WCA Flight 708 Factual (interposed commentary below)


Airliner May Have Stalled At 33,000 Feet

An MD-82 that crashed in Venezuela last August, killing 160, may have been behaving just the way Boeing had warned it might in a 2002 service bulletin. The bulletin warned that the autopilot might reduce engine power too much after a rapid climb, allowing airspeed to bleed off to the point of a stall. Pilots of the West Caribbean Airways flight, out of Panama for Martinique, may have been unaware, unnamed French investigators (Martinique is a French island) told the International Herald-Tribune. An interim report on the crash released by the Venezuelan government last November said the plane climbed from 31,000 feet to 33,000 feet and held the altitude for eight minutes before the autopilot turned itself off. The plane then descended for a minute before the stall horn sounded. It then fell to the ground at about 10,000 feet per minute, with the pilots pulling full back on the control yoke. The unnamed sources said the process would have happened gradually, with the autopilot trying to maintain altitude using pitch adjustments until shutting itself off just before the stall horn sounded. The pilots reported that both engines had flamed out but flight data recorder information indicated the engines were running normally when the descent began, although the right engine was cut back to idle shortly after. Data wasn't available for the left engine. Evidence from the wreckage shows both engines were turning at high speed at the point of impact. The recorder also shows that rather than push the nose over to recover from the stall, the pilots held the yoke to their chests all the way to the ground.


Ice on wings determined as cause behind Colombian plane crash
www.chinaview.cn 2005-11-28 12:45:07
Despite this "ice on wings" finding (see below), the cause is likely to have been a lack of thrust caused by an engine inlet icing build-up - enough to cause a loss of RPM, perhaps a generator to fall off line (under-frequency trip) and cause flight instruments to fail (leading to a loss of control). As the aircraft descended that inlet icing may have melted, allowed the engines' RPM to climb back up and the generator to reset. Unfortunately by that time the aircraft may have been locked in a disorienting deadly spiral dive. Wing icing alone is unlikely to cause a loss of control.

"The flight crew stated that they had a dual engine flameout"

For another explanation of this power-loss/regain phenomenon see this link and this pdf file
    BOGOTA, Nov. 27 (Xinhuanet) -- Ice on the wings of a Colombian plane that crashed on Aug. 16, killing the 160 people on board, has been determined as a cause behind the tragedy, according to the El Colombiano newspaper.

 The passenger plane's black box had recorded the cabin crew discussing bad weather conditions and the possibility of ice on the wings, the newspaper reported on Sunday, citing the US National Transport Security Board (NTSB).

    The results were released after investigations made by the NTSB using data supplied by Venezuela's Air Accident Investigation Committee.

    The twin-engine McDonnell Douglas MD-82 aircraft of Colombia's carrier, the West Caribbean Airways (WCA), was flying from Panama City to Martinique, the French-owned island in the Caribbean, when it crashed over Venezuela, killing the 152 passengers and its eight crew members on board.

    The NTSB report said that "both engines show evidence that the rotors were turning at high speed at the moment of impact."

    The crew had said it was not an emergency, but had requested a new flight altitude from Venezuelan controllers on the ground.

    "The report shows that what happened was an accident: not problem with the plane, as the media have said," WCA director general Captain Jorge Perez told El Colombiano   (from this link)



On August 16, 2005, West Caribbean Airways flight 708, an MD-82 (registration HK-4374X), crashed near Machiques, Venezuela while on a charter flight from Panama to Martinique. All 160 persons aboard the flight died in the crash.

The following information has been released by the Comite de Investigacion de Accidentes Aereos (CIAA) of Venezuela. All States assisting the investigation -- France, Colombia and the United States -- agree with the factual findings. The NTSB is distributing this information at the request of the Venezuelan Investigator-in-Charge.


Movement of the wreckage has been delayed due to very heavy rains in the area where the airplane crashed. However, it should be moved to a secure area in Maracaibo in the next few days. Once the wreckage has been moved, additional inspections will be completed. Initial examinations on site revealed:

--Ground scars indicate that the airplane impacted in a nose up and slight right roll attitude.
--Wreckage was distributed over a triangle shaped area that was approximately 205 meters long and 110 meters at its widest point.
--Both engines exhibited indications of high-speed compressor rotation at the time of ground impact.
--The engine inlets, empennage and wing leading edges showed no sign of pre-impact damage.
--The horizontal stabilizer was found at about the full airplane nose up position (about 12 units nose up).

Flight Recorders

The cockpit voice recorder (CVR) and flight data recorder
(FDR) were downloaded at the Bureau d'Enquetes et d'Analyses pour la Securite de l'Aviation Civil (BEA) laboratory near Paris, France, during the week of September 5, 2005.

Both the FDR and CVR casings were severely damaged due to impact forces. Both recorders operated until impact.

Flight Data Recorder

The FDR protected module was in good condition and the magnetic tape was extracted in good condition. The investigation has gained good information from the recorder.
However, several parameters were not recorded as designed, including left engine pressure ratio (EPR), pitch attitude, roll attitude, magnetic heading, and column position.

The following events are recorded on the FDR:

--The accident flight lasted about one hour from takeoff to the end of the recording.
--The flight reached its initial cruise altitude of flight level (FL) 310 at about 6:25 UTC (universal coordinated time).
--At about 6:41 UTC, about 20 minutes before the end of the recording, the airplane began a normal climb to FL330.
It leveled off at about 6:43 and accelerated to Mach 0.76. The right engine EPR was consistent with normal climb and cruise values.
--About 90 seconds after reaching Mach 0.76 (6:49 UTC), the airspeed began to steadily decrease. Timing works for a gradual build-up of engine inlet ice in heavy cloud at height. The horizontal stabilizer moved from about 2 units nose up to about 4 units nose up during this deceleration. Autopilot auto-trim (classic unnoticed loss of performance due to ice build-up and consonant engine power losses.)
--About 3 minutes and 30 seconds (6:57 UTC) from the end of recording, the Mach number reached about 0.60. The autopilot was then disengaged and the airplane started to descend from FL330.
--As the airplane descended past about FL315, the airspeed continued to decrease and the right engine EPR decreased to about flight idle.
--The airplane descent rate increased after passing through FL310.
--The airspeed reached a minimum of about 150 indicated air speed (IAS) knots at about FL250.
--Right engine EPR stayed at approximately flight idle through the descent and even increased several times, including shortly before the end of the recording. Engine stuck at idle due to combo of inlet choke ice build-up plus p1/p2 sensor iced over. As they descended into warmer air the choking inlet ice started to melt and the EPR started rising although the pilots had not noticed this "recovery" and just left the throttle back at idle.
--Once the airplane started to descend, the horizontal stabilizer moved in increments to about 12 units nose up (which is about full nose up trim) while descending through FL200. Sounds like autopilot was left engaged and Vs mode used during descent (autopilot ran hoz stab to its limits in an attempt to achieve the decreed min descent rate during the descent)

Cockpit Voice Recorder

The CVR protected module was partly opened due to impact forces. Overall, the magnetic tape was in good condition, but the tape was partially cut due to impact forces. The overall quality of the recording is poor, with many areas of static and loud background noises. However, valuable data was obtained.

Almost all crew discussions to communicate with each other and with air traffic control (ATC), in Colombia and Venezuela, were in Spanish.

The CVR recorded the last 32 minutes of the accident flight.

The following events are recorded on the CVR:

--At about 06:53 UTC (i.e. 4 minutes after the loss of cruise performance had started to occur at time 0649 - see above), approximately 8 minutes before the end of recording (while the airplane is level at FL330) the flight crew discusses weather concerns that included possible icing conditions. The flight crew also discusses turning on engine and airfoil anti-ice. .......but if you don't actually do that and the engine runs down (even just to flight idle or below without flaming out), the engine's bleed air output will then be insufficient to stop icing build-up - the only solution is then to very quickly get out of icing by diving, not just drifting down at min descent rate). i.e. over 4 mins after performance loss had been noted by the DFDR, it was probably already an irretrievable situation
--About 3 minutes and 30 seconds (6:57 UTC) before the end of the recording, the crew requests and is cleared to descend to FL310.
--About 3 minutes before the end of the recording, an audio warning similar to altitude alert is heard, followed 22 seconds later by a sound similar to stick shaker (autopilot had run out of hoz stab adjustment and was trimming a/c into a stall) (6:58 UTC) and then an aural stall warning alert.
These warnings sound continuously until the end of the recording.
--The flight crew requests subsequently lower altitudes of FL290, FL240, and finally 14,000 feet.
--The flight crew does not declare an emergency, and they do not refer to any checklist during the descent. (crew culture? afraid of subsequent sacking/punishment, already thinking about covering it all up, not realizing how serious their predicament was?)
--About 1 minute after the start of the sound similar to the stick shaker, the flight crew states that they had a dual engine flameout when asked by ATC if they had a problem. (Doubt the "dual flame-out. At least the right engine was still running and its generator was on line - or lighting and comms would've been down). The RH engine was just too choked with inlet ice to produce thrust at height - and the throttle had been left at idle, crew not realizing that thrust might be available as ice had part melted in the descent. They'd assumed a flame-out because of the lower than normal flt idle RPM (due ice-choke). They may have assumed that EGT seen was just residual jet exhaust temperature. It's possible that the overtaxed crew shut down the wrong engine (based on that LH engine probably having been available to minimize their rate of descent, although having been iced up to some extent. If they had, then the RH engine's generator would've stayed on line, even though that engine's idle was likely somewhat below flight idle (due to the inlet ice choke). But they would've had to have later restarted that engine (not shown as a CVR event) because "Both engines exhibited indications of high-speed compressor rotation at the time of ground impact." So why was there insufficient power available from the LH engine? Did it continue to ice up in their descent?

--Last radio transmission from the flight crew to ATC was at about 07:00:11 UTC. (20 secs before impact)
--About 8 seconds before the end of recording, a ground proximity warning system (GPWS) warning starts to be heard and continues to the end of recording. This would've alerted crew to prepare for a crash landing attitude but unfortunately they hit rising terrain at night, albeit in a nicely flared attitude.
--The time from first sound similar to stick shaker, to the end of the recording is about 2 minutes and 46 seconds. Autopilot was disengaged only 3.5 mins before impact (i.e. after 0657 - see above). Crew possibly overloaded and unaware of the fully autopilot N.U. trimmed hoz stab (which would've been well iced at that high AoA)
--The end of the tape occurs at about 7:00:31 UTC.


a. Is there a cross-bleed manifold valve that must be opened so that engine/airframe de-ice/anti-ice is available from either engine (or is each engine "self sufficient" only in respect of engine anti-ice?)

b. Is there an MD82 SOP (or POH advice) that says to increase Rate of Descent to clear icing conditions if a loss of performance is noted in icing conditions?

c. At what stage is the RAT manually deployed? Will the RAT work in heavy icing (or just ice up?)

d.  They didn't complete any checklist. Which checklist would've been appropriate for the scenario?

e.  What was happening to the LH engine?

For further information on the investigation and the contents of this release, please contact Tcnel. Lorllys Ramos Acevedo, Directora, CIAA, Venezuela, +58 (212) 201- 5491.

The information in this advisory has been translated into French by the BEA and can be accessed at the following URL :
www.bea.aero/francais/actualite/actu.htm The information also will be available in Spanish (from the
CIAA) on the Safety Board's website Spanish-language page:


answers on an email to james.smith@iinet.net.au