Aircraft
Type and Registration: Boeing 747-412, B-KAG
No &
Type of Engines: 4 Pratt & Whitney PW4056 turbofan engines
Year of
Manufacture: 1992
Date
& Time (UTC): 1 March 2008 at 0128 hrs
Location:
Manchester Airport
Type of
Flight: Commercial Air Transport (Cargo)
Persons
on Board: Crew - 2 Passengers - None
Injuries:
Crew - None Passengers - N/A
Nature of
Damage: Damage to Nos. 1, 2 and 4 engine nacelles, one main landing gear
tyre ruptured
Commander’s
Licence: Airline Transport Pilot’s Licence
Commander’s
Age: 48 years
Commander’s
Flying Experience: 10,800 hours (of which 699 were on type)
Last 90 days - 188 hours
Last 28 days - 55 hours
The aircraft was landing at
Manchester Airport on Runway 23R at the end of a scheduled cargo flight from
Dubai. There was a strong wind from the north-west and a number of aircraft had
earlier diverted to other aerodromes.
During final approach, the crew
received a windshear warning when the aircraft was at 500 ft agl. They carried
out a missed approach and were given radar vectors for another ILS approach.
The second approach was described as smoother but still with a strong wind from
the north-west, resulting in a crosswind from the right which was close to the
operator’s limit for landing this aircraft. During the ensuing touchdown the
aircraft
rolled right and the No 4 engine nacelle made contact
with the runway surface. The aircraft then rolled left and Nos 1 and 2 engine
nacelles also made contact with the runway and the No 2 tyre on the left main
landing gear burst. There were no abnormal indications on the engine
instruments and, after an external safety check by the Airport Fire fighting
and Rescue Service, the aircraft taxied on to a stand.
History of the flight
During the pre-flight briefing
for their scheduled cargo service from Dubai to Manchester, the flight crew
noted that there were strong winds forecast throughout northern Europe. In particular,
the forecast for Manchester Airport predicted crosswinds which would be outside
the co-pilot’s 20 kt limit at their estimated time of arrival. Consequently, it
was agreed that the commander would act as the pilot flying (PF). Also, in view
of the weather forecast, the crew uplifted an extra 2 tonnes of fuel in
anticipation of a potential diversion, should that be necessary -‘two
approaches before diverting’ being the company standard procedure.
The flight was uneventful, apart
from a minor problem with the transponder, and, en route, the crew monitored
the weather at their destination and potential alternates. Because of the
possibility that they might not be able to land at Manchester, the crew
contacted their company during the flight to confirm the preferred priority of
alternate destinations. They were advised that Nottingham East Midlands was the
number one alternate, with London Heathrow as the second.
As they neared Manchester, the
crew noted that the Airport’s Aeronautical Terminal Information Service (ATIS)
arrival information included a warning that moderate to severe turbulence had
been reported on the approach to Runway 23R at a range of 16 nm to 10 nm. They
did not experience this and established the aircraft on a Category 1 ILS approach
to Runway 23R with 30° of flap selected. During the course of the approach the
aircraft’s airspeed increased above the flap limiting speed and the aircraft’s
flap load relief system automatically reduced the flap setting to 25°, until
the airspeed had reduced sufficiently for flap 30° to be redeployed
automatically. Without delay they carried out a missed approach. (The commander
later commented that this was the first windshear warning that she had
experienced as a result of actual conditions.) The crew advised ATC and
requested radar vectors for a second ILS approach to the same runway, for which
the Decision Altitude was 450 ft amsl (airport elevation 257 ft). They
rebriefed and decided to carry out the second approach with 25° of flap
selected, a permitted setting, because of the speed fluctuations and windshear
during the first approach.
The second approach was described
as being smoother. Approaching a range of 11 nm from the runway, ATC advised
the crew that the aircraft ahead had landed successfully. It was established
subsequently that this was an Airbus A321 belonging to another operator. B-KAG
was fully configured for landing and stabilised on the ILS glideslope and
localiser by 1,500 ft aal, and the landing checklist was completed by 1,000 ft
aal.
During B-KAG’s final approach ATC
transmitted a number of surface wind readings. When the aircraft was cleared to
land, the crew were advised that the surface wind was 280°/20 KT
MAXIMUM 36 KT. Following that, ATC transmitted three instantaneous surface
wind readings; 280°/37 kt, 300°/31 kt and 290°/31kt. The aircraft again
produced a “FIVE HUNDRED” aural
annunciation, followed by another saying “ONE HUNDRED” as the commander disengaged the autopilot and disconnected
the autothrust. The co-pilot advised the commander that the aircraft was at the
correct approach speed with a 700 fpm rate of descent and ATC transmitted the
final instantaneous wind. The aural alert “MINIMUMS” was emitted on the flight deck, confirmed by the co-pilot,
and the commander called “land”. In quick succession, the
co-pilot advised her that the rate of descent was 900 fpm, an aural “SINK RATE” warning
was generated twice, backed up by the same call from the co‑pilot, a “TEN” [feet]
aural alert sounded and the aircraft landed. Later, the commander recalled
seeing a rate of descent of 1,100 fpm on the flight instruments at the time of
the ‘sink rate’ warning.
During the rollout, ATC advised
the crew that sparks had been seen coming from both wings and that the wings,
or some of the engines, had touched the runway during the landing. This was
confirmed by a member of the airfield operations staff, who was in a vehicle
positioned adjacent to the touchdown zone on Runway 23R. The rollout was
completed without further incident and the commander checked that the engine
instruments were indicating normally before taxiing the aircraft clear of the
runway. She then stopped B-KAG on a taxiway and the Airport Fire fighting and
Rescue Service (AFRS) attended the aircraft to carry out an external
inspection. They reported damage to the Nos 1 and 4 engine cowlings but no
signs of any fuel or hydraulic fluid leak. With the agreement of the AFRS, the
commander taxied the aircraft slowly on to a stand and the crew shut B-KAG
down.
Meteorology
At the time of the occurrence, an
area of low pressure lay to the north-east of the United Kingdom resulting in
strong to gale force north-westerly winds across north‑west England.
The Terminal Area Forecast (TAF)
for Manchester Airport for the period from 2200 hrs on 29 February to 0700 hrs
on 1 March 2008 forecast a surface wind from 260° at 22 kt, gusting to 35 kt,
with visibility in excess of 10 kilometres and scattered cloud at 2,000 ft agl.
Temporarily during the period from 2200 hrs to 0300 hrs the surface wind was
forecast to veer to 270° and increase to 28 kt, with gusts to 45 kt. There was
also a 30% probability of a temporary change between 2200 hrs and 0700 hrs when
the visibility would reduce to 7,000 metres in showers of rain and there would be
broken cloud at 1,200 ft agl.
The ATIS arrival information for
Manchester Airport at 0050 hrs reported the surface wind as 280°/25 gusting 42
kt, visibility greater than 10 kilometres, few cloud at 2,800 ft agl, scattered
cloud at 3,200 ft agl and a wet runway. This arrival information also included
a report of moderate to severe turbulence on the approach to Runway 23R between
16 nm and 10 nm. The next ATIS arrival information, timed at 0120 hrs, when
B-KAG was making its second approach to land, reported the surface wind as
being 300°/28 gusting 42 kt. The visibility was still greater than 10 km, there
were few clouds at 3,200 ft agl and the runway continued to be wet. The QNH
pressure setting was 994 mb.
The UK Air Pilot entry for
Manchester Airport contains the following warning:
‘Pilots
are warned, when landing on Runway 23R in strong north westerly winds, of the
possibility of turbulence and large windshear effects.’
Between 0020 hrs and 0220 hrs,
the surface wind at Nottingham East Midlands Airport was blowing down the
runway at comparable speeds. At London Heathrow Airport a similar situation
existed. At both, the visibility and cloud base were suitable for an approach
and landing.
Other aircraft
Between 2115 hrs and 0120 hrs on
the same night, ATC logged 10 aircraft, including B-KAG, which carried out
go-arounds due to windshear, turbulence or the strength of the crosswind. Of
these, five diverted to other aerodromes.
Aircraft loading
The landing weight of the
aircraft was 257.8 tonnes, which was less than the maximum landing weight of
295.7 tonnes. The centre of gravity (CG) was at 21.1% Mean Aerodynamic Chord
(MAC), towards the centre of the permitted range.
Personnel
Earlier in the day, the commander
had flown the aircraft from Hong Kong to Dubai: a Flight Duty Period (FDP) of 9
hours 35 minutes. She then had 11 hours rest, the minimum permitted, during
which she slept for 7 to 7½ hours, before reporting for duty and meeting up
with the co-pilot. The co-pilot had operated another aircraft from Manchester
to Dubai, via Amsterdam. He then had 11 hours 30 minutes rest, achieving about
5 hours sleep. Prior to reporting for duty in Hong Kong for her first sector to
Dubai, the commander had had six days off duty.
Neither flight crew reported
being fatigued during the two approaches to Manchester, nor was there evidence
that it was a factor in the occurrence.
Runway examination
Examination of Runway 23R
revealed a number of scrape marks, indicative of nacelle contact, and several
tyre track marks that, by virtue of their relative locations and by comparison
with the aircraft damage, could be matched to B‑KAG’s landing (Figure 4).
Other tyre marks from B‑KAG may
have been present but hidden by the dark-coloured asphalt forming the central
portion of the runway surface and by multiple heavy tyre smudges present in the
touchdown area. The following distances, along the runway from the threshold
and left or right of the centreline, are approximate.
The initial mark, from the right
body MLG tyres, started around 193 m from the threshold. This was almost
immediately followed by a short scrape mark, 6.1 m right of the runway
centreline, caused by momentary contact of the No 4 nacelle, and by the start
of tracks from the left body MLG tyres. The markings indicated that the
aircraft Centre of Gravity (CG) was around 15 m left of the runway centreline
at initial touchdown, with the aircraft tracking approximately 4° right of the
runway heading. Tracks from the left wing MLG started at 226 m.
The MLG tyre tracks continued,
curving to the left. A short scrape mark from the No 1 nacelle started at 359 m
and two short scrape marks from the No 2 nacelle started at 375 m. The tyre
tracks showed that by around 500 m B‑KAG was tracking parallel to the runway,
approximately 8 m left of the centreline.
Aircraft Examination
Examination of the aircraft
showed that the underside of No 1, 2 and 4 nacelles had sustained scraping and
abrasion damage, consistent with having contacted the runway. The most severe
damage was to the No 4 nacelle, where the bottom of the nose cowl, fan cowl and
translating cowl had been heavily abraded and locally deformed. Distortion of
the translating cowl also resulted in minor damage to the HP gearbox driveshaft
cover installed on the engine. In the case of the No 1 nacelle, the bottom of
the rear part of the nose cowl and the forward part of the fan cowl had been
scuffed. The No 2 nacelle suffered locally heavy abrasion damage in the same
areas as the No 1.
The direction of the scrape marks
indicated that the aircraft had been heading approximately 7° right of its
track when the No 4 nacelle contacted the runway.
No evidence of anomalies with the
landing gears was found, with the exception of rupture of the No 2 tyre
(forward right tyre of the left wing MLG). Tyre pressures measured some hours
after the incident were measured at 190‑195 psig for all MLG tyres (normally
200‑205 psig), except for the ruptured No 2 tyre. MLG shock strut pressure
readings were within limits.
The No 2 tyre rupture resulted
from a ‘flat’ worn completely through the carcass, indicative of the wheel
having been locked during part of the ground run. No faults with the
wheelbraking system were apparent and none was recorded on the aircraft’s
central maintenance computer (CMC) when the brake control system was tested
using its built-in test equipment (BITE). However, an operational check by the
operator found that the No 2 brake failed to release to prevent the wheel from
locking. Testing revealed that the No 2 anti-skid control valve (PN 39‑617, SN
059627620) was inoperative, and that the valve’s electrical insulation
resistance was below the minimum specified. The type of valve is not ‘lifed’
and the operator was unable to establish its time in service. While the brake
system BITE includes an integrity check of the servo coil in the anti-skid
control valve, it would apparently not detect all faults that might render the
valve inoperative.
A Boeing Fleet Team Digest (FTD)
(No 747‑440‑FTD‑32‑04009, issued 24 September 2004) described reports from
operators of water ingress into Anti-Skid Control Valves, resulting in
corrosion of several parts of the assembly. The FTD noted that in two cases a
tyre skid-through had resulted. In all cases the problem had concerned wing MLG
valves. These are installed facing down and appeared to be more susceptible to
water ingress than the body MLG and alternate anti-skid valves. The valve
manufacturer, Crane Hydro-Aire, had issued a Service Information Letter (SIL)
(No 39‑617‑3‑12, Revision 1 of 1 July 1994) recommending adding RTV106 sealant
at all external interfaces of the servo valve body and cover. A further SIL (No
39‑617‑2‑14, Revision 1 of 16 December 1994) provided for improvements in the
valve cover. The FTD noted that even with this configuration it was possible
that a water ingress path was present in the connector area.
Flight recorders
Recordings were recovered from the flight data recorder (FDR), a
2-hour cockpit voice recorder (CVR) and the enhanced ground proximity warning
system (EGPWS). The following is derived from these recordings and reported in
UTC.
The CVR recordings indicated that throughout the flight the crew
were communicating fully with each other, discussing the situation and
observing procedures, briefings and check lists in a professional manner.
The aircraft took off from Dubai at 1728 hrs on 29 February 2008.
The aircraft climbed to cruise at FL340 and then FL360, with a short period at
FL380. The descent started at 0049 hrs. During the descent the crew discussed
the high wind conditions enroute, at the destination and at alternate airports
with reference to the aircraft limits. Whilst enroute, it was stated that the
destination airport was at the time “right on the limits” and the others were
within limits. Windshear and go-around briefings were made. Emphasis was given
to checking the ATIS and ensuring updated surface wind conditions were
received. Passing through 9,000 ft, the
surface wind of 290°/30 kt gusting between 31 and 41 kt was reported. This was
highlighted by the crew as out of the operator’s limits. The decision was made
to continue the descent and prepare to go around if the surface wind remained
out of limits.
The final descent on the first approach to
Runway 23R commenced from 3,500 ft amsl with the three-channel autopilot
tracking the glideslope and localizer and with the autothrottle speed mode engaged.
The aircraft was flown with 20° of drift to maintain track, reducing with
altitude. The flaps were progressively deployed, with flap30 selected at 1,460
ft amsl and 3.3 nm to the threshold. Just prior to the automatic “one thousand”
altitude callout, the surface wind was reported as 290°/24 kt gusting 42 kt
which was described as being “justin”. At 490 ft agl and 1.3 nm to go, a
windshear warning was triggered by the EGPWS. The autopilot and autothrottle
switched to go-around modes, flaps were reduced, gear raised and the crew flew
a go-around. Prior to the windshear warning, the drift had reduced to 13°; this
changed to 9° in the following 3 seconds.
The aircraft climbed to 3,500 ft amsl, and then 4,000 ft amsl, and
was vectored for a second approach. During the go-around the crew elected to
make a further attempt to land at Manchester, this time with flap25. As the
aircraft captured the localizer the wind was reported as 280°/23 kt gusting
between 14 and 36 kt and the crew were informed that the aircraft ahead had
landed.
By 10.5 DME, and descending through 3,700 ft amsl, the three
autopilot channels were fully coupled to the glideslope and localizer and the
autothrust was in speed mode. By 6 DME and 2,200 ft amsl the aircraft was fully
configured with gear down and flap25. At this point the aircraft was flying
with 16° of drift and reducing. Subsequent wind checks received were 290°/24 kt
gusting 36 kt, 290°/23 kt gusting 36 kt and 280°/20 kt gusting 36 kt. At this
point the aircraft was cleared to land and the 1,000 ft automatic callout was
triggered.
The next call was 280°/21 kt gusting 36 kt. This was followed 13
seconds later by the automatic 500 ft callout .
The autopilot and autothrottle systems were disconnected at
approximately 270 ft agl and 220 ft agl, respectively. During this period “minimums”
was called and the captain responded with “land”. At this point the aircraft
was slightly to the right of the centreline and rolling left to recover the
centreline. Right pedal and roll inputs were made, checking the left bank and
motion at the same time that the wind speed started to ramp up and change
direction so it was mostly from the right of the aircraft. The aircraft started
yawing right and rolling right. Left control wheel and rudder inputs were made,
slowing the rate of roll to the right but not stopping it before touchdown.
At touchdown, the aircraft was to the left of the centreline and
had a recorded right roll of 9.7°. This exceeded the nominal ‘pod scrape’ roll
limit for the recorded pitch angle. At the point of touchdown only a small
Normal acceleration was measured. However, it is worthy of note that the sensor
position is inappropriate for measuring forces at the wing.
The aircraft rolled left and the maximum recorded left roll was
6.7°. This did not exceed the nominal 7.5° pod scrape roll limit for the
recorded pitch angle but the peak value of roll may not have been recorded and
the pod scrape limit would not account for dynamic flexing of the wing. The
rollout then stabilised. From the recorded data, it is clear that the aircraft
was being subjected to a strong crosswind component, with gusts, and a change
in wind direction and speed occurred just prior to touchdown.
An EGPWS sink rate warning was
triggered just prior to touchdown and the sink rate was reduced prior to ground
contact and only a relatively small normal acceleration was measured at
touchdown. The sink rate warning issued by the EGPWS inhibited the majority of altitude
callouts in the final stages of the landing.
The aircraft roll at touchdown
was sufficiently large to indicate a pod strike on the right engine. The data
pertaining to the subsequent left roll did not indicate a pod strike on this
side but the limits do not account for flexing and the sample rate of the roll
parameter is likely to have missed the peak left roll.
The aircraft manufacturer was
provided with flight data and asked for analysis, principally concerning
whether the aircraft responded correctly to the crew’s control inputs although
the sample rate of the wind direction parameter, once every four seconds, was
insufficient to assess dynamic wind conditions. The manufacturer used
simulation tools that calculated the wind conditions based on the recorded
parameters, which improved the understanding of the wind conditions. The
calculated wind showed large shifts in direction and magnitude just prior to
touchdown.
The simulations and modelling
indicated that the aircraft was responding correctly to crew control inputs and
that the inputs were appropriate, although their magnitude may have contributed
to the likelihood of a nacelle contact.
Discussion - OPERATIONS
As forecast, the crosswind at
Manchester Airport was outside the co-pilot’s limit when B-KAG made its
approach to land. However, the cloudbase was above 3,000 ft agl and the
visibility was good. Having gone around from the first approach, following a
windshear warning, the crew carried out a second approach. During it, they were
advised by ATC that the aircraft ahead of them had landed. The aircraft in
question was a different type and there was, therefore, no reason to link the
success of its landing with the one B-KAG was about to make.
B-KAG’s second approach, using a
flap setting of 25° for the landing, appears to have been stable until the
autothrust was disconnected at a height of 220 ft agl. The aircraft then
drifted above the glideslope, before descending through it, during the course
of which B-KAG lost 20 kt of airspeed in one second and then gained 23 kt in
the next four seconds. The pitch attitude reduced from 2° above the horizon to
0° and then increased back to 4° nose-up for the landing. The rate of descent
increased to a maximum of approximately 1,400 ft/min, resulting in a ‘sink
rate’ warning below a height of 50 ft aal, but then reduced to about 300 fpm at
touchdown. The aircraft landed at an airspeed of 163 kt.
When the aircraft rose above the
glideslope, it also started to drift to the right of the localiser. The
correction back on to the localiser coincided with the increase in rate of
descent. The aircraft continued through the localiser, to the left of the
extended runway centreline, and right roll was applied, reaching a maximum of
9.7° at touchdown. It was at that point that the underneath of No 4 engine
nacelle struck the runway. The aircraft then rolled left, to a degree that was
less than that required for a static aircraft to suffer ground contact with an
engine. However, the dynamic behaviour of the wing probably accounted for the
flexing that enabled the Nos 1 and 2 nacelles to touch the runway surface.
Following the clearance to land,
the magnitude of the crosswind, as reported, reached a maximum of 28 kt. That
was within the operator’s specified crosswind limit for the commander. However,
evidence indicates that there was significant variation in the strength and,
possibly, the direction of the wind experienced by the aircraft. This was
commensurate with the warning contained in the UK AIP entry for Manchester
Airport, regarding the possibility of turbulence and large windshear effects
when landing on Runway 23R in strong north westerly winds. It seems that the
conditions were as challenging as any the commander had experienced since converting
on to the B747.
An option existed for the
commander to initiate a go‑around and divert to an alternate destination where
the surface wind conditions were more favourable for a landing. However, the
conditions at Manchester Airport, the planned destination, were within the
aircraft’s and the commander’s limits and it was only on landing that the
aircraft rolled sufficiently for the nacelles to strike the ground, the earlier
high rate of descent having been corrected to an acceptable level.
Although the commander employed
the ‘Touchdown In Crab’ method, the aircraft had drifted from the right of the
localiser to be 15 metres to the left of the runway centre line at touchdown.
The roll to the right, which had developed just prior to touchdown, was countered
with substantial left control wheel and left rudder pedal inputs. The
combination of these, and the reactive forces on the aircraft during the
landing, resulted in the subsequent roll to the left.
In this serious incident the
landing conditions were challenging and very close to, or at, the crosswind
limit for this operator. It appears from the recorded data that the last stage
of the approach to land coincided with a change in wind direction and speed.
There is no evidence that the flight crew were fatigued.
No comments:
Post a Comment