The Reckless Final Approach of US Bangla Flight 211
The crash of US Bangla flight 211 made international headlines when it occurred on the 12th of March 2018; however, it wasn’t until a month ago that the true sequence of events was made clear.
Landing was completed in sheer desperation after sighting the runway.
This is never a phrase you want to see in the final conclusion of a report, let alone one involving a commercial airliner. But that’s a direct quote from the conclusion of the final report on US Bangla Airlines flight 211 crash. The report came out a few weeks ago, finally shedding some light on the tragic crash on the runway which killed all four crew and 45 passengers in Kathmandu, Nepal.
There aren’t a lot of accidents in Nepal so there isn’t an advisory body like the NTSB or the AAIB on hand to deal with investigations. The Government of Nepal had to set up an Aircraft Accident Investigation Commission to deal with this accident. TSB Canada took responsibility for the technical decoding and analysis, including CVR, FDR, EGPWS, EMU etc.
The commission released their full report on the accident (see note below) on the 28th of January 2019 and wow, is it not what I expected when I first heard about the crash.
At the time, the media reported that there was some confusion between air traffic control and US Bangla flight 211 and it wasn’t clear which runway the flight was cleared to land on. The flight crew, who perished in the crash, were praised for narrowly avoiding crashing into the airport tower as they tried to regain control of the aircraft.
The reality, though, is somewhat different and there’s certainly nothing to praise the crew for in this one.
The aircraft (registration S2-AGU) was a Bombardier DHC-8, better known as a Dash 8, a twin turboprop regional transport passenger aircraft. This was a 400 series, a stretched Dash-8 which allows for up to 78 passengers. On that day, there were 67 passengers who boarded at Hazrat Shahjala International Airport in Dhaka, Bangladesh, for the scheduled flight to Kathmandu.
The captain was a 52-year-old ex-military pilot with 5,518 hours, 2,824 of which were on the Dash-8. He’d retired from the Bangladesh Air Force in 1993 after being declared unfit to fly, owing to his medical condition of depression. However, he was cleared medically in 2002 and there was no evidence that he’d suffered from the condition since his retirement from the military. There is a point, though, that his self-declaration form submitted in annual medicals did not mention this medical history, nor the fact that he used to smoke, even though both were meant to be listed. As a result, his medical examinations did not look for any symptoms of his previous condition. The captain had been with US Bangla since 2015, when he joined as a Pilot-in-Command and Check Pilot for the DHC8 Q400. He was described by his colleagues and operation staff as friendly, soft-spoken and level-headed. His students spoke positively about his teaching techniques.
Two weeks before the crash, he was chosen (along with two other pilots) to ferry the operator’s latest Dash-8 from Canada to Dhakar, based on his competence, his good nature and his ability to deal with situations in a mature way. He flew to Kathmandu on the 16th of February 2018, his last flight there before the crash that took his life on the 12th of March.
The first officer was 25 years old with 390 hours, 240 on type. She had two type ratings, Cessna 152 and Dash-8. She was the airline’s first female pilot and one of only 30 female pilots in Bangladesh. She knew the captain, as she’d done simulator training with him twice in 2017. This was her first flight to Kathmandu as crew and she was focused on learning every step during the flight with the much more experienced captain.
She was probably not the right person to attempt to take control once the captain lost the plot.
US Bangla Airlines, a privately owned carrier from Bangladesh, has its operational base at Hazrat Shahjalal International Airport. At the time of the accident, their fleet consisted of four DHC8 Q400 aircraft, including the accident plane, and four Boeing 737-899 aircraft.
On that day, at 12:41 local time (06:41 UTC), shortly before flight 211 was due to take off, Dhaka Ground Control contacted the flight crew about their Bangladesh Air Defense Clearance number. This was a new reference number which the Bangladesh Aviation Authority had made mandatory just a few weeks before. The flight crew responded with the number as listed on their flight plan.
For some reason, the controller asked the crew a second time if they had their Bangladesh Air Defense Clearance number. The captain contacted Operations to ask if they had the right number. The issue was quickly resolved and the US Bangla flight 211 departed Dhaka ten minutes later.
This was notable only in that the captain appeared very agitated and stressed about what was really a minor incident.
As the aircraft climbed away to its cruising altitude of FL240 (24,000 feet), operations spoke to another flight crew about their fuel on board. The captain responded without verifying whether the message was meant for him (it wasn’t) and again sounded upset and stressed.
The captain was the Pilot Flying. During the cruise portion of the flight, he began to tell the first officer his personal issues, repeatedly bringing up an issue with a colleague, another pilot at US Bangla, who had questioned the captain’s reputation at the airline. He was going to resign over this, he told his first officer, even though he had no other job lined up or any other income.
At one point, he appeared to be having an emotional breakdown. He then lit a cigarette, strongly forbidden at the airline. The captain had not disclosed on his medical that he was an ex-smoker and no one seemed to know he had started again.
Intermittently, he would put effort into coaching the first officer in a wide range of navigation and communication issues. During these conversations, his voice was calm and professional.
The captain spoke almost non-stop during the cruise, pausing only to ask the first officer if she was comfortable with what he explained. “Yes, very comfortable Sir,” she told him reassuringly.
About an hour into the flight, the flight crew contacted Kathmandu Control and about 15 minutes later requested their descent. The captain mentioned that he didn’t have the approach charts and made do with a very short briefing, using the first officer’s charts for reference.
It’s clear from the first officer’s comments during the briefing that she was confused by the approach procedures at Kathmandu and specifically did not understand the missed approach procedure. The captain didn’t attempt to explain but instead told her that he would brief the remaining items later. He handed her the charts and told her to clip them to her side console, which meant that he had no means of referencing them while performing the approach. He then went over selecting altitudes and minima for the descent, explaining that he was setting the radio altimeter so that the Extended Ground Proximity Warning System (EGPWS) would sound with “100 feet above” before they reached minimums.
ATC cleared the flight to FL160 (16,000 feet) with an estimated approach time of 08:26 UTC, that is, 19 minutes to run.
The first officer acknowledged and then transferred to Kathmandu Approach, who instructed the flight to descend to 13,500 feet and hold over waypoint GURAS.
The flight crew discussed the published holding pattern and procedure.
The captain helped the first officer to set up the navigation, including entering the holding patterns into the Flight Management System.
Tribhuvan International Airport is at the foothills of the Himalayas and cited in pilot polls of the most challenging airport (and also appears in the most hated list in a passenger poll run by CNN). There are two standard instrument arrivals: RNAV (RNP AR) approach and VOR/DME approach.
An IFR flight may be cleared to execute a visual approach provided that the pilot can maintain visual reference to the terrain, the reported ceiling is at or above the approved initial approach for the aircraft and the visibility is 5km or more. But the important point is that the pilot reports at the initial approach level that the meteorological conditions are such that a visual approach and landing can be completed; he needs to let the controllers know that he is flying a visual approach.
As they continued towards the GURAS waypoint, the captain explained various aspects of navigational aids and approaches. The approach controller instructed them to reduce their speed to “minimum clean”, that is, approach as slowly as safe, and cleared them to descend to 12,500 feet.
Then, instead of asking them to hold at GURAS as expected, the controller told them to descend to 11,500 feet and cleared the flight for the VOR approach for Runway 02, maintaining minimum approach speed.
The flight crew acknowledged the clearance. But somehow, both of them forgot to cancel the holding pattern that they’d programmed into the Flight Management System. The captain lit another cigarette.
The cabin crew confirmed that the cabin was secured and the first officer reported to the controller that they had arrived at the GURAS waypoint at 11,500 feet.
As the controller cleared them to continue, the aircraft turned left to enter the holding pattern. Both the captain and the first officer reacted immediately. The captain corrected the approach at the same moment as the Kathmandu Approach controller contacted them to say that they had turned off course.
At this point, everyone was on the ball.
The captain overrode the Flight Management System LNAV guidance and set the aircraft up for heading mode, selecting a heading of 027° in order to continue as cleared: this heading would allow them to intercept the approach path (the 202° inbound radial of the KTM VOR) at a 5° angle.
The airborne air data computer registered winds aloft data during this phase revealed strong westerly winds from 270 to 280 degrees at an average of 28 knots, seemingly drifting the flight path towards east with heading correction set at 027 degrees by the pilot in heading mode. In the meantime, the vertical flight profile had also deviated to be high as the descent was due; the PIC was distracted and the aircraft had gone to pitch hold mode reversion due to heading selection in HSI. The PIC commanded descent inputs in the FGCP pitch wheel to approximately 1300 feet per minute (fpm) to initiate descent.
The point here is that instead of flying the final approach course using the navigation system, the Dash 8 continued in heading mode on a heading 027°, slowly deviating from the approach path. They passed the GURAS approach fix to the right of their intended track.
Because of the change in clearance, everything was happening very quickly. The aircraft should have been in full landing configuration with the landing checklist completed before they reached the GURAS approach fix. Instead, the initial flap setting of 5° wasn’t set until they had passed GURAS and flap 15° wasn’t set for another two nautical miles (13 DME on the approach).
The First Officer quickly went through the landing checklist and the captain confirmed the landing gears were configured for landing by saying “gears down three greens” without ever looking to see the status. As it happens, the gears were NOT down and there were no greens. As they entered their ten-mile final, the landing gear unsafe tone started sounding. Neither pilot reacted.
As the aircraft continued in heading mode and with a high approach speed because the landing gears were still retracted, the deviation from the approach path swiftly increased.
Kathmandu Tower cleared the US Bangla flight to continue approach. At eight miles out, the captain accidentally set the minimums to 4,688 feet instead of 4,950 feet as published, although the radio altimeter had been set to call out 100 feet above early. The captain asked again for the landing checklist and the first officer responded that it was completed. Still, neither noticed the continuous sounds of the landing gear unsafe tone.
The first officer prompted the captain that they were 500 to 600 feet high on the final approach profile, raising her voice to be heard over the warnings. The captain remained preoccupied with the interception of the final approach course, without trying to re-engage the Flight Management System LNAV mode or to select VOR mode. The Dash-8 drifted further and further off course. This time, neither the flight crew nor the ATC controllers noticed.
The captain requested the landing checklist for a third time. The first officer confirmed, again, that they had already completed it. The landing gear unsafe tone continued to sound.
By now, the Dash-8 was reaching vertical descent rates of 1,700 feet per minute, in part because the aircraft hadn’t been configured for landing. The Extended Ground Proximity Warning System (EGPWS) sounded with sink rate and too low gears warnings repeatedly, attempting to alert the flight crew that they were descending more rapidly than expected and that they were very close to the ground to have not extended the landing gear.
They were not lined up with the runway and neither crew member had the runway in sight. The landing gear was not down. The cockpit was full of warning tones. Still, neither crew member seemed to consider breaking off the approach and starting again. As they flew past the missed approach point, Kathmandu Tower cleared the flight to land on runway 02.
The first officer finally noticed that the landing gear was not down and began to lower the gear.
The captain asked for the landing checklist for a fourth time.
By now, they had already flown past the threshold of runway 02 and were flying northeast along the eastern side of the runway. However, the captain was convinced that the runway was still ahead of them and commented that he would have expected to be visual with it by now. He continued to descend as he searched for it.
They passed the VOR beacon at the far end of the runway as the captain disengaged the autopilot, still expecting to be visual with the runway at any minute.
The controller at Kathmandu Tower suddenly noticed that the aircraft, which should have landed by now, had flown off course to the northeast corner of the airfield.
He contacted the flight to say they’d been given landing clearance for runway 02 but they appeared to be heading for runway 20. It didn’t occur to him that the captain was lost, but simply presumed that the flight crew must have misheard and were setting up to land in the opposite direction. He asked their intention and the captain replied that they would be landing on runway 02.
At that moment the controller was replaced by the Tower Duty Controller who took over the microphone. Looking at the location of the aircraft, he cleared the flight for runway 20, presuming that’s where the captain intended to land despite him saying runway 02. The Tower Duty Controller, having added to the confusion, was then replaced by the Tower Supervisor Controller.
At the same time, Buddha Air flight 282 was approaching in a 2 mile final for runway 02 and had been cleared to land.
The controller asked whether the US Bangla flight crew were “on VFR” that is, was the flight crew flying a visual approach rather than following the RNAV approach as planned.
The captain confirmed they were in VFR conditions. The controller instructed him to join right downwind of runway 02 and to watch for the landing aircraft on the runway. The captain immediately acknowledged this but continued to fly on his heading.
As the Dash 8 flew past, the controller asked him to confirm he was tracking runway 20. The captain, who still did not have the runway in sight, replied “affirmative”. Somehow, he remained convinced that runway 02 was still in front of him; the exchange with ATC had done nothing to signal to him that he was lost. Although the landing gear was now down, the EGPWS continued to sound with warnings regarding the various flight parameters which had been exceeded.
The aircraft was now flying straight towards the high terrain on the northeast side of the airport.
The tower controller became concerned that the US Bangla flight would conflict with the landing aircraft, so he asked the captain to orbit, that is, to hold position so that he would stay clear of the landing traffic.
The captain, who appeared to finally understand that something was not right, acknowledged the call and began to fly a manual right-hand orbit, still trying to work out where he was.
The Dash-8 descended to as low as 175 feet above the ground, with bank angles of 35 to 40°, triggering even more EGPWS alerts.
The captain complained to the first officer that he’d made a mistake as he was constantly talking to her. This was not really the mistake that he should have been worrying about.
Both flight crew were still trying to work out why they couldn’t find the runway. The first officer believed that it might still be in front of them but the captain wasn’t sure. He climbed again to 6,500 feet and then turned into a steep right hand orbit northwest of runway 20, with bank angles reaching 45° and descent rates over 2,000 feet per minute, triggering the EGPWS warnings again.
A pilot on the ground saw the Dash-8 manoeuvring and contacted ATC to report that an aircraft seemed to be disoriented and lost above the northwest sector of the airport.
The Buddha Air flight landed on runway 02 and cleared the runway, at which point the controller contacted US Bangla Air and offered the choice of landing on either runway 02 or runway 20, as it was impossible to make sense of what the Dash 8 was doing. He asked if they needed a vector service, where the controller would give the flight crew headings to follow to bring them to the runway threshold.
US Bangla 211: Yes, we would like to land on 20.
ATC: OK, runway 20 cleared to land, wind 270 degrees, 6 knots.
US Bangla 211: 260 copied, clear to land.
ATC: Bangla Sierra 211, confirm you have runway in sight?
US Bangla 211: Negative, sir.
ATC: Turn right and you have the runway. Confirm you have runway not in sight yet?
The first officer finally spotted the runway at their 3 o’clock position. The aircraft was at 5,500 feet and about 1.8 nautical miles from the threshold of runway 20.
Instead of breaking off and setting up for a stabilised approach, the captain began to turn, hard. Throughout, he also handled all radio calls, his voice sounding calm and collected.
US Bangla 211: Affirmative, we have runway in sight, requesting clear to land.
ATC: Bangla Sierra 211 cleared to land.
Captain: Cleared to land runway 02.
Note the repeated request to be cleared to land and the reversion back to runway 02, both symptoms of task saturation. He asked the first officer again, for the sixth time, if the landing checklist was complete. He then asked her to set the heading bug to 022°, instead of the 202° which would have put them on the right heading for runway 20.
As another aircraft approached the airport, ATC immediately asked them to hold position, trying to keep the area clear for the US Bangla flight.
US Bangla 211: I said sir, are we cleared to land?
It appears to me that the captain was so focused on trying to recover this landing, he was unable to take in any new information, thus repeatedly asking the same questions without ever hearing the answer.
At the same time, he flew 180° (west) towards the threshold for runway 20, thus not lined up for landing on the runway but in the wrong position to join downwind for a circuit to runway 02.
The control tower staff all watched the aircraft descend rapidly at an angle to the runway. Alarmed by the fact that the aircraft was still manoeuvring so close to the ground, the controller shouted “Take-off clearance cancelled!” which was not helpful from any point of view.
Neither flight crew acknowledged. The Dash-8 pulled up, still heading west, and turned steeply left with a very high bank angle. It continued over the western end of the apron and then continued south east past the ATC Tower at a very low height. The controllers in the tower ducked their heads, and one dropped to below the table, frightened that the Dash-8 was going to crash into them.
The Dash-8 missed the tower and passed over the domestic passenger terminal, clearing it by only 45 feet. The controllers simply stared, aghast.
The captain turned towards the taxi track and started a right reversal turn to line up for runway 20. The tower controller called “BS 211, I say again…” but was unable to finish the sentence.
As the aircraft overflew the threshold of runway 20 at 450 feet above the ground, it was flying 150 knots indicated airspeed with a westerly heading of 255°. The CVR has not been officially released but according to the analysis, the first officer sounded distressed and panicked although she did not make any call outs to go around or discontinue.
The Dash-8 touched down on one outer main wheel at a speed of 127 knots indicated airspeed with a heading of 190° and still in the turn, with a bank angle of 15°. The aircraft immediately veered southeast off the runway onto the grass. It crashed through the inner perimeter fence and down a rough slope, leaving a path of debris in its wake, the wings breaking under the strain. The engines and landing gears were just barely attached when the aircraft came to a halt some 442 meters (1,450 feet) from the runway, where it burst into flames; the tanks had about 2,800 kg of fuel in them.
The flight crew, two cabin crew and 45 passengers were killed on impact. The fire fighting services were dispatched within 16 seconds from touchdown and reached the crash site in under two minutes. They were able to rescue 22 passengers from the burning wreckage. Two more passengers died in hospital, bringing the deathtoll to 47 out of the 67 passengers on board.
It’s clear that there was something very wrong with the captain: an experienced pilot and instructor who was known for his calm nature and grace under pressure.
VAS aviation released the ATC audio last year, shortly after the accident but before there was any understanding of what had happened; the assumption at the time was that the controllers offering either 02 or 20 had confused the flight crew. You can listen along here:
The captain’s calls to the tower are clear and do not show any signs of stress, although he is not reading back correctly and repeatedly asks the same questions. It’s clear that he is unable to deal with the situation and fixated at landing the aircraft at all costs, rather than breaking off for a better approach.
However, the investigation report focuses not on the break in personality but on the unhappy incident that the captain brought up during the flight.
The PIC who was also the pilot flying was under stress and emotionally disturbed as he felt that the female colleague of the company questioned his reputation as a good instructor. This together with the failure on the part of both the crew to follow the standard operating procedure at the critical stage of the flight contributed to loss of situational awareness to appreciate the deviation of the aircraft from its intended radial that disabled them sighting the runway.
The investigation concluded that the captain / Pilot Flying, was under stress and “emotionally disturbed” because a “female colleague” questioned his reputation as a good instructor, rather than his upset being a symptom of his emotional breakdown.
I also find it odd that throughout the final report, the colleague is always specified as female, as opposed to simply a professional colleague, and it appears to me that the investigation considers her comment to have been a contributing cause. That’s not to say that gender should never be referenced. The fact that the first officer is female and struggled to assert herself in the cockpit strikes me as very relevant and is barely referred to; the fact that she was the first female pilot at US Bangla is not even mentioned. Instead, again, her gender is used as a focus of his behaviour.
The foul language and abusive words he was using in conversation with a junior female FO was very inappropriate and certainly not expected from a level headed person.
I’m also uncomfortable with the attempt to get into his head, creating a motive for why he continued long after he should have broken off the approach.
Even in the last moments of flight, PIC had fixation to land at any cost and he never considered for go around procedure even after realizing that flight was not configured to land. One of the reasons could be him trying to prove to the FO that he is a very competent pilot and would be able to safely land the aircraft in any adverse situation.
The conclusion also mentions that the captain might have been fatigued, with a reference to a lack of sleep the previous night. This does not appear to have been followed up on.
The Probable Causes
The Accident Investigation Commission determines that the probable cause of the accident is due to disorientation and a complete loss of situational awareness in the part of crewmember.
Contributing to this the aircraft was offset to the proper approach path that led to maneuvers in a very dangerous and unsafe attitude to align with the runway. Landing was completed in a sheer desperation after sighting the runway, at very close proximity and very low altitude. There was no attempt made to carry out a go around, when a go around seemed possible until the last instant before touchdown on the runway.
The contributing factors is a long list which veers between specific actions and general issues. It appears to me that the investigative team, which was put together specifically for this investigation, put a strong effort into working through all the details of the last minutes of the flight. However, the sequence of events is not really a probable cause and the conclusion does nothing to help us see how the tragic crash could have been avoided.
The report ends with a number of recommendations which somewhat worryingly seem to cover a number of actions which should be expected, for example that the operator (US Bangla) should establish and implement a policy to de-roster any crew member found to be stressed, fatigued or emotionally disturbed, which is a given, and that the operator should reinforce firm policy regarding No Smoking in the flight, which quite frankly has very little to do with anything else, other than to highlight the captain’s state of mind and lack of compliance during the flight.
Finally, the report ends with a series of smiley faces.
I have no idea what they were thinking.
You can download the final report from the Nepalese government site. The link to the report has been blocked. I downloaded a copy at home and I will look to see what’s happened and upload it if appropriate.
Here’s my copy: Accident Report Downloaded 1 Feb 2019
I wouldn’t wish for the Nepalese to gain more experience in fatal air crashes, of course, but it does seem that the investigative committee could have used more support from the established investigation agencies than they perhaps received (or asked for).