the learning never stops!

Pan American World Airways 799

Accident Case Study

Taken in isolation, this accident is simply the case of a crew's poor checklist discipline and, perhaps, stick and rudder skills. But there is much more to it than that.

This was another incident in a series of 13, 11 of which pointed to a problem with the Crew Resource Management Culture at Pan American World Airways at the time. They were able to reverse this culture and became one of the safest airlines in the world.


images

Photo: Pan Am Boeing 707-321C N799PA, (John Smith)

Click photo for a larger image

Accident Report

  • Date: 26 December 1968
  • Time: 06:15
  • Type: Boeing 707-321C
  • Operator: Pan American World Airways
  • Registration: N799PA
  • Fatalities: 3 of 3 crew, 0 of 0 passengers
  • Aircraft fate: Damaged beyond repair
  • Phase: Takeoff
  • Airport (departure): Anchorage-Elmendorf AFB, AK (PAED), USA
  • Airport (arrival): Tokyo, Japan

Narrative

When flying oceanic, especially back then, departure timing was critical and these kinds of discussions in the cockpits were common. How late can we go and still have enough fuel? How low can we fly? At what point do we turn around for more fuel? When a captain manages the cockpit, somebody worries about these details but somebody also has to worry about flying the airplane. As a copilot (first officer) I would do my best to keep the big picture in mind and help the captain with these decisions where I could. But I always remained "plugged in" to the airplane, keeping the aircraft airworthy was always my top concern. You may argue that this had to be true with the first and second officers on this airplane too. It could be. But we see lots of evidence in other Pan American Boeing 707 accidents of the time that at least some of their non-captain pilots became passengers in the cockpit. The price for angering the captain was just too high, so you could be forgiving for "unplugging" from some of your aircraft duties.

[NTSB AAR-69-06], ¶1.1]

  • Flight 799 experienced several delays prior to departure from Elmendorf AFB. Initially, a discrepancy in the computation of mixed fuel density resulted in a requirement for additional fuel. Also, some difficulty was experienced in getting the jet starter unit to provide power for the engine start. Finally, at approximately 0555, the engines were started and the flight departed the ramp at approximately 0602.
  • Flight 799 had its void time extended six times for various reasons. The final void time, issued by the Oceanic Control Coordinator, was 0615. In this instance, the clearance void time was required to prevent a conflict between Northwest Airlines Flight 901, cruising at Flight Level 310, and Flight 799. Flight 799 had also requested FL310 until fuel burn off would permit an en route climb to FL35O. The controller stated that if Flight 799 had failed to make the final void time of 0615, they would have had to delay approximately 45 minutes. The only alternative was to accept a lower cruising altitude which would have resulted in excessive fuel consumption.
  • Although clearance to Runway 05 was issued initially, the flight requested use of Runway 23 because of the greater effective ·runway length. A "follow me" truck was used because the crew was not familiar with the airport and a portion of the lights on one of the taxiways was out. The cockpit voice recorder (CVR) revealed that, when the Elmendorf Tower controller offered to send out the "follow me" truck, the crew was going over the taxi portion of the cockpit checklist. One of the items called out during the reading of this checklist was "wing flaps." Additionally, the CVR revealed that approximately the time the "follow me" truck arrived, a discussion took place between·the captain and first officer regarding the flaps. The captain advised that he had raised the flaps. The first officer then remarked, "Oh, okay, let's not forget them." As the taxiing continued, to Runway 23, the first officer continued talking with the Oceanic Control Coordinator about extending the void time for their previously issued clearance; the flight engineer was computing burn off of fuel to determine how quickly they could climb to FL350; and the captain was absorbed with controlling the aircraft on the slippery taxiways and coordinating the efforts of the crew.
  • Flight 799 arrived at the takeoff end of Runway 23 at approximately 0610 and held, awaiting their turn in sequence. During the next few minutes, MAC 172 landed and MAC 651 departed on Runway 05. Flight 799 was then cleared ·for, " ... right turn on the east-west runway and 180 at the end for a departure to the west; taxi into position and hold." As the aircraft was positioned for takeoff, coordination between the pilots and the Oceanic Control Coordinator was still being accomplished to determine the latest possible departure time which would not conflict with other traffic. This was finally established at 0615, and at 0614:30, the flight was cleared for takeoff.
  • The crew based their takeoff speed computations on an aircraft flap configuration of 14°. Accordingly, the speeds appearing on the crew takeoff information sheet were as follows:
  • V1 148 knots, VR 154 knots, V2 168 knots, Engine pressure ratio (EPR) readings were 1.78 static and 1.82 rolling takeoff

  • The CVR indicated that during the period of time when the flight was awaiting takeoff clearance, the flight engineer challenged the pilots several times, "Gyro compass." They finally acknowledged that this had been checked. It was the last item on the pre-takeoff portion of the cockpit checklist. The captain then told the first officer, "Okay, you got it," and takeoff power was applied. Callouts were made by the captain for air speeds at 120 knots, V1, and VR as the attempted takeoff progressed. Shortly after VR, a noise identified as the stick shaker was heard on the CVR record. This noise continued throughout the rest of the recording. There were also numerous popping noises heard shortly after the stick shaker noise commenced. The recording ended approximately 59.2 seconds after the first officer called for takeoff power.
  • Statements were obtained from 41 witnesses who encircled Runway 5-23. However, the majority of the witnesses were located at the southwest end of the runway and in the vicinity of the operations building, which was located approximately 1 mile from the initial impact point. These witnesses indicated that the aircraft had an unusually long takeoff roll prior to becoming airborne. Several also observed what they described as a settling following lift-off. The rate of climb was described as slow, and estimates of the maximum altitude reached ranged from 10 to 20 feet to 150 to 200 feet. Three persons observed flames from the left engines; three saw flames but could not associate their observation with a specific side of the aircraft; and 16 saw flames emanating from the right engines. All described these flames as occurring while the aircraft was airborne and maneuvering in various combinations of noseup and wing-down attitudes. The consensus was that the initial impact was made in a steep right bank, with the nose low, and that a large ground fire broke out immediately.
  • The right wingtip of the aircraft first contacted the ground at a point just to the left of the extended centerline of Runway 23 at an elevation of approximately 207 feet m.s.1. (61° 16'N. latitude - 149°50'W. longitude). The accident occurred at nighttime at approximately 0615.

[NTSB AAR-69-06], ¶1.12]

  • The landing flaps were in the retracted position at the time of breakup. This was established by the fact that eight of the ten flap drivescrews were found with extensions equal to, or near, the extension which would be obtained if the flaps were fully retracted. Also, the jackscrew in the left-hand outboard aileron lockout mechanism was found in a position consistent with a fully locked-out aileron. This mechanism is designed to fully lockout the outboard aileron at a o0 wing flap setting.
  • Examination of the four aircraft engines revealed no evidence that would indicate there was an overtemperature on the hot section parts. In addition, there was no evidence suggestive of any abnormalities within the powerplants or their accessories, other than those attributed to impact.

[NTSB AAR-69-06], ¶1.12]

  • Flight 799, like other B-707-321C aircraft, had a takeoff warning system that was intended to provide an audible warning signal (horn) when the thrust levers were advanced (through the 42° position of thrust advancement) if flaps, speed brakes, or the stabilizer were not positioned properly for takeoff. No such warning was heard on the CVR tape.
  • Flight 799, like other B-707-321C aircraft, had a takeoff warning system that was intended to provide an audible warning signal (horn) when the thrust levers were advanced (through the 42° position of thrust advancement) if flaps, speed brakes, or the stabilizer were not positioned properly for takeoff. No such warning was heard on the CVR tape.

Analysis

As soon as I read this I pulled out my current airplane's checklist to confirm setting the flaps for takeoff only occurs once for us too, during the taxi check. You would be right to point out that modern aircraft are better at telling pilots things are left undone prior to takeoff. The accident report noted that some airlines of the time had mechanical checklists with slides that reminded them when things got skipped. All that is true. But it is still a fundamental duty on the flight deck to follow these critical checklists with some degree of precision. The Boeing 707 I flew had similar weaknesses when it came to failing to warn pilots about things undone. But we became paranoid about them as a result.

[NTSB AAR-69-06], ¶2.1]

  • An examination of a copy of the cockpit checklist used by the Pan American crew indicated that the takeoff flap item appears only on the taxi portion of the checklist. There is no provision on the pre-takeoff portion of the cockpit checklist to remind pilots that the flaps should be lowered. The Safety Board believes that the placing of a flap reminder item further down the checklist, for example, on the takeoff portion of the checklist, would be most helpful to the pilot. In addition, the use of a slide cover type checklist as used by some other air carriers would enable the pilot to see at a glance which items have not been accomplished. This type of checklist is one means of assisting the crew to accomplish the "passed over" items just prior to takeoff, in those situations where the checklist item is not accomplished on the first reading of the list, or the action taken is subsequently altered, as occurred in this accident.
  • In the case of Flight 799, the first officer apparently lowered the flaps to the takeoff position (14°) during the initial reading of the taxi portion of the cockpit checklist. However, unknown to the first officer, the captain retracted the flaps and the first officer was not aware of this action until the flap item was mentioned during a second reading of the taxi portion of the cockpit checklist. The captain had apparently retracted the flaps in compliance with the company cold weather operating procedures. The flaps remained in the retracted position during a second reading of the taxi portion of the checklist, in spite of a reminder by the first officer not to forget to lower them. Thus, since the taxi portion of the checklist is the only portion of the checklist prior to taking off that contains a reference to wing flaps, any reading of further portions of the checklist would not have alerted the crew as to the position of the flaps.
  • During the period of time that the crew were going over the cockpit checklist in preparation for the flight, they were busily engaged in taxiing the aircraft and in handling numerous radio communications with various air traffic control facilities concerning their departure. These factors, along with probable apprehension due to operating from an unfamiliar airport during nighttime, undoubtedly consumed much of the crew's attention. In addition, the crew's desire to comply with the various void times that were issued by ATC, based upon the traffic within the system, and their own estimates of their capability of meeting these void times, must have caused a considerable amount of mental stress. The CVR transcript clearly reflects the tension in the cockpit and the over emphasis on expediency by all concerned in an effort to fit this aircraft in with other aircraft in the Elmendorf area.
  • Takeoff warning system was designed to prevent. In all probability, the crew did not detect the up-flap condition as they continued their takeoff roll and, subsequently, attempted to rotate the aircraft at the precomputed 14° flap speed of 154 KIAS.
  • Boeing Company indicated that the aircraft left the ground with some margin above 1 g stalling speeds and also above Vmu speeds, but very close to, or below, stick shaker speed. Analysis of the heading changes for the period immediately preceding lift-off until the right wing made initial contact with the ground, revealed that the aircraft was experiencing a progressively increasing lateral oscillation. Calculations made by the Board indicate the aircraft was in an approximate 90° right bank at the time of initial contact with the ground.
  • The rapid changes in aircraft attitude near the stall caused momentary compressor stall(s) on one or more of the engines. Testimony of ground witnesses as to the presence of flames in the vicinity of the engines immediately after lift-off indicates that this occurred.

Cause

I would add another cause to this list: a company-wide culture of poor cockpit discipline and a climate where challenging the captain was received poorly. I've read in a few Pan American pilot biographies where some captains would tell their new first officers to put away their checklists. "Forget that damn checklist, son," the captain would say. "You're not flying for Pan American today. You're flying for me." (Gandt, p. 118)

[NTSB AAR-69-06], ¶2.2(b)] The Board determines that the probable cause of this accident was an attempted takeoff with the flaps in a retracted position. This resulted from a combination of factors: (a) inadequate cockpit checklist and procedures; (b) a warning system inadequacy associated with cold weather operations; (c) ineffective control practices regarding manufacturer's Service Bulletins; and (d) stresses imposed upon the crew by their attempts to meet an air traffic control deadline.

See Also:

Crew Resource Management Culture

References

Gandt, Robert, Skygods: The Fall of Pan Am, 2012, Wm. Morrow Company, Inc., New York

NTSB Aircraft Accident Report, AAR-69-08, Pan American World Airways, Inc., Boeing 707-321C, N799PA, Elmendorf Air Force Base Anchorage, Alaska, December 26, 1968

Revision: 20180201
Top