MyCargo B744 at Bishkek on Jan 16th 2017, impacted terrain on go around

Last Update: March 3, 2020 / 18:37:45 GMT/Zulu time

Bookmark this article
Incident Facts

Date of incident
Jan 16, 2017

Classification
Crash

Aircraft Registration
TC-MCL

Aircraft Type
Boeing 747-400

ICAO Type Designator
B744

On Mar 3rd 2020 the MAK released their final report concluding the probable causes of the accident were:

The cause of the Boeing 747-412F TC-MCL aircraft accident was the missing control of the crew over the aircraft position in relation to the glideslope during the automatic approach, conducted at night in the weather conditions, suitable for ICAO CAT II landing, and as a result, the measures to perform a go-around, not taken in due time with the aircraft, having a significant deviation from the established approach chart, which led to the controlled flight impact with terrain (CFIT) at the distance of ~930 m beyond the end of the active RWY.

The contributing factors were, most probably, the following:

- the insufficient pre-flight briefing of the flight crew members for the flight to Manas aerodrome (Bishkek), regarding the approach charts, as well as the non-optimal decisions taken by the crew when choosing the aircraft descent parameters, which led to the arrival at the established approach chart reference point at a considerably higher flight altitude;

- the lack of the crew's effective measures to decrease the aircraft vertical position and its arrival at the established approach chart reference point while the crew members were aware of the actual aircraft position being higher than required by the established chart;

- the lack of the requirements in the Tower controllers' job instructions to monitor for considerable aircraft position deviations from the established charts while the pertinent technical equipment for such monitoring was available;

- the excessive psycho-emotional stress of the crew members caused by the complicated approach conditions (night time, CAT II landing, long-lasting working hours) and their failure to eliminate the flight altitude deviations during a long time period. Additionally, the stress level could have been increased due to the crew's (especially the PIC's) highly emotional discussion of the ATC controllers' instructions and actions. Moreover, the ATC controllers' instructions and actions were in compliance with the established operational procedures and charts;

- the lack of the crew members' monitoring for crossing the established navigational reference points (the glideslope capture point, the LOM and LIM reporting points);

- the crew's failure to conduct the standard operational procedure which calls for altitude verification at the FAF/FAP, which is stated in the FCOM and the airline's OM. On the other hand, the Jeppesen Route Manual, used by the crew, contains no FAF/FAP in the RWY 26 approach chart;

- the onboard systems' "capture" of the false glideslope beam with the angle of ~9 degrees;

- the design features of the Boeing 747-400 aircraft type regarding the continuation of the aircraft approach descent in the automatic mode with the constant descent angle of 3¡ã (the inertial path) with the maintained green indication of the armed automatic landing mode (regardless of the actual aircraft position in relation to the RWY) while the aircraft systems detected that the glideslope signal was missing (after the glideslope signal "capture"). With that, the crew received the designed annunciation, including the aural and visual caution alerts;

- the absence of the red warning alert for the flight crew in case of a "false" glideslope capture and the transition to the inertial mode trajectory, which would require immediate control actions from the part of the crew;

- the lack of monitoring from the part of the crew over the aircraft position in regard to the approach chart, including the monitoring by means of the Navigation Display (ND), engaged in the MAP mode;

- the crew's failure to conduct the Airline's Standard Operational Procedures (SOPs), regarding the performance of the go-around procedure in case the "AUTOPILOT" (the AP switching to the inertial mode) and "GLIDESLOPE" (the EGPWS annunciation of the significant glideslope deviation) alerts during the automatic CAT II landing at true heights below 1000 ft (with no visual reference established with either the runway environment or with the lighting system);

- the delayed actions for initiating the go-around procedure with no visual reference established with the runway environment at the decision height (DH). In fact, the actions were initiated at the true height of 58 ft with the established minimum of 99 ft.

The MAK analysed runway 26 was approved for CAT II ILS landings, a minimum RVR of 350 meters was required. Weather forecast before departure indicate a visibility of 200 meters horizontally and 30 meters (100 feet) vertically.

While the aircraft was descending towards Bishkek the crew requested and received current weather data at the aerodrome: "Turkish six four nine one, runway in use two six, RVR, in the beginning of the runway four hundred meters, in the center and at the end of... on the runway three two five meters, vertical visibility one three zero feet." which permitted the continuation of the approach.

While on the standard arrival, although controllers complied with the procedures while providing altitude clearances, the crew developed a negative atittude towards ATC complaining they were kept high. Upon contacting the approach controller instructed: "Turkish six four niner one, Approach, good morning, radar contact, descend flight level six zero TOKPA one arrival expect ILS approach for runway two six." The MAK wrote:

In accordance with the approach chart, the TOKPA reporting point must be flown over at FL 60 or above. Although the controller's instruction was in compliance with the approach chart, the clearance for the descent to FL 60 again aroused the PIC's negative attitude: "They left us high again."

In fact, the MAK analysed that all air traffic controllers followed the requirements of their working procedures throughout the entire sequence of events leading to the accident (the particular repetitive analysis portions are omitted in the following).

At that point the aircraft overflew RAXAT waypoint at FL180. The crew read back "I will be six zero (illeg) at TOPKA".

The MAK analysed:

For performing the standard descent, "Descent, Approach and Landing Procedure" (ACT Airlines SOP p. 3.11) runs as follows:

"The distance required for the descent is approximately 3 nm /1.000 feet altitude loss for no wind conditions using ECON speed. The rate of descent is dependent upon thrust, drag, airspeed schedule and gross weight. Normally, descend with idle thrust and in clean configuration (no speedbrakes). The speedbrake may be used to correct the descent profile if arriving too high or too fast."

The distance between the RAXAT and TOKPA reporting points is 27 nm. According to the approach chart and the controller's instruction, at this distance it was necessary to descend to 12 000 ft. Thus, the given distance was not enough for the standard descent procedure, required by ACT Airlines SOP (in this case the aircraft would descent approximately 9000 ft lower and would reach TOKPA at approximately 9000 ft instead of [the required] 6000 ft). Despite this, the initial descent was conducted without the use of the speedbrakes and with the increase of the IAS. After the initial descent, the MCP IAS was reset from 262 kt first to 270 kt and then to 280 kt.

...

At 01:08:03, the FO reported the setting to RWY 26 ILS: "India Bravo Kilo ILS two hundred fifty five tuned identified". The PIC confirmed: "Check."

At 01:08:20 (1 minute 40 seconds after the start of the descent), the "V/S MODE OPER" mode was engaged in the longitudinal channel. By this time, the vertical speed of descent had been reduced to 1700 ft/min, the IAS increased to 290 kt. After the change of the autopilot mode, the vertical speed of descent was set to 2400 ft/min and the IAS was set to 280 kt. The autopilot maintained the vertical speed, set by the crew, while the IAS continued to increase (the maximum value of 317 kt was recorded at 01:09:53).

As it has been stated previously, there was a significant increase of the IAS during the descent. Judging by the report (at 01:09:00), the PIC was conducting the descent at the increased speed on purpose: "I will maintain high speed below ten thousand feet, I will correct it later." The PIC gave no explanation of his decision. According to ACT Airlines SOP, the maximum IAS below 10 000 ft is 250 kt.

At 01:09:20, at 12 200 ft, the manual deployment of the speedbrakes (spoilers) was recorded. The speedbrake control handle was initially set to 30¡ã, and after 30 seconds it was set to 36¡ã. The further descent was conducted with the deployed spoilers.

At 01:10:00, the engagement of the "FLIGHT LEVEL CHANGE OPER" A/P mode was recorded in the longitudinal channel. 26 seconds later, the IAS was set to 250 kt. The mode change caused the IAS to decrease simultaneously with the decrease of the vertical speed to 770 ft/min.

...

The flight past TOKPA was performed at 01:11:18, the aircraft reached 9200 ft in descent with the IAS of approximately 270 kt. According to the chart, the flight past TOKPA must be conducted at 6000 ft (FL 060) or above. Thus, the chart requirements were not formally violated, but the aircraft was too high for continuing the approach in accordance with the chart withoutadditional maneuvering for decreasing the altitude.

On request from the investigation team, the specialists of one airline company, operating Boeing 747 aircraft, have conducted the analysis of the aircraft's descent profile during the accident flight. According to the provided conclusion, taking into consideration the short distance available for the descent, the PIC's actions on maintaining the increased IAS and the absence of the speedbrakes application at the beginning of the descent were found incorrect. For the execution of the descent in accordance with the PIC's decision to reach FL 060 at TOKPA and the controller's instructions, the speedbrakes must have been applied at the beginning of the descent and the significant increase of the IAS must have been prevented. Even following this requirements, the deficiency of distance and the surplus of altitude are evident; thus, at a distance of 5¡­8 nm prior to the TOKPA reporting point, the landing gear must have been extended, the flaps must have been positioned at 5¡ã (with the speedbrakes remaining in the deployed position), and the IAS must have been decreased to 210 kt. This would have provided a steeper descent profile.

After passing TOKPA waypoint an "AP CAUTION" warning occurred because 17 seconds after the autopilot had issued a stab trim command no stab trim movement had occurred (previous response times had been 3-4 seconds). The AP Caution disappeared after 7 seconds when the stab trim started to move (during the remainder of flight no further trim delay occurred).

The MAK analysed:

After flying past the TOKPA reporting point, the AP R disengagement, followed by almost immediate AP C engagement were recorded. The crew's actions on autopilots' switching may have been related to the triggering of the "AP CAUTION" annunciation. The autopilots' switching occurred 17 seconds after the "AP CAUTION" annunciation disappeared. The flight was continued in the automatic mode.

At 01:11:45, the controller informed the crew on the transition flight level (FL 060), the QNH (1023 hPa) and cleared for the ILS-approach to RWY 26: "Turkish six four niner one, transition level six zero, QNH one zero two three hectopascals, cleared ILS approach runway two six, call me on localizer". The crew confirmed receiving the information. At 01:12:00, at 8300 ft (which was higher than the transition flight level), the crew set the QNH12. At 01:12:07, at the speed of 250 kt and the distance of 12.5 nm, the flap control handle was set to 1¡ã.

At 01:12:36, on request from the crew, the controller cleared for further descent to 3400 ft:

"Turkish six four nine one, continue descend three thousand four hundred feet report localizer established".

At this moment, the aircraft was passing the altitude of 7900 ft in descent with the IAS of 240 kt. On receiving the controller's clearance for further descent, the crew set the selected altitude to 3400 ft.

At 01:12:51, with the speed of 240 kt and at the distance of 9.8 nm, the crew started to extend the flaps to 5¡ã, after which the performance of the Approach Checklist was recorded. The crew identified the three-letter code of RWY 26 ILS (India Bravo Kilo), checked the selected QNH and performed the check of the altimeters¡¯ indications. The Approach Checklist was performed in full, no crew findings related to the operation of the aircraft equipment were recorded.

...

At 01:13:36, at the speed of 220 kt and the distance of 7.2 nm, the flap control handle was set to 10¡ã. When extending the flaps, the FO expressed his worries about the possibility of performing the approach: "Speed check, flaps ten. We may end up high and we have speed as well." There was no response from the PIC, and the FO did not raise this issue again. The aircraft continued the descent to the D 6.4/R-080 MNS 3400' guidepoint of the approach chart.

After the flaps were extended to 10¡ã, the descent continued with the deployed spoilers (the spoiler control handle remained in the position of 36¡ã). According to ACT Airlines SOP, to prevent buffeting, the spoilers should not be applied with the flaps extended to more than 5¡ã.

At 01:13:55, the "LOC MODE ARM" mode was engaged, and at 01:14:05 the localizer capture and the start of the automatic positioning for approach were recorded. At that time, the aircraft was positioned at the distance of approximately 6.1 nm in the descent mode, the flight altitude was approximately 5700 ft and the IAS was 200 kt.

The localizer capture was monitored by the crew. (FO: "Localizer capture." PIC: Approach mode is selected." FO: "Glideslope arm.") These actions of the crew were in compliance with the Boeing 747 FCOM and ACT Airlines SOP requirements. The crew did not report to the controller about the localizer capture.

After the localizer capture, the "G/S MODE ARM" mode was engaged, and at 01:14:08 the three autopilots were engaged. The flight continued in the automatic mode: the "LOC MODE OPER" was engaged in the lateral channel, and the "FLIGHT LEVEL CHANGE OPER" mode was engaged in the longitudinal channel. The selected altitude was set to 3400 ft (which corresponds to the altitude of the glideslope capture). According to the approach chart, this altitude must be reached at the distance of 5.4 nm and must be maintained until the distance of 3.2 n (the point of glideslope capture).

At 01:14:18, on the PIC's command, the FO started to extend the landing gear. At that time, the aircraft was positioned at the altitude of 5300 ft near the D 5.4 MNS 3400' guidepoint. Thus, the crew failed to perform the descent in compliance with the approach chart: the aircraft was positioned 1900 ft higher than required by the approach chart.

At 01:14:29, at the speed of 190 kt and the distance of 4.8 nm, the flap control handle was set to 20¡ã. During the process of the flaps extension to 20¡ã, the spoilers were retracted.

At 01:14:36, the audio annunciation about the reached height of 2500 ft AGL was recorded. The FO checked the reached height: "Checked". No report from the PIC followed. In accordance with the ACT Airlines SOP p. 4 "Low Visibility Operation" requirements, in case of this annunciation triggering, the aircraft control must be provided by the both pilots.

On flying to the glideslope capture point, the PIC commanded the FO to concentrate his attention on the environment outside the cockpit:

PIC: "Look, actually you can see the (runway/dirt) down there."
FO: "I¡¯m monitoring the instruments."
PIC: "Instruments, I monitor the instruments, you look outside."
FO: "He, he." (expressing agreement).

The PIC's command was in violation of the ACT Airlines SOP p. 4 "Low Visibility Operation" requirements and made the FO stop the continuous monitoring over the instrument flight performance, while the actual weather conditions did not allow to establish the visual contact with the RWY or with the ground references at this stage of the flight.

At it has been mentioned above, the glideslope capture point is at the distance of 3.2 nm an at the height of 3400 ft. At the above-mentioned distance, the aircraft was crossing the altitude of approximately 4000 ft, the IAS was 190 kt, with the flaps positioned at 20¡ãand the landing gear extended. No automatic glideslope capture occurred, probably, because the aircraft was positioned significantly higher than required.

According to the Boeing 747 FCOM and ACT Airlines SOP, at the point of Final Approach Fix (FAF), the crew must check the flight altitude. The onboard recorder contains no data confirming that the crew performed this procedure. For Manas aerodrome (Bishkek), this point corresponds to the Final Approach Point (FAP D 6.0 143400' and is provided in the RWY 26 approach chart of AIP of the Kyrgyz Republic (Figure 24)). In Jeppesen Route Manual, which was used by the crew, the FAP/FAF is not shown.

No report on the height of flying past the reporting point was recorded. However, judging by the internal communication, at that stage the crew members were aware that the aircraft was higher than the required glidepath: (PIC: "Fuck, he left us high, fucking faggot15." The FO made an attempt to calm down the PIC: "Come on, nothing happened.") This communication took place after the aircraft flew past the FAP at the distance of approximately 3 nm. Despite the evident deviation from the approach pattern, again, no corrective actions were performed by the crew.

During the standard approach, the glideslope capture occurs during the horizontal flight with the aircraft positioned below the glidepath. According to the Boeing 747 FCTM, the glideslope capture from above (in a descent mode) should be applied only in some cases and requires the crew to exercise higher level of attention and control.

In fact, the Boeing 747 FCTM recommendations were not followed by the crew during the accident flight. The aircraft had already passed the FAF.

At 01:15:06, at the speed of 190 kt and the distance of 2.7 nm, the flap control handle was set to 25¡ã.

At 01:15:25, the aircraft performed the descent to the altitude of 3400 ft, the autopilot "ALT HOLD OPER" mode was engaged in the longitudinal channel. Thus, the altitude of glideslope capture was reached only at the distance of ¡Ö1.7 nm (while the FAP is located at the distance of 3.2 nm). The further flight was performed at the constant altitude almost along the RWY 26 centerline. The aircraft was positioned considerably higher than the glidepath; according to the recorded information, the value of the "deviation from the equisignal glidepath zone" was +4¡­+5 dots, ( the digit "+" corresponds to the position of the aircraft above the glidepath). The "G/S MODE ARM" was engaged; however, no glideslope capture occurred.

At 01:15:31, the aircraft's crossing of the outer marker (OM) in cruise flight at 3400 ft was recorded (according to the approach chart, the altitude for the OM crossing is 2008 ft). Neither audio annunciation nor comments from the crew on flying past the OM were recorded on the CVR. The crew can disengage the audio annunciation; however, in any case, the crossing of the marker beacon is indicated on the PFD. The disengagement of the audio annunciation on crossing the OM violated ACT Airlines SOP and contributed to the fact that the crew did not monitor the aircraft altitude on flying past the OM.

The tower controller is not required to monitor the aircraft position in related to the approach pattern via radar surveillance, he is only required to visually monitor the aircraft within his visibility range.

The MAK wrote:

In this case, the controller could not visually monitor the
aircraft. Moreover, according to the controller's explanatory report, he had not been trained and authorized for the operations with the data provided by the radar surveillance system. According to his statement, at the time of the aircraft approach, the data on the current flight plans and of the aircraft parking areas were displayed on the screen, no data from the radar surveillance system were displayed. The investigation team points out that, as the on contacting the controller, the aircraft had already passed the LOM, but had not start the descent along the glidepath. The presence and use of the radar surveillance system data and the presence of the corresponding requirements in the controller's working instructions could have provided the Tower controller with the opportunity to reveal the non-landing position of the of the aircraft and to inform the crew about this issue.

At 01:15:50, at the speed of 175 kt and the distance of approximately 0.4 nm, the flap control handle was set to 30¡ã. At the time of the flaps were extending to 30¡ã, the FO was conducting the radio communication with the Tower controller. Probably, the flaps had been extended by the PIC himself (PIC: "Thirty", with no response from the FO) or by the FO, but without the appropriate callout.

The flap extension into the landing configuration was started after the pointer of the glidepath deviation indicator started to move from its end position, but before the actual capture of the signal from the glideslope beacon. According to ACT Airlines SOP p. 3.11, "Descent, Approach and Landing Procedure" and Boeing 747 FCOM, "Landing Procedure ¨C ILS", the flap extension to the landing configuration must be performed after the glideslope capture.

The capture of the signal from the glideslope beacon occurred at 01:15:52, at that time the aircraft was positioned at the distance of approximately 1.15 nm form the RWY 26 threshold, the angle of the caught glideslope beacon signal was ~9degrees, the flight altitude was 3400 ft. The nominal glidepath angle, required by the approach chart is 3¡ã. Thus, in the accident flight, the capture of the "true" glideslope signal could have been performed only "from above", that is the aircraft must have been in the descent mode. In fact, the glideslope signal capture was performed in the aircraft in cruise flight; however, the crew did not pay proper attention to this fact.

The analysis of the disposed information may lead to a conclusion that the onboard system of the aircraft captured the "false" glideslope signal with the angle of ~9degrees. Besides, the aircraft design company confirmed that the glideslope signal capture was conducted in a "standard" way, as all the conditions, under which the on board equipment determines that the aircraft is approaching the equisignal zone, were fulfilled. The feature of monitoring the aircraft position in relation to the RWY is not provided in the glideslope capture algorithm by design.

The PIC's and the FO's navigation displays (NDs) were operating in the "MAP" mode with the scaling of 10 nm. The aircraft was in the landing configuration: the landing gear was extended, the flaps were positioned at 30¡ã. The recorded internal communication confirm that at the glideslope signal capture and the start of descent, the crew did not monitor the distance to the RWY threshold and the distance from the VOR/DME beacon. Moreover, it must be pointed out that the Boeing 747 FCOM and ACT Airlines SOP do not require any monitoring of altitudes or distances (the altitude check is required only during the FAF crossing). The crosscheck of the altitude and distance is required only by 747 FCTM in case the crew suspect they are descending along the "false" glideslope.

After the glideslope signal capture, despite the fact that the "false" glideslope has the "reversed polarity", the aircraft was automatically set to the descent mode with the vertical speed of up to 1425 ft/min.

At 01:16:01, at the altitude of 3300 ft, the inner marker (LIM) beacon crossing was recorded (the altitude of the LIM crossing in accordance with the approach chart is 2290 ft). Neither the audio annunciation nor the crew's report on the LOM crossing was recorded by the CVR, while the LIM crossing (as the LOM crossing earlier) is indicated for the crew on the PFDs. The crew did not check the LIM crossing altitude.

Six seconds after the glideslope capture, the triggering of the "LAND 3" annunciation was recorded, which indicates that the autoland system was engaged in a triple redundant fail-operative configuration, suitable for use during the landing approach in the ICAO CAT III conditions.

The crew checked the capture of the glideslope beacon signal; the PIC's callout: "Glideslope... capture." and the FO's response: "Check. Four thousand four hundred" are recorded. The altitude of 4400 ft was selected on the MCP in accordance with the operational procedure and was in compliance with the go-around altitude, required by the approach chart. The PIC also checked the annunciation on the armed automatic landing mode: "Land three." At this stage of flight, ACT Airlines SOP p. 3.11.4, "Landing Procedure ¨C ILS" and the Boeing 747 FCOM, "Landing Procedure ¨C ILS" require the performance of Landing Checklist. This checklist was not performed by the crew.

During the time period of 01:15:56¡­01:15:59, the aircraft crossed the glideslope beacon, which, however, did not cause the loss of the stable signal receiving on board the aircraft (which is confirmed by the "GLIDESLOPE DEVIATION VALIDITY" discrete signal, recorded on the FDR), thought the signal itself was changing within the range of -4 to +4 dots (Figure 35). In fact, during the entire subsequent flight, the values of the glideslope deviation were considerably higher than one dot. The similar situation (the fluctuating trend of the changes in glideslope deviation up to the maximum values) was observed during this flight phase on the instruments onboard the Da-42 aircraft, which performed the flight tests in accordance with the special program in the frame of accident investigation.

The recorded internal communication confirm that the crew did not monitor the values of the glideslope deviation. According to ACT Airlines SOP, one of the requirements for performing the ICAO CAT II approach is the glideslope deviation of no more than one dot.

At 01:16:07, 15 seconds after the glideslope signal capture at the altitude of 3150 ft (the radio height was approximately 1000 ft), the FDR recorded triggering of the "AP CAUTION" (the autopilot caution annunciation) and "FMA FAULT 2" discrete. This discrete signal corresponds to ¡°AUTOPILOT¡± caution-level EICAS message and glideslope mode fail indication (amber line drawn through the glideslope mode annunciation on the FMA) respectively. The triggering of these discrete signals was accompanied by the corresponding audio alerts (4 ¡®beeps¡¯ in a row). The recordings of these discrete signals lasted almost up to the end of the flight (up to the time the "FLARE" mode of the autopilot was engaged.

No reports from the FO related to the triggering of the above-mentioned annunciations were recorded, which was in violation of the ACT Airlines SOP requirements. As it has been mentioned above, the constant instruments and annunciations monitoring could have been prevented by the PIC's direct command to the FO to concentrate mainly on the environment outside the cockpit: "Instruments, I monitor the instruments, you look outside".

According to the explanation, provided by the Boeing Company, the "FMA FAULT 2" discrete signal recording indicates that the automatic system has determined the "PITCH MODE FAILURE" occurrence, that is the failure to maintain the capture of the glideslope beacon signal.

In this case, the following annunciation is provided:
- the Flight Director pitch bars are removed from the PFDs;
- the "G/S" indication (about maintaining the glideslope) on the PFDs (FMA) are crossed with an amber line;
- the two "MASTER CAUTION" lights come on;
- the "MASTER CAUTION" audio alert is triggered;
- the amber "AUTOPILOT" caution annunciation is triggered on the EICAS display.

It should be pointed out that the Boeing 747 FCOM and FTCM do not provide a complex and complete description of the "PITCH MODE FAILURE" annunciation triggering situation; the data about the light and audio annunciations are generalized and are provided in different sections (for example, in the "Automatic Flight ¨C Controls and Indicators, PFD Flight Mode Annunciations (FMAs)" of FCOM. No direct instructions for the pilots' actions in case of this situation are provided either.

At the same time, the autopilot continues to operate. In the pitch control channel, irrespectively to the factual glideslope angle at the exact aerodrome, the autopilot will maintain the descent track with an inertially derived approximate angle of 3¡ã ("INERTIAL PATH"), its calculation starting from the point at which the incapability of following the glideslope signal had been detected for the first time. The flight with this constant angle of descent will continue until either the glideslope beacon signal, which the autopilot may follow, appears or the crew override the autopilot control either by the autopilot disengagement or by the go-around actions (by pressing the "TO/GA" pushbutton). With no crew's overriding actions, the aircraft will follow the inertial path until the "FLARE" mode is engaged. The armed "LAND 3" (or "LAND 2") automatic landing mode will also continue to be displayed. According to the designer's information, the abovementioned feature which allows the autopilot to continue the approach in case of the loss of the valid signal of the glideslope or localizer beacon is common for the following Boeing aircraft types: 737 (CAT IIIb Fail Operational aircraft only),747-400/-8, 757, 767, 777 and 787
.
The analysis of the recorded data confirm that almost immediately after the glideslope signal capture, the onboard system of the aircraft determined the incapability to follow this signal. Besides, according to the information, provided by the Boeing Company specialists, the 15-second delay in triggering of the caution annunciation is purposely designed. The behavior of the alert is described in the FCTM, including the fact that anomaly detection is not annunciated to the crew if it is of short duration. However, the information about the exact the length of the delay (which differs depending on whether the airplane is above or below Alert Height), is found missing in any documents available for the crew or the airline (for further information refer to the note below).

According to the Boeing 747 FCTM recommendations, in this case (the loss of the glideslope signal while descending along the glidepath), no immediate action are required from the crew, except for the situations of the autopilot abnormal operation. In the accident flight, the aircraft was following a stable descent track at the angle of approximately 3¡ã with the maintained annunciation on the armed automatic landing mode. Besides, the Boeing 747 FTCM sets the following limit: no continued approach below the weather minimums unless the adequate visualreference with the runway environment is established.

The MAK stated with reference to algorithms used by the ILS explained in the following paragraph:

Thus, the investigation team points out to some logical contradiction between the aim (to reduce the quantity of go-around events in case of temporary loss of the ILS signal) for which the aircraft designer (the Boeing Company) has developed the algorithm of transition to the flight along the inertial path, and the pilots' CAT II and CAT III landing training program, authored by the Boeing Company.

The MAK quotes the FCOM with respect to ILS interference and temporary loss of signal:

5.19. The AFDS includes a monitor to detect significant ILS signal interference. If localizer or glide slope signal interference is detected by the monitor, the autopilot disregards erroneous ILS signals and remains engaged in an attitude stabilizing mode based on inertial data. Most ILS signal interferences last only a short time, in which case there is no annunciation to the flight crew other than erratic movement of the ILS raw data during the time the interference is present. No immediate crew action is required unless erratic or inappropriate autopilot activity is observed.

"5.20. If the condition persists, it is annunciated on the PFD. If the autopilot is engaged, annunciations alert the flight crew that the autopilot is operating in a degraded mode and the airplane may no longer be tracking the localizer or glide slope. When the condition is no longer detected, the annunciations clear and the autopilot resumes using the ILS for guidance."

The MAK analysed that after descending through 300 feet AGL the crew received a total of 5 glideslope hard and soft warnings. The MAK annotated: "Besides, as in the case with the glideslope capture algorithm, the aircraft position in relation to the RWY is not assessed. In fact, at the moment of the alert generation, the aircraft was positioned above the RWY closer to the RWY end."

The MAK analysed:

Despite the constant receiving of the significant deviation from the glideslope beam, the "GLIDESLOPE" caution annunciation was aborted at the AGL height of approximately 200 ft, which is higher than the lower threshold of the annunciation triggering. The alert generation was aborted because the onboard system detected that the signal about the glideslope deviation was no longer valid (this is confirmed by the fact that the "GLIDESLOPE DEVIATION VALIDITY" discrete signals, recorded by the FDR, were aborted). Until this moment, the signal received by the onboard equipment was considered valid. The direct indication on the signal validity/invalidity is not provided for the crew. On the other hand, in the FCOM section entitled "Flight Instruments, Displays ¨C Control and Indicators ", p. 10.10.25, it is stated that the glideslope pointers (indicating the aircraft position relative to the glideslope) are in view only when the glideslope signal is received. Thus, the FCOM contains information only on receiving the glideslope signal, not on the signal validity/invalidity.

The crew did not respond to the "GLIDESLOPE" caution annunciation. It should be pointed out that, during the Approach Briefing, the PIC made a briefing on the necessary actions in case of the ILS signal deviation: "If we get an ILS deviation above one thousand feet, sorry, below one thousand feet do a go-around in case of ILS deviation."

Despite this, the crew performed no go-around procedure. Later on, the EGPWS provided the information only about reaching the specified altitudes and minimums.

...

Thus, during the accident flight on 16.01.2017, the Boeing 747-412F TC-MCL EGPWS was serviceable and was operating normally, in accordance with the designed algorithm.

At 01:17:04.6, the EGPWS aural alert on reaching the radio height of 100 ft was triggered, at this time the decision height was 99 ft. 2 seconds later, the FO reported: "Minimums."

At 01:17:07.7, the PIC reported negative visual reference with the RWY environment and commanded to perform the go-around: "Negative, go around."

At 01:17:09, at the height of approximately 60 ft, the autopilot "FLARE" mode was engaged, and half a second later, at the height of 58 ft, the pressing of the go-around pushbutton was recorded (Figure 35).

The engagement of the go-around mode caused the engine thrust increase, which, together with the control column pitch-up input, resulted in the vertical load factor of 1.4¡­1.5 g and almost inhibited the aircraft descent rate. However, 3¡­3.5 seconds after the go-around pushbutton pressing aircraft touched down past the runway end (the FDR discrete signals for the landing gear compression were recorded). At the moment of the touchdown, the aircraft ground speed was approximately 165 kt.

The first touchdown occurred with the insignificant vertical speed and almost without a banking angle at the distance of ~930 m from the RWY 26 end. Directly prior to the aircraft touchdown, the right wing tip impacted trees. As a result of the aircraft impact with the trees, the RH wing flight control devices started to disintegrate. After moving 20 m further on the ground surface, the aircraft main landing gear and engines impacted the aerodrome concrete barrier. At the distance of approximately 160¡­170 m from the place of the first touchdown, the second aircraft touchdown occurred, during which the aircraft right wing and engines impacted the buildings. As a result of the impacts with the obstacles, the aircraft disintegrated, a significant part of the aircraft structure was destroyed by the occurred ground fire.

The MAK stated: "It should be pointed out, that the crew noticed and responded to all the changes in the PFD indications, related to the "desired" parameters. At the same time, they did not react on the "undesired" changes, even if these changes were accompanied by the audio annunciations."
Incident Facts

Date of incident
Jan 16, 2017

Classification
Crash

Aircraft Registration
TC-MCL

Aircraft Type
Boeing 747-400

ICAO Type Designator
B744

This article is published under license from Avherald.com. © of text by Avherald.com.
Article source

You can read 2 more free articles without a subscription.

Subscribe now and continue reading without any limits!

Subscribe now
Are you a subscriber? Login
Subscribe

Read unlimited articles and receive our daily update briefing. Gain better insights into what is happening in commercial aviation safety.

Compare our plans

Send tip

Support AeroInside by sending a small tip amount.

Send tip

Newest articles

Subscribe today

Are you researching aviation incidents? Get access to AeroInside Insights, unlimited read access and receive the daily newsletter.

Pick your plan and subscribe

Partner

Blockaviation logo

A new way to document and demonstrate airworthiness compliance and aircraft value. Find out more.

ELITE Logo

ELITE Simulation Solutions is a leading global provider of Flight Simulation Training Devices, IFR training software as well as flight controls and related services. Find out more.

Blue Altitude Logo

Your regulation partner, specialists in aviation safety and compliance; providing training, auditing, and consultancy services. Find out more.

AeroInside Blog
Popular aircraft
Airbus A320
Boeing 737-800
Boeing 737-800 MAX
Popular airlines
American Airlines
United
Delta
Air Canada
Lufthansa
British Airways