SU95, Moscow Sheremetyevo Russia, 2019
SU95, Moscow Sheremetyevo Russia, 2019
On 5 May 2019, a Sukhoi RRJ-95B making a manually-flown return to Moscow Sheremetyevo after a lightning strike caused a major electrical systems failure soon after departure made a mismanaged landing which featured a sequence of three hard bounces of increasing severity. The third of these occurred with the landing gear already collapsed and structural damage and a consequential fuel-fed fire followed as the aircraft veered off the runway at speed. The subsequent evacuation was only partly successful and 41 of the 73 occupants died and 3 sustained serious injury. An Interim Report has been published.
On 5 May 2019, a Sukhoi RRJ-95B (Superjet 100) (RA-89098) being operated by Aeroflot on a scheduled domestic passenger flight from Moscow Sheremetyevo to Murmansk as SU1492 failed to complete a normal landing after returning to its departure airport in day VMC following significant electrical systems failure and after a series of bounces of increasing severity, landing gear collapse and structural damage resulted in a fuel fed fire developing as the aircraft veered off the runway and stopped. An emergency evacuation commenced but 41 of the 73 occupants died before leaving the aircraft, three sustained serious injuries and seven others sustained minor injuries.
An Investigation is being carried out by the Air Accident Investigation Commission of the Interstate Aviation Committee (MAK). Both the FDR and the 2 hour CVR were recovered from the wreckage with the former having sustained "substantial damage" due to the high temperatures in the fire but the data from both was successfully downloaded. Recorded data from ATC and airport ground-based facilities including surveillance cameras was also available.
It was noted that the 42 year-old Captain who was PF for the accident landing, had a total of 6,800 total flying hours of which 1,570 hours were on type with 1,428 of the those on-type hours having been recorded when in command. Prior to joining Aeroflot in 2016, he had been employed as a pilot by Transaero on the Boeing 737 for five years after leaving Russian government employment as a pilot in 2011 where he had mainly flown the IL76. The 36 year-old First Officer had a total of 765 flying hours of which 615 hours were on type and achieved after joining Aeroflot in 2018 following 150 hours of initial pilot training at flight school which had been completed in 2016.
It has been established that the flight had departed from Sheremetyevo runway 24C on a KN 24E SID. The crew realised as they continued to climb on the SID track that there was significant convective weather ahead and the CVR recorded the Captain saying “we’re going to get shaken" to which the First Officer responded with “damn it" and the Captain replied "nothing to worry about". Soon after this, as the turn to track inbound track to the KN (Kostino) NDB began five minutes after takeoff (see the illustration below) with the aircraft about to pass FL090 at 250 KCAS, it entered an area of significant convective activity.
As the crew were acknowledging a re-clearance to FL110 and continuous engine ignition was being selected on, the aircraft was reported by the crew to have received a lightning strike. They responded by replacing ‘CLIMB’ mode with ‘VS’ mode and setting a 3280 fpm rate of climb but almost immediately an uncommanded disconnect of the AP and an audio alert of AFCS reversion to ‘DIRECT MODE’ occurred. The A/T remained engaged. Ground-based doppler weather radar confirmed that at this time, the aircraft had been flying through the centre of an active thunderstorm area which was moving north-east at a little over 20 knots.
A few seconds after the reversion, “the Captain switched to manual control”. His first control input was to reduce the bank angle and lower the nose. FDR data then showed further side stick control inputs “of an abrupt and intermittent character” in both roll and pitch channels and as the aircraft initially continued to climb on the SID track, the A/T was disengaged by thrust lever movement. After a short discussion with the First Officer, the Captain decided to return to Sheremetyevo and instructed the First Officer to declare a PAN. When several attempts to make contact with the controller using COM 1 failed, the transponder squawk was set to 7600 to indicate a loss of radio communication before contact was re-established on 121.5 MHz using COM 2. Contact with the Moscow APP controller was then made advising the request to return. An initial descent to FL 080 from the maximum altitude reached (FL 106) maintaining present heading was given and acknowledged and further descent and radar vectors towards the ILS for runway 24L followed.
The Captain advised the Senior Cabin Crew Member (SCCM) that the aircraft was returning but emphasised “no emergency, we are simply going back". The controller asked the crew if any assistance was required and received the response “No, so far everything is normal, standard” but the crew then responded to the controller’s request for further clarification by saying that they were having problems with the radio communication and that the automatic flight control system had been lost. The First Officer began reading out the QRH drill for ‘F/CTL DIRECT MODE’.
The first attempt to make the ILS approach as cleared resulted in the Captain deciding that he was not yet ready to complete it and ten minutes after the lightning strike had occurred, at 2600 feet QNH (790 metres, which was equivalent to 600 metres on the Sheremetyevo QFE), a right hand orbit at that altitude was requested and approved. FDR data showed that during this orbit, the Captain was not able to maintain the target altitude accurately. Whilst it was continuing, the First Officer read out the QRH drill for an overweight landing (the ELW was about 1600 kg above the 41,000 kg MLW). This included switching off the engine bleed air supply to the air conditioning and pressurisation system and ensuring that the rate of descent for touchdown must not exceed 360 fpm which was still in excess of the FCOM-recommended 150-200 fpm for normal operations.
After completing the orbit, the aircraft was established on the localiser at more than 11 nm. Descent on the ILS GS began at 1800 feet QFE and was flown mainly at or slightly above the calculated Vapp of 155 KIAS and consistently slightly below the GS but, with one brief exception, within half scale deviation. FDR data recorded a predictive wind shear warning as the aircraft descended through 1600 feet QFE (1100 feet agl) which lasted for 11 seconds but was ignored without crew discussion. When the aircraft auto callout of 1000 feet agl occurred whilst the warning (which could be disregarded if the crew assessed that it was safe to do) was still active, the Captain responded with “proceed” and the First Officer replied “check”. The approach thereafter was normal until immediately after the auto call at the DH of 270 feet (82 metres) when a rapid increase in the rate of descent increased the ILS GS deviation beyond half scale. This triggered a TAWS "GLIDESLOPE" alert which continued for 4 seconds to which the Captain responded with “Advisory" whilst he simultaneously increased thrust. This in turn increased the airspeed which reached 164 KCAS as the runway threshold was crossed at 40 feet agl and continued to increase to 170 KCAS - although this speed was still less than the FCOM maximum of Vapp+20 required to remain within stabilised landing approach criteria.
Once past the runway threshold, the rate of descent was decreased and at 17 feet agl, the auto callout "RETARD. RETARD" prompted the crew to set the thrust to Idle during the flare. The Captain began to flare the aircraft only after setting the thrust levers to Idle and thereafter his side stick inputs began to oscillate between the maximum pitch up and maximum pitch down positions “with a relatively long holding period in both maximum positions”.
The first touchdown on the 3700 metre long 60 metre-wide runway 24L occurred approximately 890 metres beyond its threshold at Vapp +3 and at a pitch attitude close to zero with all three landing gear legs making near-simultaneous runway contact. At touchdown, recorded as at 2.55 g, the side stick was in the maximum pitch up position but within 0.4 seconds, this had changed to maximum pitch down. It was noted that in ‘DIRECT MODE’, automatic ground spoiler deployment does not occur and FDR data showed that the manual deployment did not occur.
After this first touchdown, the aircraft bounced no more than 6 feet but contrary to the required response for this relatively modest bounce, the side stick was held at the maximum pitch down position which resulted in a nose down attitude and a recorded increase in the rate of descent to over 600 fpm. A nose-gear-first second touchdown approximately 1070 metres beyond the runway threshold occurred just 2.2 seconds after the first at a similar forward speed. The side stick was returned to the fully nose up position immediately before the 5.85 g touchdown after which the aircraft became airborne again. This impact resulted in the main gear collapsing rearwards in accordance with the significant exceedance of the design limits for vertical acceleration at touchdown.
Although the Captain had selected maximum reverse during the first bounce, this initially had no effect since the required main gear weight-on-wheel (WOW) sensors were not active again until the second bounce occurred which delayed the associated engine thrust increase. After the second bounce during which the aircraft reached the height of between 15 and 18 feet it was found that the thrust levers had been briefly set to "Takeoff" and the side stick position moved to maximum pitch up. It was considered by the Investigation that these actions may be interpreted as an attempt to commence a go-around.
Three seconds after the second touchdown at 140 KCAS, a third touchdown occurred at approximately 1360 metres beyond the runway threshold at a minimum of 5 g with the right wing low. Runway markings confirmed that by this time, both main gear legs had already collapsed so that the final touchdown caused structural disintegration of the airframe which led to fuel spillage. As the aircraft continued along the runway resting on the engine nacelles and the rear fuselage, an “aft cargo compartment fire" event was recorded as the ground speed reduced through around 100 knots (although given the nature of the sensors involved, this was considered to have indicated the presence of smoke rather than fire).
Approximately 130 metres beyond the position of this final touchdown, an intense fuel-fed fire was observed to begin. The TWR controller called for "Emergency Services to the runway" and as the aircraft continued to slow down, members of the cabin crew were recorded reporting fire. With the aircraft ground speed down to around 25 knots, an APU Fire event was recorded and the aircraft was observed to turn left towards a nose into wind orientation, finally coming to a stop off the side of the runway some 38 seconds after the first touchdown. A flight crew PA of "Attention crew! On station” was made twice after which the Captain “called for the Emergency Evacuation Checklist" although there was no subsequent CVR record of it being completed. Eleven seconds after the flight deck “on station” PA to the cabin crew, a cabin crew PA of "undo the seatbelts, leave everything, evacuate” was recorded and the Captain called out "evacuation".
Initial and Intended Actions of the Investigation
The Investigation has already carried out preliminary comparative analysis of manually flown approaches previously made by the Captain with the flight control system in ‘NORMAL’ mode with the approach made prior to the accident landing. The accident flight approach was found to be characterised by side stick movements of a significantly wider amplitude and degree of oscillation than flights in ‘NORMAL’ mode with these causing “significant changes of longitudinal motion parameters”. An examination of a series of manually flown approaches in ‘DIRECT’ mode which had been made by other Aeroflot pilots in the same aircraft type showed that side stick inputs were similar and “reasons for these peculiarities are being analysed”.
The aircraft designer has been requested by the Investigation to perform mathematical modelling of the accident flight in order to confirm that the aircraft performed in accordance with its type certification and to make an assessment of the possible influence of external disturbances on the performance. The Investigation also intends to conduct experiments in a full flight simulator to assess crew performance during the accident flight.
The damage consequent upon a total of 16 previously recorded lightning strikes to the aircraft type has been examined and it was found that nine of these events had resulted in the aircraft electrical systems being compromised or reversion of the flight control system to ‘DIRECT’ mode for any other reason. Eight previous reversions of RRJ-95 flight control systems to ‘DIRECT’ mode during the four years prior to the accident under investigation have been identified, as have 6 hard landings, none of which were as severe as the one which led to the accident.
It was noted that certification of the aircraft type for severity of a landing was based on a design maximum vertical acceleration of 3.75 g beyond which shear pins in each assembly were designed to fracture to allow the gear leg(s) involved to move upwards and rearwards in such a way that no damage would be caused to the wing box. Accordingly, the available evidence shows that the first touchdown did not compromise the integrity of the main landing gear legs but the 5.85 g second touchdown led to fracture of the shear pins as per design so that the final 5 g touchdown occurred with no structure which would have allowed the main gear to absorb the load and the adjacent wing structure “was destroyed in the areas of the hinges of the hydraulic cylinders” causing the aircraft to drop onto the engine nacelles and the rear fuselage. It was noted that in accordance with certification requirements, “evaluation of secondary impacts of the airframe on the ground after the destruction of the landing gear is not required”. Further work including but not limited to that mentioned above is ongoing.
An Interim Report, on which this summary is based, was released on 14 June 2019.