system failure.

•Training Score: Moderate – conceptually high for crew knowledge of software; risk assessment; recognition of single-engine reverse thrust. •But offset by MEL & dispatch procedures (A320 to short, wet runway without T/R) & by airport operations (un-grooved, short runway).


14

• FO (PF) recently completed IOE in CRJ2; this was his second flight on this route (PHL-PVD)

• ILS approach to runway 5 (7,000 feet, snow-covered in rain & mist, in darkness).

• Winds aloft at initial descent from 220 at 100 knots (large tailwind component).

• 2 miles out at 700 AGL, FO disconnected the A/P & FD to “get the feel of the airplane.”

• A/C drifted left & above glidepath.• Broke out at 300 & saw approach lights at 2

o’clock. PIC took control.• FO thought PIC had called for power to idle &

FO reduced power without PIC’s knowledge. • A/C reached max 22-degree bank below 100

AGL & descent rate of 2,000 FPM. • “Porpoised” in the flare & landed 1,200 feet

long in 9-degree bank; main gear collapsed.• Ran off 3,700 feet on snow-covered grass.

Causes, Factors & Issues: • Unstable approach, high sink rate, stall, &

hard landing.• FO’s poor execution of ILS approach• Poor communication in cockpit.• Inadequate FO training & experience.

12/16/07: Air Wisconsin as US Airways Express, CRJ-200 at Providence (No Injuries to 3 Crew & 31 Pax)

Training Score High: time in A/C; Go-around gates; CRM; winter ops.


30 May 2008, TACA A320-200 at TegucigalpaOf 135 onboard, 3 fatal, 60 serious; 2 fatal on ground

•Landed near max weight (63.5 versus max 64.5t) on short, wet runway in light drizzle & 12-knot tailwind from nearby tropical storm “Alma.•Runway: 5,410 feet available landing distance at 3,300 MSL; 1% downslope & ungrooved).•A/C was configured for landing in Speed Mode – (above Vref); landed at 139 IAS. •Immediately selected MAX REV. Nose touched down 7 seconds after mains. •Applied manual braking 4 seconds later & max pedal braking 10 seconds later. •Selected IDLE REV at 70 knots & 625 feet remaining.•Overran at 54 knots & dropped down 65-foot embankment onto street.

Training Score: Moderate. LOFT, risk assessment & go-around gates score high. Offset by national airport policies, dispatch & crew’s failure to monitor.


16

Training scores range from moderate at TEG & CLE to high at Sao Paulo & PVD.

Some high scores:More time learning the airplane;Stable approach & go-around gates (all 4 accidents);Risk Assessment (all 4 accidents)LOFT & winter operations (CLE & PVD).

Offset in some cases by: Airport configurations;Crew fatigue & violation of established SOPs;Carrier policies & guidelines (unclear or inappropriate).

Findings from 4 Runway Excursions on Landing(CLE, PVD, TEG, SPO)


17

Nine Undershoot Accidents,2008 & 2009


18

17 January 2008, British Airways 777 at Heathrow (1 serious among 152)

Cause: No final report yet.Issue: Rarely identified fuel-heating issue unique to one system.

AAIB recommends all CAA’s require the use of anti-icing fuel additives, such as FSII, used in military and high-end business jets.

•Normal approach until 600 feet AGL & 2 miles out, when auto throttle demanded an increase in thrust.•Both engines failed to respond.•Crew moved levers manually & again no response.•A/C lost speed & landed short.Ice deposits of water & fuel (soft & mobile) accumulated in fuel lines.•Restricted flow in fuel/oil heat exchanger (FOHE) in Trent-800.•This led to thrust rollback in flare.

Training Score: zero.


21 Feb. 2008, Santa Barbara Airlines ATR42 at Merida, VZ (All 46 fatal)

•Airport closes at sundown. Accident A/C last scheduled T/O for day.

•Clearance delayed for inbound aircraft.

•On taxi, FO (PF) notes gyros had failed.

•PIC says equipment is “crap” & says he has had to operate w/o gyros before.

•Crew: “Go visual” & “try to reset it in flight.”

•Start T/O roll; no brief of visual departure.

•On climbout, PIC tells FO “a hair more to the right.”

•Then confusion: FO & PIC read different headings, then “PULL UP.”

•PIC takes control but confusion continues about heading, with more “PULL UP.”

•PIC starts right turn early & strikes 80-degree rock face.


Broad Issues: SOPs; Maintenance; Corporate cultureDecision making; Situation AwarenessCAA (Carrier’s AOC later revoked).

Training Score: zero. Lots of conceptual opportunities, but plausibility is over-whelmed by more basic issues.


6 July 2008, USAJET (Cargo) DC-9-15 at Saltillo, MX (1 of 2 pilots fatal)

Reuters

•Crashed ILS approach at night. •Airport at 4,646 MSL & surrounded by mountains.•Visibility 1 mile in fog.•Crew never checked in with tower & did not have current weather.•Video shows A/C flying low, wings rocking, then wing dropped (stalled).•Crashed onto loop road next to major highway.Cause: No report yet.

Possible Issues: Go-around gates, current weather information, SOPs.

Training Score: High, depending on strength of SOPs.


14 Sep. 2008, Aeroflot-Nord B737-500 at Perm, RU; (All 82 pax & 6 crew fatal)

•Late night approach in rain & fog; A/P & A/T off due to long-recurring A/T problem.•Crew recently transitioned from T-134 & AN-2 (Reversed ADI); both low-time M/M•On final, ATC tells crew they are right of course. •Corrected but climbed from 600m to 900m instead of descending to 300M to land. •ATC advised crew of climb: “Affirmative; we’re descending, then climbed to 1,200m.•ATC instructed right turn for go-around.•Acknowledged but turned left & rapid descent. Impacted industrial area & rail line.

Causes:Spatial disorientation (ADI)Pilot workload (independent throttles)Pilot’s BAC, fatigue & lack of CRMMaintenance - - carrying faulty throttle.Carrier’s transition training to 737 fleet.Carrier’s operation of 737 fleet.

Training Score: Despite extreme corporate issues, crew pairing issues & maintenance practices, net training score is high because a single item (ADI transition) comes close to a knock-out.


27 Jan 2009, Empire at Lubbock; 1 of 2 crew serious)

•Crashed short of 17R on ILS in night IMC. (Wind 350 at 10), visibility 2, light freezing drizzle, mist, 500 overcast.•On descent from FL100 to FL80, ATC advised crew of wind shift. •PIC acknowledged & noted 8-degree drop in outside air temp. •At 0434, cleared to land 17R. •Set flaps 15 but got asymmetric flaps.•Shedding speed on final. •Landed short, struck approach lights, & skidded off right, into grass.

Cause: Still under investigation.

Broad Issues: •Possible flap damage from past events (Engine/wing fire & bird strike).•Asymmetric flaps?•Monitoring airspeed


12 Feb. 2009, Colgan Air DHC-8-400 at Buffalo (All 49 & 1 fatal on ground)

BBC & AP

•FO arrived EWR on red-eye from West Coast via MEM at 0623. PIC with significant sleep deficit.

•Accident flight delayed; T/O EWR at 2120.

•Newly upgraded PIC (110 hours in M/M); FO had 700 hours in type.

•Steady chatter throughout flight.

•Cleared to descend & maintain 2,300. Had been bleeding off airspeed & 20 knots slow.

•Failed to note low-speed cues.

•A/P disengages; A/C stalled in turn & struck home in dense area, 45 degrees wing low, 30 degrees nose low, & little forward speed.

Causes & Factors: PIC’s incorrect response to stick shaker & then stall; failure to monitor airspeed & low-speed cue; sterile cockpit; PIC failed to manage the flight; & inadequate procedures for airspeed selection in icing conditions.

Training Score: Lots of conceptual opportunities (transition to DHC-8-Q400, stall recognition, CRM & flight monitoring-SOPs). But effectiveness reduced by constant chatter, lack of professionalism & especially by fatigue. Yet, net score still earns a “moderate.”


25 Feb. 2009, THY B737-800 at Amsterdam

(9 fatal & 28 serious, 134 onboard)

Reuters

•Crashed 1km short of Rwy 18 in day-time mist & low ceiling, wind 200 at 10.•At 1,950 feet on coupled approach, left altimeter suddenly read 8 feet. •Crew noted faulty altimeter but did not consider it a problem. •Faulty altimeter reading caused A/C to assume landing logic, so throttles went to retard & A/C lost altitude.•When crew selected 144 knots airspeed but with thrust levers at idle.•A/C commanded more pitch.•Crew did not notice loss of altitude until stall warning at 150 AGL. •Added power & pulled up but impacted before spool-up.

Broad Issues: (No report yet.) Faulty altimeter;•Knowledge of A/C (A/P reads #1 altimeter);•Flight monitoring & SOPs

Training Score: Moderate to high for software & recognizing that A/C reads left altimeter, & that A/C assumed landing logic; offset by maintenance issues.


22 March 2009, Fedex MD11 at Narita (Both Pilots Fatal)

•Flared late & touched down flat.•Bounced nose high & PIC pushed the nose over. •Second touchdown landed sharply on nose gear, at 30 degrees nose-down in moderate left roll. •Left wing failed & instant fire ball. •Right wing continued flying; A/C rolled inverted.


Broad issues: •Familiarity with MD-11 (tendency to pitch up after ground spoiler deployment & crew tendency to over-control).•(Newark & Air China accidents).

Training Score: Moderate. Crew experienced in MD-11; had the appropriate training many times.


26

•Overran shortly after midnight in “fierce rain.’”•Paxs said turbulence forced crew to halt cabin service 3 times, then terminated service. •Before descent, pilot warned of more turbulence but said it likely would not be much worse. •Crew requested Runway 12 with 14-knot tailwind from 310 degrees. ATC offered Runway 30 but crew repeated request for 12. •Cleared to 12; ATC added that runway was wet. • After descent through clouds, crew made visual contact with runway at @ 800 AGL. •Landed 4,000 feet long on 9,900-foot runway at 162 knots. Overran at 73 mph, through perimeter fence, crossed road & came to rest 175ft past runway end, 12 meters from water line.

Under Investigation Possible Issues: •Unstable approach, SOPs.•Risk assessment; go-around gates; approach briefing; CRM.

23 Dec. 2009: American 737-800 at Kingston, Jamaica (4 pax serious, minor-no injury to 144 pax, 4 FA & 2 pilots)

Training Score: High – if good SOP


27

On balance, high scores.

Zero to little chance in 2 cases (LHR & Santa Barbara).

Moderate or moderate-high at BUF, Saltillo & Narita)

High at Perm, Amsterdam & Kingston

Training issues focus on aircraft characteristics, SOPs, risk assessment & go-around gates. Offset by absence of good SOPs & more basic corporate short-comings in some cases

Findings from 9 Undershoots


28

Training is most effective as an intervention when: (1) we establish good standard operating procedures ; (2) we train repeatedly to those good procedures; (3) we enforce those good procedures; and(4) we train to the aircraft characteristics.

Conclusions from 19 Recent Accidents

Though training is not a key issue in every accident, recent major accidents illustrates that training can significantly reduce risk in the large majority of cases.

27 th International Aircraft Cabin Safety Symposium Orlando, Florida, USA 27-29 April 2010 Recent Major Accidents & the Capacity of Training To Avoid Them.

Documents

capacity of training

training programs

runway end

shorter runway

runway ac

flight crew

correct runway

good sops