ACCIDENT OF AN IBERIA AIRBUS A320 IN BILBAO

 

Airbus Industrie initiates a move to increase pilots authority over their automated safety system

by Tim van Beveren, Miami    2001-06-01 

The landing accident of an Iberia Airbus A 320 on March 7 in Bilbao, Spain, shows vital consequences for Airbus Industrie. The European aircraft manufacturer recently surprised the aviation community by announcing that it is going to revise the software programming of its automated Angle-of-Attack (AoA) protection, also known as “alpha-protection”. This highly sophisticated safe-guard system has been installed on all of Airbus’ fly-by-wire aircraft, dating back to the first delivery of an A320 in 1988. The alpha protection has always been regarded as one of the prestigious, outstanding and truly safety-enhancing features of the modern Airbus products. The computer guided system prevents the aircraft of entering a stall, resulting from a excessive angle of attack. If, for example, the pilot inadvertently commands too high a nose-up attitude in a climb, which would lead to a subsequent loss of lift, the system will automatically lower the nose by acting on the elevators and thereby prevent the aircraft from entering into a stall. The Airbus A 320 Flight Crew Operation Manual (FCOM) defines it as a system “which provides protection against stall and windshear, (and) has priority over all other protections.” (Emphasis added)

             However, according to first released findings of the Bilbao accident investigation, the ‘activity’ of this safety feature was a contributing factor in the event: the alpha-protection contradicted the desired pilots action. During the final approach to runway 30, the Ground Proximity Warning System “sink rate” warning was triggered and the crew applied TOGA-power (Take-Off/Go-Around power) to abort the landing.

             Yet, special attention should be given to the specifics of the scenario: At around 23:10 local time Iberia flight 1456 was approaching Bilbao’s Sondica airport. The flight originated in Barcelona. On board were 136 passengers and a crew of seven. As the flight was a training flight for the First Officer, there were three pilots in the cockpit. At the time of their ILS approach the crew encountered a thunderstorm and was advised of light turbulence and surface wind speeds from 240 degrees of only 8 to 9kts, but no windshear. The airport is dreaded by pilots for its critical conditions, - especially in the winter, and is not equipped with improved weather measuring equipment or modern windshear detectors. The airport was the scene of two other weather related accidents that occurred during the preceding 15 days and another three in the previous five months.  

Bilbao’s Air Traffic Control did not mention to the Iberia crew that, just shortly before the A320's approach, three other aircraft had tried unsuccessfully to land at Sondica and had finally decided to divert. According to statements of airport personnel to local media after the event, other flights also diverted directly to their alternate, without even trying to land in Bilbao.  

            During the final approach the A320 encountered heavy turbulence at about 200 ft AGL, with gusts up to 65mph, an 1.25g- updraft, followed by a downdraft and tailwind gusts at an altitude of 70-50ft. The associated change of wind direction during the event clearly points to a windshear encounter. 

             According to information released by Airbus’s deputy director of flight operations support, Cpt. Michel Brandt, the flight crew applied a forward sidestick input during the updraft, then an aft input to reduce the subsequent increasing sink-rate. When the GPWS alerted the crew about their unusual increased sink rate, the pilots decided to perform a go-around and applied TOGA-power. But the crew’s desired and commanded action was not performed by the aircraft. As the alpha-protection was triggered during this event, the system commanded a nose down signal, which was performed, even though both pilots had their sticks full backward, commanding a “climb”. Nevertheless the airplane touched down with all three gears struts almost simultaneously and with an estimated vertical speed of 1,400ft./min. The nose gear subsequently collapsed and the plane slid along the runway for about 3,280 ft. before coming to a stop. During the emergency evacuation, four passengers and some of the crew received minor injuries, among those, one passenger, a 75 year old female, was hospitalised.  

The barely six month old aircraft received substantial damage, including the wing structure and the engine nacelles. It is beyond economical repair and therefore should be regarded as a total loss. The accident has been under investigation by the Spanish CIAI (Comisión de Investigación de Accidentes e Incidentes). For Spain’s national carrier Iberia this accident represents the first loss of an Airbus A 320. The company operates 85 modern Airbus fly-by-wire aircraft (A319, 320, 321 and 340) and last year transported over 30 million passengers. The Spanish company CASA is part of the European Airbus consortium.  

Back in early April, the French Civil Aviation Authorities (DGAC) had issued an airworthiness directive (AD) for the A319/320 aircraft. It ordered the crews to fly at least 10 kts. faster and use only “CONFIG 3” (flaps 3) setting on approach in conditions with gusts greater than 10kts reported wind increment (max. wind minus average wind), or when moderate or severe turbulence on short finals has to be expected. In such events, the crew must select no more than flaps 3 and maintain a minimum approach speed of Vls (= “lowest selectable speed”) + 10kts. If the GPWS “sink rate” warning occurs below 200ft, an immediate go-around is required. Operators incorporated this AD-note into special bulletins for their pilots, but no additional information about the nature of this special procedure was given so far.  

            In the light of the accident occurring only four months ago, and the sometimes lengthy ‘normal’ timeframe for implementation of safety revisions after an accident, it appears to be of “amazing speed” how Airbus Industrie, - even in the absence of a final report, has already decided and performed a modification on the alpha- protection control laws. This was done in an approach “to increase the flight crew’s authority”, - as Cpt. Brandt was quoted saying by media. - A step applauded even by staunch Airbus critics among the international pilot community. A revised software version is expected to be validated this month and has already received certification by the French Civil Aviation Authority (DGAC) and the European Joint Aviation Authority (JAA). Airbus plans to implement a “rapid retrofit program” for its entire A319/320 fleet.

(sources: AW&ST, IBERIA, others & own research)

© TvBeveren 6-2001

AAIB Logo

United Kingdom
Air Accidents Investigation Branch

Bulletins (May 2001)


AAIB Bulletin No:

Ref: EW/C2000/6/8

Category: 1.1

Aircraft Type and Registration:

Airbus A321-211, EI-CPE

No & Type of Engines:

2 CFM56-5B3/P turbofan engines

Year of Manufacture:

1998

Date & Time (UTC):

21 June 2000 at 1200 hours

Location:

London Heathrow Airport

Type of Flight:

Public Transport

Persons on Board:

Crew - 9 - Passengers - 180

Injuries:

Crew - None - Passengers - None

Nature of Damage:

Aft fuselage damaged between frames 63 and 67

Commander's Licence:

Airline Transport Pilot’s Licence

Commander's Age:

51 years

Commander's Flying Experience:

13,190 hours (of which 700 were on type)

 

Last 90 days – 130 hours

 

Last 28 days – 58 hours

First Officer's Flying Experience:

274 hours (of which 82 were on type)

 

Last 90 days – 82 hours

 

Last 28 days – 58 hours

Information Source:

AAIB Field Investigation

The aircraft was operating a scheduled service flight from Dublin Airport, Ireland, to London Heathrow Airport. The crew had operated one round trip earlier in the morning, this flight was thus their third sector of the day. The 1050 hrs ATIS report received by them was as follows: Surface wind 210°/16 kt, visibility 40 km, scattered cloud at 3,000 and 4,500 feet, temperature 17°C, dewpoint 9°C, QNH 1009. The ATIS also included information regarding windshear on the approach to Runway 27R, of plus or minus 10 kt at 400 feet agl. The last reported surface wind, 20 seconds before the aircraft landed, was 220°/18 kt.

The aircraft received radar vectors to an ILS approach on Runway 27R with the first officer (FO) as the pilot flying (PF). When established on the ILS approach, and in visual contact with the runway, he disconnected the autothrottle and the autopilot. The remainder of the approach was flown manually with reference to managed speed and using the flight director. From 300 feet agl until landing the aircraft became destabilised in pitch. At around 200 feet agl the rate of descent reduced significantly and the aircraft began to deviate above the ILS glideslope. The FO attempted to correct and in so doing a rate of descent of 800 to 900 feet per minute developed. He then started to apply aft sidestick at a height of 60 feet agl and continued the input until touchdown but the rate of descent was not fully arrested and the aircraft landed firmly.

The main wheels contacted the runway and then the aircraft rebounded into the air with both wheels lifting off again. Ground spoilers deployed at the initial touchdown and the pitch attitude continued to increase. The aft fuselage made contact with the runway at the same time as the second mainwheel touchdown occurred. The remainder of the landing was carried out uneventfully with the crew being unaware of the tailstrike until advised by ATC on vacating the runway. The cabin crew also reported to the commander that there had been an unusual noise at touchdown.

Damage to the aircraft

The rear fuselage lower skins were heavily abraded between the aft potable water and toilet service panels over an area extending some 2.5 metres longitudinally and 0.4 metres laterally, and the lower segment of the aft galley drain mast was bent and partly ground away by contact with the runway. A scrape mark approximately 0.4 metres wide and 8 metres long was found on the surface of Runway 27R in Block 18, slightly to the left of the centre line, consistent with the damage observed on EI-CPE.

The lower skin damage was bounded approximately by frames 63 to 67 and stringers 43L to 42R respectively, implying a small amount of right bank at the instant the tail contacted the ground, and comprised localised penetrations through the full thickness of the skins where these were supported by frames and stringers, together with minor damage to underlying structure including localised bending of butt-straps and frame flanges. The fuselage skins immediately surrounding the penetrated regions, which were not directly supported by underlying structure and consequently had been able to deflect partially away from the runway surface, suffered less severe abrasion damage together with inwards buckling at the centres of some skin panels.

Whilst any penetration of the pressure hull has potentially serious implications, in this case the structural integrity of the fuselage had not been compromised.

The aircraft

The aircraft landed at a weight of 70.7 tonnes, the maximum allowable being 75.5 tonnes. A landing wind of 240°/15kt had been entered on the performance page of the Multipurpose Control and Display Unit (MCDU). The aircraft was flown with reference to managed speed displaying to the pilot a computed variable speed target according to the actual wind, with a minimum target of 141 kt (VAPP). The A321 pitch attitude when on a stable approach at VREF (136 kt in this case) is 4°.

Ground spoilers deploy when both main landing gears touchdown, and remain extended during a bounce unless either thrust lever is advanced. Automatic braking commences either 2 or 4 seconds after ground spoiler deployment, dependant upon the level pre-selected.

The manufacturer produced a Flight Crew Operators Manual Bulletin No 22/2 dated April 1999 in which the issue of tailstrike avoidance was addressed. Deviations from normal landing techniques were cited as being the most common cause. The bulletin included a note: "it is imperative to reach the flare height at the appropriate speed and flight path angle (-3°)".

The two sidestick controls on the aircraft have an equal priority during normal operations. There is no physical movement or any other mechanism by which the pilot non-flying (PNF) can know what the PF demanded input is at any particular moment. When the PNF operates his sidestick in the same or opposite direction as the PF, both pilots inputs are algebraically added. Thus, the procedure for the commander to take over control was to press the takeover pushbutton on his sidestick, to establish priority, and to announce "I have control".

Pilot experience

The aircraft commander had been flying this type of aircraft for approximately one year. The nature of the scheduled service operation ensured that he had carried out a large number of sectors in this time. He had no previous experience of a bounced landing in this type of aircraft, it being difficult to reproduce in an aircraft simulator for a pilot to practice. The FO had recently completed his line training on the type.

Flight recorders

The Cockpit Voice Recorder (CVR), an Allied Signal Solid State A200S was removed and replayed by the AAIB. The two hour recording began prior to the previous approach into Dublin, and covered the entire incident flight, including the approach and landing at Heathrow. The Flight Data Recorder (FDR), an Allied Signal Solid State Recorder was also replayed and parameters from the entire incident flight and 12 previous flights were obtained.

Figure 1 shows some of the recorded parameters from 300 feet agl to touchdown. The aircraft was on the ILS glidepath with a rate of descent of 700 ft/min. At around 200 feet and an airspeed of 142 kt CAS, there was an increase in engine power from 53% N1 to around 67%N1 for 4 seconds. The rate of descent reduced to 450 ft/min and the aircraft began to deviate slightly above the glidepath. There was a forward movement of the FO's sidestick and the aircraft pitched down to 0°; airspeed increased to 148 kt CAS and the rate of descent increased to 800-900 feet/min. The aircraft returned towards the glidepath and pitch was increased to 4.5° nose-up. At 90 feet agl the FO made two brief nose-down demands on the sidestick which caused the aircraft to pitch down and reach an attitude of 0.5° nose-up at 50 feet agl. The rate of descent increased and at 60 feet agl the FO’s sidestick began to be moved aft to demand a nose-up elevator position for the flare. Two seconds later the pitch attitude had reached 2° nose-up and continued to increase but the rate of descent reduced only slightly. From 90 feet agl thrust was gradually reduced and the thrust levers were closed by 20 feet.

The aircraft touched down at an airspeed of 130 kt CAS, with a pitch attitude of 7.4° nose-up and a normal acceleration of 2.0g. The FO’s sidestick position was 92.5% nose-up demand with an up-elevator angle of 12.3°. The FO's sidestick demand then reduced, towards 46.8% nose-up demand. The ground spoilers deployed automatically; this is designed to occur when both the main landing gear oleo switches are compressed. The FDR showed that these switches then ‘unmade’ indicating that the aircraft had rebounded into the air. The pitch attitude continued to increase to a maximum of 9.8° nose-up, which was reached just as the aircraft mainwheels touched the ground again. The tailscrape occurred at this point. The second touchdown recorded a normal acceleration of 1.6g at which time the commander's sidestick moved forward to a 56.3% nose-down demand.

Simulation

Tests were conducted in a simulator by the aircraft manufacturer using recorded data from the flight. It was found that to match the aircraft performance to the achieved flight path a corrective wind input was required. The magnitude of this could then be measured. A significant vertical turbulence was observed at 300 feet agl. Coincident with this the angle of attack changed and the aircraft pitched up to 4° nose up for about 4 seconds. Below 100 feet agl there was a sustained loss of headwind of 10 kt which was considered to be the principle cause of the limited change of flight path with the increasing pitch attitude during the flare.

Previous incidents

The manufacturer had details of 9 tailstrikes on the A321 aircraft before this event, all of which occurred during the landing phase. Two of these events have been reported on previously by the AAIB. In the incidents for which data was available to the AAIB it could be seen that, after an initial touchdown, the aircraft all continued to pitch up, resulting in the tailstrikes after touchdown or during the resultant bounce. Subsequent to this event there have been two further instances, one of which (D-AIRE) is also reported on in this bulletin.

Aft fuselage contact will occur on the A321 at a pitch attitude of 9.7° with the main landing gear oleos compressed, as compared with 11.7° for the A320. The rate of tailstrike incidents per aircraft landing for the A321 is 4 times greater than that for the A320 probably as a result of the smaller margin from the normal flare attitude. Damage to the structural integrity of the fuselage should not pose a significant flight risk to the aircraft provided the tailstrike is detected during or after landing. In this case although it was not detected by the flight deck crew, both ATC and the cabin crew reported it to the commander.

Analysis

This incident is similar to others reported on previously in that a combination of three effects result in the tailstrike. These were: the pitch up effect of the automatic ground spoiler deployment; the elevator position, which, although the sidestick moved forward, would still produce some nose up demand; and the pitch inertia which had developed during the landing flare.

A comparison was performed using data from the previous incidents investigated by the AAIB and 12 other landings on EI-CPE which were recorded on the FDR. The data showed that in the normal landing a flare would be initiated at a height of around 60 feet, the sidestick position would then be checked to maintain a sidestick position between 25% and 37.5% nose up demand below 40 feet. The average sidestick position at touchdown was 37.5% nose up demand. All the recorded landings showed a similar flare initiation height and rate of application of aft sidestick input. However, in the tailstrike incidents the sidestick continued to move aft and achieved a stick position of between 75% and 100% nose up demand at touchdown. This additional aft sidestick movement produced higher pitch attitudes at initial touchdown, between 5° and 7°, as compared with an average of 3.5° nose-up for the normal landings. It would seem that the tailstrike becomes very likely with such an aft sidestick position at the initial touchdown, due to the three effects discussed earlier.

All four tailstrike incidents examined showed a greater variation of rate of descent through the approach, suggesting a less stable approach. The average rate of descent at the flare initiation was 700 ft/min, reducing to around 300 ft/min at touchdown. In the tailstrikes the rate of descent did not reduce as much during the flare, leading to a rate of descent at touchdown between 700 and 800 ft/min.

In the case of the incident involving EI-CPE there were some quite large variations in rate of descent. The destabilisation of the approach during the later stages was probably as a result of local turbulence arising from the gusty crosswind conditions. The approach however was not sufficiently destabilised for the commander to have had concerns about the eventual outcome of the landing.

Below 100 feet the aircraft experienced a loss of headwind and by 70 feet the airspeed had reduced to 136 kt, which was 5 kt below VAPP. The margin between the more positive pitch attitude required to maintain the flight path and the 9.7° at which tailstrike would occur was therefore reduced. As a result of destabilisation resulting from turbulence and sidestick inputs the aircraft pitch attitude was 1.5° decreasing to 0.5° nose-up at the beginning of the flare. Inertia from an increased rate of descent and the low pitch attitude would have delayed the change of flight path angle at the flare.

After touchdown, the pitch inertia from a late and large aft stick input, the continued presence of a nose up elevator demand, and the pitch up effect of spoiler deployment all contributed towards the aircraft continuing to pitch up. It is also worth noting that in the event of a bounced landing the aircraft would not experience the normal pitch down effect as a result of braking action. The pitch up effect at touchdown would have made it difficult for the pilots to determine whether or not a bounce had occurred.

The commander did not anticipate a problem until after the aircraft's initial touchdown. He could not have been aware of the control inputs applied by the FO, in particular the continued aft sidestick input late in the landing, because his own sidestick showed no movement. It is unlikely that he had sufficient information available to be able to evaluate and carry out any corrective action at this stage. When he applied an input after the second touchdown any effect would have been reduced because he did not use the takeover pushbutton.

A review of previous events showed that a significant proportion of tailstrike incidents occurred when co-pilots recently qualified on type were handling the aircraft. There are several possible explanations for this. When a deviation above the glideslope occurs at low level a choice arises as to whether it is preferable to maintain a stable approach and accept a greater landing distance, or to destabilise the approach. A greater level of experience may be required to make this judgement. It is also possible that some less experienced pilots may follow the ILS glideslope too closely below 100 feet thereby diverting their attention from the visual landing phase leading to a late recognition of a low pitch attitude and high rate of descent. Once the high rate of descent has been detected, this then leads to an input of aft stick being sustained until after touchdown has occurred.

The sidestick control authority logic requires a different method of intervention by commanders from that which they may have experienced on other aircraft types. Because of the difficulty of detecting the inputs made by the other pilot early takeover of control based on flight characteristics is required. On a number of tailstrike incidents commanders have made control inputs without using the takeover pushbutton and have thereby limited the effectiveness of those inputs. There is therefore a need for commanders to be well practised in a strategy for taking over control when necessary during the landing phase.

It is considered that pilots need more guidance towards detecting the flight characteristics which may lead to a tailstrike to enable them to take effective preventative action. Flight path destabilisation with increased rates of descent below 150 feet are an indication and particular attention needs to be given to abnormally low pitch attitudes at a late stage of the approach. Airbus Industrie have indicated that they are reviewing the Operator's Manual Bulletin dealing with the subject of tailstrikes. It is therefore recommended that Airbus Industrie should include this information in a revised Bulletin.

Summary

This incident occurred when a first officer who had recently qualified on type was flying the approach. In its latter stages the approach became destabilised, probably because of local turbulence arising from the gusty crosswind conditions and sidestick inputs which resulted in some quite large variations in rate of descent. Inertia from the increased rate of descent and the low pitch attitude of 0.5° at the beginning of the flare delayed the response of the aircraft to the aft stick input at the flare and resulted in a heavy touchdown. After touchdown, the pitch inertia from a late and large aft stick input, the continued presence of a nose up elevator demand, the pitch up effect of spoiler deployment, and the lack of the normal pitch down effect as a result of braking action all contributed towards the aircraft continuing to pitch up and bounce. The aft fuselage made contact with the runway at the same time as the second mainwheel touchdown occurred.

Safety recommendation

Recommendation 2001-46

It is recommended that Airbus Industrie should reissue the Operator's Manual Bulletin dealing with the subject of tailstrikes. This should include further guidance regarding flight path destabilisation with increased rates of descent and abnormally low pitch attitudes at a late stage of the approach. It should also re-emphasise the need to use the takeover pushbutton to achieve effective intervention.


Published May 2001
Return to May 2001 Index
Return to Bulletins Index
Return to Air Accidents Investigation Branch Index
Return to DETR Aviation Index
Return to DETR Home Page
Web Site Terms

http://www.aaib.detr.gov.uk/bulletin/may01/eicpe.htm 

AAIB Logo

United Kingdom
Air Accidents Investigation Branch

Bulletins (May 2001)


AAIB Bulletin No: 5/2001

Ref: EW/C2000/10/04

Category: 1.1

Aircraft Type and Registration:

Airbus A321, D-AIRE

No & Type of Engines:

2 IAE V2530-A5, turbofan engines

Year of Manufacture:

1995

Date & Time (UTC):

9 October 2000 at 1545 hours

Location:

Runway 27 Left, London Heathrow

Type of Flight:

Public Transport

Persons on Board:

Crew - 8 - Passengers - 170

Injuries:

Crew - None - Passengers - None

Nature of Damage:

Tailscrape between frame 64 and frame 68 in the area of the rear fuselage

Commander's Licence:

Airline Transport Pilot’s Licence

Commander's Age:

46 years

Commander's Flying Experience:

7,390 hours (of which 2,700 were on type)

 

Last 90 days -150 hours

 

Last 28 days - 65 hours

Information Source:

AAIB Field Investigation

History of the flight

The flight crew were on the second day of their rostered cycle. The previous day was the first occasion on which the commander and First Officer (FO) had flown together. Prior to the first departure a normal briefing and discussion took place, which included the operating experience of the FO, who had acumulated some 1,236 hours on the A300 series operated by the company. This included the A319, A320 and A321. The FO and commander both had approximately one third of their hours on each variant of the type.

On 8 October 2000, the day before the incident, the crew flew an A320 on three sectors, departing from Frankfurt for Oporto, returning to Frankfurt and then continuing to their overnight stop in Stuttgart. The FO flew the first and third sectors with the commander conducting the second. The flights were uneventful and the commander considered the FOs’ handling of the aircraft to be of a good standard.

On 9 October 2000 the crew flew an A320 from Stuttgart to Frankfurt with the commander handling. At Frankfurt the crew changed aircraft to the A321 for the flight to London. The flight was uneventful with the only difference from that planned was a change of cruising level due to turbulence. Prior to arrival at the Lambourne hold the commander obtained ATIS information ‘India’ which for Runway 27 Left was: surface wind 180°/18 gusting 27 kt, 5,000 meters in rain with cloud FEW at 1,100 feet, BKN at 1300 feet, OVC at 1,800 feet, temperature +12°C, dew point +11°C and QNH 992 runway wet. The FO briefed the approach and the crew discussed the conditions. It was agreed that the FO would fly the approach and make a positive landing given the wet runway surface. The aircraft’s landing weight was 72,999 Kg.

Radar vectors were provided by ATC to position the aircraft onto the final approach for Runway 27 Left. The crew used the autopilot coupled to the Flight Director in managed speed mode. At about 1,000 feet the FO de-selected the auto-throttle and, with the approach stabilised at 570 feet, the auto-pilot was disconnected. At approximately 500 feet the commander called that they were ‘coming low’ meaning below the glide path, to which the FO reacted by slightly increasing nose up pitch and thus reducing the rate of descent. The commander estimated the height to have been 300 feet when he saw the Visual Approach Slope Indicator (VASI) displaying three white lights and one red light and shortly afterwards he noticed the rate of descent was high. He looked inside the flight deck and saw a rate of descent of 1,100 feet per minute and called ‘sink rate’. Almost immediately the Ground Proximity Warning System (GPWS) sounded a Mode 1 alert of ‘sink rate’. Seeing the rate of descent was still high the commander applied full aft sidestick to cushion the landing. He did not depress the takeover push button on his sidestick but, as the aircraft bounced, he neutralised his sidestick and as the aircraft touched down a second time, he again applied a large amount of aft sidestick to lower the nose landing gear gently. At that point he noticed that the Primary Flight Display (PFD) showed a nose up attitude of some 10°. Neither pilot recalled hearing the normal voice height calls from 50 feet or the voice call of ‘RETARD’.

Believing there may have been a tail scrape the commander asked the rear cabin attendants if they had noticed anything. They described a rattling noise but thought that it had come from the galley. After shut down the commander inspected the tail area and discovered the damage caused by the tail scrape.

Engineering investigation

Examination of the damage to the aircraft was carried out at London Heathrow Airport whilst the aircraft was being temporarily repaired prior to an unpressurised ferry flight to a maintenance facility in Ireland. Damage was confined to severe abrasion of the aircraft’s skin longitudinally from just forward of frame 64 to just aft of frame 68 and laterally over the lowermost four stringers and drain mast close to this area (see Figure 1). There was no apparent damage to the fuselage frames. The damage was biased slightly to the right side of the fuselage, indicating that the aircraft had been rolled slightly to the right at the time of contact. Also, geometrical considerations indicated that the damage had occurred over a pitch attitude range of 9.65° to 10.65° referenced to the horizontal aircraft datum. Data from the aircraft manufacturer indicated that tail scrape damage would occur over a pitch attitude range of 9.7° (with the main landing gear oleos fully compressed) to 11.4° (with the main landing gear oleos uncompressed, but with the tyres in ground contact).

The aircraft’s maintenance computer was read out but no defects of any significance had been recorded. However, during the inspection of the aircraft by the operator’s maintenance personnel at London Heathrow after this event, slight leaks from two fuel couplings in the wing root area were noticed. This is a known problem that can occur following a firm landing. Rectification of this defect was deferred in the aircraft technical log and was carried out at the repair station.

Flight Recorders

Recorded information was available from the solid state Flight Data Recorder (FDR) fitted to the aircraft. The 30 minute tape based Cockpit Voice Recorder (CVR) had recorded over the period of the landing.

Accident Landing

An ILS approach was made to Runway 27L at Heathrow with both the autopilot and autothrottle engaged. The aircraft was established on the localiser at 3,200 feet agl and glideslope established at 3,000 feet agl. Landing configuration was achieved by 1,400 feet agl with full flap and gear down selected.

A stable ILS approach was flown with airspeeds varying between 150 kt and 160 kt down to 1,000 feet agl at which point the autothrottle was disconnected. Glideslope and localiser stability was maintained as the aircraft descended with an approximate pitch attitude of 1.5° nose up and at a manually controlled airspeed of approximately 150 kt. During this phase of the approach the autopilot had maintained a drift angle in the order of 12° to counter the effects of a crosswind from the left. Typical wind values were recorded as being 185° / 32 kt during the latter stages. At 570 feet agl the autopilot was disconnected and movements in the right sidestick position only indicated that it was the First Officer flying the manual section of the approach.

Throughout the approach and landing phase, sidestick movements resulted in corresponding changes in elevator deflection. As the aircraft descended through 500 feet agl it became low on the glideslope and reached a maximum deviation of 0.76 dots low by 320 feet agl before the nose was raised to regain the correct profile. Movement of the power levers ceased at this point with N1 values on both engines remaining constant at 68% until just before touchdown.

By 174 feet agl, airspeed had reduced to 141 kt as the aircraft flew through the glideslope neutral point and became progressively higher than the desired flightpath. By 145 feet agl the aircraft was 0.6 dots high and forward sidestick was applied to lower the nose towards 1.75° nose up. At 100 feet agl the aircraft was at its maximum above the glideslope by 1.8 dots and had a calculated vertical speed in the order of 900 ft/min. Forward sidestick was applied again, lowering the nose to a level pitch attitude and increasing the descent rate to 1,000 ft/min. At 62 feet agl, aft sidestick movement (25% aft of neutral) was recorded at the start of the flare and, one second later at 41 feet agl, an additional small amount of aft movement of the commander’s sidestick was observed. By 21 feet agl, the aircraft was descending at 1,200 ft/min and the activation of a GPWS Mode 1 (‘Sink Rate’) warning was recorded. The aircraft was pitching up through 2.1° nose up with the commander’s and the first officer’s sidestick at 56% and 79% aft of neutral respectively at that time. Both power levers were closed at approximately 15 feet agl and the commander’s sidestick position was recorded as being 96% aft of neutral; that of the first officer was 69% aft.

One second later, with the commander’s sidestick neutralised and the first officer’s at 53% aft, the aircraft touched down at a pitch attitude of 7.4° nose up and 1.4° right wing low. A normal acceleration of 2.08g was recorded at this first touchdown, a descent rate for which was calculated to be approximately 850 ft/min. The sampling rate of normal acceleration was eight times per second and, with the sharp rise and fall of the ‘g spike’, it is possible that the peak value was slightly higher than that recorded. Both left and right main gear ‘weight on wheels’ switches closed and ground spoiler deployment was initiated. Immediately following the first touchdown, pitch attitude reduced to 5° nose up and the aircraft became light on the oleos, momentarily rolling 4.5° right wing down and opening the left main gear ‘weight on wheels’ switch. During the low bounce, the application of 89% aft commander’s sidestick and 52% aft first officer’s sidestick resulted in the aircraft beginning to pitch up.

At the second touchdown, which was two and a half seconds after the first and during which a normal acceleration of 1.94g was recorded, pitch attitude was increasing rapidly through approximately 8° nose up. The first officer applied 27% forward sidestick and the commander’s sidestick was neutralised over the next second as the aircraft pitch attitude reached a maximum of 10.5° nose up. The aircraft’s roll attitude at that point was between 1° and 2° right wing down. During the initial stages of pitch attitude reduction (from 10.5° to 9.8° nose up), as the nose gear was lowered towards the runway, a normal acceleration spike of 1.3g was recorded. It is likely that this was the point that the tail came into contact with the runway surface.

The aircraft was slowed using reverse thrust before it vacated the runway. No evidence was found of any significant windshear during the landing although it was noted that the wind parameters recorded on the FDR indicated an almost pure crosswind from the left of 15 kt.

Weather

At 1529 hrs the weather at London Heathrow was given on the ATIS with Runway 27 Left in use for landing with wind shear forecast. The surface wind was 180°/16 kt, gusting 26 kt, visibility 5km in moderate rain with cloud, FEW at 1,100 feet, BKN at 1,300 feet and OVC at 1,800 feet. The temperature was +12° C and dew point+11° C, QNH 992 mb. The surface wind passed by ATC to D-AIRE with the landing clearance was 180°/13, gusting 27 kt.

Sidestick authority

Each pilot has a sidestick, which is used to control the aircraft manually in pitch and roll. The left seat pilot has the sidestick on the left side and the right hand seat pilot has the sidestick on the right. The sidesticks are not mechanically linked and do not both move together. They are spring loaded to the neutral position. When only one pilot operates the sidestick, it sends control signals to the flight control computers. If both pilots move their sidesticks simultaneously in the same or opposite direction and neither takes priority, the system adds the signals of both pilots algebraically. The total is limited to the signal that would result from the maximum deflection of a single sidestick. A green takeover light is located on the glare shield in front of each pilot that flashes when both pilots operate their sidesticks together. A ‘DUAL INPUT’ voice message is also given every 5 seconds as long as both pilots operate their sidesticks simultaneously.

A pilot can deactivate the other sidestick and take full control by pressing and keeping pressed the takeover button on the sidestick. For latching the priority condition, it is recommended to press the takeover button for more than 40 seconds, this allows that pilot to release the push button without losing priority. Either pilot can reactivate or deactivate the priority by momentarily pressing the takeover push button on either stick. If both pilots press their takeover buttons, the pilot that presses last gets priority. In a priority situation a red light comes on in front of the pilot whose stick is deactivated and a green light illuminates in front of the pilot who has taken control. This only occurs for as long as the non-priority sidestick is not in the neutral position.

Aircraft tail strike limits

Flight Crew Operating Manual (FCOM) Bulletin Number 22/2 dated April 1999, promulgated by the manufacturer, gives guidance for the avoidance of tail strikes. In the case of this incident where the landing gear oleo was fully compressed, a tail strike limit is given of 9.7° nose up aircraft attitude. The limit with the main landing gear wheels on the ground and landing gear oleo fully extended is 11.4° nose up aircraft attitude.

The landing technique requires a stabilised approach as essential to achieve consistent successful landings and it is described as imperative to reach the flare height at the appropriate speed and flight path which is given as -3°.

The flare should be started at approximately 20 feet and thrust reduction should be between 20 feet and 15 feet. It should be co-ordinated with pitch rate especially during cross winds. The attitude increase between the start of the flare and touchdown is given as approximately 4.5°, assuming the correct airspeed decay of 8 kt and the reduction in flight path of -1° (3 feet/second) is achieved.

Discussion

The crosswind at the time of the landing was within the aircraft operating limits. Although the crew had experienced some moderate turbulence in the early stages of the approach, they considered that after passing approximately 1,000 feet the turbulence reduced significantly. Whilst there had been no reports of windshear experienced by landing aircraft prior to the approach of D-AIRE, an A300, which was the seventh aircraft on the approach after D-AIRE, carried out a go-round due to windshear. The landing winds advised to the intervening six aircraft were similar to that passed to D-AIRE and were generally 180°/13kt, gusting 27. The A300, which carried out the go-round was advised of a wind of 190°/12 kt. Given that the anemometer display only records gusts of 10 kt or more above the mean wind speed over the previous two minutes it was not known what, if any, gusts were being experienced by the A300.

The approach flown by the FO was initially stabilised with auto thrust and autopilot selected, these were manually disconnected. The FO continued to fly the aircraft manually carrying out an ILS instrument approach to Runway 27 left. At 500 feet the aircraft descended below the glideslope, and this was corrected. The aircraft gradually regained the centre of the glidepath by 174 feet but then passed through it to become high. The FO had flown solely by reference to instruments down to 200 feet and then used a combination of visual and instrument cues referring to the ILS display until commencing the flare using visual reference. On passing through the glideslope the FO progressively increased the nose down pitch inputs in order to recover the normal approach path, which led to the high rate of descent and the ‘SINK RATE’ warning.

The commander considered that the change in aircraft attitude was subtle and, given the rain and visibility, he first became aware of the developing situation by noticing the apparent increase in the rate of descent which was confirmed on his PFD screen. Whilst he estimated this was at about 300 feet it was probably much lower at about 100 feet. At 55 feet the commander applied aft sidestick, which progressed to nearly full aft sidestick by a height of 10 feet, in order to arrest the rate of descent but he did not activate his sidestick takeover push button. The only voice warning recalled by both pilots was the Mode 1 ‘SINK RATE’ heard immediately after the commander had said it. The normal voice height and ‘RETARD’ calls were not heard during the final part of the landing.

Conclusions

Following the incident, the manufacturer’s technical experts agreed that when the aircraft had descended below the glide path the approach had become unstable in its final stage. There was evidence of turbulence during the latter stages of the approach and a variation of headwind component from -4 kt loss at about 500 feet followed by a 10 kt increase at 400 feet. Whilst the Calibrated Air Speed (CAS) shown on the FDR read out provided by the manufacturer averaged 144 kt in the last 400 feet, the air speed fluctuated by some 4 kt either side of this due to the gusty wind conditions.

The primary reason for the high rate of descent was the attempt by the FO to regain the normal glideslope when 1.8 dots above at 100 feet. Two dots above the glideslope represents a height difference of 26 feet from normal and, given the inertia of the aircraft, once the nose down pitching moment was initiated at such a low height, a correspondingly large movement aft of the sidestick would be required to arrest it.

As with other such incidents the commander could not see the control inputs of the FO and his first indication was a high rate of descent at about 40 feet shortly after the flare was initiated. He did not activate his sidestick takeover button and, given the circumstances, this action would not have prevented the tail of the aircraft contacting the runway. This occurred following the bounce on the second touch down when the commander was using aft sidestick to prevent the nose wheel coming down heavily.

Summary

The aircraft suffered a tailscrape whilst landing in gusty cross wind conditions. The latter stages of the approach included significant changes in pitch. The rate of descent resulting from a final pitch down correction applied by the FO could not be arrested in time, although the commander applied almost full aft sidestick.

A similar incident involving an A321, EI-CPE, occurred on 21 June 2000 at London Heathrow. A report on this incident containing a safety recommendation that is relevant to both occurrences is published elsewhere in this bulletin.


Published May 2001
Return to May 2001 Index
Return to Bulletins Index
Return to Air Accidents Investigation Branch Index
Return to DETR Aviation Index
Return to DETR Home Page
Web Site Terms

   http://www.aaib.detr.gov.uk/bulletin/may01/daire.htm

 

Interesting to note that Airbus are admitting a software error in the A320 heavy landing that led to the "err on the safe side" caution to crews approaching in gusty conditions (and otherwise risking the loss of urgent pitch control). The fugoid (phugoid) effect is the natural tendency to continually overshoot a rapidly selected pitch attitude selection and can lead to a divergent out-of-sync condition known as a PIO (pilot induced oscillation). PIO's can quickly lead to a divergent phugoid with destructive results (as in the MD-11's high-altitude upsets). In the F4 Phantom and early Sabre-jets the only real solution was to let the stick go.

So they (Airbus) had tried to follow their policy of eliminating pilot error by "programming out" any possibility of PIO's - but had inadvertently gone too far. Their software-imposed limit was limiting controllability in precisely the "near the ground" conditions where the pilot needs to cope with gusts and apply his reflexes instinctively, earning his money by avoiding precisely what happened to that aircraft. It's the Law of Unintended Direct Consequences. One has to wonder why it's taken so long for the circumstances to defeat the pilots (albeit in a series of incidents). Perhaps it's a case once again of the pilots becoming consciously (or subconsciously) attuned to the limitation and habitually ensuring that they avoided the threat area of limited controllability by always making early judicious inputs.

IASA Safety


Subject: Software error in A320

Refering to E.Jarabo request , Oct/10/01 , on the subject matter` follow up actions,this is translation of some points of the preliminary joint report :
The manufacturer admits that the unwanted activation of the high angle-of-attack protection law, in the landing phase, under turbulent conditions (in this case as in previous similar incidents) was due to the fact that the risk assumed in implementing the fugoid effect was underestimated.

The IBERIA A320 fleet uses flight controls computer type ELAC L80 with manufacturer' modification nº27-276.
Details can be obtained from a OIT and a FOT,Ref.AI/SE 999.0036/OI/CL,from the manufacturer, reflected on the official Telegraphic Aeronavigability Directive nºT2001 -106-B Flight Control-ELAC computer-(ATA2)

From the FAA Airworthy Directive AD 2001-08-26.

Revision of Airplane Flight Manual (AFM)
(a) Within 10 days after the effective date of this AD: Revise the
Limitations Section of the FAA-approved AFM to incorporate the following
procedures. This may be accomplished by inserting a copy of this AD into
the FAA-approved AFM.

"FOR APPROACH TO RUNWAYS WITH KNOWN GUSTY ENVIRONMENT, ESPECIALLY IF
THESE CONDITIONS GENERATE VERTICAL GUSTS DUE TO THE SURROUNDING TERRAIN,
OR

-REPORTED GUST WIND INCREMENT (MAX. WIND MINUS AVERAGE WIND) HIGHER THAN 10 KT,
OR

-EXPECTED MODERATE TO SEVERE TURBULENCE ON SHORT FINAL,

THE FLIGHT CREW SHOULD STRICTLY ADHERE TO THE FOLLOWING PROCEDURE:

-USE CONF 3 FOR APPROACH AND LANDING,

-MINIMUM VAPP IS VLS + 10 KT; THE RECOMMENDATION TO USE MANAGED SPEED
REMAINS VALID,

-CORRECT THE LANDING DISTANCE FOR THE SPEED INCREMENT,

-IF "SINK RATE" GPWS WARNING OCCURS BELOW 200 FT, IMMEDIATELY INITIATE A
GO AROUND."



"EXPECTED MODERATE TO SEVERE TURBULENCE ON SHORT FINAL,"
(see above) - I always found that it's the UNexpected severe turbulence on short finals that's the real bummer.
I'd tend to call the design airworthiness of the aircraft into question with this sort of limitation - particularly after so many incidents.