British Airways B744 near London on Jun 9th 2019, unreliable airspeed and stick shaker

Last Update: April 30, 2020 / 16:10:07 GMT/Zulu time

Bookmark this article

Photo of British Airways G-BNLN, Boeing 747-400 — British Airways G-BNLN, Boeing 747-400 (Photo credit: Aero Icarus / Flickr / License: CC by-sa)

Incident Facts

Date of incident
Jun 9, 2019

Classification
Report

Airline
British Airways

Flight number
BA-289

Departure
London Heathrow, United Kingdom

Destination
Phoenix, United States

Aircraft Registration
G-BNLN

Aircraft Type
Boeing 747-400

ICAO Type Designator
B744

A British Airways Boeing 747-400, registration G-BNLN performing flight BA-289 from London Heathrow,EN (UK) to Phoenix,AZ (USA) with 320 passengers and 18 crew, had just reached the top of climb out of Heathrow, autopilot and autothrottle were engaged, when the crew received a Master Warning together with a red overspeed warning on the EICAS, Altitude disagree, IAS disagree, Rudder Ratio indication, airspeed low and altitude alert indications as well as the VNAV path on the Flight Mode Announciator being crossed out. The crew felt ear discomfort from a change in the cabin pressure. The first officer, pilot flying, called unreliable airspeed, noticed the engine thrust had reduced to about 1.2 EPR and invoked the relevant procedures. He set the pitch to 4 degrees nose up and thrust to 80% N1 as required by the drill. The crew declared PAN and were cleared for a block altitude. The captain (left seat) and training captain (observer seat) consulted the Unreliable Airspeed Table in the QRH, determined the required pitch was 3.5 degrees nose up at 87.5% N1. While that data were still being agreed on, the stick shaker activted with pitch at 4 degrees nose up and 80% N1. The crew quickly discussed whether to continue to refer to the airspeed unreliable procedure or invoke the stall procedure and agreed, the stall procedure needed to be executed. The first officer pushed the nose down to 1 degree nose down, the stick shakers ceased reaffirming the belief of the crew, that the stick shaker activation was genuine. The first officer increased pitch slowly, the stick shaker activated again, the pitch was reduced, airspeed increased and the procedure repeated until the QRH pitch and N1 could be maintained without stick shaker activation. The aircraft had lost about 2800 feet during the maneouvers. With the aircraft now stable the crew continued the Unreliable Airspeed checklists, changed the Air Data Source from the right hand to the center Air Data Computer (ADC), which permitted to re-engage autopilot and autothrottle. The creew consulted with maintenance who confirmed the right ADC had failed. After verifying the aircraft was still compliant with required navigaton performance and minimum navigation performance specifications the captain decided to continue the flight to Phoenix where the aircraft landed without further incident.

The UK AAIB released their bulletin reporting the aircraft was operated for 5 more sectors under MEL before the technical problem was fully addressed. The AAIB wrote:

System redundancy enabled continued operation to take place without compromising safety after the right ADC had been isolated and the centre ADC activated to function in its place (in conjunction with restrictions on low visibility operation). A series of rectification actions was proposed, including leak and drain checks on the relevant pitot-static systems, and replacement of the right ADC. These actions were carried out at the conclusion of the five sectors following the event.

The aircraft then flew a further eight sectors, with no subsequent problems encountered. Following these sectors, the aircraft was grounded and permanently removed from service.

The AAIB described the analysis of the right ADC:

The removed right ADC unit was forwarded to the operator’s avionics overhaul base for inspection. Unfortunately, the non-volatile memory incorporating fault codes was mistakenly erased from the unit as the first action. The unit was then subjected to a series of checks which did not reveal any abnormalities. It was subsequently forwarded to the manufacturer where extensive functional tests were carried out. Despite prolonged operation and testing, the unit did not exhibit any incorrect functioning.

The data plate on the relevant ADC indicated that a number of modifications had been incorporated, one of which was not amongst those applicable to this unit type. It was further determined that a non-mandatory modification, applicable to this ADC type and devised to overcome a previously identified periodic malfunction, had not been incorporated.

Stall warning, in the form of stick shaker operation, requires inputs of vane angle and Mach number to compute proximity to the stall and to operate the stick shake when a predetermined threshold is passed. If the malfunction which the non-incorporated modification was designed to address occurred, it would have resulted in an incorrect Mach number being supplied to the stall warning system. This would then cause the stall warning system to operate erroneously at a safe airspeed (ie a speed corresponding with a safe angle of attack).

Assessment of the recorded data suggested that the malfunction may have occurred on the incident flight.

The AAIB analysed:

No evidence was found to account for the initiation of the event sequence, but the sequence of events was consistent with a known fault mode of the model of ADC which was fitted to the aircraft, for which a modification was available but had not been incorporated.

The malfunction of the right ADC was not identified despite extensive functional testing. It is likely that the false warnings had been generated erroneously as a result of an incorrect Mach number being supplied by the right ADC. This would then also have caused the stall warning system to operate erroneously at a safe airspeed.

The identification of the recorded faults within the right ADC unit during the flight was not possible as the fault codes had been deleted after the unit had been received into the operator’s avionics workshop.

The QRH procedure applicable at the time of the incident noted that ‘overspeed warnings and AIRSPEED LOW alerts may occur erroneously or simultaneously’. Stall warnings were not mentioned specifically as the aircraft manufacturer considered that crews would understand this was included. It is apparent this was not however the case with the crew involved who considered they must react to the stall warning when it occurred. The AIRSPEED LOW alert is a specific warning and the crew considered that as the stall warning was not mentioned separately in the procedural note, operation of the stick shaker should not be considered erroneous. This seemed to be confirmed to them when the stick shaker operation ceased when pitch was reduced, as they would expect after a genuine stall warning. This highlights the importance of clear, unambiguous information being readily available to crews at times of high workload when dealing with potentially critical incidents. It also reinforces the need for crews to understand the protection afforded by adopting the pitch and power settings provided as part of the procedure.

The AAIB reported following safety actions were taken, while the investigation is continuing:

The aircraft manufacturer is planning to update the QRH procedure to specifically include stall warnings as part of the note. This update is due to be included in the block revision to the B747-400 FCOM in April 2020. They are also considering similar action with other relevant types.

Since the event, the operator has taken steps to identify the process shortcomings that permitted the loss of the fault codes to occur following arrival of the ADC in their avionics workshop. As a result, procedural changes are being introduced aimed at preventing future loss of troubleshooting and fault data that can assist incident investigations.