Sichuan flight 8633 blown windshield

18 May 18 17 Comments

It seems that physical issues with aircraft are all over the news right now, with the latest being Sichuan Airlines flight 8633. On the 14th of May 2018, Sichuan Airlines flight 8633 departed Chongqing, China for a scheduled passenger flight to Lhasa, the capital of the Tibet Autonomous Region (the area of Tibet in the People’s Republic of China).

Scheduled flights in the area didn’t start until 1965, when navigation aids and flight computers became reliable enough to allow for safe crossing. But flying to Lhasa is still a challenge, even with modern aviation technology. The Tibetan Plateau, located to the west of the Sichuan province, is the highest region on earth, with an average elevation of 16,000 feet (Mount Everest has an elevation of 29,000 feet and is located on Tibet’s border with Nepal). The region is particular dangerous for flying, with high mountains, deep gorges, violent turbulence and 200mph (320 km/h) winds.

Lhasa Gonggar Airport is the largest airport in the region, serving as an aviation hub for Tibet with daily flights serving major Chinese cities. The airport elevation is 11,710 feet (3,570 meters) and is surrounded by mountains ranging from 17,000 to 20,000 feet (5,362-6,126m). The runway is 4,000 meters (13,000 feet), allowing it to handle wide-bodied aircraft in the thin Tibetan air. As the wind picks up in the afternoon, most scheduled flights into the airport are in the morning. So it’s no surprise that pilots flying in and out of Lhasa Gonggar Airport require special training in handling manoeuvres for high altitude.

Sichuan Airlines (Chinese: 四川航空) is one of the seven airlines (all based in China) that fly in and out of Lhasa Gonggar. Their major hub and headquarters is in Chengdu, the Sichuan capital, at Chengdu Shuanliu International Airport. The airlines also offer scheduled domestic flights from Chongqing Jiangbei International, which is located to the southeast of the Sichuan province; that is to say, a scheduled flight from Chongqing to Lhasa passes over the Sichuan province, flying through Chengdu’s airspace.

Sichuan Airlines has 23 Airbus A319-100s in their fleet. The aircraft for the scheduled flight from Chongqing to Lhasa that day was serial number MSN4660, registration B-6419. The aircraft was built in 2011 and delivered to Sichuan Airlines in July of the same year. It had 19,942 flight hours.

I was lucky enough to discover Shimin on PPRunE who translated a number of articles from Chengdu Business News into English. Shimin works in the aviation industry and, although his English may not be as fluent as an official translator, he understands aviation terms, so I found his quotes much more readable than those from the mainstream British and American press. I have taken the liberty to fix some minor grammatical issues while taking care to ensure that Shimin’s original intent remains.

Click through for playback of Sichuan Airlines flight 8633 on the 14 May 2018

The scheduled flight 8633 departed on time from Chongqing with 128 souls on board. The weather was clear and visibility was good. The flight proceeded normally as they reached their cruising altitude of 9,800 metres (32,000 feet). Forty minutes into the flight, passing through the Chengdu air traffic control zone, it was clear that something had happened. The cabin crew were serving breakfast and passengers were dozing. Suddenly there was a loud bang heard throughout the aircraft. The aircraft shook and the oxygen masks dropped.

In the cockpit, Captain Liu was faced with an unexpected emergency. The inside layer of the right windshield cracked and an error illuminated showing an anti-icing failure.

There was no warning; suddenly, the windshield cracked and burst, making a loud noise. I looked up to see the co-pilot was hanging halfway out the window. Luckily, he had fastened his seatbelt. The cockpit items all flew into the air. Much of the equipment failed and the noise was so loud that I couldn’t hear the radio. The whole plane was vibrating so hard that I wasn’t able to read the instruments.

The right main windshield had suddenly cracked with a bang and then separated from the aircraft, sucking the co-pilot out of his seat. The pressure dropped and with it the temperature. The three pilots were in short sleeves and now suddenly it was -40°C in the cockpit; cold enough that the captain knew they were at risk of frostbite. The sudden loss of pressure had also caused multiple instruments to fail, including both Flight Control units (FCU). According to China Aviation review, the FCUs failed because they were sucked out.

I wasn’t sure about why there were three pilots in the cockpit but Shimin was kind enough to explain that in western China, high plateau operations are always flown by three-man crews, where two are qualified captains.

Chinese airlines keep the three-man system in many daily routes, especially in remote, bad conditions and long-haul operations. It does cost a lot and the practice is often mocked but it is safer in my opinion. In such a three-man system, one of the key considerations is the ranking among the three men. In Chinese airlines, the practice is that the top one is designated as the Lead Captain, even if he is in the observer’s seat when any accidents happen.

The three-man system was one of key factors for this successful landing. With the help of the second captain, many checklists were implemented before landing while Liu was left to concentrate on “Aviate” and make every decision timely and correctly. As the Lead Captain, Liu is the highest with over 13,000 flight hours, the secondary captain had about 8,000 flight hours and FO was a young man with about 2,800 hours.

The state of the cockpit (photo unattributed from Evening Standard)

Captain Liu punched 7700 into the transponder, alerting air traffic control to the fact that he had an emergency. The captain descended rapidly from 32,000 to 23,500 feet — passengers in the cabin described it as a free fall or suddenly becoming weightless. The aircraft continued to shake violently. I’m sure the captain wanted to get down more quickly but the area safety altitude was 22,100 feet. He had to wait until they were clear of the high ground and on the approach to Chengdu in order to descend to 10,000 feet.

The press asked him more details and I love his matter of fact answers (again translated by Shimin)

Like in the cabin, when the cockpit loses pressure, the oxygen mask will automatically fall off. Hypoxia is not a big issue. Also, the cockpit and the cabin are sealed and insulated, so the loss of pressure and the cold in the cockpit did not affect the passengers.

Yes, it was a completely manual operation, I had to visually rely on my own judgement. Much of civil aviation is automatic operation, but this time the equipment was of no help. I’ve flown this route over a hundred times, I should say everything about flying here is familiar to me.

At the time, I just wanted to operate the aircraft safely, I was unable to pay attention to my and my co-pilot’s physical condition. In order to avoid further damage to him and to the passengers, the first thing I had to do was reduce the speed and then try to descend as quickly as possible. The noise was great and the automatic equipment can’t help. Completely by and visual, I had to rely on perserverance to grasp the stick to complete the return [to Chengdu] and the safe landing. My body should have been a very big wobble.

I feel like I’m a very big wobble just reading about it! Once the aircraft had lost speed, the first officer was able to drag the second captain back in from the cockpit.

There was no power in the cabin and the PA system didn’t work. Although the aircraft was in control, the passengers panicked, with some beginning to vomit. A flight attendant began to shout to be heard over the passengers.

Please believe us, we have the confidence that we are able to take you to the ground. We have the confidence to land safely!

Although initial reports said that the tyres had burst on landing, the Captain was clear that this wasn’t true.

The aircraft was overweight and the thrust reverser was not working. We needed longer than the normal taxing distance, which means that the tire rubs longer and causes the temperature too high. The tire automatically deflated–this is a protection, not a flat tire.

When compared to Captain Sully, Captain Liu pointed out that he had much more in common with the British Airways 5390 incident, in which the captain was partially blown out of a newly installed windscreen which had not been bolted in correctly.

However, that was in a BAC One-Eleven, a British short-range jet airliner where the windscreen is attached from the outside with bolts. The Airbus A319 windshield is installed into a frame from the inside, so it can’t detach but must have failed completely across all layers after the crack.

This photograph appeared with only the attribution “as seen on social media” and I haven’t found a way to verify it; however apparently it is the co-pilot’s uniform, or at least the remains of it, as recovered after the incident.

Second Captain's Uniform

I don’t usually include a lot of local news reports but I love the fact that these are available because Shimin took the time to translate them. The following is a passenger’s note to friends that was published on Chinese social media and translated by Shimin. The language and metaphors are quite beautiful, even in translation.

In May 14, 2018, 3U8633, Chengdu Shuangliu Airport, I just experienced a life and death.

From Chongqing to Lhasa, the route that I had taken countless times, I slept on my seat as usual. The stewardess woke me up to the breakfast time. As usual, I reckon how long it will take to get to Lhasa. The shock always happens in a sudden break. Suddenly, there was a loud noise at the top of the plane, the cabin suddenly darkened, the oxygen mask fell in front of me, and the plane began to fall as fast as it was, just for a moment. My mind was blank. After seeing around, it was confirmed that it was not my dream. Then I wore mechanically the oxygen mask according to the instructions of the stewardess. After the blanks, there was a surge of fear. I was clearly know I was going to experience an bad accident, and I had no way to do it. When I realized that I couldn’t do anything about survival, depression, fear and panic began to fill me. I always thought I was not afraid of death, but when you really faced death and could do nothing, I was like a fish that had just been put into the boiling water and was attacked by little bits of death. I can see clearly that the icebergs beneath the plane and they are only about a kilometer away. Real depression just started. I feel the despair in the eyes of the passengers around, perhaps they can also feel my despair.
Of course, I am still alive and lovely now. Just like fairy tales, pilot has become hero in the twinkling of an eye. He gradually controlled the situation, changed direction, flew to Chengdu, and then landed on the runway of Shuangliu Airport. Hanging heart also slowly landed, trance seems to have gone through a dream, to be thrilling, and finally returned to peace. I am seating in the cabin now, waiting for the trailer to pull the plane to safe area and all the cabin is just back to life.
It’s not clear how I feel now. To escape or to die? Thank you for pilots and stewardess. Save me alive without death at such a young age. I can only say one thing: it’s nice to be alive.

The majority of the passengers were put on the next flight to Lhasa, arriving at 12:09 local time. Captain Liu’s wife said that she phoned him to see that he was all right but he told her that he was busy and that she shouldn’t go to Chengdu to meet him.

I’m not very worried about him. He has been flying in the Air Force and Sichuan Airlines for many years. I really believe in his technology. His character is calm and steady.

The first officer was treated for scratches and a sprained wrist but suffered no serious injury.

The Civil Aviation Administration of China (CAAC, the Chinese aviation authority) will be leading the inquiry with teams from the French BEA and from Airbus already in China to support the investigation. They’ve already confirmed that the right cockpit windshield on the Sichuan Airlines Airbus A319 was an original part of the aircraft which was assembled at the Tianjin plant in China in 2011 and came into service later the same year. According to the logs, the windshield has never been replaced or undergone any maintenance.

I’m looking forward to finding out more (and to read more translations by Shimin!).


  • A very interesting story.
    Reading this, it is obvious that Sichuan Airlines operate at a high professional level with a competent, well trained crew.
    Most aircraft I flew, like the Fokker F27 and the BAC 1-11 had flat windshields. They were built-up and consisted of layers, there may be a very thin sheet of gold in between. So thin that it is invisible, but it allows it to be heated electrically. One layer usually is a tough flexible film or membrane that should bond them in case of damage e.g. bird strike.or a crack. That can be caused e.g. when the crew forgets to switch on the windshield heat until it is already cold-soaked. But even a crack should not cause the windshield to fail totally.
    The idea is that even with one layer cracked or broken the remainder should have sufficient strength to prevent decompression.
    Of course, more modern aircraft may have windshields that may be curved and / or composed of different materials.
    But I always was under the impression that there was no time limit, nor was there a limiting number of cycles. Glass does not suffer from fatigue. There were limits to the maximum area of delamination, which was quite visible and usually started in corners or edges. The mechanics would trace the delamination with marking pens so that they could keep track of eventual progress, and speed, of the ageing. Exceedance of the maximum allowable area of delamination determined when the windscreen had to be replaced.
    Operating in and out of high-altitude airports has its challenges. Performance will be seriously degraded, even flap settings may have to be adjusted, i.e. reduced in order to reduce drag.
    Should an engine fail during take-off, a flap setting, normal for the type of aircraft (at a guess probably 2) may at high altitude generate enough drag to prevent the aircraft from accelerating to flap re traction speed. A reduced or even zero flap setting at take-off, under the circumstances means a considerably increased take-off distance. Take-off weight may have to be reduced, but there will be RTOW tables the crew can consult.
    Take-off has its challenges, so does approach and landing.
    The oxygen masks must be inhibited prior to landing, otherwise they will deploy when the aircraft is depressurized on the ground.
    If ever I were to take an airline flight in China, Sichuan Airlines would seem to be a safe choice.

    • “Glass does not suffer from fatigue” – that’s a very good point. I think with the recent accidents and the BA flight, there’s a tendency to fall back on fatigue/age of aircraft but that doesn’t seem to be the case here at all.

    • “Glass does not suffer from fatigue” – Incorrect. EVERY non-self-healing material will fatigue eventually – the only question is “how quickly”.

      • Actually, some materials have stresses below which fatigue will not occur – the endurance or fatigue limit. Steels, for example, have a fatigue limit; aluminium alloys don’t. Not sure about glass, but equally, there wasn’t a fatigue limit on glass components when I worked in aircraft maintenance (and I managed fleet structural fatigue for my aircraft type)

        • OROWAN, E. The Fatigue of Glass Under Stress. Nature 154, 341–343 (1944).

          GLASS under stress shows a characteristic fatigue phenomenon of practical importance: it can be broken by stresses far below its ordinary breaking stress as measured in short-time tests, provided that the load is applied for a sufficiently long time. About one third of the short-time breaking stress is sufficient to produce fracture if it is sustained for a number of weeks. In contrast to metals, the stress need not fluctuate periodically in order to develop this fatigue phenomenon, of which the most spectacular everyday manifestation is the sudden spontaneous cracking of glasses or bottles under internal stresses which they may have withstood for many years.

          However, the final report shows that this is not what happened here. See my comment below.

          • Cool! Suggests that in effect glass does have an endurance limit, but it’s not really fatigue, more a stress-time envelope. Very interesting!

    • I don’t know from where you picked up those absurd info, like the flap setting thing or the oxy mask inhibition. Airbus A320 family aircraft all have a normal operationg limitation for max field altitude of 9200 feet. Mask automatically deploy at 14000 feet cabin altitude.
      Regarding the “professionality” of our heroes, rumors in the aviation community (but which are somehow more than just rumors) suggest the windshield blew out as the result of lacking to apply the correct procedure following an “arching”. Arching occurs upon a failure of the heating elements among the transparent layers that leat to a short circuit, causing an electric glow. That can drive parts of the windshield to temperatures as high as 1000 decgees celsius, and such an extreme temperature unbalance within the part may lead to shattering. That is why, for Airbus, there is a “paper procedure” (that is it is not displayed/suggested on the ECAM screen) aimed at removing electricity from the device so to stop the arching. Apparantely the hero was not familiar with that particular section of what should be his bible resulting in what we all know. To summarize: fantastic job and great airmanship bringing the plane to the ground without any significant injuries but the whole could have been easely avoided without any fuzz. Provided the above “rumor” is true. Something very similiar to the exeptional and effective mesures put in place by the authorities in the Wihan virus outbreak.

      • The aircraft was flying to Lhasa; this airport is at 14,315 feet altitude (second highest airport in the world), which matches the criteria for the deployment of the oxygen masks.
        Has the accident report been released yet?

  • I loved this article. I’m so glad you were able to find someone to translate the news and the first hand accounts. A slight nitpick, however: I believe that the pinyin romanisation for the origin city is ‘Chongqing’ instead of ‘Chongquing’. The Chinese don’t have our funny rule of always putting a ‘u’ after ‘q’!

    • Oh, argh! I noticed that I’d done that and fixed it (I thought) but only in one paragraph. I hadn’t noticed that I regularly inserted that u from the start. Thank you and now fixed!

  • You did wonderful writing! Appreciate you so much for your work, but also for correcting my English!
    At the beginning of the incident, there are lot of busy and messy information, including the reporter’s own understanding. As a result, early information will be inevitably with some mistakes. I am so sorry for not detected it in a timely manner and have made mistaken translations. This may confuse you. One example is that the oxygen mask in the cockpit does not fall off automatically. It is stored in a side box. For Captain Liu, the challenge is that his left hand grasp the sidestick and use his right hand cross his body and over the left arm to get the oxygen mask out of the box at the same time. This action must be completed within a few seconds. Otherwise, he may be deprived of oxygen. Under the explosive environment, he was able to control the plane first and obtain his own oxygen mask before the out of oxygen. This is a very remarkable thing in my view.

  • Shi Min makes a point that is obvious to pilots, but not to people who only fly as passengers.
    In many aircraft, the crew oxygen masks are indeed stored in a box at the side console. They do not drop automatically when the cabin pressure drops below a certain value, usually calibrated to correspond to a pressure altitude of 10.500 feet.
    Of course, this can be higher or lower than the elevation, that is fixed, a geological datum. The pressure altitude can vary with the barometer (= altimeter setting).
    Since the masks in the cabin drop when the cabin pressure drops to a value that corresponds to a pressure altitude, the masks must be inhibited prior to landing at an airport at high elevation.
    As Shi Min points out, the cockpit masks are deployed manually.
    The pilot takes a masks between his (her) fingers, there is a toggle switch that must be grabbed and squeezed. By doing so, oxygen under pressure will be forced into tubes. These tubes form the harness of the mask. Oxygen under pressure will open the mask that now will fit effortlessly over the head. When in place, the pilot will release the toggle switch and the harness will contract to form a tight fit. The pilot now will open the oxygen flow, switch the audio to “mask”. That will switch the audio to the microphone built in the mask. The pilot will also put the headset over his / her head. These actions will ensure that the pilots will be “on oxygen” and that communications will be restored. Of course, what must follow is the “emergency descent” drill.
    This is all practised in the simulator many times over, but if it happens in real life there is of course a natural delay, especially if the windshield breaks and the copilot nearly sucked out of the windscreen.
    On the other hand, may I reassure all who fly: The crew has ample time to react. Loss of consciousness is NOT nearly instantaneous, there is enough time to react. I have encountered a rapid decompression twice, at FL 370 and FL 390. We functioned rationally without problem for what may have been a few minutes.
    This event demonstrated a problem with the design of the cockpit masks in a certain type of aircraft. They were of the “halter” type: hanging ready from a hook at the side of the cockpit, they had to be put over the head. A tug, a pull if you like, freed an elastic band in the rigid harness, which was supposed to pull the mask on to the pilot’s mouth. Supposed to, because always being stretched, ready for the eventuality, the elastic could deteriorate over time and lose its elasticity. Which had happened in our first decompression. I had to perform the cockpit tasks holding the mask with one hand to my face. The next time was better: the elastic bands had been replaced and we had a new item on the checklist: On the parking, we had to release the rubbers, pre-flight we had to arm it again. This way, it was not permanently stretched.
    I am not sure that his is clear, I guess that pilots will recognise and understand my explanations.
    What is clear, though, is that the captain did an outstanding job – and so did the rest of the crew.

  • Sichuan airlines 3U8633 flight from changing to lhasa it was an immediate incident but the real hero’s the pilots who had saved all life’s I salute you all captain and all 6 crews you are the real hero’s of the real life I love you👁💌u me prahlad from India🇮🇳

  • From Avherald (excerpted):

    On Jun 2nd 2020 China’s CAAC released their final report in Chinese only.

    The report concludes the probable cause of the accident was:

    The seal of the right hand windshield of B-6419 became damaged, cavities inside the windshield permitted moisture to accumulate at the bottom edge of the windshield. Due to long term immersion the windhsield heating’s power wires’ insulation low but continuous arcing in a humid environment occurred. The temperatures caused by arcing caused the double glass layers to fracture, the windshield could no longer withstand the pressure difference between outside and inside pressure and burst.

    The CAAC analysed the cockpit resource management stating that initially the captain responded to the emergency by himself only and initiated an emergency descent. Subsequently the first officer was able to return into his seat, the second captain entered the cockpit. Although no verbal communication between the crew members was possible, they communicated through gestures, the first officer set the squawk to the emergency code after returning into his seat, the second captain reminded the captain to don his oxygen masks (which however was never donned, the captain was still not wearing his mask after landing) and indicated that the conditions in the cabin were normal. Using the navigation charts on the EFB the crew tracked safe altitudes. During the descent the hand held microphone was used to declare Mayday and advise ATC of the aircraft condition and crew intent. The second captain patted the shoulders of both captain and first officer to encourage them and rubbed shoulders and arms of both captain and first officer to ease discomfort caused by the cold. Although the crew transmitted the aircraft status and their intentions to ATC, they never reported the key information that the windshield had burst in conditions of hypoxia, freezing temperatures, strong wind and high noise level.

    The explosive decompression causes sudden expansion and separation of dissolved gasses in human body fluids to form bubbles and cause tissue damage. Both captains reported they did not feel their bodies after the loss of cabin pressure. The first officer reported pain in his left arm after colliding with hard objects such as instrument panels in the cockpit. He was later diagnosed with skin contusion of his left arm in hospital. Audiometric examination showed both captain and first officer had reduced hearing with no impact to the hearing of the second captain, the first officer was subsequently diagnosed with “high frequency mild sensorineural deafness”.

    While the captain remained without oxygen masks from the burst of windshield to landing, the first officer was able to don his oxygen masks about 3 minutes 12 seconds. Both reported never losing conscience, the flight data confirm they performed continuous and correct flight maneouvers. There is no evidence that any of the flight crew suffered a decline in understanding, analysis and judgement. It thus can be determined, that neither captain nor first officer became (partially) incapacitated.

    The tripping of the 17 circuit breakers happened as result of a safety feature of the cockpit door. During a rapid decompression the cockpit door will automatically unlock and open in order to prevent a pressure differential between cockpit and cabin. The design however did not consider the force and impact the door can produce when it impacts the 120VU panel in such a decompression.

    The captain was unable to reach his oxygen mask with his right hand due to the shoulder harness already secured and the mask too far away for this scenario. The captain would either have needed to stop controlling the aircraft for some brief time while fetching the mask with his left hand, or unfasten his shoulder harness – however, the first officer at this time was not fully in his seat, still climbing back, and due to his injuries the first officer was deemed incapacitated, the shoulder harness of the only remaining pilot thus had to remain secured.

Post a comment:

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.