Dissection of a Hard Landing at London City
It was the morning of the 18th of August 2007 in London. The aircraft was HB-IYU, an Avro RJ100, the improved version of the British Aerospace 146 short-haul airliner, with four turbofan engines and a retractable tricycle landing gear (the most common landing gear for modern aircraft). The captain had over 9,000 flying hours, of which 1,340 were on type. The first officer was pilot monitoring.
The flight was a scheduled passenger service from Zurich to London City Airport. The weather forecast was for short periods of rain and southerly winds of 10 knots, so pretty much as good as it gets in London. The runway length required for the aircraft that day using a steep approach was 1,066 metres (3,500 feet). London City has a concrete runway of 1,500 metres (4,900 ft).
As the RJ100 approached London City using the ILS approach, everything was normal. At 3,000 feet, the flight crew initiated the glideslope capture, lowered the landing gear, set the flap to 33° and deployed the airbrake.
The captain disconnected the autopilot at 1,300 feet. As they came in, the flight crew monitored their approach using the PAPI lights and noted that they were slightly high.
The first officer was monitoring the airspeed relative to the target speed at which the aircraft should cross the runway threshold (VREF). As they passed through five hundred feet, he called out the relative speed: +7, +3, +1, +3 and +4.
These figures meant that they were below the target speed as they passed through 500 feet above ground level. Please see John’s comments below regarding the call outs, which at this stage appear to have shown them as coming slightly fast.
The Extended Ground Proximity Warning System (EGPWS) called out FIFTY: the altitude of the aircraft was 50 feet above ground level. The first officer called out “+2” and then “+1” right after the EGPWS announced 50 feet above ground level. The power levers were pulled back to idle.
As a part of the final approach, the Pilot Flying executes the “landing flare” where the thrust levers are at the idle position and the nose of the plane is raised slightly. This slows the descent rate and positions the aircraft in the proper attitude for the landing. For a tricycle landing gear, this means that the aircraft touches down on the main landing gear with the nose wheel touching down after the aircraft is stable and reduced in speed. For a steep approach, which is required for London City, the pitch attitude of the aircraft would be around 4°. The flare height for an aircraft of that size is around 30 feet, so the Pilot Flying was either flaring or about to flare.
The Automatic Terminal Information Service (ATIS) at the time of the landing cited the wind as “surface wind from 190° at 11 knots” however, an aircraft which landed twenty minutes later said that he found the approach much more turbulent and difficult than he expected.
Just before the EGPWS call for 30 feet, the flight crew felt the aircraft suddenly drop. The captain immediately responded by pulling back on the control column to avoid a hard landing.
As a result of the increased pitch, the rear of the aircraft touched the runway. The flight crew were not aware that they’d suffered a tail strike and were not initially clear on what had destablised the aircraft.
Here’s the video taken by a plane spotter filming at London City.
An inspection of the runway showed that the first contact was the rear galley drain pipe, which left a five-metre long scrape starting at the PAPIs and just to the right of the runway centreline. Then the lower rear fuselage scaped the runway for another eleven metres. Based on the touchdown speed of 113 knots, the rear section of the aircraft was in contact with the runway for 0.24 seconds. It was enough.
The aircraft came to a safe stop. There were no injuries.
The flight data recorder showed that there was a slight headwind until approximately 50 feet above ground level, at which point it became a variable and slight tailwind. It only captures the information every four seconds but the data did not appear to show evidence of gusty conditions. The aircraft was descending at 900 feet per minute and reducing smoothly. The flight data recorder did not show any sudden drop in altitude.
This was not the first Avro RJ100 to suffer a tail strike at London City. The manufacturer carried out an investigation after this incident and determined the following key factors.
- The aircraft speed was below the final landing approach airspeed target (VREF), which meant the final approach required a high angle of attack.
- A high rate of descent which meant that a higher pitch attitude was required in the flare.
- Excess speed leading to the aircraft floating and a high pitch attitude on touchdown.
The accident investigation looked at these factors and that the tail strike events are clearly a result of various circumstances which lead to excessive pitch attitudes at touchdown. They concluded that for a successful steep approach for a relatively short runway such as the runway at London City, a high degree of accuracy is required.
Although the aircraft was stable and had made a good approach, the call outs from the Pilot Monitoring made it clear that the aircraft was generally below the target speed by one to five knots. The Flight Data Recorder confirmed this, showing the aircraft at four knots below target speed at 50 feet and 35 feet above ground level. As the thrust levels were pulled back to idle, the airspeed decreased again caused by the shift of the headwind to a tailwind.
These minor changes all reduced the energy of the aircraft. This was the apparent drop that the pilots felt. The commander pulled back to avoid the hard landing, which brought the pitch attitude to 9.3° as they touched down, steep enough to cause the tail to strike the runway.
From the report:
Another operator of this aircraft type, who had previously experienced several tailstrikes at LCY, introduced revised training and procedures for their pilots. One element of this was to introduce an SOP monitoring call of ‘ATTITUDE’ if a pitch angle of 5º or greater is seen during the flare. If this call is made, then the pilot flying must not increase pitch but is required either to accept the pitch attitude for landing or to go around.
Since this accident the operator has undertaken a re‑assessment of the risk level of its operations into LCY. A further review of procedures and training requirements for LCY has also been completed. Some changes to SOPs have been implemented and an additional training programme for LCY has been incorporated into the recurrent simulator schedule.
In other words, this mistake was way too easy to make in that aircraft at that airfield.
In my days, a call like “+ 7” etc, would tell the PF (the handling pilot) that (s)he would have been ABOVE Vref, not below. Another company with other cockpit drills and call-outs perhaps? A friend of mine once said that after his company (British United) had been subject to take-over by BA, he had to re-learn flying the BAC 1-11 all over again, so different were the SOP’s.
Anyway, a steep approach into a runway surrounded by built-up areas does have an increased risk factor and requires additional procedures and training. It also requires performance criteria that will enable an aircraft to avoid the obstacles in the even of N-1, an engine failure. Hence the AVRO RJ with 4 engines (once nicknamed four vacuum cleaners flying in close formation). The Swiss called it the ‘Jumbolino’, a relatively small aircraft but with four engines. They nicknamed the SAAB 2000, a fast turboprop, the ‘Concordino’
There is no reason to blame the crew for the incident. High obstacles in close vicinity can cause turbulence and unexpected downdrafts. The video, obviously taken with a long lens, shows a distorted view of the landing. But nevertheless, it is apparent that there was a substantial crosswind. The main thing is: only pride was hurt. Damage can be repaired.
The Fokker F27 had a tailskid, but in case of a tail strike the skid would be pushed into the fuselage. It would compromise the pressurized section and the damage would be considerable.
Anyway, in the scenario described, in my opinion no pilot could say that under the same conditions – and with the same training – it would not have happened to him or her.
I’m now a bit confused about the calls. The report says:
“The co-pilot made a number of calls in the latter stages of the approach with reference to the target VREF. These indicated that the aircraft was generally below the target speed and this is confirmed by the recorded data.”
Both you and John have made the point that his calls are signalling that they are too fast. I will read the report again tomorrow and see if I can make sense of it.
The video shows it much more vividly than the write-up. He lands so hard the wings are flapping. He bounces two or three times, lands on the nose gear once, and spends a bit of time sliding down the runway sideways (on two sets of tires). I imagine the passengers were doing some serious praying during that landing.
The video was taken with a (very) long lens which distorts the picture.
Yes, it obviously was not a smooth landing but the wings bending is not really that abnormal, taking into account the weight of 4 engines attached to them. In order to make an accurate assessment, it is necessary to have a transcript of the CVR and the ATC conversations. E.g. what was the wind strength? What was the crosswind component? Did the crew get a proper read-out? Did they exceed the crosswind limitations? Was there a wind shear? Considering the obstacles (buildings) in the vicinity, in my opinion a wind shear was very probable. Once things do get wrong, they tend to get nasty very quickly. The crew nearly lost control, that much can be seen. A result of an unexpected wind shear? Or the result of the hard landing? Many questions and the outcome will be different each time, and also each time the sequence changes: What came first, the incident? Or was the near crash the result of the crew losing control?
Without access to data that rightfully are for the investigators, it is not possible to give an adequate assessment, other than that it looked scary.
I give the crew the benefit of the doubt and leave the conclusions to the experts.
Looking at the video again, still the same: a windshear possibly resulting in a drop of airspeed just before the flare. The sink rate may have increased a bit too. At such a critical moment there is hardly time to take the crab off, but the aircraft should be able to take that. But to arrest the sink rate the PF overcompensated the flare, resulting in the tail strike. And yes the aircraft bounced and the nosewheel appeared to hit the deck a few times before the mains. The wings flexing does not look excessive. The long lens is distorting the picture, making the crab look more dramatic and the runway seem shorter. Okay, this is my interpretation. This MUST NOT be seen as an attempt to come to a (premature) conclusion; that, I repeat, is for the investigators. Nobody was injured, that is the main thing.
I did a similar thing in Microsoft Flight Simulator once! I was landing a Rilo Airvan (a funky looking aircraft that doesn’t exist in real life!) at North Weald when I experienced a sudden updraft (I always have real weather turned on) that must have taken me about 50ft into the air! I should have pushed the throttles up and went around, but I figured that I could still make the landing so I raised the nose and stalled! I hit the ground to hard and smashed the main gear, skidding off the runway! I was actually surprised at how realistically the aircraft behaved!
I suppose it’s better that I learn that lesson in a game than end up doing it in real life! 0.0
Wise words. I looked at the video a few times and still think that the pilot was taken by surprise by a downdraft just as he started the flare. In my book, the PNF will read out the speed relative to the Vref. The transcript makes little sense: a call like ‘plus five’ to me would mean five knots above the speed, NOT below. This also could explain why the PF thought that he had a little bit of extra speed in hand. When looking at the video, it is visibly windy. The definition of Vref is 1.2 x Vstall, so the margin is small in the best of scenarios. In gusty wind conditions, it is normal to add a few kts to Vref ‘for wife and children’.
But London City Airport is a special case: the runway is not a long one, the airport is surrounded by buildings and the final approach is steep.
Meaning that the rate of descent is also a bit higher than normal. The extra few knots will bleed off surprisingly quickly. As the nose goes up, drag increases exponentially and the speed decreases correspondingly.
All factors combined will pose a higher risk of a hard landing. The steeper approach in itself causes a greater downward momentum, so there is more energy required to arrest a sink rate.
The buildings themselves would have caused low level turbulence.
My guess remains the same: the normal reaction of any pilot, even an experienced one, to arrest a sudden increase in sink rate just during the flare would be to hold off a bit more. It usually works but not in this case.
What is the big deal anyway? These things can happen. The lawyers will have days if not weeks to consider the outcome of the reports and to blame the crew. The pilot had but a split second to make his decision.
Nobody was injured, the aircraft no doubt went back into service after repairs. The airline will no doubt add a few lines to the AOM and the SOP’s no doubt will be amended. The whole she-bang will no doubt be programmed into the simulators and chances that it will happen again will no doubt be reduced correspondingly.
The official report gives some more information.
There is an indication that the increase in rate of descent in the last 50 ft was most likely or at least partly caused by shear from head to tailwind (according to the report).
The call out +7, +5 etc refer to speeds above VREF.
It is an interesting read. Many factors contributed to the tailstrike.
I’ll look at it again tomorrow when I have more time, but I understood the report to be making the point that they were incoming below VREF.
Edit: Actually, as per my response to Rudy above, this is what I was reacting to:
The co-pilot made a number of calls in the latter stages of the approach with reference to the target VREF. These indicated that the aircraft was generally
below the target speed and this is confirmed by the recorded data.
But dinner is burning so I haven’t had a chance to unravel it yet.
Thank you for linking the report; I thought I had done so and am embarrassed that I didn’t cite my references!
It appears that we are both correct.
Unfortunately there is no CVR with a good time line available but looking at the FDR data the airplane appears to be above VREF till about 50 ft AGL and about 3 seconds before touchdown. Up untill then it is likely that the PNF had made call outs of being too fast (the +7,+5 call outs ) during gusts which increased the IAS. This appears to be followed, after decreasing call outs (+2,+1) by a shift to tailwind around 50 ft altitude.
I assume (dangereous, I know), that the remark in the report, as mentioned by you above, refers to the last part of the approach where the airspeed decreases below VREF (ie below 50 ft).
I assume, (dangereous, I know) that the PNF would have made some additional call outs at that time when he saw IAS decreasing. Unfortunately the CVR is absent.
In my mind the FDR reads like a classic windshear trap;
First a performance increasing shear, resulting in too high an IAS, accompanied by the call outs from the PNF.
Likely, the autothrottle, or the pilot, might have reduced thrust to compensate for the high IAS. Then, at 50 ft, the shift to tailwind with a rapid decrease in IAS resulting in the too high a rate of descent with the tailstrike as a result.
The last 50 ft were likely to be accompanied by extra callouts of the copilot as mentioned in the sentence of the report you mentioned.
Unfortunately none of the CVR data (callouts) nor the throttle settings are provided against the FDR data so all this remains a bit vague.
Also worth mentioning is that London City Airport does not have any windshear detecting systems (lidar or other). The airplane might have doppler radar but as far as I know requires some form of droplets in the air to give a reliable windshear warning.
The windshear might have been all but invisible to the crew. This, combined with the steep approach required, and the short runway, appear to have accumulated in this scrape.
The quick reaction of the Pilot, trying to raise the nose at the last moment, probably saved the airplane from an even harder landing..
Everybody walked away, which is always a good result.
Thank you for the detailed analysis – I think your assumptions must be right, they certainly make more sense, and I misinterpreted the report. I’ve added a note in the post pointing to your comment.
Your analysis is excellent and a very good summing up of the events as they probably unfolded in the cockpit.
There are a few uncertainties to consider. We do not know for certain if the aircraft was caught in the “windshear trap”, but it is very likely.
In such a situation, a few knots extra are lost literally “in the blink of an eye” and the aircraft may well have been below Vref just as the pilot had started the flare. But I think that we all, given the limited information, have come to the same conclusion.
Do some more research before you post: the aircraft on the video is not HB IYU. The one on the vid did not suffer a tail strike but “only” a (very)hard landing!
And yes, IYU flew again after repair, I actually performed the flight LCY back home.