TUI B738 at Aberdeen on Sep 11th 2021, deviation from flightpath during go around

Last Update: August 18, 2022 / 19:23:22 GMT/Zulu time

Bookmark this article

Photo of TUI Airways G-FDZF, Boeing 737-800 — TUI Airways G-FDZF, Boeing 737-800 (Photo credit: Aircaft @ Gloucestershire Airport By James / Flickr / License: CC by-sa)

Incident Facts

Date of incident
Sep 11, 2021

Classification
Report

Airline
TUI Airways

Flight number
BY-1665

Departure
Palma Mallorca, Spain

Destination
Aberdeen, United Kingdom

Aircraft Registration
G-FDZF

Aircraft Type
Boeing 737-800

ICAO Type Designator
B738

A TUI Airways Boeing 737-800, registration G-FDZF performing flight BY-1665 from Palma Mallorca,SP (Spain) to Aberdeen,SC (UK) with 67 passengers and 6 crew, was on approach to Aberdeen's runway 34 in instrument meteorologic conditions when descending through 5000 feet the air traffic controller advised the crew they might need to abort their approach, a search and rescue helicopter about to depart would receive priority once airborne, in this case they would be required to climb to 3000 feet at runway heading. The aircraft established on the ILS approach to runway 34 and was descending through 2600 feet on single autopilot, flaps at 15 degrees and gear down already, ATC instructed the crew to abort the approach, climb to 3000 feet and turn left heading 270. The crew pressed TOGA, initiated the climb and the left turn. Just before reaching 3000 feet MSL the aircraft began to descend again reaching up to 3100 fpm sink rate, tower controller noticed the aircraft was descending again and notified approach control who contacted the crew just as the crew was pulling the nose up and returned the aircraft into a climb at 286 KIAS (speed selected by the crew was 200 KIAS). The aircraft levelled off at 3000 feet, the crew re-engaged the autopilot. The aircraft positioned for another approach to Aberdeen and landed safely on runway 34.

The AAIB released their special bulletin stating:

The Boeing 737-800 is a dual autopilot, CAT III capable aircraft. Normal procedures, as outlined by the manufacturer, require the use of a single autopilot on an ILS approach unless the intention is to conduct a CAT II or III approach and landing. Automatic go-arounds are only available from a dual autopilot approach. The autopilot/flight director go-around mode is engaged by pressing the Takeoff/Go-around (TO/GA) switches. Pressing either of the switches when the engagement criteria are met will disconnect the single autopilot (if connected) and place the flight directors in go-around mode. The autothrottle (if engaged) will move to go-around thrust, and the flight directors will then command 15º nose-up pitch until the aircraft reaches a programmed rate of climb. Flight director pitch commands then target airspeed for each flap setting, based on a maximum takeoff weight calculation.

The AAIB annotated: "The pilots of G-FDZF, like many other pilots, had not flown for significant periods during the 18 months before this incident. Although the investigation has not established a link between this incident and a lack of recent line flying, it is clearly a possibility."

The investigation is ongoing.

On Aug 18th 2022 the AAIB released their final bulletin concluding the probable cause of the serious incident was:

The crew of G-FDZF were instructed to go-around by ATC. After initially climbing towards the miss approach altitude, the aircraft began to descend. The descent continued for 57 seconds reaching a minimum of 1,565 ft agl before the aircraft was recovered to a climb. A combination of an unexpected large increase in thrust when the go-around was initiated, instructions from ATC to fly a heading, a lack of manual pitch trimming, and the changes in the flap configuration, caused the crew to become overloaded, allowing the aircraft to descend unnoticed for a significant period. Both pilots had experienced significant periods away from flying the aircraft type during the pandemic.

The AAIB analysed:

The aircraft descended from close to 3,000 ft amsl for 57 seconds before a climb was re-established, and this represented a significant deviation from the crew’s expected flightpath. The rate of descent peaked at 3,100 ft/min before the aircraft began to climb having descended to 1,565 ft agl, significantly reducing the aircraft’s separation from terrain. During the descent and subsequent recovery, there was an uncommanded and undesirable increase in airspeed to 286 kt that was not corrected in a timely manner.

Having pressed the TO/GA switches once for the go-around, the crew expected the engaged autothrottle to select power for a climb rate of between 1,000 – 2,000 ft/min. However, the aircraft was above 2,000 ft radio altitude and as a result, unexpected by the crew, the power advanced towards the full Go-Around N1. With the underslung engines, and at an approach speed for the flap selected, this large increase in power meant that the aircraft pitched up significantly and climbed towards the selected altitude of 3,000 ft amsl very rapidly. The autothrottle remained engaged, and as the aircraft approached the level-off it reduced the thrust towards that required to maintain the selected speed in level flight. The reduction in thrust caused the aircraft pitch attitude to reduce, and this was exacerbated by trim changes due to the retraction of the flaps from Flap 15 to Flap 5. The aircraft then began a descent, and since it had not reached the criteria for ALT HOLD, the AFDS remained in ALT ACQ. The retraction of the flaps from Flap 5 to Flap 1, and then from Flap 1 to Flap up during the descent also further decreased the pitch attitude. As the aircraft was descending, the speed increased despite the selected speed remaining at 200 kt.

The crew were assigned several heading changes both before and during the aircraft descent. These instructions placed an additional burden on a crew that was already working hard. The heading instructions had to be acknowledged and actioned by the co-pilot, which could have distracted him from his monitoring tasks. The commander, who was manually flying, had to manoeuvre the aircraft in roll during a very dynamic period in pitch control.

Although the crew seemed unaware of the descent for a significant period, there remained further barriers to a continued descent that might have alerted them to the situation. These were the aircraft’s Terrain Avoidance and Warning System and the ATC radar system alerts. In this instance, the ATC radar system alert that was supposed to warn of an aircraft with a rapid rate of descent failed to recognise that G-FDZF’S descent rate exceeded 2,500 ft/min for a total of approximately nine seconds. This barrier, therefore, did not function as expected. However, the crew became aware of the descent and began to correct it at the same time as the tower controller noticed their descent on the radar repeater in the tower - both the crew and ATC therefore acted to correct the flight path as soon as it was noticed.

The investigation looked at the possibility that the crew were affected by a somatogravic illusion as the aircraft accelerated, but although this could not be completely dismissed, an analysis of the FDM data showed it was unlikely. Any nose-down force on the controls during the initial part of the go-around was most likely due to the aircraft being out of trim, with the large increase in thrust causing a pitch up that the commander countered by pushing forward on the control column. There were no abnormal nose-down forces on the controls during the subsequent acceleration during the descent.

The COVID 19 pandemic led to most pilots flying significantly less than normal. This presented challenges to operators and crews in remaining current and maintaining skill levels to levels equivalent to when the flying intensity was greater. These same challenges applied also to those providing a service to the aircraft, such as ATC. The operator of G-FDZF had a plan for both aspects of the lack of flying. The simulator program was designed not just to maintain crews’ legal currency requirements, but also to allow them to maintain their skills in both the normal and emergency phases of flight. Whilst simulators provide an excellent environment for practising operations, they do have some limitations in reflecting real-world experience.

Two engine go-arounds in day-to-day flight operations are rare. With a go-around rate around three per 1,000 flights in the UK, the average crew from the operator might have expected to experience one a year when flying at the pre-pandemic rate. Regular practise in the simulator is usually conducted from the approach minimums for regulatory compliance, either single engine or with an autopilot-coupled go-around available. Go-arounds from higher altitudes on the approach are less regularly practised. It is unlikely that either crew member had conducted a go-around in the aircraft in the previous two years.

Whilst the go-around should have presented little problem to the experienced crew, the combination of less than average flying in the recent period (and very little flying in the case of the co-pilot), the unexpected large increase in thrust and the changes in heading given by ATC probably combined to overload the crew. Subsequently, they were unable to retain their situation awareness. The changes in thrust generated corresponding changes in the pitch of the aircraft, which together with the pitch changes generated as the crew changed the flap configuration were not dealt with through manually trimming the aircraft. The pitch of the aircraft was not managed effectively by the commander and the aircraft began to descend.