Lufthansa A321 near Bilbao on Nov 5th 2014, loss of 4000 feet of altitude

Last Update: March 24, 2015 / 20:43:43 GMT/Zulu time

Bookmark this article

Photo of Lufthansa D-AIDP, Airbus A321 — Lufthansa D-AIDP, Airbus A321 (Photo credit: tagsplanepics-lhr / Flickr / License: CC by-sa)

Incident Facts

Date of incident
Nov 5, 2014

Classification
Incident

Airline
Lufthansa

Aircraft Registration
D-AIDP

Aircraft Type
Airbus A321

ICAO Type Designator
A321

On Mar 24th 2015 Germany's Büro für Flugunfall Untersuchungen (BFU) reported in their November 2014 bulletin, that the first officer observed an irregularity in the properties of the speed indication just prior to reaching FL310 and disengaged the autopilot, the aircraft in response began a descent that lasted for about one minute before the crew was able to stop the descent at FL270.

The BFU reported Spain's CIAIAC delegated the investigation to the BFU on Nov 11th 2014.

The BFU reported that according to flight data and cockpit voice recorder the first officer (35, ATPL, 6,473 hours total, 5,179 hours on type) was pilot flying, the captain (52, ATPL, 16,384 hours total, 12,414 hours on type) pilot monitoring. After the aircraft climbed clear of top of clouds at about FL200 the flight data recorder recorded a fixed value of +4.2 degrees for the left hand AoA sensor, less than a minute later the FDR began to record a fixed value of +4.6 degrees for the right hand AoA sensor. The aircraft subsequently turned to fly direct to LATEK waypoint, during this turn the captain noticed the Alpha Protection Band had unusually and significantly increased. The first officer therefore reduced the climb rate from 800 to 500 feet per minute to enable the aircraft to accelerate. A short time later the first officer disengaged the autopilot and gave a brief nose down input.

The aircraft however continued to pitch down, inputs to counter the pitch down remained without effect. About 45 seconds after the nose down began the first officer alerted the captain who took control of the aircraft, that at this time had reached a rate of descent of 4000 feet per minute and a pitch of -3.5 degrees. The captain provided a maximum nose up input which caused the aircraft to pitch up again and the rate of descent decreased and the aircraft entered level flight.

The captain was able to maintain altitude by providing a continuous nose up input deflecting the side stick about 50% of its travel. The autopilot could not be engaged again, and a manual nose up trim was not possible.

The crew checked for related checklists but did not find any. The crew reset the Flight Augmentation Computers 1 and 2 in sequence with no effect.

8 minutes after the aircraft began its descent the Aircraft Communications Addressing and Reporting System (ACARS) issued an automated information to dispatch showing the three AoA sensor values amongst other data.

21 minutes after the aircraft began its descent the crew sent a message to maintenance checking whether a simultaneous reset of all FACs would be possible. Maintenance replied in the positive stating that the aircraft would revert to alternate law as result. Another 7 minutes later the crew reported they needed to constantly pull on the sidestick, trim was inoperative and autopilot could not be engaged and the Alpha Prot Band came up extremely quick. In addition the crew received a message "PH6 AOA3" on the centralized fault display system (CFDS). Upon suggestion by maintenance the crew switched off the air data reference unit (ADR3), however, without effect. ADR3 was reengaged. Another 12 minutes later maintenance wrote a message to the cockpit along the lines "after review of the data we found the values for AoA 1 and AoA2 appear to be frozen and report too high an angle of attack. If the problem persists, disengage ADR1 and ADR2 which will cause the aircraft to revert to Alternate Law however." then followed up "perhaps it is sufficient to just disengage ADR2".

The crew disengaged ADR2 which immediately prompted the aircraft to revert to Alternate Law and it was no longer necessary to pull the nose up.

The crew decided to use the remaining hour of flight time to verify the system status and to prepare for landing and landed safely at the destination.

The BFU reported that the aircraft features three Angle of Attack sensors consisting of a heated movable vane, the movement of the vanes is converted into electrical signals and the actual angle of attack computed by the related air data reference unit.

If the measured/computer Angle of Attack exceeds the value of Alpha Prot by one degrees, the autopilot is automatically being disengaged. In manual flight if the Alpha Prot Angles is exceeded, the Alpha Protection activates, the position of trim is stored and used as maximum nose up trim, the function of the side stick changes to command a specific pitch angle with the most nose up angle being Alpha Max which can be reached by full nose up deflection of the side stick.

The BFU reported that all three AoA sensors were examined by the manufacturer, no damage, malfunction or anomaly was identified with either of the sensors.

Airbus analysed the data and stated: "all three sensors worked normally until about 8 minutes into the flight, when the aircraft climbed through FL195. At that point, at an ambient temperature of -35 degrees C, AoA sensors 1 and 2 froze up at a position of approximately 4.5 degrees nose up and remained in this position until the aircraft descended towards the destination airport. At the time, when the autopilot disengaged the aircraft was flying at 0.675 mach, the Alpha Prot angle was 4.2 degrees, the Alpha Max 5.8 degrees. Within 15 seconds the first officer made increasing nose up input until reaching 75% of the maximum travel of the side stick, the attitude however changed from 4.5 degrees to -3.5 degrees against this input. The system disregarded/turned off the AoA 3 sensor because it disagreed more than the permitted value with the other 2 sensors.

When later ADR2 was disengaged, the system immediately reverted to Alternate Law because ADR3 had already been disengaged by the system and now two ADRs were offline.

The BFU reported that they are working to establish how reliable AoA sensors are but annotated: "The algorithms and boundary conditions differ from each other and are not entirely known to the BFU. The investigation is aiming to establish the probability of a repeat of this occurrence."