There is much stupidity in this accident. The crew used an inappropriate autopilot mode in the climb, made a hasty (and wrong) decision to shut down an engine, and then made a poor decision to continue their flight across the Atlantic. The fact nobody was hurt (except the airplane) is due to the robustness of the airplane, despite the ineptness of the crew.
— James Albright
Aeromexico 945, View of left outboard elevator tip,
NTSB AAR-80-10, figure 1.
I am not going into a lot of detail here, if you would like that please read the NTSB report, which is quite good. I am more interested in the lessons learned as far as autopilot usage.
- Date: November 11, 1979
- Time: 2138
- Type: McDonnell-Douglas DC-10-30
- Registration: XA-DUH
- Fatalities: of crew, 0 of 311 passengers
- Aircraft fate: Damaged
- Phase: Climb
- Airport (departure): Frankfurt, Germany
- Airport (arrival): Mexico City, Mexico
I cannot help but read this accident report and wonder about the passengers who experienced all this and then had to sit in the cabin of this airplane heading across the Atlantic Ocean. But there is even more for us to contemplate. First, if your autopilot defaults to holding a vertical speed, you ought to be extremely paranoid about having a vertical mode selected. Most of my autopilots will hold a pitch in the absence of a selected vertical mode and that can be just as deadly. So let's revise this: be paranoid about your autopilot's vertical mode!
What else can we be paranoid about? How about airspeed? If you have three perfectly good engines your first reaction to airframe vibration should not be the engines!
But let's leave paranoia aside and talk about arrogance. How can you possibly make the judgment that the airplane remains airworthy after losing control of it and losing 10,000 feet?
- While climbing through 14,000 feet m.s.l., the AP became disengaged. The captain reengaged it by positioning the No. 1 AP lever to the CMD position and then pressed the inertial navigation system (INS) selector in the heading (HDG) panel. No other FD/AP modes were reselected. The pilot commented to his crew that if the AP disengaged again, they would write it up in the aircraft logbook.
Source: NTSB AAR-80-10, §1.1
Reading the analysis section of the report (see below) it seems that with this autopilot selecting the autopilot without a vertical mode places it in the default mode of maintaining a vertical speed.
- According to the crew, while climbing through 27,500 ft about 100 miles west of the departure airport, they felt a vibration which, within seconds, increased in intensity. The crew suspected an abnormal vibration in engine No. 3 and elected to reduce its power and then to shut it down. The crew also stated that, upon reducing power on engine No. 3, the aircraft assumed a pitch down attitude, the AP became disengaged, and the aircraft rolled to the right and then to the left and started to lose altitude.
Source: NTSB AAR-80-10, §1.1
As the remainder of the flight and post-flight maintenance would reveal, there was nothing wrong with the engine. The vibration was caused by the wings approaching the stall buffet. What about stall warning? The report says the stick shaker was going off but the crew was too distracted to notice it. The loss of thrust only aggravated the stall.
- The Digital Flight Data Recorder (DFDR) revealed that, after the No. 3 engine power was reduced, the aircraft decelerated into speeds that were below the stall buffet speed and the design flight envelope. Shortly thereafter, the nose dropped and the aircraft entered into a stall while at 29,800 ft and an IAS of 226 kns.
- The captain said that as the aircraft nose dropped, the spoilers were deployed to arrest the impending overspeed condition that could have been created by the aircraft's nose low attitude. About 10 seconds later, the autoslats extend system became active. The DFDR readout showed the recovery started at 23,900 ft. At that time, the airspeed increased to a value above the calculated stall speed. The vertical acceleration reached a maximum of 1.68 g's during the recovery process which ended at an altitude of 18,900 ft, and the crew regained full control of the aircraft about 18,000 ft.
- Shortly after recovering control of the aircraft, the crew airstarted engine No. 3. It appeared to be functioning normally with all parameters indicating within normal limits with no indication of vibration.
- The captain stated that since all systems appeared to be functioning normally and since a landing at Madrid would require dumping 140,000 pounds of fuel he elected to continue the flight to Miami, Florida. According to French ATC, shortly after issuing clearance to Madrid, the flight requested and received clearance to proceed to Miami.
- The flight climbed in visual meteorological conditions to 28,000 ft and then to 31,000 ft, ending at 33,000 ft near Bermuda while cruising at Mach .82 (385 kns) in an ambient temperature of international standard atmosphere (ISA) +9° C (static air temperature -41°C) The flight landed at Miami, Florida, on November 12, 1979, at 0705 without any further problems.
- Upon arriving at the passenger terminal gate, the captain requested that maintenance personnel give the aircraft a visual exterior inspection. It was found that about 4 ft of each outboard elevator tip, including the corresponding counterweights and the aircraft's tail area lower access door were missing.
- The aircraft was grounded at Miami, Florida, where it underwent a detailed inspection, a thorough examination, and a functional test of all flight control systems, ATS, AP/FD, engine No. 3, and all other related systems that could have induced the condition experienced by the crew during the incident. No discrepancies were noted.
Source: NTSB AAR-80-10, §1.1
The analysis speaks for itself, though I must say the NTSB was overly charitable to the crew.
- The actions described by the crew regarding the ATS and AP selections are not compatible with the system design. The system design is such that airspeed can be controlled by the ATS through modulation of thrust level while the pitch attitude of the aircraft is controlled by criteria other than airspeed; or airspeed can be controlled by the AP through variation in pitch attitude while thrust is maintained at a constant level or is controlled to a maximum limit. The system design will not permit the simultaneous selection of airspeed control on both the ATS and AP. Thus, the crew's recollections of the ATS and AP selections must have been incorrect and imply that they were not completely knowledgeable in the use of the DC-10 flight guidance and control systems.
- For an initial climb to cruising altitude, normal autoflight procedures would be to select the N mode of autothrottle operation and the IAS (or Mach) Hold mode for autopilot operation. With these selections, the engine thrust would be continually modulated to the maximum allowable (continuous) level as determined by the thrust computer. The pitch attitude of the airplane would vary to maintain the AP selected airspeed. The aircraft's vertical speed would also vary during the climb as the engine thrust decreases with the changing ambient environment. The vertical speed would begin high and decrease as altitude is gained.
- The DFDR data, however, do not substantiate this type of climb profile. Rather, the data show that the aircraft, as it climbed through 25,000 ft, was maintaining a nearly constant rate of climb of about 1,200 ft per minute at an IAS of 318 kn. During the subsequent 4 minutes, the rate of climb was a constant 1,200 ft per minute while the airspeed decreased to 226 kn and the pitch attitude increased from 4.5 to 11 nose up. This performance is most consistent with that which would be produced with the ATS engaged in the airspeed mode and the AP engaged in the vertical speed mode. With these selections, a constant vertical speed would be maintained by AP pitch attitude control and a constant airspeed would be maintained by engine thrust modulation. This is contingent however, on the relationship between thrust required and thrust available. As the climb progresses, the aircraft will reach an altitude where the ATS system would be commanding the maximum continuous thrust level. Beyond that altitude, the aircraft would be unable to maintain both the AP selected vertical speed and the ATS selected airspeed because of a thrust deficiency. The AP however, would continue to command the increasing pitch attitude necessary to achieve the selected vertical speed, regardless of the aircraft's airspeed or angle of attack. There are no angle of attack limits in the AP circuitry to prevent the aircraft under these circumstances from entering a stall.
- The Safety Board thus concludes that the crew erred in both their actions and recollections regarding the AP mode selection. It is probable that the flightcrew did begin, or intended to begin, the climb with the ATS N mode/AP IAS mode selections. However, when the captain selected 320 kn into the ATS speed window he may have either intentionally or unintentionally pulled the ATS speed selector knob. This action would have changed the ATS selection from the N1 mode to the airspeed mode. This in turn would have caused the AP IAS Hold mode to disengage and revert automatically to the vertical speed mode of operation.
- Although the crew failed to recognize the approach and entry to the stall, they did, after approximately 1 minute, recognize the aircraft's stalled condition and responded with proper control to recover. A full minute for stall recognition is excessive. However, the DC-10's stall warning system consists only of a stickshaker, the operation of which might be misinterpreted by an inattentive or distracted flightcrew, particularly when the aircraft is controlled by the autopilot rather than a pilot.
- The Safety Board views with concern the decision of the flightcrew to continue on to their scheduled destination after the incident occurred. The violent, as well as the unexpected, nature of the incident and the flightcrew's initial lack of understanding of the reason for the occurrence should have been sufficient reason to terminate the flight and land as soon as practicable. Therefore, the Safety Board believes that a more prudent judgment would have been to land and assess the reason for the loss of control as well as possible damage to the aircraft.
Source: NTSB AAR-80-10, §2
I would have rewritten the cause statement as follows: The airline failed to select and train their pilots to fly their aircraft with an appropriate attention to detail, or with the necessary judgment needed to place the lives of the paying public in their charges.
The National Transportation Safety Board determines that the probable cause of this incident was the failure of the flightcrew to follow standard climb procedures and to adequately monitor the aircraft's flight instruments. This resulted in the aircraft entering into a prolonged stall buffet which placed the aircraft outside the design envelope.
Source: NTSB AAR-80-10, §3.2
NTSB Aircraft Accident Report, AAR-80-10, Aeromexico DC-10-30, XA-DUH, Over Luxembourg, Europe, November 11, 1979