Photo: Boeing 737-73V, from Konstantin von Wedelstaedt, 9 Mar 2002 (Creative Commons).
The basic problem with the airplane was a faulty adjustment on the elevator balance tabs that caused the airplane to pitch nose down when flown without hydraulic power. More about this: Basic Aerodynamics / Balance Tabs. Since the airplane is almost never flown under "manual reversion," crews may have never known about the adjustment problem except for the fact the airplane's lease had expired and a test of this condition was required by the lease agreement.
The real problem demonstrated by this incident is that aircraft manufacturers are reluctant to share their test procedures, understandably, because they don't have control over the companies and pilots attempting to use them. That leaves these companies to come up with test procedures on their own. Failing that, it can also force pilots to "wing it" and invent their own. In this case, the company had their own test procedure to test the airplane under manual reversion; but there was a mistake. The pilot did a great job under the circumstances, but was not suitably trained to deal with the situation.
If you are ever called upon to fly a functional check flight, study the actions of those who came before you. If your procedure calls for doing something you wouldn't do on a normal revenue flight, be very careful. More about this: Normal Procedures & Techniques / Functional Check Flights.
What follows are quotes from the relevant regulatory documents, listed below, as well as my comments in blue.
- Date: 12 Jan 2009
- Time: 1545
- Type: Boeing 737-73V
- Operator: easyJet
- Registration: G-EZJK
- Fatalities: 0 of crew, 0 of passengers
- Aircraft Fate:
- Phase: En route
- Airports: (Departure) Southend (SEB/EGMC); (Destination) London Stansted (STN/EGSS)
[AAIB, G-EZJK, pg. 1]
- The aircraft had reached the end of its lease agreement with its current operator and required a combined maintenance check and demonstration flight to confirm its serviceability before being transferred to another airline. The checks to be carried out, which were agreed between the operator and the owner, were detailed in a Customer Demonstration Flight Schedule (CDFS). The commander involved in the incident had flown the aircraft to Southend Airport for maintenance the previous month and during that flight carried out checks, in accordance with the agreed CDFS, to identify any existing defects.
- The commander returned to Southend on 12 January 2009 to conduct the post-maintenance check flight and the customer demonstration flight. Before the flight he discussed with the crew chief from the maintenance provider the work that had been carried out. He recalled being told that an adjustment had been made to the elevator balance tab setting and was given extracts from the Aircraft Maintenance Manual (AMM) to assist him in conducting an in-flight elevator power-off test and to identify any asymmetrical flight control forces; both were required as part of the maintenance procedures. prior to departure he checked the aircraft’s technical log and confirmed that arrangements had been made with ATC for the flight to be conducted in the east Anglia Military Training Area (MTA). The commander and co-pilot, a first officer from the operator, were accompanied on the flight by two observers representing the aircraft owner and the airline due to take delivery of the aircraft. no problems were identified during the pre-flight preparation and the aircraft departed at 1400 hrs with the commander as the handling pilot.
- The commander climbed the aircraft to Fl410, conducted a series of checks and, after about 45 minutes, descended to Fl150 where an Apu bleed check was performed and the aircraft was configured for the flight control manual reversion check. The aircraft was flown at Fl150, 250 kt3, with the fuel balanced, the autopilot and autopilot selected off, the STAB TRIM MAIN ELEC and AUTOPILOT TRIM switches set to the CUTOUT position, and the aircraft in trim. The CDFS also required Spoiler A and B switches to be selected off. All these checks were conducted using the operator’s CDFS and not the AMM extracts as the guiding reference.
- Before the manual reversion check began, the individual hydraulic systems were isolated in turn by placing the FLT CONTROL switches A and B to the OFF position individually and reinstating each in turn enabling the flight controls to be checked for normal operation on a single hydraulic system. Operation was confirmed as satisfactory on both systems. The commander then released the controls and the co-pilot selected both FLT Control switches (A and B) to the off position, removing all hydraulic assistance from the primary flight controls. As he did so the aircraft pitched rapidly nose-down. The commander pulled back on the control column with considerable force but was unable to prevent the aircraft from maintaining a nose-down pitch attitude of 2.8° and descending at up to 3,100 ft/min. The commander decided to abandon the check and reinstate the hydraulics. However, he did not wish to re-engage them immediately as he stated that he had been trained that, should the aircraft pitch up or down uncontrollably during a manual reversion check, the aircraft should be rolled to unload the pressure on the elevators and the controls released before the hydraulics are reinstated. It was his understanding that not releasing the controls prior to reinstating the hydraulics could overstress the airframe or cause serious injury to the handling pilot. He therefore rolled the aircraft left to 70° before releasing the controls and calling for the co-pilot to re-engage the flight control switches. The aircraft continued to roll to 91°.
- The recording from the Cockpit Voice recording indicates that at this point there was confusion between the two pilots. The commander believed that hydraulic power had been restored to the flight controls although there is no evidence that the FLT Control switches had been moved from the OFF position. The commander retarded the thrust levers and selected the speed brakes but the spoilers had been selected off as part of the test procedure and the speed brakes, therefore, did not deploy. He then rolled the wings level and attempted to arrest the rate of descent. This had peaked at 20,000 ft/min with the aircraft pitched 30° nose-down after the aircraft had been rolled to the left. The control forces remained high but the commander considered this to be due to the aircraft’s speed, which he observed at a maximum of 447 kt.
- The commander continued to maintain backpressure on the controls and made a PAN call to ATC. The aircraft eventually recovered from the dive at about 5,600 ft amsl having entered a layer of cloud. The pilots reviewed the situation and selected the FLT CONTROL switches, which had remained OFF throughout the flight excursion, to the ON position. The control forces returned to normal.
- The commander stated he had considered repeating the test, but was concerned that, as a result of the incident, the aircraft might have sustained damage. The check flight was abandoned and the aircraft returned to Southend. Considering possible structural damage, the commander kept the speed below 250 kt and configured the aircraft for landing early during the approach. The aircraft appeared to operate normally and landed without further incident at 1606 hrs.
- The aircraft was inspected after landing for damage or deformation in accordance with AMM task 05-51-04 titled ‘severe or unusual turbulence, stall, or speeds more than design limits – maintenance practices (conditional inspection)’. No evidence of damage or deformation of the structure was found.
Figure: G-EZJK radar track with extracts from CFR, from AAIB, G-EZJK, figure 1.
[AAIB, G-EZJK, pg. 3]
- The graphical presentation starts just after the aircraft was trimmed, at FL150, at a computed airspeed of 245 kt, the STAB TRIM MAIN ELECT and AUTO PILOT TRIM switches were selected to their CUTOUT positions, and Spoiler A and B switches were selected to OFF In this trimmed condition (with zero stick force) the elevator position was about 5° trailing edge down.
- The FLT CONTROL B switch was then put in the OFF position and the flight controls were moved slightly. The switch was then put back to the ON position and the FLT CONTROL A switch was then put in the OFF position and the flight controls again moved slightly. The co-pilot sought confirmation from the commander that he was prepared before each selection and verbally confirmed each action.
- With the FLT CONTROL A switch still in the OFF position, the co-pilot asked the commander if he was ready for the FLT CONTROL B switch to be selected OFF (putting the aircraft in manual reversion). The commander replied “YES GO AHEAD”. The hydraulics to the B system were turned off (time 15:36:47). The elevator position rapidly moved to just over 8° trailing edge down (an increase in 3° trailing edge down from the trimmed position), and the aircraft pitched from 2° nose up to 2° nose-down. The commander immediately pulled back on the control column, reaching full aft stick five seconds later. This was enough to return the elevators to their trimmed position. He was, however, unable to maintain this control column position and the aircraft remained in a nose-down attitude. The amount of column movement and force required with the hydraulics off increases to produce the same elevator deflection with the hydraulics on. The column force being pulled throughout this time was in excess of 170 pounds-force.
- Ten seconds after system B had been selected off (time 15:36:57), the aircraft had descended almost 300 ft and was continuing to descend at a rate of about 2,200 ft/min. The airspeed had increased by 10 kt to 255 kt and was still accelerating. The commander then rolled the aircraft to the left at a little under 8°/second and, as the aircraft passed through 50° bank angle (time 15:37:04) said “GET READY TO PUT IT BACK”. The co-pilot responded with “HEY” and “SAY AGAIN” to which the commander said “AND BACK”. Then, as the bank angle passed through 70°, the commander released the pressure on the control column, allowing the elevator to move back to just under 8° trailing edge down. The descent rate was now 6,000 ft/ min and increasing, and the aircraft had descended a further 400 ft to FL143 while accelerating to 270 kt.
- The commander pulled back on the control column, reducing the elevator deflection a little and, as the bank angle approached 91° (maximum recorded), put in a wheel input to roll the wings level. The engine thrust was also reduced and the speed brakes selected but the speedbrakes did not extend as SPOILER A and B switches were still in the OFF position.
- The aircraft’s descent rate increased considerably, reaching a maximum of 20,000 ft/min as it descended through about FL110. At this point the bank angle was reducing through 40°, the pitch attitude was 30° nose-down, and the airspeed was 320 kt (Mach 0.60) but still accelerating. As the aircraft’s descent rate started to reduce, the commander made a PAN call (time 15:37:20). This call coincided with the sounding of the aural overspeed warning which remained active for the next 48 seconds.
- The maximum recorded airspeed during the recovery was 429 kt (Mach 0.719). The maximum recorded vertical acceleration was 1.6 g and the minimum recorded altitude was 5,655 ft amsl. The Mach trim was in operation above Mach 0.615 making pitch-up commands to the elevator in addition to deflections demanded by the commander’s control column inputs.
- There was no recorded discussion between anyone on the flight deck during the event. The first comment was recorded 76 seconds after the commander called “AND BACK” and shortly after the aircraft had levelled at 7,000 ft amsl. The commander then said “ARE THEY ALL BACK ON – PUT ALL THE [unintelligible] CONTROLS BACK ON”. Both flight control system A and B hydraulics were then reinstated (time 15:38:27). A transmission was made by the commander 15 seconds later cancelling the PAN.
[AAIB, G-EZJK, pg. 15]
- The commander of the flight stated that he found the AMM to be poorly constructed and difficult to follow. He also believed that the operator’s own check schedule (CDFS) encompassed the requirements of the check and therefore chose not to use the AMM, either before or during the flight check.
- Comparison of the AMM with the operator’s CDFS identified a number of differences.
- The most significant related to the AMM procedure requiring the selection of the cockpit switches for hydraulic systems A and B to be placed directly to the STANDBY RUDDER position during the test. This removes hydraulic power to the flying controls, except the rudder as it requires control forces that are too great for manual control.
- In contrast, the operator’s CDFS called for the switches to be selected to the OFF position, as part of an additional, unrelated test. The CDFS did not then require them to be selected to the STANDBY RUDDER position prior to conducting the manual reversion test, thereby rendering the rudder inoperable by the pilots during the incident.
- In addition, the AMM procedure called only for the autopilot stabiliser trim cutout switch to be selected to the CUTOUT position, whereas the operator’s CFDS called for the main autopilot switch to be selected to the CUTOUT position as well. Operation of the stabiliser trim autopilot function during the test would interfere with the manual wheel crank procedure. The main trim, however, should be available to assist recovery during an upset.
- The aircraft was ferried to Southend for its end-of-lease maintenance on 1 December 2008. During this flight the crew used their CDFS to carry out checks to identify any previously unrecorded defects and allow rectification work to be planned into the forthcoming maintenance input. One of the checks was a manual reversion test. During the test the commander identified that 11.5 turns of nose-up trim were required to trim the aircraft for level flight in the test configuration.
Because the limit was 12, the pilot chose not to formally enter the discrepancy in the maintenance log but sought to have it corrected. Several work orders were transcribed outside the normal system and the pilot's "nose-down" write up was eventually written incorrectly as "nose-up" and the balance tab was adjusted incorrectly.
[AAIB, G-EZJK, pg. 27] During the incident flight, the elevator response was always normal when the hydraulics were selected on. The control force provided by the hydraulic system was easily sufficient to overcome the aerodynamic force generated by the elevator tab, giving a level flight position of the elevator of 5° trailing edge down. However, when both hydraulic systems were selected off, the control force applied to the control surface was reduced to that provided by the pilot on the control column. This was insufficient to resist the aerodynamic load caused by the incorrectly rigged balance tab, which subsequently moved the elevator to a zero hinge moment position of 7° to 8° trailing edge down, creating a nose-down pitching moment on the aircraft. Consequently, the aircraft settled at a constant -2.8° pitch attitude with a corresponding increasing airspeed, despite the commander applying as much back column force as he was able.
There was no formal "cause" section, but safety actions were listed.
[AAIB, G-EZJK, pg. 32] The following safety action has been taken since this incident occurred:
- Boeing has amended the wording of the flight test task in the latest revision of the AMM designed to improve ease of use and reduce the likelihood of incorrect interpretation.
- Boeing has issued Service letter 737-Sl-27-211 which provides further advice on rigging the elevator tabs and conducting post-adjustment check flights.
- The U.K. CAA has published Airworthiness Communication (AIRCOM) 2009/03 to raise awareness of the issues relating to the coordination between operators and maintenance organisations surrounding the conduct of maintenance check flights. it has also issued Flight Operations Division Communication (FODCOM) 15/2009 regarding the definition, preparation and conduct of check flights. it also advises on crew qualification requirements and the need for coordination with relevant maintenance organisations to ensure information is formally documented and distributed.
- The operator has carried out an internal investigation into the incident. This identified the causal and contributory factors discussed in this report and made 38 safety recommendations.
- The UK CAA has re-written Section 3, Tech 2, Part 10 of the CAA Check Flight Handbook to ensure its previous advice in dealing with a pitch down incident is not misinterpreted.
- MRO(B) conducted an internal investigation which addressed the formalisation of the customer work request procedure and introduced a procedure to improve flightcrew/maintenance interface.
Basic Aerodynamics / Balance Tabs
Normal Procedures & Techniques / Functional Check Flights
Air Accidents Investigation Branch, Serious Incident, Boeing 737-73V G-EZJK, 9/2010.