Landing Gear Retracted: PIA flight PK8303
On the 22nd of May 2020, an Airbus A320-214 flying as Pakistan International Airline (PIA) flight 8303 crashed short of Jinnah International Airport in Karachi, killing all but two passengers on impact.
The accident is being investigated by the Pakistan Aircraft Accident Investigation Board. The investigation is led by the AAIB president and includes their own specialists in addition to two investigators from the French BEA, one investigator from the NTSB and five “co-opted members”: two A320 pilots, one doctor, one aviation psychologist and an A320 engineer. On Tuesday, the team briefed the government as to their progress and on Wednesday a preliminary report was released to give a brief overview of the process so far and the facts established in the first month of the investigation.
The report has been posted online as a PDF: ACCIDENT OF PIA FLIGHT PK8303 AIRBUS A320-214 REG NO AP-BLD CRASHED NEAR KARACHI AIRPORT ON 22-05-2020. It is a very accessible and relatively short document. I will only be covering the highlights.
Flight PK 8303 departed Allama Iqbal International Airport in Lahore at 13:05 local time for the short domestic passenger flight to Karachi. On board were two flight crew, six cabin crew and 91 passengers.
The departure and cruise were uneventful, however the preliminary assessment from the Cockpit Voice Recorder (CVR) shows that the crew did not follow standard callouts and that they “did not observe CRM aspects (Cockpit Management Resources) during most parts of the flight.”
There are no cockpit voice excerpts in the report and a full analysis will need to wait until the CVR is released. But I think we can see pretty clearly that something went wrong during the approach phase of the flight.
Area Control Karachi East cleared the flight for the Nawabshah 2A approach procedure and advised the flight crew to expect an ILS approach for runway 25L. An ILS approach is based on the instrument landing system where the flight crew are given a specific course and glideslope to follow, which will lead them directly to the runway threshold, at which point, if everything goes well, the Pilot Flying will flare and land the aircraft.
Soon after, the aircraft was given the option to fly directly to the waypoint MAKLI with a descent to FL200. The MAKLI waypoint is around 50 miles east of the Karachi airport. They were then cleared to descend to FL50, or 5,000 feet above average mean sea level. The elevation at Karacha’s international airport is 100 feet, so height above the amsl and height above the runway are about the same.
As they drew near to the MAKLI waypoint, the aircraft changed frequencies to Karachi Approach, who cleared them to descend further, crossing MAKLI at 3,000 feet.
However, the aircraft did not manage to maintain this descent. As they crossed MAKLI, they weren’t at 3,000 feet, they were at 9,780 feet and travelling at about 245 knots indicated airspeed.
They attempted to lose the additional height by selecting OP DES mode on the FCU. This gets a bit weird because the report identifies this as the Fuel Control Unit but I am taking that to be an error, as the approach mode would be set on the Flight Control Unit on an Airbus. A fuel control unit controls the amount of fuel supplied to the combustion chambers which has very little to do with the approach mode.
The Airbus A320 has a managed descent mode where the flight management and guidance computer (FMGC) calculates a descent profile by working backwards from one thousand feet over runway height. The Flight Management System then uses pitch and thrust to follow that descent profile. The managed descent needs to be monitored and, if it needs adjusting, the flight crew can change to an open descent (OP DES mode). When OP DES mode is set, the managed speed target range is removed and the thrusts are set to idle. This mode would be used if, for example, ATC has required you to reduce speed to below the calculated speed of the vertical descent profile (this guide on Airbus Descent Monitoring goes into great detail). It is effectively somewhere between a managed descent and manually flying the aircraft.
The flight crew clearly selected OP DES mode in order to lose the additional height and manage the descent. At 10.5 nautical miles from the runway and a height of 7,221 feet, both autopilots were disengaged and the speed brakes were extended as well as the landing gear. The point of extending the speed brakes and the landing gear is to add drag which will slow down the aircraft during the steep descent.
Karachi Approach contacted flight PK8303 to “confirm track mile comfortable for descent”. The controller then asked if they wished to “take an orbit” in order to lose the height safely and then rejoin the descent.
The report says that Karachi Approach controller repeatedly warned the flight crew that they were too high and twice advised them to discontinue the approach. He offered them a left turn to heading 180° which I suspect was an attempt to interrupt pilot fixation and get the flight crew to reconsider the unstabilised approach.
The flight crew continued their steep descent. Flaps were set to 1 while travelling at 243 knots indicated airspeed.
I found these limit speeds (Vfe) for the Airbus A320 flaps extension in a document called A320 Limitations:
- Flaps 1 (10°): 215 knots
- Flaps 2 (15°): 200 knots
- Flaps 3 (20°): 185 knots
- Flaps full (40°): 177 knots
Overspeed warnings and ground proximity warnings began to sound in the cockpit.
At 1,740 feet, the aircraft had lost enough height to intercept the localizer — still travelling too fast but they were now able to follow the glideslope for a descent at 3°. The Flight Data Recorder (FDR) shows that at this point, five nautical miles from the runway, the speed brakes were retracted and the landing gear was raised.
Italics mine. We’ll come back to this point.
As they continued on finals for runway 25L, they would normally have changed frequency to talk to the tower, in this case Aerodrome Control, for runway information and landing clearance. However, the Karachi Approach controller opted to contact Aerodrome Control himself rather than ask the flight crew to change frequencies — to my mind, clearly worried about overloading the flight crew. The controller phoned Aerodrome Control, who confirmed that the Airbus A320 was clear to land. They would have had the aircraft in sight but they did not seem to realise that the landing gear of the Airbus A320 was not extended.
The Karachi Approach controller (who could not see the aircraft) contacted the flight to tell them that they had been cleared to land.
At three miles out, the aircraft was some 400 feet above the glideslope again, travelling at 225 knots. One of the flight crew extended the slat/flaps to the third setting, 45 knots over the maximum speed for this configuration. The FDR shows that at 500 feet above the runway height, they were descending at 2,000 feet per minute.
Neither seemed to notice the alarms sounding, perhaps because they had already decided to ignore the overspeed warnings.
The aircraft touched down. The flight crew applied reverse engine power and initiated a braking action. Neither of these did anything, as they had no weight on the wheels and no brakes to brake with. The engine nacelles scraped along the runway, sparks flying.
The controllers at Aerodrome Control must have been staring aghast as they saw the aircraft skidding along the runway at full speed. They didn’t tell the flight crew (who were not on frequency anyway, as far as I can tell) but instead relayed the information to Karachi Approach. Karachi Approach didn’t bother to contact the aircraft with this information, presumably on the basis that by now, it was obvious. Similarly, the cabin crew did not call forward to inform the captain that they were scraping the runway because how could anyone not have already known? However, there have been hints in the press briefing that ATC and cabin crew actions were found wanting, presumably because of their silence at this stressful time. I’m not sure that I agree that this is fair; however there may be more information in the final report.
I would expect, at that point, that the flight crew would be praying for the aircraft to stay on the runway and that it would stop before they ran out of runway. However, somehow, presumably assisted by their high speed on impact (I hesitate to call it a landing), they managed a go-around.
The flight data recorder shows a brief action to select the landing gear to the down position, immediately followed by moving the lever back to the up position. As they climbed away, the flight crew called that they intended an ILS approach to runway 25L. They turned into the circuit for a second attempt but, while they were in the downwind leg, the right engine failed. A moment later, the left engine also stopped producing thrust.
At this stage the Airbus A320 was going 200 knots indicated airspeed and flying at 2,700 feet.
The Ram Air Turbine (RAT) deployed to power the essential systems at which point the FDR stopped recording; it is no longer essential as there is now only limited information that it can record. The flight crew called ATC to say they were proceeding direct to the runway, by which they mean rather than fly the standard circuit.
This is where the controller asks the question that we discussed at the time: “Confirm you are carrying out belly landing?”
At this stage, the controller knew they had scraped the runway and he must presume that the landing gear had failed or collapsed as they touched down. He clearly doesn’t know that they have extended the landing gear. The flight crew’s response is unintelligible and ATC simply clears them to land on two five.
No longer able to maintain height, the flight crew declared an emergency. The controller responds by confirming that they can land on either runway, basically offering them the freedom to just get the plane down where they can.
It didn’t help; the aircraft continued to lose altitude and crashed at a slow speed with a high angle of attack about 1,340 metres short of runway 25L, in the middle of a populated residential area.
The aircraft caught fire after impact. There were only two survivors on the 99 souls on board. Four people on the ground were also injured in the crash; one of those was reported to have died in hospital.
The wreckage was spread out over some 75 metres on a single street with some debris on the roof tops of adjoining houses.
The layout of the wreckage was consistent with the Airbus A320 crashing into buildings on both sides of the street at a low speed. The aircraft looked to be travelling below 150 knots indicated airspeed at the time of impact. The slats were in position 1 and the flaps were fully retracted. The landing gear was extended.
There is a “free fall” mechanism for lowering the landing gears in an emergency, which basically uses gravity to let the wheels fall into place. From what they found, the landing gear has been extended but the free fall mechanism was not used — a sign that the landing gear mechanisms were working and support for the idea that the landing gear had simply not been extended before the first attempt to land.
Both engines showed evidence of external fire. The fan blades on both engines were in good condition, which is consistent with engine being at low rotational speed at time of impact, probably not producing thrust. The right engine had clearly failed; the left engine requires further examination. This probably isn’t meaningful but please don’t let this be a case of shutting down the wrong engine in addition to everything else.
The transfer gearbox and the drain mast reservoir, both of which are located on the underneath of the engine (the engine lower part) had friction marks consistent with “scrubbing the runway”.
In a surprising break from standard crash response, local media claims that the controllers were left on duty after the crash and the airport did not immediately close the runway. Normally, the controllers would be removed from active duty immediately and interviewed while the incident was still fresh in their minds. More importantly, parts of the engine and A320 nacelle parts were recovered from the runway but allegedly after an additional two aircraft had landed.
If you want to read the original report, you’ll find the the full document in English on the CAA Pakistan website.
Although the FDR is still with the French BEA for advanced analysis, the Aviation Minister stated in his press conference that there was a clear lack of CRM in the cockpit. “In the last half hour, the pilots’ discussion was about coronavirus, they were not focused as their families were affected.”
Chatter in the cockpit is one thing but that in no way explains how either or both of the flight crew could possibly believed that this approach was worth continuing.
For me, the biggest question is why both pilots were happy to continue with this completely unstabilised approach where it was clear from waypoint MAKLI that they were in a mess. And yet, they dismissed the offer of an orbit to lose speed and ignored the request for a left turn to heading 180°, which seems to have been a clear attempt by the controller to help them lose height in a safe manner. They clearly never ran through their checklists and at the last minute, they believed that they were correctly configured for landing, despite the alerts and warnings.
The second question, however, is somehow more compelling. At five miles out and a height of 1,740 feet, why did they retract the landing gear? I can’t help trying to find logical reasons for this single action in what was undoubtedly a botched and ill-considered approach throughout. Initially, I thought that there might be a problem with the landing gear but having retracted it in an attempt to recycle the gear doesn’t fit in with continuing the flight and applying brakes as they skidded along the runway.
Here’s three theories:
- The Pilot Monitoring presumed that they were going to go around and configured the aircraft with this in mind; however he did not call a go-around or the Pilot Flying did not hear him do so.
- Once the aircraft intercepted the localiser, they no longer needed the increased drag of the speed brakes and the landing gear to slow their acceleration, so one of the flight crew instinctively retracted both, without considering the imminent landing. As they then had no time for a checklist, the fact that the landing gear had been retracted was then forgotten about.
- The Pilot Flying forgot that they had extended the landing gear and called out “gear”, meaning for it to be extended. As it was already extended, the Pilot Monitoring did the only thing he could think of and retracted the gear rather than ask for clarification.
I don’t like any of them, to be honest, but I don’t like any of the facts to do with this case either.
From the report:
The aircraft was reportedly serviceable for the said flight; necessary scrutiny of the aircraft maintenance records / documents is under way. Captain and First Officer were adequately qualified and experienced to undertake the said flight; necessary scrutiny of the aircrew records / documents is under way.
This sounds like a normal update but is thrown into sharp relief by the comments made by the Aviation Minister of Pakisan, who said in advance of a final report that he held the pilot, the cabin crew and Air Traffic Control responsible for the plane crash.
Yesterday, he added that almost 40% of pilots in Pakistan had fake licenses. In an address to Pakistan’s National Assembly, the minister said, “The inquiry which was initiated in February 2019 showed that 262 pilots did not take the exam themselves” out of a total of 860 active pilots in Pakistan. “They don’t have flying experience.”
A spokesperson for Pakistan International Airlines said that PIA has grounded 150 pilots (out of 434) with either bogus or suspicious licences. This follows an aircraft skidding off the runway in November 2018 where the pilot was discovered to have questionable credentials.
How much of this scandal relates to the crash of PIA flight 8303 is yet unknown and other than the single line regarding the scrutiny of the aircrew records and documents, the preliminary report makes no mention of this.