Fatal Go Around at Tatarstan
On the 17th of November in 2013, the Boeing 737-500 registered in Bermuda as VQ-BBN departed Kazan Airport for scheduled flight TAK 363 to Moscow’s Domodedovo Airport (the largest airport in Russia and, as of 2014, the largest airport in Eastern Europe). The aircraft departed normally at 11:22 (UTC) and landed at Domodedovo at 12:43. After the passengers disembarked, the aircraft was prepared for flight TAK 363 back to Kazan.
Both of the flight crew had the required pre-flight rest period. However, looking at the roster for the eighteen months before the accident, it is clear that both pilots had flown a great amount of overtime without enough rest and without a real break (the captain had 111 days leave and the first officer had 275 days leave). It’s likely that both pilots were suffering from accumulated fatigue at the time of the flight, which means that they would have been quicker to feel fatigued and prone to absentmindedness.
At 14:15 the flight crew started up the engines and the flight data recorder began recording the flight parameters.
The aircraft departed Domodedovo at 14:25 with 7,800 kg of fuel, which was more than enough for the planned flight. It appears that initially, the First Officer was the Pilot Flying, with the Captain likely taking over as Pilot Flying at 14:47, when Autopilot B was disconnected and Autopilot A engaged.
The flight crew listened to the ATIS after they left the Moscow area control centre and entered the Kazan ATC control area. The ATIS information Juliet, recorded at 14:42, was as follows:
Wind 220 degrees 9 m/sec gusting 12 m/sec, at 100 m 230 degrees 8m/sec, at circuit height 250 degrees 16m/sec, visibility over 10 km, moderate sleet, overcast (8 oktas), cloudbase 270 m, temperature 3 degrees, dewpoint 2 degrees, QFE 735 mm/980 hPa, moderate icing in clouds, no significant changes.
A rough summary of this would be that the weather was not great but the visibility was good and there was no reason for concern.
At 15:00, the aircraft passed the SHUMERLYA waypoint and contacted Kazan Control.
Tatarstan 363: Kazan Control, Tatarstan 3-6-3, good evening, Shumerlya, flight level 2-9-0, ready for descent.
ATC: Tatarstan 3-6-3, Kazan Control. Descend flight level 7-0, identified.
The crew conducted a landing briefing and agreed that they would make an ILS approach to runway 29. The actual weather was compatible with the captain’s weather minima and suitable for landing. They did not discuss the aerodrome navigation aids and lighting. The captain didn’t distribute duties to ensure it was clear who was doing what, which should be standard procedure as a part of the landing briefing. However, it was clear that the captain was the Pilot Flying and the first officer would cover the radio communications.
The first officer interrupted the captain’s briefing to contact Kazan Transit to report their arrival; another deviation from standard procedure.
The aircraft began to descend at 15:01. Local time at Kazan was 19:01 and night was falling.
The Boeing 737-500 was not equipped with a GPS receiver. On the Boeing 737, the Flight Management Computer determines the aircraft’s current position using inertial reference system (IRS) data as well as data from ground navigation aids. The inertial systems provide data on current position which does not depend on ground navigation. Before the flight, the crew enters the current aircraft position. The important thing to understand here is that the inertial reference system shifts during flight, leading to an error accumulation of up to two nautical miles per hour of flight.
In order to correct this error accumulation, the Flight Management Computer uses signals from ground navigation aids. These corrections are done continuously with signals from DME beacons, VOR beacons or localizer beacons at an aerodrome (which must be set manually). Signals received from two or more DME beacons at the same time are treated as the most accurate for corrections. The correction is applied to the Flight Management Computer data, not to the inertial reference system.
This means that the accuracy during different phases of flights depends on the correct IRS alignment before the flight and continued updates to the aircraft’s actual position using radio navigation aids.
The Flight Crew Operations Manual states that a single Flight Manage Computer is not certified as an autonomous source of navigation information; in order to achieve the required navigation accuracy, the system should be used in an “accurate radio navaid environment.” Thus, if there are not enough DME and VOR beacons on the flight path to correct the data, the accuracy of the Flight Management Computer may not be sufficient for safe navigation. The system offers warnings to the crew if the required accuracy is not achieved, in which case the crew must use all onboard and ground means of navigation available rather than rely on the Flight Management Computer data.
For this aircraft type, flights through Russian airspace are only conducted along routes with permanent Secondary Surveillance Radar.
At 15:06, the Kazan air traffic controller noticed that the blip on his radar was to the left of the flight track centreline. He asked the flight crew to report their heading. The first officer responded that they were on heading 068°.
The Kazan controller responded: “Tatarstan 363, Kazan, for your information, according to my data, you are going 4 km left of the track.”
The first officer acknowledged the information and thanked the controller.
The crew discussed the fact that the Flight Management Computer was inaccurately showing the aircraft position, an issue known as map shift. They changed from Lateral Navigation (LNAV) to Heading Select (HDG SEL) mode. The heading was changed from 68° to 73°. This correction was not enough to capture the track axis; they continued the flight with an offset 4 km to the left.
Four minutes later, the first officer commented to the captain that a warning had lit up saying IRS NAV ONLY. At that moment, the controller asked them to contact the adjacent sector.
The first officer contacted Radar Control and the flight was cleared for a descent to 500 metres (1,640 feet) above the aerodrome (QFE) for the Instrument Landing System approach to runway 29.
The crew set their altimeters for the destination aerodrome and completed their checklists. Then they made another corrective left turn of 8°. This turn made things worse.
The controller noticed the turn but he misunderstood their intentions; it made no sense based on the standard approach pattern. He came to the conclusion that the flight crew were starting to turn base and immediately called them to say, “Tatarstan 3-6-3, early for base turn.”
The aircraft continued. The crew were presumably equally confused, as they weren’t turning base yet, they were just correcting their downwind heading. The flaps were set to 10° as they descended through 800 metres (2,600 feet).
The flight crew called that they had reached the base turn point. The controller cleared them for the base and final turn. The turn was initiated at 15:18 in the HDG SEL mode with a significant deviation from the standard pattern, which would have been visible to ATC on radar.
For some reason, ATC did not comment that the aircraft was not where it should have been.
It also would have been obvious to the crew if they had been using the navigational aid at the aerodrome. The VOR/LOC is the VOR/Localizer Capture Mode, which is used to intercept the localizer in an instrument approach.
The crew did not engage the VOR/LOC mode as they crossed the localizer centreline, presumably because they couldn’t because they were too far off course. To make matters worxse, they were now also suffering from a strong tailwind.
As they turned base, they reached the circuit height of 500 metres (1,700 feet) which was maintained by the autopilot. However, the turn was completed with significant “overturn” — rather than making a clean 90° turn, the aircraft continued to turn.
Based on the conversation in the cockpit, the crew only became aware of their position in the approach pattern at this point.
Captain: What’s this the fiddle?
First Officer: Wait, I’ll VOR/LOC it, what’ll that give?
Captain: Something strange. I’ve VOR/LOC’ed it
First Officer: 111 decimal 7, 111 decimal 7, so what? What’s that?
This sounds to me like they are both engaging VOR/LOC mode and not liking the results. It is failing because they are deviating too far; by the time they finished the turn, they were about four kilometres to the right of the approach course.
(I have no idea what the captain means by the fiddle.)
Radar Control handed the aircraft over to Tower Control, who advised them that they were 14 kilometres out and offset to the right of the approach course.
The controller didn’t specify how far to the right and the crew did not appear to have worked it out for themselves using on-board equipment. The deviation pointer (HSI LOC) was pegged hard to the left.
Nevertheless, the flight crew continued to prepare for landing.
The crew completed their Landing Checklist, configuring the aircraft for landing including turning on the landing lights. But those lights reflected off the clouds, producing a light screen, so they disengaged the lights again.
At 15:20:49, the captain said, “Well, we neither have the landing position anywhere, nor anything else. Just 4 miles to go… Now it’s going to appear… we’ll press the Go… Go-around”.
First Officer: Aha. Yes.
Captain: Nothing bad about it.
That’s right, there’s nothing bad about going around. Or at least, there shouldn’t be, even taking into account that they hadn’t briefed for a missed approach. They continued.
The flight crew continued to discuss their deviation from the glide path which could no lonbger be ignored. The first officer said, “It asks more to the left, that’s why I’m watching.”
The captain grumbled in response. “It has not even started to be alive, damn.”
The deviation pointers remained at their limits.
Kazan Tower asked if they were ready for landing. The captain responded in the cockpit, “Tell him we are ready.” The first officer did so.
They were most certainly not ready for landing. The aircraft was still four kilometres from the runway threshold and still at circuit altitude (500 metres/1,700 feet).
The first officer said, “It feels as if we are going the wrong way.”
The captain responded, “If something goes wrong, we can go around now.”
The PIC’s decision to continue the approach was apparently caused by his natural wish to land from the first approach. In spite of the lack of accurate aircraft position information with reference to the approach pattern and the runway he probably hoped they would finally be able to capture the glide slope and land.
Besides, there was a psychological aspect to the flight that has to be taken into account: the crew was aware of a VIP passenger on board, which according to the opinion of the pilots conducting the assessment of the accident flight could have added to crew’s stress. A go-around at the home base aerodrome in relatively plain meteorological conditions could have become a ground for a debriefing in the airline that might have revealed the erroneous actions of the crew during the approach. And though the PIC was not aware why they failed to intercept the LLZ timely, he could have very well supposed that it was due to some of their erroneous actions.
Meanwhile, it was apparently quite clear to ATC that the aircraft was nothing like in the right position for the landing. Someone could have informed the crew of their actual position and recommend that they go around.
Instead, the tower controller cleared them to land.
The flight crew engaged the vertical speed mode to descend at 1,200 feet per minute to 270 metres (900 feet), which the ATIS weather information had stated was the cloudbase. This means that the autopilot would descend and then maintain this height. It’s likely that the captain was hoping to break through the clouds and regain situational awareness by being able to see the ground and, hopefully, the runway.
Although they continued to discuss going around, described by the report as complacency than real readiness, the flight crew never really seemed to prepare for the reality of going around: the primary task remained to find the runway and land on it.
They captured the localiser at 15:22:17 at a distance of approximately two kilometres from the runway threshold.
First Officer: It’s unclear. I can’t see…
Captain: Wait, we’ll descend 270.
Captain: Is it there? Can you see anything, eh?
First Officer: Can’t see it yet.
At 1,000 feet above the ground level, the enhanced ground proximity warning system sounded with One thousand.
This is a cue, described in the standard operating procedures.
Landing procedure – ILS SOP B737 of Tatarstan Airlines Extract
At 1000 ft radio altitude
14. ANNOUNCE:“1000 STABILIZED NO FLAGS” or “1000 NOT
In other words, this is the moment where the pilots decide whether the aircraft is stabilised and either continue or, if not stabilised, break off the approach.
There’s some variance depending on the operator but generally to continue an approach, at 1,000 feet the aircraft must on the correct flight path flying at the correct speed and descending at the correct speed. If any of these elements are not established and maintained, the flight crew should abort the landing and go around.
The aircraft was significantly higher than the glideslope, turning right with a bank angle of 29°. This was not a stabilised approach by anyone’s definition.
The captain replied to this call-out with, “One thousand, stabilised, no flags.”
The first officer, as pilot monitoring, is obligated to verify that the aircraft is stablised and if not, to inform the captain. He didn’t.
It can be well assumed that during that flight segment the crew were not aware of the aircraft position with reference to the runway and glideslope as accurate as required at that phase. Moreover, part of their mind and attention was devoted to solving the appearing problems. This was reflected in deviations from the SOP. Trying and hoping to recover on their own and land the pilots did not realize that position can be monitored using VOR or NDB or that they could ask the ATC officer for assistance.
At that moment, the first officer spotted the runway. “Huh, that’s it, here, the runway underneath us. No, we’re too high. Four white lights, we’re too high.”
The four white lights are a reference to the PAPI lights, which in a correctly configured approach should show two red, two white lights. Three white lights means the aircraft is above the descent path. Four white lights means they are much too high.
The captain still didn’t have it in sight. “Where do you see it? I don’t. Where is it?”
First Officer: “Here it is, the runway. No. Go around. Go around.”
The captain never saw the runway and he followed the first officer’s advice. “Go around, report we are going around. Position unsuitable for landing.”
This decision was the right one, although late. The investigators discovered later that this is the first go-around the captain had done since he’d been upgraded to Pilot in Command.
As he made the call to go around, the aircraft was configured for landing (gear down, flaps set to 30°), travelling at 130-135 knots (IAS) level at 270 metres (900 feet) above the runway maintained by the autopilot. The autopilot was in the process of an S-shape manoeuvre to attempt to capture the landing track.
In a normal story, this would be the point when the catastrophe was avoided. The aircraft would pull away from the ground and follow the missed approach procedure. With more time to consider their situation, the crew may have noticed the map-shift or they may have moved on to another airfield with better weather where they could be visual with the runway at circuit height.
This is no normal story. As you can see, there’s already a lot of information as to how they got into this situation. Next week, we’ll look at what happened when they tried to get themselves out of it.
Happy Friday! You can read the rest of this incident here: