West Atlantic ATP at Jersey on Aug 18th 2020, difficulties with roll control

Last Update: April 19, 2021 / 10:30:52 GMT/Zulu time

Bookmark this article

Photo of West Atlantic SE-MAO, BRITISH AEROSPACE ATP — West Atlantic SE-MAO, BRITISH AEROSPACE ATP (Photo credit: Alec BHX/KKC / Flickr / License: CC by-sa)

Incident Facts

Date of incident
Aug 18, 2020

Classification
Report

Airline
West Atlantic

Flight number
PT-424

Departure
Jersey, United Kingdom

Destination
Guernsey, United Kingdom

Aircraft Registration
SE-MAO

Aircraft Type
BRITISH AEROSPACE ATP

ICAO Type Designator
ATP

A West Atlantic British Aerospace ATP, registration SE-MAO performing flight PT-424 from Jersey,CI to Guernsey,CI (UK) with 2 crew, was enroute at 2000 feet MSL with the autopilot engaged at 200 KIAS when ATC instructed a right turn onto heading 340, when a jolt occurred followed by a left roll which was corrected by the autopilot. After a second jolt the autopilot disconnected, the first officer, pilot flying, took manual control of the flight but found it difficult because the aircraft was pushing left. According to the CVR she commented: "“i’m just giving everything i have to just hold it steady", however, she did not appear strained. The captain requested delay vectors to sort a minor issue. The captain took temporary control but experienced the same issue needing a lot of right input though the aircraft remained fully controllable. The first officer took control again, the captain verified that all other indications were normal and the roll trim was neutral, then requested only left turns on the ILS runway 27 at Guernsey reporting a slight control problem, nothing major. The captain operated the aileron trim, the first officer commented she thought that was feeling better, the captain performed an approach brief and mentioned the ramifications of abnormal control forces during a go around. The aircraft descended out of cloud at 1500 feet MSL, became fully established on the localizer when the flight directors failed, the first officer continued flying the ILS using raw data and the aircraft landed safely on runway 27 after a total flight time of 14 minutes. After engine shut down the first officer commented that the flight controls felt normal.

On Apr 19th 2021 the UK AAIB released their bulletin concluding:

As the aircraft levelled-off at 2,000 ft and 200 KIAS, there were two uncommanded left rolls and the autopilot automatically disengaged. The crew found it harder to turn the control wheels to the right, but they maintained control of the aircraft and, although the flight director failed during the approach, they made an uneventful landing in Guernsey.

Extensive testing on the aircraft did not identify the cause, but the operator replaced several components as a precautionary measure. Subsequent component testing found no anomalies that could be definitively associated with the incident, although it did identify issues relating to equipment maintenance and testing. The operator has addressed these through appropriate safety action, and they reported that there been no recurrences since the aircraft returned to service.

The AAIB described the aircraft examination:

Initial examination and system testing was carried out under the supervision of the AAIB.

There was no evidence of anything untoward when the aircraft was checked externally and, with electrical power switched off, the control wheels were free to operate throughout their entire range of movement without any restrictions. The ailerons, balance tabs, and aileron trim were all observed to operate normally.

The rear avionics compartment, where the No 2 autopilot is located, was checked and found to be dry. All the equipment in the compartment was found to be secure.

The aileron control system mechanical runs were visually checked and operated throughout their range of movement. No anomalies were identified, and the control cable tensions were confirmed to be correct.

The autopilot, SCS and AHRS were functionally tested and no faults were apparent, although the right aileron synchro required a small adjustment to bring it into the limits defined in the Aircraft Maintenance Manual.

The operator decided to remove the following components as a precautionary measure before the aircraft was returned to flight:

- Aileron servo (serial number 391)
- Autopilot computer No 1 (serial number 311)
- Autopilot computer No 2 (serial number 218)
- Autopilot controller (serial number 158)
- Accelerometer (serial number 152)

After reviewing the FDR data and the CVR audio recording, AHRU 1 (serial number 1124) and AHRU 2 (serial number 1281) were also removed. In the meantime, the aircraft had been returned to service and flown without further incident.

The AAIB described that the laboratory examination of the components found no anomalies except for autopilot computer #2. The unit had been repaired in 2001 after a problem with overshooting selected headings. It then had been stored for 3 years, then fitted to another aircraft and in September 2010 fitted to SE-MAO. The AAIB wrote: "It was fitted to SE-MAO in September 2010 and the operator’s maintenance history showed that it had been transposed with the No 1 computer several times whilst troubleshooting various autopilot anomalies." The AAIB annotated that the overhaul agency tested the autopilot computer at normal ambient (room) temperatures and certify the unit airworthy if the tests were passed. Tests of the unit under supervision by the AAIB also did not find any malfunction.

The AAIB subsequently wrote with respect to the laboratory examination:

The printed circuit boards (PCBs) were removed and visual examination found localised corrosion, white translucent staining, and an accumulation of detritus on three of them. The affected PCBs were associated with the SCS logic and autopilot servo operation.

The manufacturer responsible for the autopilot design reviewed photographs of the PCBs and stated that, in their opinion, the corrosion was probably caused by trapped flux, possibly left after a previous repair. They noted an area adjacent to one corrosion site where there was a visible difference in the conformal coating. The repair records from 2001 were no longer available so the detail of the work carried out is unknown.

There was no requirement to check the internal condition of a computer returned for test or repair, so, under the normal test regime, the computer from SE-MAO would have been recertified as airworthy because it passed the requirements of the automated test.

The AAIB analysed:

Initial “jolts” and autopilot disengagement

The FDR data showed that the two unexpected left rolls occurred whilst the No 2 autopilot was engaged. Both rolls were preceded by the ailerons deflecting and the aircraft manufacturer determined that the rate of aileron deflection matched the performance capability of the aileron servo. When the second roll occurred, the autopilot disengaged with an audible continuous cavalry charge. This indicates that the autopilot logic detected a condition that met the requirements for the autopilot to disengage automatically. It was not possible to determine the reason for the automatic disengagement because no faults were identified during the investigation and the autopilot computer does not record a ‘fault log’.

Increased force when turning the control wheel to the right

The crew reported that it was harder to turn the control wheels to the right than to the left after the autopilot disengaged. They perceived that this force decreased as the aircraft slowed down, which might indicate an aerodynamic influence, but no anomalies were found in the aileron control system and the ailerons, balance tabs, and trim worked normally.

The flight data showed that the autopilot disengaged when the continuous ‘cavalry charge’ started, so the aileron servo clutch should have de-energised at that point. If the clutch failed to de-energise, the aileron servo would have remained connected to the control system, but any increased resistance caused by this should have been apparent in both directions of control wheel movement. The crew reported that the aileron out-of-trim indicator remained neutral after the autopilot disengaged, indicating that electrical power had probably been removed from the autopilot servo as designed.

When the servo was tested and dismantled, there was no evidence of any fault that could have caused or contributed to the incident, and the reason for the increased force to turn the control wheel could not be established.

Loss of the flight director

Flight director status and autopilot modes are not recorded on the FDR. No anomalies were identified during functional testing after the incident and the No 2 autopilot computer passed the automated test in accordance with the recertification requirements. The reason for the loss of the flight director modes during the approach to land at Guernsey could not be established. The operator reported that there have ‘not been any significant autopilot issues’ since the aircraft was returned into service after the incident.

No 2 autopilot computer and corrosion

The No 2 autopilot computer was active when the incident occurred and would have been responsible for automatic flight control, the SCS, and the flight director functions. The computer passed the requirements of the recertification test which was conducted at ambient temperature.