Aeroflot SU95 at Moscow on May 5th 2019, aircraft bursts into flames during rollout and burns down

Last Update: May 5, 2022 / 17:11:25 GMT/Zulu time

Bookmark this article

Photo of Aeroflot RA-89098, SUKHOI Superjet 100-95 — Aeroflot RA-89098, SUKHOI Superjet 100-95 (Photo credit: Anna Zvereva / Flickr / License: CC by-sa)

Incident Facts

Date of incident
May 5, 2019

Classification
Accident

Airline
Aeroflot

Aircraft Registration
RA-89098

Aircraft Type
SUKHOI Superjet 100-95

ICAO Type Designator
SU95

On Jun 14th 2019 the MAK released their preliminary report (editorial note: in English!!) reporting the captain (42, ATPL, 6,800 hours total, 1,570 hours on type thereof 1,428 hours in command) was pilot flying, the first officer (36, CPL, 765 hours total, 615 hours on type) was pilot monitoring.

The MAK reported that examination of the fuselage revealed typical traces of "lightning impact" was found on the right hand angle of attack sensor, the right temperature probe and right ice detector.

The MAK reported the sequence of events stating the crew performed a normal departure though noticing flashes to their right, had cleaned up the aircraft and were cleared to climb to FL070. The MAK wrote:

At 15:07:10, the «HEADING» / «HDG» mode was engaged in the lateral control channel, and the selected heading was set to 327°. According to the Vnukovo doppler weather radar, the aircraft was flying through the active thunderstorm area (Fig. 3), that was moving from the southwest to north-east with the speed of 40-45 km/h. The transition to the selected heading mode caused the aircraft to initiate the right turn earlier than it is prescribed by SID KN 24E (Fig. 4). The crew did not requested the active thunderstorm area avoidance clearance.

On contacting the Approach controller, the crew was instructed to climb to FL 90. At 15:07:21, the crew set the new selected altitude of 8992 ft (2741 m) and activated the "CLIMB" / "CLB" mode.

During the period from 15:07:30 to 15:07:33, there is the following conversation in between the crew members was recorded: PIC: "We are going to get shaken", FO: "Damn it.", PIC: "Nothing to worry about".

At 15:07:34, the Approach controller instructed the crewto climb to FL 100. At 15:07:39, the selected altitude was 9984 ft (3043 m), with "CLIMB" / "CLB" mode continued to be engaged in the longitudinal control channel.

At 15:07:43, the "LATERAL NAVIGATION" / "LNAV" mode was engaged again in the horizontal plane. By that time, the aircraft was flying with heading of 317°, the right roll of 24° and the indicated speed of 250 kt (463 km/h).

At 15:08:03, the Approach controller instructed the crew to climb to FL 110. When the FO confirmed receiving the instruction, recorded by the CVR, the sound effect lasting for 1.5 seconds, starting from 15:08:09.7, was recorded. Most probably, at this stage, an atmospheric electricity impact affected the aircraft. 3 seconds prior, the FDR started to record the single commands, indicating the engagement of the permanent ignition on the both engines.

At 15:08:11, the crew set new selected altitude of 11008 ft (3355 m) and the new selected vertical speed of 3281 ft/min (16.7 m/s), and the "VERTCIAL SPEED" / "VS" mode was engaged.

At 15:08:11.9, the disengagement of the autopilot occurred, accompanied by the corresponding aural alert and the switching of the automatic flight control system into the "DIRECT MODE" (see Section 1.18.% of the present Report) with the audio alert "DIRECT MODE. DIRECT MODE". After the sound effect, simultaneously with the autopilot disengagement, the emotional exclamation of one of the crew members was recorded: "Wow!". The auto throttle continued to be engaged. The aircraft was in 20° right roll, crossing FL 89 (2700 m) in climb.

Starting from 15:08:12, during 15 seconds, the FDRs incorrectly recorded single commands and the analog parameters, which are usually recorded via the electronic interface unit, EIU-100. Two single commands and , "EIU1 FAULT" and "EIU2 FAULT", were recorded simultaneously.

The conducted examinations (Section 1.16.2 of the present Report) have shown that, most probably, at that stage, a reboot of the data concentrator units occurred, which caused the switching of the automatic flight control system into the «DIRECT MODE». According to the "List of special situations for the RRJ-95B aircraft" during the certification this event had been classified as a "complicated situation" (Section 1.18.14)

After the switching to the "DIRECT MODE", the "FLAPS ICE" configuration (slats 0°, flaps 1°) was engaged automatically.

Starting from 15:08:16, the aircraft control was switched to manual control of from the lefthand side pilot station. During the period of approximately 4 seconds, starting from the moment of the autopilot disengagement and till the PIC took over the aircraft control, the right roll decreased to 13°, pitch attitude increased (from 11° to 12°). The PIC's first control input from was affecting the roll channel, the sidestick was declined to 11.7° to the left (which is for more than half of its travel range), after that the sidestick was advanced to 6.8° (which corresponds to the half of its pitch-down travel range). Further sidestick control movements both in the roll and in pitch channels was of an abrupt and intermittent character. At 15:08:22, the aircraft was turned to the right via multiple impulse deflections of the sidestick ranging from 30% to 65%. To set the roll of ≈20° the pilot performed more than 10 roll deflection movements of the sidestick during the period of 18 seconds. Moreover, at different moments of the further flight the PRIORITY / APOFF8 button was pressed momentarily (for approximately 1 second) on the left sidestick 6 times.

The aircraft continued to develop the right turn in accordance with the KN 24E chart and continued to climb. At 15:08:47, the auto throttle was disengaged by the overriding movement of the thrust lever. At 15:09:17, the aircraft was put out of the right roll mode and was set to the heading of about 60°.

After short discussion with the FO, the PIC made the decision to return to Sheremetyevo and ordered the FO to report PAN-PAN. After several failed attempts to contact the controller on the working frequency via VHF-Station 1 (the one that was used from the beginning of the flight), at 15:09:32, after the discussion, the crew set the 7600 squawk code (for the lost radio communication).

At 15:09:35 radio-contact was restored at the emergency frequency (121.5 MHz) via VHFStation 2. After that, at 15:09:39, the FO reported the Approach controller: "Moscow-Approach, request return of 14-92, radio lost, aircraft in direct mode". The controller issued an instruction to descend to FL 80. Maximum altitude the aircraft gained was 10600 ft (3230 m) (QNE). The crew responded: "Aeroflot 14-92, heading 0-57, descending to FL 8-0". The controller confirmed the clearance for the descent with the maintained present heading. Further flight till the moment of the glideslope capture was performed by means of vectoring.

At 15:09:52, the crew set the selected altitude of 8000 ft (2438 m) and disengaged the FMS speed control mode, set the selected speed of 250 kt (463 km/h) and put the aircraft into the descent mode. In descending, the PIC maintained the indicated air speed by changing the pitch angle and the thrust levers' positions.

Later into the descent while intercepting the localizer the MAK writes:

At 15:15:34, the controller cleared the crew for the ILS Yankee approach, RWY 24 left.

After the permission confirmation, the FO started reading section QRH "F/CTL DIRECT MODE", inter alia: "… Auto throttle should not be used, fly smoothly. … Balance manually. … Speed brake no less than a half. … use Flaps 3. TAWS, landing gear, Flaps 3 on. V approach, V reference plus 10. Landing distance – increase by 1-34. … Speed brake set to full after touchdown. Go-around lever to NTO". At 15:16:54, the controller instructed: "… proceed right heading 210 till localizer capture …".

At 15:17:39, the FO, after the command from the PIC, reported to the controller that the crew was not ready to perform landing approach and requested an «orbit», but then corrected himself: "the circuit" and the controller responded: "… heading 360 to the right".

At 15:18:53, the PIC tried to contact the controller himself: «Aeroflot 14-92, holding area above Kilo November, if possible». The message was not recorded by the controller's recorder. The PIC did not raise this question again.

When performing an orbit flight at 600 m QFE, the PIC could not maintain the altitude precisely. During right turns with roll up to 40°, the deviation from the selected altitude exceeded ±200 ft (60 m), and that triggered multiple aural alerts. The PIC was aware of that, at 15:22:53: "How come… plus-minus 200 feet".

During the orbit flight performance, the FO was reading out the QRH "OVER MAXIMUM LANDING WEIGHT LANDING" section (1.18.7). The weight of the aircraft was about 42 600 kg which exceeded the maximum landing weight by 1600 kg. To gain the maximum thrust in order to be able to perform the go-around, the crew switched off the engine air bleed supply to the conditioning system. The FO also read out that the maximum vertical speed of descend before the touchdown must be no more than 360 ft/min (1.8 m/s). During the standard approach and landing (the Standard Procedures)) the vertical speed recommended by FCOM is 150-200 ft/min (from 0.76 to 1 m/s).

At 15:20:41, the crew started the APU.

At 15:21:38, at 225 kt (415 km/h) the crew initiated the flaps deployment to FLAPS 1 (slats 18°, flaps 3°).

At 15:23:03, after the PIC's order, the FO reported the controller that the crew was ready for landing.

At 15:23:19, at 190 kt (350 km/h) the crew initiated the flaps deployment to FLAPS 2 (slats 24°, flaps 16°).

At 15:23:58 the crew started the landing gear extension.

At 15:24:02, the crew armed the speedbrake system.

Note: Automatic speedbrake deployment mode which can be engaged during the landing and which is inactive when the Remote Control system is in the "DIRECT MODE"». At 15:24:20, at 170 kt (315 km/h), the crew set the flaps into the landing configuration, FLAPS 3 (slats 24°, flaps 25°).

During the landing gear and flaps deployment, the PIC performed stabilizer trimming manually. When the aircraft was set to the landing configuration, the stabilizer deflection angle was ≈3.5° pitch-up and it almost did not change until landing (one short pitch-up pressing was recorded during the glideslope descent).

The aircraft was put to the landing heading at the distance of more than 20 km from the RWY 24L threshold. The final approach fix is located at the distance of 9.3 km.

At 15:24:38, the crew responded to the controller's request that they would perform the ILS approach (the QRH the ILS approach is required with the use of the instrument-landing horizondirector indicator). By 15:25:57, the crew had finished the "DURING LANDING" section checklist.

The crew performed neither the before-landing briefing nor the "APPROACH" checklist.

At 15:26:05, the crew had the following conversation: the FO "Should I set the transponder to 7700? Or leave as it is? Ok, roger.". The PIC: "We could have done it long before, actually." The CVR record does not provide us with clear understanding of the PIC's decision. Nevertheless, at 15:26:31 the crew set the transponder to 7700.

At 15:27:20, the glideslope descent was initiated. The go-around altitude was not set by the crew.

During the approach about 2 minutes prior to landing the MAK describes:

At 15:28:26, at 1600 ft (490 m) QNH (at the radio altitude of 1100 ft (335 m)) the "WINDSHEAR" warning was recorded, it is generated by the crew alerting system based on the wind shear forecast data of the weather radar, accompanied by the voice message "GO-AROUND, WINDSHEAR AHEAD". This warning informs the crew about the possible wind shear enroute. The alert lasted for 11 seconds and included 2.5 warning cycles (5 voice messages, each cycle consisting of 2 messages one after another, with the 1 second interval in-between). The crew had no discussion of this warning triggering.

Note: In the QRH «W/S AHEAD» Section (1.18.8) states that the triggering of this warning during the approach indicates that the crew must initiate the go-around procedure. At the same time in the beginning of the section, there is the notice that if the crew is sure that there is no wind shear hazard and there are no other signs of wind shear, and the RWS is operative, this warning may require no responsive actions from the crew. The same provisions are incorporated into the Flight Operations Manual.

The MAK continued:

At 15:29:22, the FO called out that the landing decision height was close. The PIC confirmed continuing the approach.

At 15:29:31, at 270 ft (82 m) the aural annunciator system informed the crew on reaching the landing decision height minimum, it was repeated by the FO. Starting from that moment, there was noted fast increase of the downward glideslope deviation (up to minus 1.4 dots), which caused the TAWS "GLIDESLOPE" alert triggering to indicate the glideslope deviation The alert was on for 4 seconds. The PIC confirmed hearing the alert by saying "Advisory".

The MAK added that FCOM states the GLIDESLOPE warning below decision height is "advisory", hence a go around was not required.

The MAK then continues the sequence of events:

Simultaneously with the warning triggering, the PIC increased engines power (the trust levers were set to 24° to 23°, which caused the N1 increase up to 77 to 74%). The increase of the engine power caused the increase of the indicated air speed: when the aircraft was passing RWY threshold: at 40 ft (12 m) it was up to 164 kt (304 km/h), and by 16 ft (5 m) it was up to 170 kt (315 km/h). According to FCOM, the required landing approach speed for the actual conditions was 155 kt (287 km/h). This speed was set by the crew as the selected speed. The airline's Flight Operations Manual contains the value plus 20 kt as the criterion for the stabilized landing approach (Section 1.18.9).

Over the RWY threshold, the glideslope deviation was minus 0.9 dot (i.e. below the glideslope). After passing the RWY threshold, the the vertical speed of descent was decreased. The flight parameters during landing are provided in Fig. 8.

At 15:29:54, at 17 ft (5.2 m) aural alert "RETARD. RETARD" informed the crew on reaching the height recommended for the thrust levers to be set to "IDLE" during the flare. The PIC started to change thrust levers positions simultaneously with the alert triggering.

Almost simultaneously with the thrust levers positions changed to "IDLE", the PIC initiated the flare by pulling the sidestick by 8.8° (65% of its travel range). Further on, the PIC's everamplitude increasing pitch idestick inputs were recorded up to the both maximum advanced and maximum retarded positions with a relatively long holding period in the both maximum positions. These inputs caused counter-reversal pitch change (+6…-2º). At 15:30:00, at a distance of ≈900 m from the RWY threshold at indicated air speed of 158 kt (293 km/h), first touchdown occurred. The touchdown occurred at the pitch angle close to zero ( 7 pitch up rate; at the moment of the touchdown, the sidestick was in the maximum retarded position; and during the landing within 0.4 seconds, its position was changed for the maximum advanced), for "threedots", with vertical g-force of no less than 2.55 g.

In "DIRECT MODE", the automatic deployment of the speedbrakes (spoilers) is not provided by the design, and the manual speedbrakes deployment was not engaged by the crew. After the touchdown, the aircraft bounced up for the he height of no more than ≈ 5-6 ft (2 m).

After the aircraft was bounced up, the sidestick continued to be held in the maximum advanced position, which caused the pitch downward rate of up to 10.5°/s, the rapid downward pitch angle decrease to 4° and the repeated touchdown with the nose landing gear touching the ground first.

The second touchdown occurred 2.2 seconds after the first touchdown, at the indicated air speed of 155 kt (287 km/h).

Landing performed with the nose gear touching the ground first, with the high vertical speed and the fully retarded sidestick immediately before touchdown, caused the intensive pitch-up rotation of the aircraft. The maximum pitch rate recorded was ≈25°/s and the vertical g-force was no less than 5.85 g. As the result, the angle of attack and the pitch angle increased, which, due to the remaining significant value of the indicated air speed, caused the repeated bouncing ("progressing bouncing"), despite the sidestick fully advanced position at bouncing up.

During the first bouncing up, when the aircraft was in the air, the PIC switched the thrust levers to the "Maximum Reverse" mode but the thrust reverser doors did not deployed, as there were no weight-on-wheel (WOW) signals. Upon receipt of the left and right main landing gears WOW signals, the reverser doors started to deploy, which was completed after the second aircraft bouncing up movement off the RWY. There was no engine power increase as at this moment, there was no WOW signals again.

During the second bouncing up, the aircraft reached the height of ≈ 15 - 18 ft (5-6 m). In 2-3 seconds after this repeated bouncing up, the thrust levers were set to "Takeoff" mode and the sidestick was pulled to the maximum retard position. This actions may be interpreted as an attempt to perform the go-around, as before that the thrust reverser system was engaged (the reverser doors were in the open position, though starting to close), the engine thrust did not increase.

At 15:30:05, with the indicated air speed of 140 kt (258 km/h), the third touchdown occurred with vertical g-force of no less than 5 g. The type of the markings, left on the RWY after the third touchdown, indicate that the main landing gears were by that moment already partially destroyed ("weak links" were cut off, see Sections 1.3, 1.12 and 1.18.12 of the present Report). The main landing gears collapsed, further on, the aircraft structural disintegration with fuel spillage and fire occurred.

With respect to the oscillating control inputs the MAK wrote:

The Commission carried out preliminary comparative analysis of the approaches performed by the PIC manually (with disengaged autopilot) in «NORMAL MODE», with the emergency flight (Fig. 42). The figure shows that the sidestick movements for the pitch in the emergency flight are characterized by significantly wider amplitude, and are oscillatory, which caused significant changes of longitudinal motion parameters. Similar "sweeping" movements were observed during landing performed in «DIRECT MODE» by other crews of the airline (Fig. 43). Reasons of these peculiarities are being analyzed.

The MAK reported with respect to weak links in the airframe design:

There are safety pins in the RRJ-95 landing gear design which are the elements of a certain level of strength, and they are the first to be destroyed upon the impact of a single design load, determined basing on the load application conditions given in aviation regulations, it is proved by calculations and tests. Destruction of the safety pins causes the gear to move up and backwards and does not cause the damage to the wing box.

The RRJ-95 safety analysis concerning the fuel leakage from the wing box upon the main landing gear destruction was carried out during design studies by modeling of the process of gear destruction when the calculated loads on the landing gear are exceeded. The conditions of application of the load listed in the certification requirements were considered. The certification report was issued. Basing on the calculations the strength and design requirements for safety pins in the gear hinge fitting were determined. In production the safety pins are particularly important parts, the blanks are under special control, and the parts themselves undergo periodic testing.

During operation, the confirmation of the compliance of the serial design to the design parameters was established as a result of an investigation of a serious incident - a very rough landing of the aircraft MSN 95032 with a vertical acceleration of about 4g, as a result of which the safety pins of the right main landing gear were destroyed. The corresponding report has been issued on this incident. In the course of the accident with RA-89098 there were three consequent touchdowns to the RWY with accelerations not less than 2.55g, 5.85g, 5.0g. The airframe endurance design vertical acceleration is 3.75 g.

The acceleration of 2.55 g did not exceed the value, calculated for the cut off of the safety pins of the main landing gear hinge brackets on the rear spar, so at the acceleration of 5.85 g the cut off of the safety pins of the right and the left main landing gear to the rear spar of the wing occurred. During the repeated aircraft lift of the landing gear were in the loose position towards the rear spar. The wing box was not destroyed, according to the records and the fuel gauge data, there were no fuel leakages, there were no structure fragments found on the RWY. At the third runway touchdown, with the allowable vertical loads exceeded, the structure condition did not allow the gear to absorb the loads of the landing hit and they broke down, wing structure was destroyed in the areas of the hinges of the hydraulic cylinders. The aircraft went low and moved on the nacelles and the rear part of the fuselage. The situation described above is not provided for in the current airworthiness standards.

In accordance with the certification rules, the evaluation of secondary impacts of the airframe on the ground after the destruction of the landing gear is not required.

On Apr 15th 2020 the Investigative Committee of Russian Federation (ICR) released a video of the final stages of roll out, evacuation and scenes of board after the fire was extinguished (see ICR Video below). The ICR reported a criminal case against the pilot was sent to the General Prosecutor General of the Russian Federation. The ICR argues: "Investigators and experts studied the data of the flight recorders of the crashed aircraft, according to which during the flight the aircraft adequately responded to the pilot's control actions. This information refutes the accused’s version of a possible aircraft malfunction."

On May 4th 2020 the MAK released a statement regarding the the investigation progress as required by ICAO annex exactly 12 months after the occurrence. The MAK reported following investigation steps amongst others have been completed:

- decryption and analysis of ground and airborne data (FDR)
- assessment of simulators used for pilot training by Aeroflot and characteristics of the aircraft
- mathematic modelling of flight including confirmity of aerodymanic characteristics with the accident aircraft as well as possible external influences onto the aircraft
- Analysis of previous rough landings in DIRECT MODE and transition into DIRECT MODE
- analysis of fuselage for signs of lightning strike

Three researches into the investigation are not yet completed, a final reported will be drafted upon completion of these studies.

On May 11th 2021 the MAK released another interim statement stating, that the final report has not yet been completed. However, in addition to the works completed until May 4th 2020 (the last interim statement) following works have been completed:

- analysis of load modelling acting on the chassis elements
- analysis of the characteristics of the safety pins at the attachment points of the main landing gear
- evaluation of the sequence of destruction of the aircraft structure, in particular landing gear and wings at the location of the fuel tanks
- analysis of the weather radar of the aircraft and the weather radar image of Vnukovo
- assessment of flight training of flight and cabin crew
- assessment of training and actions of ATC
- assessment of rescue operation
- materials of the accident of another SU95, RA-89011, were studied (see Accident: Yakutia SU95 at Yakutsk on Oct 10th 2018, overran runway on landing)

Currently not completed are the works on:

- assessment of engine control after structural failure of landing gear and the aircraft lowering onto the engine nacelles
- studies of the fire protection units to be done by the NTSB that were halted by the Corona Pandemic

On May 5th 2022 the MAK reported that all remaining research tasks have been completed, however, one of the reports have not been forwarded by the manufacturer of the fire protection unit about the data recorded in the non-volatile memory of that device. The MAK is going to prepare the draft final report without these data. The MAK writes:

The study of the FPC (fire protection unit) block was carried out only on August 12, 2021 at the premises of the Curtiss-Wright Defense Solutions enterprise (Santa Clarita, California, USA). According to the report received by the IAC dated September 29, 2021, information was uploaded from the non-volatile memory of the FPC unit, as well as the conversion and analysis of the uploaded information. At the same time, representatives of Curtiss-Wright did not provide the IAC with the results of the decryption of data downloaded from the non-volatile memory of this device, referring to the fact that this information is the intellectual property of the company. For the same reason, Curtiss-Wright representatives did not provide answers to additional questions from the commission of inquiry sent in a letter dated 08.10. 2021 to the NTSB and then forwarded to Curtiss-Wright (information was requested on all parameters of the fire system, which are recorded in the NVM of the FPC unit). In further correspondence through an NTSB authorized representative, Curtiss-Wright did not forward any further information to the inquiry committee and reaffirmed that the requested data is intellectual property. In the current situation, the Accident Investigation Commission is forced to prepare a draft Final Report without the results of the interpretation and analysis of the data downloaded from the non-volatile memory of the FPC unit. that the requested data is intellectual property. In the current situation, the Accident Investigation Commission is forced to prepare a draft Final Report without the results of the interpretation and analysis of the data downloaded from the non-volatile memory of the FPC unit. that the requested data is intellectual property. In the current situation, the Accident Investigation Commission is forced to prepare a draft Final Report without the results of the interpretation and analysis of the data downloaded from the non-volatile memory of the FPC unit.

Based on the results of all the work, the commission will soon prepare a draft Final Report on the results of the investigation, which will be sent to authorized representatives of the countries participating in the investigation of the accident for comments. According to Annex 13, comments on the draft report must be sent no later than 60 days from the date of its receipt.