E50P, Berlin Schönefeld Germany, 2013
E50P, Berlin Schönefeld Germany, 2013
On 15 February 2013, an Embraer EMB-500 Phenom 100 crew lost control of their aircraft shortly before touchdown at Berlin Schönefeld when it stalled and crash-landed. The Investigation was not completed for almost six years but concluded that the stall was a result of ice accretion during an approach in icing conditions without activation of the airframe de icing system. It found poor crew awareness of both the ice and stall protection systems and, suspecting that this may be true of other type-rated pilots, accordingly made Safety Recommendations to key regulatory authorities concerning the type rating syllabus.
On 15 February 2013, the crew of an Embraer EMB-500 Phenom 100 (OO-MAS) being operated by Abelag Aviation on a non-scheduled flight from Kortrijk-Wevelgem to Berlin Schönefeld with one passenger on board lost control of the aircraft just before touchdown at destination in day VMC and it crashed on the edge of the runway ending up just off it. None of the three occupants were injured but the aircraft was severely damaged and damage was caused to both the runway surface and runway lighting.
An Investigation was carried out by the German Federal Bureau of Aircraft Accident Investigation (BFU). The aircraft was not fitted with an FDR or a CVR and was not required to be but it did have a CVDR (Cockpit Voice and Data Recorder) installed from which relevant voice and flight data was recovered. In the case of the voice recording which contained a mixture of English and Dutch, an Investigator from the Belgian AAIU assisted with transcription into English where necessary (direct reference to any Dutch language exchanges quoted below will appear as translated into English). ATC radar data for the approach was also available.
Both pilots held CPLs issued by the Belgian CAA. The 48 year-old Captain had a total of approximately 4,500 flying hours which included 800 hours and about 400 landings on type. The 22 year-old First Officer, who had been PF for the accident flight, stated that she had a total of approximately 260 flying hours which included 32 hours and about 20 landings on type.
It was established that from about 45 minutes after takeoff, whilst still in the cruise at FL 390, the crew had tried several times to copy the destination ATIS but had not been successful. Ten minutes later, when approximately 140 nm from destination they began descent and two minutes later, the Captain advised that he considered it likely that runway 25R would be in use. However, a few minutes later, they were cleared for a runway 07L STAR and on subsequent transfer to Bremen Radar about 70 nm from destination, were re-cleared to fly direct to a twelve mile final for runway 07L at Schönefeld.
Passing FL 200 the PF asked for the Descent Checks but was told by the Captain to wait whilst he copied the ATIS which included a cloud base of 1,400 feet, visibility 4,800 metres in mist with “moderate icing reported below 3,000 feet”. The two pilots “then conducted a two-minute approach briefing” during which details of the approach and go-around procedures were discussed. As the aircraft left FL 100, the First Officer again asked for the Descent Checks but was initially asked to “stand by” until after receipt of a re-clearance to descend to 4000 feet QNH, when they were completed. The Captain then said "landing speeds?" and the First Officer replied “set by me" (a VREF of 96 KIAS was subsequently found to have been written on the operational flight plan). The Captain then said "icing conditions?" and the First Officer replied “negative, no visible moisture" - which was true for the conditions at that time but not in respect of the ATIS weather received. Shortly afterwards, the Captain said "below ten thousand [...] temperature is not negative [...] I am going to wait a little with the anti-ice". The flight was re-cleared to descend to 3,000 feet QNH and for the ILS ‘Y’ 07. The Captain then requested that the PF keep the speed up saying that "ATC won't like it if you will slow down everything behind us".
According to recorded radar data, the 07L localiser was intercepted at 3000 feet QNH at a range of about 13 nm upon which the controller requested 170 knots or more until 4nm and got the response that the current 175 KIAS would be maintained until 4 nm. The Captain then advised that he was selecting the engine icing on because of the ATIS information and the fact that once they left 3000 feet on the ILS GS they would be in a cloud layer with a sub zero temperature.
Shortly after becoming fully established on the ILS, landing clearance was given with the spot wind “100°/4 now 090°7” with a request to vacate to the right after landing. The First Officer asked the Captain if she could reduce speed to 130 and was reminded that 170 or more to 4 nm had been requested and accepted and that “then you still have enough time to position the flaps - no problem". It was noted that in the five minute period prior to this exchange, the CVDR recorded eight ‘proactive’ reminders by the Captain to “keep high speed”.
The Captain began to run the Approach Checks and on reaching the item “icing conditions" he said “it is still on". At around 4 nm, the Captain advised the First Officer that she could start to reduce speed and soon after, with the speed at 163 KCAS, she called for flaps 2 and 11 seconds later, with the runway in sight, she disconnected the AP. Passing 1000 feet agl at 151 KCAS with the aircraft well below the cloud layer which it had only been in for about 3 minutes, the Captain switched off the engine and windscreen anti-ice systems. At 800 feet agl, full flap was requested and set and the PF set the 96 knot Vref on the AFCS. By approximately 640 feet agl, the flaps were fully extended and the speed was reducing through 121 KCAS. As 500 feet agl was passed, the speed of the aircraft had settled at around 114 KCAS where it temporarily remained.
The 500 foot call prompted the Captain to quickly voice some of the landing checks which had still not been completed and was recorded as saying “gear down three greens, flaps are full, landing clearance update” before asking the controller "just to confirm cleared to land?" During this radio call, at an altitude of about 250 feet agl, the airspeed began to decrease again. Immediately after the auto callout of the DA at 200 feet agl, the Captain, having had no response to his request for confirmation that a landing clearance had been given, repeated it and received this as the aircraft passed 162 feet agl with the speed decreasing through 106 KCAS.
The aircraft then went slightly below the ILS GS and began to pitch up and over the next nine seconds, by which time the aircraft was at 30 feet agl, the speed decreased to about 90 KCAS, the recorded pitch angle increased to about +6°and the recorded AoA to between 16 and 17°. The aircraft then “began to roll left and within two seconds reached a bank angle of 30° to the left (whilst) at the same time, normal acceleration decreased from approximately 1 g to approximately 0.8 g”. Two seconds later, the CVDR recorded impact noises before the recording ended.
Both pilots subsequently stated that “during the flare above the threshold the left wing had suddenly dropped and touched the runway” after which the aircraft rolled right and the right main landing gear made a “hard” contact with the runway and fractured. The aircraft then slid towards the right edge of the runway and came to a stop approximately 447 metres past the displaced landing threshold just off the runway but still on its asphalt-surfaced margin.
Subsequent examination of the aircraft found the outermost 2 metres of the left wing were bent upwards by about 10° and its lower surface, aileron and flap showed “scratch marks and deformations”. The right main landing gear had fractured and folded rearwards with parts of it penetrating the upper surface of the right wing and its wheel had been destroyed and part of the wheel rim was abraded. The right flap was deformed and parts of the flap track and its running channel had been abraded. The airframe was subsequently declared a hull loss due to the extent of the damage.
Subsequent examination of the accident site (see below) showed that the left wing tip had made contact with the ground about 7 metres ahead of the displaced threshold about 4 metres left of the centreline. In the same area, rubber deposits made by the nose landing gear wheel were present about 4 metres ahead of the threshold and then 6 metres further on there were rubber deposits from the right main landing gear wheel.
Photographs of the aircraft taken by some of those to first reach the accident site (see the first illustration below) showed that at that time, with the surface temperature at 0° C, there was still up to 10mm of ice on the aircraft nose and on the full length of both wing leading edges, on the leading edges of the horizontal stabiliser and on the front end of the landing gear assembly.
Sometime later (see the second illustration below) there was still ice on the leading edge despite the temperature not being negative and when the BFU Investigation Team examined the wreckage three hours after the accident, they found leading edge ice accretion still visible and “about 4 cm wide, with a milky, rough, crystalline appearance and a slightly concave form with strongly developed feathering at the upper and lower edges”. In addition, they observed that “the upper and lower surfaces of the two wings showed a lamellar ice accretion (except that) there was no ice accretion on the left wing, where the impact had caused a bend”.
The photographic evidence was considered to conclusively demonstrate that the aircraft had made its final approach whilst accreting ice on critical parts of the airframe with the airframe de ice system not operating. The German Meteorological Service (DWD) subsequently confirmed that “the prevailing closed stratus clouds had been supercooled and therefore carrying ice” and it was also noted that moderate icing conditions had been included in the forecast issued.
Why It Happened
Central Maintenence Computer (CMC) data which included activations of (and faults in) the airframe de icing system was accessed and it was confirmed that it had not been activated during the flight. Apart from the consequent absence of airframe icing protection, this also had the effect of making the stall warning system ineffective. With full flap extended and the ice airframe protection system switched off, an aural warning is generated if the AoA reaches 21° and the stick pusher operates if it then reaches 28.4° whereas with the same flap extension and the system switched on, the corresponding figures are 9.5° and 15.5° which were both exceeded during the final stages of the accident flight.
The accident flight Captain said in his statement to the Investigation that “his reason for not activating the wing and horizontal stabilizer de-ice system was that he had not seen any ice accretion and further explained that the activation of the system would result in significant limitations of aircraft performance and there are discussions among pilots whereby the system should only be activated at ice accretion of at least 0.25 inches”. He also claimed, although there was no confirmation of it on the CVDR, that when the aircraft was in the cloud layer below 3000 feet, “he had checked the part of the left wing, about the outer third, which he could see from his point of view, for ice accretion and not seen any”. The Investigation was of the opinion that “with only a fleeting look at the wing it is possible that he could have overlooked the milky-white ice accretion on the silver-grey de-ice boots (but that) it cannot be ruled out that he did not take a look at all”.
To show that either pilot looking at the wing leading edge when seated in the flight deck in flight could have readily observed ice of the leading edges, a photograph of this view was taken and another photograph of the wing as observed an hour after the accident taken from “a similar viewing direction” was superimposed on it to and showed that the remaining ice could still be easily seen.
For a wider perspective on the Captain’s remarks, the Investigation looked at the CVDR data for previous flights by the aircraft and found that in a period of just over three months prior to the investigated accident, the engine anti-ice system had been in use on 22 occasions but during only 8 of these had the airframe de-icing system also been on and that “during 6 of the remaining 14 flights the OAT was less than 5°C”. Whilst it was noted that this finding might have led to a Safety Recommendation directed to the aircraft operator to ensure that their EMB 500 pilots were better acquainted with use of the airframe de-icing system and its effect on the stall protection system, since they no longer operated the type this was not appropriate.
It was noted that as the aircraft passed approximately 150 feet agl about 15 seconds prior to impact, the First Officer had begun to flare the aircraft and as pitch increased, the aircraft descended below the ILS GS. Neither pilot mentioned this and the First Officer was reportedly looking out and concentrating on reducing the speed Vref. The Investigation was of the opinion that “this shows that the Captain concentrated on the communication with the controller (to confirm the landing clearance) and was distracted from monitoring the actions of the First Officer”. It was considered that the continuously increasing AoA and the reduction in airspeed not only below the VREF but also significantly below the VMC specified in the AFM for flight in icing conditions (97 KIAS) had led to an aerodynamic stall such that recovery from the sudden roll movement and accelerated descent was no longer possible.
It was noted that such “sudden roll movement with subsequent loss of altitude at an AoA, where normally the aircraft should still fly” was typical of stall behaviour with significant rime ice on the leading edges and clear (runback) ice on both the upper and lower wing surfaces as had clearly been the case.
Whilst not directly contributory to the accident, the overall performance of the flight crew was assessed from the CVDR voice data to have lacked many of the features of effective CRM. The Investigation took the view that “the working relationship between the very experienced Captain and the First Officer who was at the beginning of her flying career, resembled more a relationship between a flight instructor and a student pilot” but that this reality had not been managed properly. Amongst other things, “the recording showing that the First Officer had been highly stressed during the approach, especially during the part where she flew the aircraft manually”. It was also noted that the recording “did not contain any indication that the pilots had discussed whether the First Officer had viewed herself capable of flying the approach, given that she was inexperienced on type and with flight in icing conditions”. It was considered that part of good CRM would have been the Captain observing his junior colleague and making a timely decision should a change of PF or special support during the approach have been appropriate.
It was noted that a fatal loss of control accident almost two years later to the same aircraft type in the USA had been investigated by the NTSB and found to have also been the result of failure to activate the airframe de-icing system prior to making an approach in icing conditions in the context of a similar lack of appreciation of its importance.
Given the forgoing findings, the Investigation examined the FAA 2010 aircraft type ‘Flight Standardisation Board’ (FSB) Report and the equivalent EASA 2012 (before the accident) and 2013 (after the accident) ‘Operational Evaluation Board’ (OEB) Reports which included pilot training requirements. It was found that these documents identified the aircraft as in ICAO approach category ‘B’ and that their ‘Training Areas for Special Emphasis (TASE)’ for Initial Type Rating Training included “operation in icing conditions and the stall warning and stick pusher system”.
It was noted that the aircraft had not flown a stabilised approach as required by SOPs and that the very inexperienced First Officer was quite likely to be (and apparently became) somewhat uncomfortable with the Captain’s willing acceptance of the request made by ATC to keep at least 170 knots to 4 nm in the prevailing conditions. It was also clear that the Captain became overloaded in the latter part of the approach due at least in part to this choice but showed no sign of having anticipated the potential consequences of it, having even offered to exceed 170 knots on acceptance.
However, although a pilot is not obliged to accept such a request to “keep up the speed” (and must accept the possibility of a delay to the approach if declined), the Investigation examined the context of the ATC request made. It was discovered from DFS, the ANSP involved, that “controllers generally instructed approach speeds of 160 - 170 KIAS because pilots could easily adhere to them” but that the applicable section of the Manual of Air Traffic Services ( MATS) merely stated that “between 12 nm and about 4 nm from touchdown, instructed speeds should not be lower than 150 KIAS”. It was also noted that ICAO approach category ‘B’ to which the accident aircraft was assigned was for “small multi-engine aircraft with final approach speeds of 85 - 130 knots”.
The Immediate Causes of the accident were formally determined to have been that:
- The crew conducted the approach under known icing conditions and did not activate the wing and horizontal stabiliser de-ice system, which was contrary to the applicable Standard Operating Procedures (SOP).
- Due to ice accretion on wings and horizontal stabiliser and infringement of the required approach speed the aircraft entered an abnormal flight attitude during the flare phase and crashed.
One Contributory Factor was also identified as “the crew had insufficient knowledge of the connection between the ice protection system and the Stall Warning Protection System (SWPS)”.
Three Safety Recommendations were made as a result of the Investigation as follows:
- that the DGAC Belgium should, as part of their oversight responsibilities, ascertain that the operator takes measures which ensure the consequent adherence to Standard Operating Procedures and the application of the criteria for a stabilised approach. [9/2018]
- that the European Aviation Safety Agency (EASA) should, in cooperation with the Brazilian aviation authority, the Agência Nacional de Aviação Civil (ANAC), ascertain that the aircraft manufacture of the EMB-500 renders the syllabus for the acquisition of the type rating more precisely to the effect that pilots unmistakably understand the importance and operation of the ice protection and the stall warning protection systems of the EMB-500. [10/2018]
- that the Agência Nacional de Aviação Civil (ANAC) - the Brazilian aviation authority should, in cooperation with the European Aviation Safety Agency (EASA), ascertain that the aircraft manufacture of the EMB-500 renders the syllabus for the acquisition of the type rating more precisely to the effect that pilots unmistakably understand the importance, function and operation of the ice protection and the stall warning protection systems of the EMB-500. [11/2018]
It was also noted that in connection with another Investigation still in progress, that the BFU has also issued the following Safety Recommendation which has relevance to this Investigation. It is as follows:
- that the Deutsche Flugsicherung (DFS) should amend the Air Traffic Services Manual of Operations (MO-ATS) to the effect that the assignment of a final approach speed takes into account the separation requirements, the approach classification of the respective airplane and the Safety Gates the pilots have to adhere to. This amendment should read “Prior to assigning a speed, the pilot should be asked which speed could be maintained up to which distance to the touch-down point” and “Single engine turboprop aircraft should not be given any speeds after initiation of the descent”. [12/2018]
The Final Report of the Investigation was completed on 17 December 2018 and subsequently published online.