Description

On 1 October 2014, an Embraer 190-100 (PH-EZV) on a scheduled international passenger flight from Prague to Amsterdam made a very hard touchdown at Amsterdam off a Cat 1 ILS day approach in normal ground visibility. None of the 90 occupants were injured but the aircraft sustained damage requiring component replacement before further flight. The hard landing was automatically and manually reported to maintenance. The crew then continued their flying duty on another aircraft.

Investigation

An Investigation was carried out by the Dutch Transport Safety Board but its commencement was delayed because notification of the event was not given by the Operator until 20 days afterwards. The airline subsequently stated that the Investigation had failed to demonstrate that the damage found to the aircraft after the investigated hard landing was a result of it and expressed the opinion that "the classification of this occurrence and the obligation to investigate this occurrence is debateable". The Investigation noted that "the procedures for reporting occurrences in the airline's Operations Manual leave room for interpretation". It was also noted that as a consequence of the delay neither DFDR nor CVR data were available. Although the QAR data provided was equivalent to that which would have been recorded on the DFDR and was used to assist the Investigation, it was considered possible that because of the delay "the crew's recollections were possibly not as sharp".

It was noted that the 40 year-old Captain, who had been PF for the investigated flight, had accumulated 3,667 total flying hours and the 32 year-old First Officer had accumulated 4,939 flying hours. However, both pilots had only 157 hours experience each on the aircraft type involved, having completed their type conversion training earlier in 2014.

Based on the QAR data and the recollections of the two pilots, it was established that the weather forecast for Amsterdam prior to the flight indicated that around the ETA, the prevailing surface visibility at Amsterdam would be 3000 metres, temporarily 1200 metres in fog patches and with a 30% probability of a further reduction in visibility to 600 metres in fog with 100 feet vertical visibility before some clearing expected to begin around the ETA. The approach was briefed as a Cat 1 ILS with autoland - the ‘AUTOLAND1’ mode. Prior to positioning for the approach the ATIS gave the runway as 18C and the weather as 190/6 knots with visibility 1400 metres in fog and 1 okta of cloud at 1300 feet. As the aircraft was radar vectored onto final approach, the 18C IRVR was continuing to increase and was 2000m/1300m/550m two minutes prior to landing. The applicable ILS minimum altitude was 225 feet and the applicable Cat 1 RVR minima were 550m/125m/75m.

Both pilots “assumed” that they had correctly configured the aircraft for a Cat 1 automatic landing as briefed and intended. The Captain reported that he saw the runway from a distance of approximately 4 nm and that, “at approximately 50 feet above the runway” (the height where the AP-commanded flare to land should begin for autoland), he “noticed that the aeroplane was continuing to fly towards the runway at a constant rate of descent and did not perform a flare”. QAR data confirmed this, recording a constant pitch attitude of +1.6 degrees. He stated that in an attempt to reduce the rate of descent, he had “pulled back on the control column at a low altitude” but could not remember whether he had disconnected the AP which QAR data showed had occurred at less than 9 feet above the runway but before touchdown. It was considered that this disconnect had been a consequence of the Captain’s control column input rather than a pilot-commanded disconnect. The effect of the control column input was to increase the pitch attitude to a maximum of +8.6 degrees by the time the MLG touched after which the Captain then lowered the NLG to complete the landing. It was noted that in a correctly configured autoland, the A/T brings the thrust lever to idle at 30 feet agl; in this case, idle was selected between 48 and 34 feet above the runway.

After the flight had arrived at its parking stand and the engines had been shut down, the CMC printed an alert message warning that the touch down had been at 2.78g. The printed alert was left in the aircraft Technical Log along with an entry that a hard landing had been made. A copy of the corresponding EICAS message generated by the hard landing was automatically sent to the airline maintenance office by ACARS. The crew then continued their duty with two further flights in a different aircraft. Some 12 hours after the hard landing had occurred, the Captain submitted an Air Safety Report (ASR) which apart from recording a hard landing and stating that Flaps 5 had been set (the QAR data showed that actually, FULL flap was the setting used) and some weather details, it “did not contain any additional background information” about the event. The flight recorders were not secured and “the crew were not grounded for a safety investigation”. It was then two weeks after the occurrence before the airline flight safety department “established that this hard landing required further investigation” and a further week before “the airline formally decided to institute an investigation” after which “the pilots could not be scheduled for an interview with investigators any earlier than 4 November 2014” - more than a month after the occurrence.

It was noted that the aircraft was approved for ILS CAT I, II and IIIA Approaches with CAT I and II approaches followed by either a manual or an automatic landing but CAT IIIA approaches requiring an automatic landing. It was established that in order to perform an automatic landing, the rotary ‘minimums’ switch must be set to ‘RA’ rather than ‘BARO’ and the flaps must be set to a final position 5 before the selection of an AUTOLAND, which must be done between 1500 feet and 800 feet aal. For a Cat 1 approach, the rotary ‘minimums’ switch must then be changed to ‘BARO’ so that the correct minimums call is annunciated. The FMA at the top of each pilot’s PFD provides system status indications which, for a Cat 1 autoland, include the indication ‘AUTOLAND1’. Where a manual landing is to follow an automatic approach, the AP must be disconnected no lower than 50 feet agl.

Contrary to the above, however, the QAR data showed that although APPR1 had been selected after passing 1500 feet, the flaps were set to FULL at 1400 feet which on its own would have prevented operation of the automatic landing mode. The QAR data also showed that the BARO/RA minimums switch was in the BARO position throughout the 1500 feet/800 feet window where the Autoland arms, which again would, on its own, have prevented APPR1 mode activation. It was noted that “the FMA indications had still not changed when passing 150 feet above the runway threshold” which meant that a Cat 1 ILS with manual landing had been set up. Since the aircraft was in an allowable configuration, no error messages were generated. However, it was noted that the FMA indications for ‘AUTOLAND1’ and ‘APPR1’ are significantly different besides the presence of either ‘AUTOLAND1’ or ‘APPR1’ at various stages during the approach.

The Investigation noted similar occurrences involving the same aircraft type operated by the same airline both before and after the one under investigation. In a 2009 event, the BARO/RA switch was not initially in the RA position and so the intended autoland was not flown and when the crew didn’t notice, a hard landing followed. The airline did not investigate this. In a 2015 event, a Cat III approach was initially briefed with reversion to a Cat 1 manual if the weather improved. Having reverted to the Cat 1 manual, a weather update from ATC prompted them to change to a Cat 2 autoland but they did not notice that their DH change resulted in the FMA ‘APPR2’ indication changing from white (signifying autoland) to green (signifying a manual landing) and again a hard landing followed. The aircraft manufacturer advised that they were not aware of other similar events.

The formally stated Conclusions of the Investigation were as follows:

The crew were incorrectly under the impression that they had configured the aircraft for an automatic landing. The indications of the automatic pilot did not lead the pilots to suspect that the aircraft was actually configured for a manual landing. The FMA indications that they saw during the approach were what they were used to seeing. Moreover, the aircraft was in a valid configuration, which meant no error messages were generated. As a result, both pilots had no reason to think that the aircraft was not flying in the correct mode for an ILS Category I approach followed by an automatic landing. The aircraft did not perform a landing flare and made a hard landing.

The fact that the Cockpit Voice Recorder was no longer available has had adverse effects on reconstructing events and gaining insight into the crew’s considerations prior to the hard landing. The crew’s recollections of the incident had faded and/or may have been influenced by more recent flight experiences. The procedures for reporting incidents described in the airline’s operations manual leave room for interpretation regarding which incidents should be reported and what follow-up actions are required. This results in the loss of important sources of information for the investigation of incidents.

The Final Report of the Investigation was completed on 31 May 2016. No Safety Recommendations were made.

Related Articles

• Autopilot
• Autoland
• Cockpit Automation - Advantages and Safety Challenges
• Portal:Automation and Safety