West Sweden ATP at Isle of Man on Apr 9th 2021, autopilot refused to disconnect

Last Update: August 25, 2022 / 17:42:47 GMT/Zulu time

Bookmark this article

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

Incident Facts

Date of incident
Apr 9, 2021

Classification
Report

Departure
East Midlands, United Kingdom

Destination
Isle of Man, United Kingdom

Aircraft Registration
SE-LPS

Aircraft Type
BRITISH AEROSPACE ATP

ICAO Type Designator
ATP

A West Air Sweden BAe ATP, registration SE-LPS performing freight flight PT-522 from East Midlands,EN to Isle of Man (UK) with 2 crew, was on final approach to Isle of Man's runway 26 when the first officer, pilot flying, attempted to disconnect the active autopilot 2, however, no autopilot disconnect wailer occurred and there was resistance on the flight controls although both primary flight displays showed the autopilot disconnected. The captain took control, pressed the "SYNC" button, felt no longer resistance to flight control inputs and cotninued for a safe landing.

On Aug 25th 2022 the AAIB released their final bulletin concluding the probable causes of the incident were:

The investigation concluded that the anomalies with the audio warnings were probably associated with corrosion and moisture caused by water ingress through the DV windows.

The AAIB was unable to replicate the reported anomalies in the autopilot system, but theoretical analysis by the autopilot manufacturer identified a scenario involving the autopilot disconnect button and the associated debounce circuit that could lead to a partial disengagement of the autopilot. This scenario could explain the previous occurrences on G-BUUR and SE-MHF but would only partially explain the occurrence to SE-LPS.

Although the aircraft abnormal and emergency checklist did contain a procedure for the failure of the autopilot to disengage, the commander reacted instinctively due to his awareness of a previous occurrence that he had read about in an AAIB report. While the slipping clutch is designed to allow the crew to overpower an autopilot that does not disengage, a substantial increase in the force required to operate the flying controls when late on the approach could present an increased safety risk to the aircraft. The commander used the syn button which allowed him to operate the controls without hindrance and land the aircraft without further incident.

The AAIB analysed:

The investigation could not identify any faults with the autopilot system that could explain the resistance to movement in the flying controls. However, the autopilot manufacturer identified a possible scenario involving an intermittent problem with either the autopilot disconnect switch or the associated debounce circuit in the autopilot controller. Corrosion was found in the audio warning unit which may have caused the extended duration of the altitude alert warnings.

Audio warnings

The crew reported anomalies with the altitude alert during the flight and the cavalry charge sounded repeatedly on SE-LPS after it landed; it also sounded repeatedly after the previous flight. The aircraft electrical connector to the audio warning unit was found to be corroded and corrosion was also found inside the warning unit.

Corrosion was also found within the warning unit fitted to SE-MAJ and the PCBs inside were wet. The combined effect of the corrosion and moisture cannot be readily assessed.

However, given the similarities between the symptoms on both aircraft, it is likely that this contributed to, or caused, the extended duration of the altitude alerts and the repeated cavalry charge warnings on SE-LPS after it landed. The DV windows, which can be opened when the aircraft is on the ground, are known to be susceptible to water ingress. Previous SBs have been issued to reduce the likelihood of aircraft equipment being affected by moisture ingress, but SE-LPS did not have these SBs embodied. The SBs embodied on SE-MAJ were found to be incomplete, and damaged, and would not have prevented water leaking through the DV window reaching the audio warning unit.

Autopilot

The autopilot manufacturer suggested that rapid electrical pulses might have occurred when the disconnect button was pressed, resulting in the autopilot computer disengaging whilst the servo solenoids remained energised. While the FDR and PFD would show that the autopilot was disengaged, the rapid pulses would cancel the cavalry charge before it had time to sound. The forces at the control column would be greater than normal, and the only way the crew could overcome these forces would be to either over-power the slipping clutch, press the syn button, or remove electrical power from the autopilot by pulling the CB.

Whilst this scenario could explain the two similar occurrences on G-BUUR and SE-MHF, it could not fully explain the crew’s account of the event on SE-LPS. In the case of SE-LPS, the commander reported that the excessive forces dissipated when he pressed his syn button.

The modification standard of the aircraft, however, meant that this should not have been possible.

The AAIB considered the possibility that the autopilot disengaged normally when the co-pilot pressed the disconnect button, but that the lack of the audible cavalry charge led the pilots to believe that it remained engaged. However, since both pilots felt resistance in the controls, and the commander reported that the resistance stopped when the syn button was pressed, the possibility that the autopilot disengaging normally is considered unlikely.

The autopilot manufacturer identified additional testing to substantiate their theoretical failure scenario. If successful, this could result in changes to minimise the future arising rate. Safety action was taken by the CAA to include the proposed testing as part of the ongoing continued airworthiness of the ATP fleet.

The operator has removed SB 22-14 from all their aircraft, so that that both syn buttons will override either autopilot.

Crew response to autopilot anomalies

The commander’s significant experience on type as well as his knowledge of the previous AAIB report into a similar event meant that he was prepared to use the syn button. The co-pilot’s effective communication that he was experiencing difficulties with disengaging the autopilot also meant the commander understood rapidly what the issue was and was able to take appropriate action. As a result, the aircraft landed safely from the approach.

The crew tried both autopilot disconnect buttons and electric trims in accordance with the abnormal/emergency checklist without success. Whilst they did not try either the go around button, or the AP system selector button, it is considered unlikely that these actions would have released the servomotor solenoids. Whilst the design of the aircraft autopilot system allows the pilot to gain control of the aircraft through the use of the slipping clutches, this requires them to apply significant force on the flying controls. The significant forces involved could have presented the pilots with a demanding manoeuvre as the aircraft was close to the ground.