Transcription
[Music] This video is brought to you by War Thunder.
An Air India Boeing 787 lines up on the runway in Ahmedabad. 9 hours to London ahead. Air India 171, clear for takeoff. The plane is cleared for takeoff and it accelerates down runway 23. But seconds after takeoff, something inside the cockpit goes terribly wrong.
"Why did you turn it off?" "I didn't do it." And the flight is doomed before it even climbs out of sight.
To understand what really happened, we've built a full 3D reconstruction of the final moments. We've recreated the city, the airport, the crash site, and the aircraft's trajectory by stabilizing and matching every frame of available footage. And we found something interesting in the simulation that doesn't match what the prelim report says happened that day.
This is a reconstruction of Air India flight 171. It's 12:30 p.m. 12th June 2025, Ahmedabad. Boarding begins for Air India flight 171 to London Gatwick. This is one of just five weekly flights connecting Gujarat to the UK's huge Indian diaspora. On board today are 230 passengers and 12 crew, business travelers, students, and entire families. But this plane will never make it to London because by 1:00 p.m. the cabin doors will close and by 1:40 the aircraft will be gone.
The crew signs in at 11:55 a.m. for mandatory breath tests and they both pass without issue. In command is Captain Sumit Sabarwal, 56 years old and with more than 15,600 flight hours, over half of them on the Boeing 787. To his right sits first officer Clive Kunda, just 32 with 3,400 hours in total and 1,100 on the Dreamliner. Aviation runs in his blood. His mother flew for Air India for three decades. It's an experienced crew flying a familiar route, but in less than 15 minutes, a small action between their seats will decide the fate of this flight.
By 1:10 p.m., refueling is done, baggage is loaded, boarding is complete, and the cabin doors are closed. The crew have taken on 54,000 kilos of fuel for the 9-hour flight to London. The plane weighs over 200 tons. It's heavy, but still within operational limits.
Today is a hot day. The temperature is 36°. Hot days like today make the air thin. There are fewer air molecules available to lift the wings and there is also less oxygen for the engines to use as well. What this means is that the plane needs more speed to be able to take off. And given that the plane is heavy, Air India 171 will need to use the entire 3,500 m length of runway 23. But that said, this is also well within the capabilities of the Boeing 787.
The plane pushes back at 1:18 p.m. On the bay, the captain moves the two engine switches to run. Both engines pull up with their unmistakable wine. "Air India 171, taxi via papa, Romeo 4, line up and wait runway 23." As the aircraft rolls along the taxi ways, the crew complete the before takeoff checklist. The speeds to note are V1 at 153 knots, VR at 155, and V2 at 162 knots.
After backtracking on runway 23, first officer Kundar performs a 180° turn to position the plane for takeoff. At the far end of the runway is the airport perimeter wall. And beyond that is a dense neighborhood of Meghani Nagar. Visibility is 6 km. Winds are light at 240° and 6 knots. It's a perfect day for flying.
"Air India 171. Winds 240 at 6, runway 23, clear for takeoff." "Clear for takeoff. Runway 23. Air India 171." Takeoff clearance is issued at 13:33 and 4 seconds later, first officer Kundar advances the thrust levers. The plane begins rolling down the runway and the countdown to disaster begins.
What you're about to see is a detailed 3D construction of flight 171's final moments recreated second by second, frame by frame. This is Sardar Vallabhbhai Patel International Airport, here built in 1:1 real-life scale. This is runway 23 here, and the crash site is 1.7 km away over here. These two structures in red are the student canteens, referred to as buildings A and B in the interim report. Behind them are the four student dorms, called building C through F in the report. The tall chimney here is the Army Medical Corps incinerator. And this small patch of trees right next to the chimney would be the first point of contact.
It's 1:38 p.m. and the canteen of BJ Medical College is packed with students having lunch. At the airport nearby, Air India 171 has just taken off and is now at its highest point, just above the airport perimeter wall. As it crosses the wall, the plane begins to lose altitude and it starts deviating towards the left. This camera here is placed exactly on the extended centerline, and the deviation to the left is quite clear. I'll explain my theory about this deviation later in the video.
Just a minute later at 13:39:09, the plane is almost near the ground and it makes first contact with a bunch of trees inside the army core compound. Almost immediately, the nose gear hits the incinerator chimney and it tears away. The main fuselage is still in one piece and it continues moving forward at 150 knots. Notice the plane's attitude here. It's at around 8° nose up. What this tells us is that the pilots, even at the very last minute, were pulling back on the control column, trying to get the plane to regain altitude. They were trying to clear the buildings, but unfortunately, they couldn't.
At 13:39:10, the right side of the fuselage strikes building A's northeast wall. The right landing gear and the tail section take a brunt of the impact. Here, the vertical stabilizer separates instantly and is flung more than 200 ft away. The tail section containing the rear flight recorder and the emergency locator transmitter embeds in the wall of building A. Almost at the same time, the right wing clips the roof of the nearby building B, causing it to tear away from the fuselage. Pieces of this wing are found scattered near buildings A and B. The fuselage continues moving forward and the right engine impacts a solid water tank on the roof of building A. The engine mount fails and the engine lodges beneath the tank. Most of the right side of the plane is now gone. What's left of it keeps moving forward. The left main landing gear and the left wing impacts the main wall of building C, causing them to shear off from the fuselage. The left engine breaks free from the housing due to this impact and is flung forward to near building D's ground floor. Since the plane was full of fuel, the impact causes an almost instantaneous explosion, reducing a perfectly functional aircraft to mere fragments within 2 and 1/2 seconds.
So, this is the crash sequence. But the big question is, what really happened during the takeoff phase that caused the plane to crash? And we have simulated that frame by frame as well. And there's something that we found that doesn't match what the prelim report says. But before I show that to you, let's pay some bills with a quick word from this episode's sponsor.
War Thunder is the most comprehensive vehicle combat game ever made, and it's available free on PC, consoles, and mobile devices. With over 2,500 tanks, planes, helicopters, and ships from 10 major nations, ranging from biplanes of the 1920s to modern fighter jets, War Thunder offers something unique for aviation enthusiasts like you to immerse yourself with highly detailed vehicles, realistic graphics, and authentic sound effects. There are no simple hitboxes here. Every aircraft is intricately modeled down to individual components. Engines, fuel systems, control surfaces all behave like their real-world counterparts. In War Thunder's X-ray View, you can see precisely how damage affects aircraft systems, which components fail, and what leads to loss of control. The game offers three distinct modes for every type of player. Arcade mode for fast-paced action, simulator mode for the ultimate realism challenge, and realistic mode, which strikes a perfect balance. Join over 95 million players worldwide in discovering why War Thunder is simply the best game for fans of military and aviation history. You don't even need special hardware. You can fly any aircraft with just your mouse and your keyboard. War Thunder is also available on mobile now for iOS and Android with the same massive multiplayer battles and vehicle variety. Play War Thunder for free today using the links in the description below. New players and the ones who haven't played in the last 6 months on PCs and consoles that sign up through my links will receive a massive bonus pack with premium vehicles, silver lions, and 7 days of premium account time. Everything you need to start your journey. Thank you for sponsoring this video.
And now, let's get back to the story. The first step in any air crash investigation is locating the two black boxes. The Boeing 787 carries two enhanced airborne flight recorders. One mounted on the front and the second mounted in the tail. Although they were both recovered within days of the crash, the tail unit, which absorbed the full force of the building A impact, was destroyed due to extensive thermal damage and the data was unrecoverable. The forward recorder survived.
In order to extract data from the damaged forward recorder, the AIB needed an identical working chassis to place the memory module in before reading it. This is called a golden chassis. However, the AIB didn't have a golden chassis nor the cables to do the job, and so they had to import it from the NTSB in the US. Now, I want you to think about this for a second. India operates dozens of 787s. Air India alone has over 30 in its fleet. Yet, the country's aircraft accident investigation bureau doesn't have the ability to download data from a flight recorder of one of the most common aircraft types in the world. Make of that what you will.
Okay, so now it's time for the simulation. I've already mentioned that we found something that doesn't match what the prelim report says. But in order for you to believe the data, I need to earn your trust that the simulation is actually correct. And the best way to do that is to show you how we simulated the flight path. I had a lot of fun doing it. So whether you like it or not, I'm going to show you how I did it.
We've already modeled the entire airport and the runway and the airplane in 3D. Now we need to know what the actual flight path is. In order to recreate the flight path, we need to know how the actual AI71 flew on that day. And for that, we'll use this CCTV footage as our reference. Problem is, it's very shaky. So to make it usable, we'll first need to stabilize it. I've done that using Photoshop, frame by frame, which gives me the actual position of the plane at every single frame. But for me to now match the 3D plane with this image, I need to see through a virtual camera that matches this camera. And so I need to know three things: Where was the CCTV camera placed in the real world? What type of camera was this? So that I can find the sensor size. And what focal length was the camera filming at?
To figure out where the camera was placed in the real world, I used this watchtower and these two road markings and this gate and cross-referenced it with Google Maps. If you look closely at this footage, you can see a security camera mounted on a pole at a height of around 15 ft. Our camera is probably mounted on a nearby pole at a similar height. I then use a tool called F-Sp to triangulate the exact location of the camera to right here.
Second, I need to know what type of camera it is so that I can find the sensor size. Again, from this footage, we know that we are working with a PTZ camera. And after searching on YouTube for a long time, I finally found this footage from Elite Aviation's channel showing a close-up of the camera used at Ahmedabad airport. Searching more on Google, I found that the cameras used in Ahmedabad airport is made by a company called Infanova, and the sensor size that they use is approximately 1x2 in.
The third unknown is the focal length, and getting that was the tedious part. Since this is a zoom camera, it could be anything. So, it was a manual process of locking the camera in place and playing with the virtual camera's focal length until the 3D environment matched the source footage. And once the camera was locked, I matched the 3D airplane at each frame with the position of the real AI71 in the footage. And this gave me the reconstructed flight path.
To validate the flight path, I use this second image, the R-A-T photograph from the interim report. From this camera's position, I again have a good match with the ground plane, the background terminals, and this tower. And if I scrub through, you can see that the 3D plane almost exactly matches the plane in the image. We now have everything we need to know how this plane flew on that day.
This simulation is running at 30 frames per second, and so every frame is 1/30th of a second. We'll start at the takeoff roll. "Air India 71 wins 240 at 6. Runway 23 cleared for takeoff." "Cleared for takeoff. Runway 23. Air India 171." The takeoff roll begins at 13:37. First officer Kundar advances the thrust levers, and both engines pull smoothly to takeoff power. The flaps are set to 5°, the standard configuration for a Dreamliner takeoff. This is confirmed from the wreckage as well.
From the simulation flight path, we can calculate the ground speed and the altitude of the plane at any frame. And since the wind on the day was 240° at 6 knots, we can also calculate the indicated air speed at each frame. I've exported all this to an Excel sheet and used it in this simulation. We'll start by going through the timeline as per the official report, and then later I'll introduce my findings and open up some questions.
So, the plane begins a takeoff roll at 13:37 and it continues to gain speed for the next 56 seconds until frame 539, at which point it hits the V1 speed of 153 knots. V1 is a speed at which the plane now has to take off, as there is not enough runway left to stop safely. 2 seconds later, frame 599 is VR, and the front nose wheel begins to lift. The prelim report says that the speed at this time is 155 knots. Our simulation says 154 knots. So, till now, the data agrees. It takes another 4 seconds until frame 668 for the main landing gear to lift off the runway. The weight on the wheel sensor now transitions the plane from ground mode to air mode. The plane then continues to climb for the next 3 seconds normally, and then at 13:38:42, the plane reaches its max speed of 180 knots. Immediately after that, the fuel cutoff switches for engine 1 and engine 2 move to the cutoff position. 5 seconds after this point, at time 13:38:47, the ram air turbine starts producing hydraulic power. Somewhere between these two points, one pilot asks the other, "Why did you turn it off?" And the other pilot says that he did not. Five more seconds later at 13:38:52, the left engine switch goes back to the run position. And 4 seconds later, the right engine switch goes back to the run position as well. And a Mayday call was given at 13:39:05.
So, this is what the prelim report says has happened. I want to direct your attention to this frame here at 758, specifically to these three events. The report says that the plane hits a maximum speed of 180 knots and immediately afterwards the two engine switches are switched off. But intuitively, you would expect the speed to stop rising once the engines are shut off, not before it. Right? The plane is accelerating at takeoff thrust. Takeoff thrust is the maximum power that the engine can produce. In that configuration, I find it odd that the report claims that the max speed was hit before the engine cutoff. And this was the whole reason why I wanted to simulate the crash in the first place. And the simulation seems to confirm my suspicions.
As per this simulation, as the plane rolls down the runway, the sim speed matches what the interim report says until frame 758 at 180 knots. But after that, instead of slowing down, the sim plane actually keeps accelerating for the next 5 seconds until frame 900, up to a speed of 196 knots. The momentum of the takeoff causes the plane to gain altitude until frame 1100, up to a height of 230 ft. Inside a jet engine, once the fuel is cut off, the thrust will collapse almost instantly. Any high-pressure gases that remain in the engine will be flushed out in under a second. So, this cannot be any residual thrust that's pushing the plane forward. I've thought about this a lot, but I've not been able to come up with an explanation as to why the plane would keep increasing in speed for 5 seconds after the engine shut off and why the prelim report claims that the max speed was hit before the engine shutting off. So, I'm putting this to the experts watching. If you know what could create a 5-second rise in indicated air speed after cutoff, I'm very curious to hear from you.
And now, this brings us to the fuel cutoff switches itself. What exactly happened here? To understand this better, we need to address four different theories. First, the switch was intentionally moved to the cutoff position by someone in the cockpit. Second, the switch was accidentally moved by someone in the cockpit. Third, a software feature caused the two switches to trip and move to the cutoff position automatically without pilot input. And fourth, the switches never moved to the cutoff position and some other system in the plane caused the engines to flame out.
Let's start with theory 4 and work our way backwards. This theory claims that the switches never actually moved to the cutoff position and that the engine simply failed for some other reason. To explain why this doesn't hold up, let's use a simple analogy. Your car has black boxes, too. Imagine your car is running. If you turn the ignition switch off, the black box records two things: Ignition off is equal to true. Engine off is equal to true. But if your car runs out of fuel and the engine dies on its own, only one thing gets recorded: Engine off is equal to true. There is no entry for the ignition because you never touch the key. The same logic applies to the Boeing 787. The black box from the flight shows an entry for the fuel cutoff switches being moved to cutoff. And here is the important part. If you look at the wiring diagram, there is just a single data line labeled RDC that runs straight from the physical switch to the flight recorder. Its only job is to record the real physical position of the switch. The way it does this is that every 1 second the flight computer will check what the status of the switch is and record it. That 1 second is called the sampling frequency. So if the recorder shows cutoff is equal to true, that means that the switch itself was physically moved there. There is no software trick or no false signal. This is a direct mechanical reading, which means that we can discard theory 4. The switches really did move to cutoff.
Now let's visit theory 3, that a software feature in the plane caused the two switches to trip to cutoff on its own without any pilot input. This is specifically referring to the TCMA, the Thrust Control Malfunction Accommodation. This is a software feature in the plane that will automatically reduce the thrust in an engine or even cut it out entirely if it believes the engine's power is exceeding safe limits. This has happened before to an All Nippon Airways 787, but that was during the landing phase and the plane landed safely. The question to answer is, could the TCMA have triggered the engines to shut down on Air India 171? Here's the wiring diagram for these cutoff switches. Again, apart from the data connection, the only other circuit to the two switches is a relay with just a true or a false status. There is no wiring to the switches that is able to trigger it automatically to move to the cutoff position based on a software trigger. Now, the other thing about the TCMA is that the TCMA doesn't work through the cockpit fuel switches at all. Instead, it physically closes a high-pressure shutoff valve deep inside the engine, cutting off fuel supply directly. Once this valve closes, it cannot be reversed by the pilots and it requires maintenance action to reset the system. This is exactly what happened to that ANA 787. The pilots couldn't restart the engines, and the plane had to be towed to the gate. But in the case of Air India 171, we know that the engines restarted successfully later in flight, based on the events here. If the TCMA was the root cause, the engines could not have restarted. So, this isn't a TCMA failure, and we can discard theory 3 as well.
Now, this brings us to theory 2, accidental activation. The switches have three levels of safeguards to prevent accidental activation. First, the switch is spring-loaded, and so it needs to be pinched and lifted against spring pressure to move it. Second, there are raised guards near the switches to prevent the hand from accidentally touching it. And third, there are mechanical stops near the base, which also prevents accidental activation. Now, somehow, even if an accidental movement occurred, I'd expect one switch to have accidentally moved. I struggle to see how both switches accidentally activated within a 1-second gap. So, I'm going to place theory 2 as very unlikely.
And this brings us to theory 1. The most prevalent claim in the media against this theory is that it is impossible for these switches to be moved manually within 1 second of each other. I am not sure where this theory began, because to the contrary, here's a video by Fad Nim showing that it is definitely possible to do just that. This is the only theory that I have no arguments against.
Let's also quickly discuss why the landing gear wasn't retracted, which seemed to be a main topic of contention in the media. For the landing gear to be retracted, the pilot monitoring needs to check for a positive rate of climb. This means that the pilot monitoring has observed that the plane is gaining altitude without issue. Only after positive rate of climb is ensured is the gear up call given, and the gear is then retracted. Now, if the first engine began to spool down at frame 758, and the plane at this time is just at 55 ft off the ground, this is just not enough time to verify that a positive rate of climb has been achieved. So, a gear up call would never have been given, and so the gear remained out for the rest of the flight.
Okay, I'm recording this much later, and this might sound like a ramble, but I initially didn't want to include this, but I'm going to include it anyway. I wanted to quickly talk about the two black boxes and the electronic locator transmitter. Now, the ELT is a device that sends out radio signals when there is a crash, and it is used by the rescue teams to identify where the crash happened. So, look back to something like Germanwings 9525 or MH370. In those cases, they were trying to locate where the crash happened based on this signal. But in the case of Air India 171, the electronic locator transmitter just didn't work. And the prelim report just has a single line which says that the ELT never activated.
If the ELT didn't work, there are two reasons why it might not have. One is that the pilots never armed the ELT. There is a switch on the top panel, uh, above the first officer, where the ELT can be in a reset position, an armed position, or an on position. If the ELT is placed in the reset position, then it won't work. But it is quite unlikely that they left it at reset. The more plausible explanation is that the ELT is actually mounted right next to the rear black box, and the ELT suffered a very similar damage to what the rear black box suffered. What bugs me here is that the black box and the ELT are designed for one specific purpose. They are designed to survive a crash. This impact was a low-speed, low-altitude crash. And still, the black box and the ELT, whose only job it is to survive a crash, didn't survive. And I really, really want to know in the final report why that happened.
This simulation is an honest attempt to piece together what might have happened on that fateful day, given the limits of the footage and the prelim report that is available to us. If anything, it was a learning exercise for me and something I wanted to share with the community.
Before we close, a quick thanks to War Thunder, who helped make this reconstruction possible. Remember, you can play War Thunder for free today using the links in the description below. New players and the ones who haven't played in the last 6 months on PCs and consoles that sign up through my links will receive a massive bonus pack with premium vehicles, silver lions, and 7 days of premium account time. It's everything you need to start your journey.