Description

On 12 January 2023, an Airbus A330-300 (N802NW) being operated by Delta Air Lines on a scheduled international passenger flight from Detroit to Amsterdam with an augmented crew touched down in night VMC just before the landing runway threshold. The event was reportedly not recognised by the flight crew and evidence of it was only discovered during a chance over-flight by a police helicopter two hours later following which damaged runway lights were discovered and an inspection of the aircraft found minor landing gear damage and mud on the lower rear fuselage.  

Investigation

Once notified, the Dutch Safety Board began a Serious Incident Investigation. The CVR and FDR were downloaded but because the 2 hour CVR had not been isolated or the aircraft electrically de-powered after the event, relevant data on the CVR were overwritten. ATC data showing the touchdown points of aircraft which landed on the same runway during the period when the landing under investigation occurred was also obtained.

The nine hour flight had a flight crew consisting of a Captain and two First Officers. The 57 year-old Captain had been employed by the operator for 31 years and was acting as PF and had a total of 18,236 hours flying experience including 601 hours on type. In the twelve months prior to the investigated event, he had flown for a total of 444 hours which had included into Amsterdam on three occasions, each time acting as PF and landing on runway 18R. The 59 year-old augmenting First Officer had been employed by the operator for 23 years and was acting as PM for the for the second half of the flight. He had a total of 11,228 hours flying experience including 1,083 hours on type. In the twelve months prior to the investigated event, he had flown for a total of 738 hours. He was familiar with Amsterdam but had not previously landed on runway 22. The 49 year-old First Officer, who had been employed by the operator for 12 years, was occupying a supernumerary crew seat on the flight deck having acted as operating crew for the first half of the flight. He had a total of 6,366 hours flying experience including 1,622 hours on type. He was also familiar with Amsterdam but had not previously landed on runway 22.

What Happened

The flight crew were aware prior to departure that their landing runway at Amsterdam was likely to be either 18R or 22 for both of which a minor crosswind component was forecast. The operator had a requirement for a minimum 20 knot headwind component for landings on the significantly shorter runway 22 which, based on the forecast, could easily be met. The conditions when the flight arrived in Amsterdam airspace were similar to those forecast with both the expected runways still in use. However, there was now almost no crosswind component and a significantly increased mean, although variable, headwind component for the much shorter (2,020 metre long) runway 22 compared to the 3,800 metre-long runway 18R and it was decided to use runway 22 for landing. Whilst not a precision runway, this runway was stated in the national AIP as suitable for aircraft approach categories A-D, the A330-300 is a category C aircraft. However, it does not have centreline lights and had a reduced approach lighting system compared to precision runways which were normally used by the operator’s aircraft.

Radar vectors to the ILS approach were provided and the aircraft was fully configured for landing with approximately 5nm to go. ATC instructed the flight to fly at 160 KIAS to 4nm and once fully configured, the Captain activated the “managed speed” which resulted in the A/THR commanding the aircraft to fly at 165 KIAS (VAPP + 24 knots). This 5 knot speed increase was stated to have “surprised the pilots” and resulted in the Captain deploying the speed brakes for 20 seconds in an attempt to reduce the speed which triggered a master caution which ceased when he stowed them.

Landing clearance was given at around 2nm with a wind check of 240° 30-41 knots which was unchanged just after touchdown. The Captain again briefly deployed the speed brakes before stowing them with about 1.2nm to go. The AP was disengaged at around 240 feet agl and almost immediately FDR data recorded a nose-down input from the Captain’s sidestick which changed the aircraft pitch attitude from 1.4º nose up to 0.3º and caused the rate of descent to increase to 1,000 fpm as the aircraft reached a half scale fly-up glideslope command on the EADI and a position which would have corresponded to four reds on the PAPI. Both operating pilots recalled having seen the PAPI indications changing to three reds but neither the operating nor observing First Officer had made the company-required SOP call for excessive descent as a half dot deviation below the glidepath had occurred which had continued despite the Captain’s nose-up sidestick input increasing the pitch attitude to +2.5º. The operating First Officer subsequently stated that he had been “confident that the Captain was in (appropriate) control of the flight’s trajectory”. 

Six seconds before touchdown, at 59 feet agl, the aircraft experienced a very small (5 knot) downdraft which was accompanied by a recorded full scale deviation below the ILS GS. The Captain responded with a sidestick input which increased the aircraft’s pitch attitude to 6.3º nose up as the right main gear touched down on the grass-surfaced Clearway 11 metres prior to the beginning of the runway with a recorded vertical load corresponding to a “firm” touchdown. The left main gear followed 3 metres further on and both main gear assemblies almost immediately hit a traverse concrete slab located 5.5 metres before the runway. As the aircraft reached the runway, the landing gear destroyed two threshold lights. Once on the runway, the rest of the landing roll was normal and the aircraft turned off the runway at intersection G7 which was 1,290 metres down the 2,000 metre long runway. The FDR record of the significant deviation below the PAPI indications visible to the pilots when the aircraft was approximately 650 metres from the beginning of the runway is shown below.

The vertical profile flown once the autopilot was disengaged showing an uncorrected descent below the correct PAPI indication below 130 feet agl. [Reproduced from the Official Report]

On the stand once the passengers had disembarked, the two operating crew “performed an exterior post flight inspection together with an aircraft technician (but) did not notice any abnormalities”. Approximately two hours after the landing, the crew of a police helicopter overflying the Runway 22 threshold reported to ATC that they had noticed debris on the runway. The airport authorities reviewed video footage of aircraft landing on this runway and identified the operator’s Airbus A330 as the aircraft involved. Delta Air Lines were advised and arranged for an aircraft inspection which found that minor damage to the main gear and observed that mud was on the lower rear part of the aircraft. Impact with the concrete slab was found by the landing gear OEM not to have exceeded the landing gear design limit load.

Why It Happened

Overall, it was assessed that the reason the aircraft descended below the ILS GS when the AP was disconnected was because the flight crew focused more on speed management than on flightpath management, thereby compromising safety margins for preventing undershoots.

It was found from an examination of the touchdown points of the 27 other aircraft (11 wide body and one narrow body) which landed during a 2½ hour period around the time of the undershoot touchdown, the other three Delta Air Lines (wide body) aircraft which were included in the total had been those which had touched down closest to the runway threshold, in two cases very close to it (see the illustration below). It was observed that despite the limited dataset, there may be a case for “a detailed review to determine the consistency of this pattern and to explore potential operational implications”.

The touchdown point of all aircraft landing on runway 22 between 0600 and 0830 on 12 Jan 2023. [Reproduced from the Official Report]

The following findings and assessments were also made:

• The PAPI configuration for runway 22 did not align with the current ICAO and EASA criteria for the threshold crossing height of larger aircraft such as the A330-300 and considered that this discrepancy contributed to a reduced safety margin.
• The fact that the runway 22 ILS antenna was 3½ feet lower than ICAO recommended minimum height had resulted in a threshold crossing height of 20 feet agl rather than the ICAO recommended 31 feet agl.
• The minor downdraft encountered just before touchdown further altered the already well-below-profile trajectory sufficiently to cause the undershoot touchdown.
• The pilots’ reaction time to respond to flight path control priorities was reduced due to concern about the correct functioning of the A/THR, the operator’s classification of the runway as short and their limited experience landing on short runways. As a result, they focused more on speed management than flight path management. 
• Aiming for an earlier touchdown is not an uncommon practice on shorter runways and the deviation when the aircraft descended significantly below the glide path was not vocalised.
• The ambiguities in the operator’s SOPs, especially regarding descent below the glidepath, coupled with the crew’s limited experience with Runway 22 and limited specific training for landing on short runways, led to deviations from standard procedures.
• Rest break scheduling is typically focused on needs of the pilot flying for landing, underestimating the need for managing fatigue of all flight crew members.

The Conclusion of the Investigation was, in summary, as follows:

Whilst the landing runway was adequately long, an attempt to mitigate the perceived risk of an overrun caused the Captain to increase the risk of an undershoot and this was followed by the aircraft touching down prior to the threshold. The psychological impact of landing on a seemingly short runway under relatively adverse conditions without specific training, likely led to a subconscious effort to land as early as possible. The Captain and the two junior pilots with him focused more on speed control than maintaining the ILS glide path and the aircraft was descended below it despite corresponding ‘four-reds’ PAPI indications. It was considered that this imbalance between the risk of a runway overrun and an undershoot and additional operator-imposed constraints had skewed their risk assessment and eliminated critical safety margins.

Safety Action taken whilst the Investigation was in progress was noted as having included the following:

Delta Air Lines:

• Added short field landing techniques to their recurrent training curriculum
• Prohibited the use of Amsterdam runway 22 by their Airbus A350, Airbus A330 and Boeing 767 aircraft

The Luchtverkeersleiding Nederland (LVNL) corrected the Dutch AIP entry for Amsterdam runway 22 to show the Minimum Eye Height over Threshold (MEHT) as 48 feet instead of 64 feet and also corrected the recorded lighting intensity for the same runway from medium to high.

One Safety Recommendation was made as a result of the findings of the Investigation as follows:

• that Amsterdam Airport restrict the use of Runway 22 for Eye-to-Wheel Height Category 4 aircraft landings until adjustments have been made to ensure the minimum threshold clearance for such aircraft can be achieved.

The Final Report was published on 18 July 2024.

Related Articles

• Controlled Flight Into Terrain
• Touchdown Zone (TDZ)
• Runway Lighting
• Approach Lighting
• Autothrottle/Autothrust
• Instrument Landing System (ILS)
• Visual Approach Slope Indicator Systems (VASIS)
• Foreign Object Debris (FOD)