This Should NOT Have Been POSSIBLE! | LATAM 8073

0:01 How could an aircraft find itself trying to take off with a 100,000

0:05 kilos more weight on board than what the pilot had calculated with?

0:09 And exactly what would happen if it did?

0:12 Well, that's the fascinating story that I will share with you today.

0:15 So, stay tuned.

0:19 When you check in your bag at the airport,

0:21 you've probably noticed signs talking about overweight luggage and maybe you

0:26 wondered what would happen if you would go slightly above that limit.

0:30 Of course, on an airliner,

0:32 one overweight bag doesn't make much of a difference at all,

0:34 but what if you forgot something that weighed more than,

0:38 let's say, a fully [snorts] loaded Boeing 737?

0:41 Well, surprisingly, this has actually happened multiple times, in fact,

0:46 and the reasons behind are as simple as they are complicated.

0:51 This particular story started on the 9th of July,

0:54 2024, at Milan Malpensa Airport in Italy,

0:58 where a flight crew from LATAM Airlines Brazil had just

1:01 arrived to operate a long-haul flight over to São Paulo.

1:05 Now, as a subsidiary of the Chile-based LATAM Airlines Group,

1:09 LATAM Brazil is the largest international airline in Brazil,

1:12 with a fleet of 170 aircraft, including 10 wide-body Boeing 777-300s.

1:18 Now, one of those was waiting at the gate in Malpensa,

1:22 ready to operate LATAM flight 8073 over to São Paulo,

1:26 with a scheduled departure time of 13:30 local.

1:30 The crew that had been planned to operate

1:32 this flight included no less than three captains:

1:35 a newly promoted trainee captain,

1:38 a line training captain who was providing the training,

1:40 and a relief captain whose job it was to take over as they rotated

1:44 to get their mandatory rest breaks during

1:46 the 11-hour flight over to South America.

1:49 Now, the trainee captain who was going to be sitting in the left

1:52 seat was 53 years old and had an impressive 16,689 total hours.

1:59 He had earned those partly during his 23 years with LATAM,

2:02 but since he had just started on the Boeing 777,

2:05 he had only flown about 95 hours on that type so far.

2:09 Instead, he had spent most of his career on the much smaller Airbus

2:13 A320, and this was actually going to be his first stint on a wide-body,

2:18 as well as his first time flying a Boeing,

2:21 except a few hours on the 737 almost a quarter of a century earlier.

2:26 On this day, he had only completed nine flights in the 777 and was

2:30 therefore still doing his line training

2:32 under supervision of a line training captain.

2:35 Now, that captain was 54 years old and had accrued an even

2:38 more impressive 18,921 flight hours over a 28-year career with the company,

2:46 which included 1,333 hours on the Boeing 777.

2:51 He was qualified as a line training captain on both the 777 and the 787,

2:56 which he flew interchangeably,

2:58 and that's worth keeping in mind because of what's about to happen.

3:03 You see, normally, we pilots can only fly one aircraft type

3:06 or group of aircraft with a common type rating at one time,

3:10 but even though the 777 and the 787 doesn't really have a common type rating,

3:16 they're similar enough that many airlines around

3:18 the world let their pilots fly both simultaneously.

3:22 And for LATAM Brazil,

3:24 that was ideal because they only had one 787 in their fleet,

3:28 so it was very efficient for them to be able to use their 777 pilots to fly it.

3:33 Anyway, the third and final pilot was the relief captain,

3:36 who was 56 years old and had 17,530

3:41 total hours across a 27-year career with the airline.

3:45 But out of those, only 418 hours had been flown on the 777,

3:50 which he had only started flying a few months earlier

3:53 after 11 years on the little bit smaller Boeing 767.

3:58 So, taken together, these three pilots made up an incredibly experienced team

4:04 with a combined total of more than 50,000 hours and 78 years of experience.

4:10 All three pilots had worked well together on the outbound flight to Milan,

4:14 which had arrived on the morning of July the 8th,

4:17 and they had then spent 24 hours resting at a hotel before

4:21 taking a shuttle back to the airport ahead of this coming flight.

4:25 And while they had been sitting there on the bus,

4:27 the pilots had received their flight plan,

4:29 fuel requirements, weather information, NOTAMs, etc.

4:32 from a proprietary LATAM Airlines smartphone app,

4:36 and they had therefore taken the opportunity to review

4:40 all of this information while they were just sitting there.

4:43 And they saw that there was nothing really unusual

4:46 at all about the flight that was coming up.

4:49 Maybe it was a little bit hot for departure,

4:52 but an outside temperature hovering at around 30° C,

4:56 but the weather was otherwise fine with totally

4:58 clear sunny conditions and a slight variable breeze.

5:03 So, by the time the pilots arrived

5:04 at the terminal and boarded their 12-year-old Boeing 777,

5:08 they were probably expecting this to be a normal and uneventful flight,

5:13 except for the slightly unusual combination of three

5:16 captains on the flight deck, of course.

5:20 As they started their preflight setup,

5:22 the only dispatch document that they were waiting for was,

5:25 as usual, the load sheet.

5:28 That's a document which lists the exact weight of the fuel, baggage,

5:32 and passengers on board for the purpose

5:34 of making performance and weight and balance calculations,

5:37 and it always tends to come a little bit

5:40 late as there might be changes in passengers and cargo,

5:44 which is only really found out during the check-in and boarding process.

5:48 So, whilst they were waiting for that, the pilots decided that the trainee

5:52 captain would be pilot monitoring for the initial part of the flight,

5:55 while the line training captain in the right seat would be pilot flying.

5:59 And this would suit the training plan fine since the new captain was required

6:04 to act as pilot monitoring during at least one leg of this supervised trip.

6:08 And on top of that, the training captain also wanted

6:10 to give the trainee captain some focused practice on the pre-flight procedures.

6:15 So, as this pre-flight got on the way,

6:17 he was mentally very much in instruction mode.

6:22 Meanwhile, as the two other captains were setting up the flight deck,

6:25 the relief captain would handle the exterior inspection,

6:28 liaise with the cabin crew, and conduct passenger announcements.

6:32 After that, he was supposed to take up his position

6:34 on the cockpit jump seat for departure before heading

6:37 back to the crew rest area until it was

6:39 time for him to take over the controls later on.

6:42 Now, whilst he would be on the jump seat,

6:45 he was also supposed to keep an eye on his two

6:47 colleagues as he was officially part of the operating crew.

6:51 But, I can say from experience that it is quite easy to let

6:55 the attention drift a bit when you don't have any real active duties to do.

6:59 In any case, fueling had already been on the way

7:02 when the pilots and their cabin crew arrived to the aircraft.

7:05 So, after having checked that they had their planned fuel quantity on board,

7:10 109,625 kilos to be exact,

7:14 the pilots together concluded that they wouldn't need any

7:16 additional fuel beyond that, which was a little bit

7:19 unusual but welcome since it meant that the fueler

7:22 could now leave and the load sheet could be completed.

7:26 Now, the cockpit voice recording began

7:28 sometime after that decision had been taken.

7:31 So, the first words on the recording

7:33 showed the pilots performing the takeoff briefing,

7:36 which involved discussing the planned taxi, takeoff,

7:39 and departure, as well as the expected take-off performance.

7:44 But because they still hadn't received the load sheet at that point,

7:47 they now deferred that last item,

7:49 the performance, and instead just continued the briefing.

7:53 Now, there will soon be a lot of things to keep track

7:55 of here in order to understand how this was allowed to happen.

7:59 And when we're on that subject,

8:01 actually writing these episodes and keeping track

8:04 of all of the info is not easy.

8:08 I often find myself with dozens of different tabs open

8:11 on my screen showing everything from the final report to ATPL booklets,

8:15 performance data, and historical cases.

8:19 And that's why I am using today's sponsor, Opera.

8:21 [music] I just love how Opera is basically built

8:26 for this kind of deep dive research and scripting.

8:29 For example, I can group everything I use into tab islands,

8:33 one for aircraft performance, one for historical and current procedures,

8:37 and one for relevant accidents.

8:40 And then, there's also the split screen,

8:42 which let me compare things side by side,

8:45 so I can look at company and generic Boeing procedures next to each other,

8:49 and at the same time see weather data and airport layouts.

8:54 Now, [music] if things would still become a little bit chaotic, which they do,

8:58 then tab trace will show me which tabs that I have used the most recently.

9:02 The darker the underscore, the more recently I've looked at it.

9:06 Which might sound simple,

9:08 but you'd be amazed at just how much time and energy it actually saves.

9:14 But probably the most useful feature is Opera AI,

9:17 which is built directly into the browser,

9:19 so I don't need any extra tabs or tools or anything like that.

9:24 I can just ask it to summarize a diagram,

9:26 explain a concept, or analyze a page that I'm looking at.

9:30 And that helps a lot during major investigations like this one.

9:34 Now, you can try Opera out for free yourself

9:37 by using the link here in the description below, which I highly recommend.

9:41 And I'd love to hear what you think of them, as well.

9:43 Thank you, Opera.

9:44 Now, let's get back to the cockpit.

9:47 Around 6 minutes later,

9:49 the relief captain then received the load sheet on his cell phone,

9:52 which he forwarded to his iPad.

9:54 But the trainee captain couldn't get his download to work.

9:58 Back in the day at LATAM Airlines,

10:00 the pilots used to receive an actual physical copy of the load sheet.

10:05 But now, this was all sent via this proprietary app,

10:08 which everyone wasn't that happy with since it meant that there was no longer

10:12 any possibility to put the load sheet

10:14 up on the center pedestal for easy reference.

10:18 Anyway, the relief captain just gave the training captain his iPad

10:21 to read off of, and he then dictated the zero fuel weight,

10:25 takeoff gross weight, and center of gravity to the trainee captain,

10:29 who now entered them into the CDU.

10:32 The zero fuel weight is pretty much what it sounds like.

10:34 It's the weight of the airplane, including all passengers,

10:37 baggage, and cargo, but without the fuel.

10:40 And in this case, after accounting for all 398 occupants and their bags,

10:46 the zero fuel weight was calculated to be 219,460 kg,

10:52 with the projected takeoff weight, including the fuel, being 328,425 kg.

11:00 Now, this takeoff weight was delivered by adding

11:03 the weight of the fuel to the zero fuel weight,

11:05 and then subtracting the weight of the taxi out fuel that was

11:09 expected to be burned during the startup and taxi out to the runway.

11:13 And knowing the precise takeoff weight is crucial,

11:17 because that's what we use to determine both takeoff thrust setting,

11:20 flaps, and also the takeoff or V speeds.

11:24 You see, on modern airliners, we very rarely use a full thrust for takeoff,

11:30 since that causes a lot of excess wear and tear on the engines,

11:33 it uses more fuel and also causes more noise.

11:37 So, instead, we tend to only use the exact amount of thrust that we

11:40 actually need in order for the aircraft to get into the air safely.

11:44 That thrust value is going to depend on several things,

11:48 like the length of the runway, the temperature, and the altitude.

11:51 But, of course, especially on the takeoff weight of the aircraft.

11:55 Now, a takeoff thrust that's lower than the takeoff go

11:58 around or toga thrust is referred to as a derated thrust.

12:02 And on most modern aircraft, that's achieved by inputting something known

12:06 as an assumed temperature or flex temperature on airbus,

12:10 which is higher than the actual outside air temperature.

12:14 You see, the engine electronic control system

12:16 or EECs varies the amount of thrust that can

12:19 be delivered based on the ambient temperature in order

12:22 to avoid going outside of its calculated limits.

12:25 So, if you enter a higher than actual temperature,

12:29 this basically tricks the EECs into using a lower thrust limit,

12:33 thus achieving that derate.

12:36 So, that's the first thing that I wanted to mention.

12:38 But, then we also have the calculation of the V speeds.

12:42 These speeds consist of V1, which is the decision speed above which

12:47 you're committed to continue the takeoff, VR,

12:50 which is the rotation speed where you start pulling back

12:53 in order to lift the nose of the aircraft off the runway,

12:56 and then V2, which is the takeoff safety speed,

12:59 the minimum speed you have to achieve to maintain

13:01 a safe climb gradient in the event of an engine failure.

13:05 Now, these speeds are very heavily dependent on the takeoff

13:09 weight as well as the length of the runway.

13:11 And if they're not calculated correctly, the consequences can be very serious.

13:17 Remember that, because I'm going to dive deeper into that a little bit later.

13:23 Now, I need to also talk a a bit here about how Boeing

13:26 777 pilots at LATAM Airlines calculated their V speeds and their takeoff thrust,

13:31 as well as the checks that were built into those procedures

13:34 in order to verify that those calculations were correct.

13:38 After receiving the load sheet,

13:40 the standard practice was to enter the zero fuel weight and the center

13:43 of gravity into the 777's flight management

13:46 computer using the control display unit or CDU,

13:50 which is that little computer interface that we pilots

13:53 use in order to interact with the flight management system.

13:57 And with those values entered, the flight management computer would

14:00 then automatically calculate [music] the gross

14:02 weight of the aircraft by adding the automatically detected weight

14:06 of the fuel in the aircraft's tanks and then [music] display

14:09 this figure on the performance initialization page in the CDU or perf init.

14:16 The pilots were then supposed to check the fuel on board against the flight plan

14:19 to make sure that it made sense and then

14:22 check the gross weight against load [music] sheet.

14:25 But remember, the load sheet doesn't show the gross weight.

14:28 It only [music] shows the takeoff weight

14:30 and the CDU doesn't display the takeoff weight,

14:33 which excludes the taxi fuel, only the gross weight.

14:38 [music] Now, when I'm talking about this subject,

14:40 I always get asked why the plane doesn't already know its own weight

14:44 or why most planes don't have weight sensors in their landing gear struts.

14:50 And the simple answer is that it's actually surprisingly hard to design a system

14:55 that's capable of estimating the weight more

14:57 accurately than the load sheet math already does.

15:00 So, there isn't actually that much value in such an expensive retrofit

15:05 and also just to keep it [music] calibrated would be a nightmare.

15:09 But some investigations have recommended the installation

15:13 of a weight sensing system anyway in order

15:16 to provide a gross error check against data [music] entry error by the pilots.

15:21 And we will soon see why.

15:24 In any case, after entering the set of fuel weight into the CDU,

15:27 [music] the pilots are then supposed to calculate the performance

15:30 figures using the Boeing onboard performance tool [music] or OPT,

15:34 which is loaded into two iPads that are stored in the cockpit of every aircraft.

15:39 Each pilot is supposed to independently enter the takeoff weight,

15:44 runway information, weather, and so on into the OPT,

15:47 which then produces a recommended thrust setting [music] in the form

15:51 of an assumed temperature as well as the V speeds.

15:54 After that, the two pilots should then cross-check

15:57 each other's results to make sure that they match.

15:59 Or if they don't, then they should clarify any differences.

16:03 Now, this procedure didn't define where

16:06 the pilots should get the takeoff weight from.

16:09 They could get it from the load sheet, which includes the takeoff weight,

16:13 or they could get it by mentally subtracting [music] the startup

16:17 and taxi fuel weight from the gross weight displayed on the CDU.

16:21 Remember that, because it's about to become really important.

16:25 Anyway, after cross-checking the results of the iPad calculations,

16:28 the [music] non-flying pilot should then enter the assumed

16:31 temperature into the thrust limit page of the CDU,

16:35 and then enter the V speeds on the takeoff reference page.

16:38 And on that page, the CDU will then display

16:41 reference V speeds [music] in a column labeled ref.

16:44 And those speeds are derived automatically

16:46 from the aircraft weight entered in the performance

16:49 initialization page and the thrust information

16:51 entered on the [music] thrust limit page.

16:53 So, those are basically the speeds that the aircraft

16:56 itself suggests [music] the pilots to use.

17:00 When the pilots manually enter the V speeds, which they get from their OPTs,

17:04 they should therefore compare those with the aircraft's

17:07 reference speeds as a form of gross error check.

17:10 Basically, those two sets of speeds should be

17:13 within a couple of knots of each other.

17:15 And if they're not, well, then someone,

17:18 somewhere probably have made [music] a mistake.

17:21 Now, after that check,

17:22 the pilots should then complete the rest of the before start

17:25 checklist and finally ask for pushback and start up their engines.

17:28 [music] So, as you can see,

17:31 this procedure has a lot of built-in cross-checks and redundancies, but [music]

17:35 there are still some areas where potential errors could creep in.

17:40 So, let's now return to the cockpit of flight 8073,

17:43 where these three very experienced pilots were

17:46 just about to begin this very same procedure.

17:48 [music] Immediately after having received the load sheet,

17:53 the trainee captain preliminary ran through the performance

17:56 numbers using his iPad and noticed nothing unusual.

18:01 Also, after he and the training captain had entered their airway into the CDU,

18:05 he saw that it produced a gross weight figure, but he soon forgot what it was.

18:11 In fact, just seconds later,

18:13 the Malpensa clearance delivery controller suddenly called

18:17 the aircraft up and informed the pilots that they

18:19 were number two for pushback and that their scheduled

18:22 departure time had been moved up.

18:24 Now, this introduced some unexpected time pressure here and gave the pilots

18:28 very limited time to finish the performance

18:30 calculations and get everything else ready.

18:34 Now, both interruptions and time pressure are factors that increase

18:37 the risk of errors and reduce the effectiveness of cross-checks,

18:41 which is always worth keeping in mind in whatever job you have.

18:45 If this happens, ask yourself,

18:47 do I really need to speed up or can I slow things down somehow?

18:52 And in most cases, the answer will be, "Yeah, I can slow things down." Anyway,

18:57 while some chatter from the clearance delivery

18:59 frequency continued to play over the cockpit speakers,

19:02 the pilots now hurried to finish the performance calculations on their iPads.

19:07 But by the time the trainee captain opened his OPT application,

19:11 he had forgotten the gross weight that was calculated on the CDU.

19:15 So, in order to obtain the takeoff weight to put into the OPT,

19:19 he needed to either navigate back to the CDU's performance initialization page,

19:24 obtain the gross weight, and then subtract the startup and taxi fuel,

19:28 or look at the load sheet in the company

19:30 smartphone app and [music] get the takeoff weight from there.

19:33 But, remember, he hadn't received the load sheet on his smartphone yet.

19:40 At that same time, the training captain decided

19:42 to look at the gross weight from the CDU

19:45 and and mentally just subtract the 400 kilos

19:48 of taxi fuel from it to get the takeoff weight.

19:51 And he did so, and then he said the result out loud.

19:55 Um one moment, OPT, 228.8.

19:59 [music] Now, it's not completely clear from the final

20:03 report whether the training captain said that number out loud

20:06 because he was trying to provide that information to the other

20:09 captain or whether he was just narrating his thoughts.

20:13 Though, it seems likely that it was probably on purpose.

20:18 But, in either case, he had just made an absolutely huge error.

20:24 Because as he now entered the figure of 228.8

20:27 metric tons into the takeoff weight field in his OPT,

20:31 the actual takeoff weight was 328 tons, not 228.

20:38 So, how was that possible?

20:40 Well, most likely, whilst mentally subtracting the taxi fuel,

20:44 the training captain just lost a digit in the hundreds place.

20:48 You know, the kind of brain fart that make you

20:50 groan when you get your math test handed back in school.

20:54 But, of course, the procedure still required both pilots

20:58 to perform the OPT calculations independently and then cross-check,

21:02 specifically to catch this type of error.

21:06 But, instead of looking at the CDU and obtaining the takeoff weight himself,

21:10 the trainee captain now just listened to what

21:13 his colleague was saying and entered that same thing.

21:17 Maybe to save time or maybe because he was

21:19 on the training and therefore just trusted his instructor.

21:23 Now this clearly violated the requirement

21:26 for independent calculations because now

21:28 the same wrong figure had been entered into both pilots' iPads.

21:32 And that obviously then led to the pilots receiving the same flap setting,

21:36 assume temperature, and V speeds back from it,

21:39 all of which were now intended for an aircraft

21:42 that was 100,000 kg lighter than their actual weight.

21:47 And that's a lot, by the way.

21:49 100,000 kilos is substantially more than the weight

21:52 of a fully loaded Boeing 737 NG.

21:55 Or put it another way, about 17 adult male African elephants.

22:02 Now if you forgot to add the weight of 17 elephants in the cabin,

22:05 someone would likely notice that.

22:07 But in this case, the pilots just cross-checked the results and moved on.

22:13 So with that, I feel that I have to address the elephant in the room.

22:18 Which is why didn't this number just feel wrong to them?

22:23 Is it generally, after you've been flying an aircraft type for a while,

22:26 you start to get a ballpark sense

22:28 of the performance numbers that you usually get,

22:31 so you can normally pick up a faulty number quite easily.

22:35 But in this case, the training captain might not have felt that 228 tons was

22:41 an abnormally low number because that is

22:44 a perfectly ordinary takeoff weight on the Boeing 787-9,

22:49 which he was also flying at the time.

22:52 And at the same time,

22:53 the training captain actually did recognize that the results were different

22:57 from the numbers that he had gotten during his quick preliminary calculation,

23:01 but those numbers had now been overwritten.

23:05 And because he was performing only his 10th flight on the 777 after having spent

23:09 many years on the Airbus A320, he probably

23:12 hadn't fully developed that instinctive number sense yet.

23:17 And on top of that, even though he was nominally the pilot in command,

23:21 the instructional tone of his interactions with the training

23:23 captain during the preflight might have predisposed

23:26 him to just accept what he was told

23:28 instead of taking ultimate responsibility for that himself.

23:33 Now, I have been training new captains in similar scenarios, and it is really,

23:39 really hard as an instructor to get the feeling in the cockpit

23:42 right where the trainee captain feels like they are in charge,

23:46 but whilst you still can instruct them.

23:49 It is super, super easy to just transform back into a captain,

23:52 in which case you basically end up with a first officer in the left seat,

23:56 and stuff like this can happen.

23:59 And finally, the relief captain wasn't paying

24:01 attention to the performance calculation at all,

24:03 since at that point he was busy coordinating with the cabin crew.

24:08 Now, even with these factors,

24:10 someone might still have second-guessed the numbers if it hadn't

24:13 been for that time pressure that had just been introduced.

24:17 And what all this meant was that after agreeing that the result matched,

24:21 the pilots now just jumped straight into entering

24:24 the performance data into the CDU without a second thought.

24:28 Remember, the OPT results only had to match each other.

24:31 There was no cross-check against the gross weight already calculated by the CDU.

24:37 In any case, for the thrust D-rate,

24:39 the pilots' OPTs now calculated an assumed temperature of 56° C,

24:44 which would reduce the thrust output during the takeoff quite a bit.

24:48 As you would expect [music] for a very light aircraft,

24:53 which, according to OPT, now only weighed 228 [music] tons.

24:58 It also called for a flap setting of 5°, which is not unusual for takeoff,

25:03 but was the lowest setting available, producing the least amount of lift.

25:08 So, after having entered the assumed temperature

25:10 into the thrust limit page of the CDU,

25:12 the pilots then flipped over to the takeoff reference page

25:16 and entered the calculated flap setting followed by the V speeds.

25:20 145 knots for V1, 149 for VR, and 156 for V2.

25:28 Now, those values were obviously much, much too low.

25:32 At the weight that they were actually

25:34 at, the speeds should have been 173 knots for V1, 181 for VR, and 186 for V2.

25:42 So, about 30 knots higher across the board.

25:46 And at this point, they were now coming up against the next cross-check.

25:50 The comparing of the OPT-derived V speeds against

25:53 the reference V speeds produced by the CDU,

25:56 which should have made this error obvious.

25:59 But, to the pilots' great surprise, those reference speeds were now missing.

26:05 To be perfectly honest, they shouldn't even have been able to enter

26:08 these faulty V speeds into the CDU at all,

26:10 since the system normally prohibits entering V speeds

26:13 that are lower than the CDU calculated reference speeds.

26:16 But, in this specific case, there weren't any reference speeds shown at all,

26:21 so that protection didn't apply.

26:24 Instead, the CDU scratchpad now displayed a message that read,

26:28 "V speeds unavailable." With no further explanation than that.

26:33 Now, the reason for this was actually their selected assumed temperature of 56°.

26:38 You see, at the actual aircraft weight known to the FMC,

26:42 there was no way that the aircraft could take off with that D-rate.

26:46 The runway just wasn't long enough with that thrust and flap setting,

26:50 so the aircraft responded to that by not showing any speeds at all.

26:55 But, of course, that message, V speeds unavailable,

26:58 was a pretty obscure way of communicating this.

27:02 Now, none of the pilots could remember having ever seen that message before,

27:06 and it was only mentioned in one brief

27:09 passage in the flight crew operations manual as well.

27:12 It stated that this message could occur

27:15 with certain combinations of high weight and low

27:17 thrust or low weight and high thrust

27:20 where the performance margins were basically exceeded.

27:25 [music] The manual then gave some fixes to this, but it didn't suggest

27:28 that the pilot should recheck the performance data that they had entered.

27:32 But, it also clearly said that a takeoff

27:35 was not allowed with this message shown.

27:39 Now, none of this helped in this case,

27:40 since the pilots never checked their manual.

27:43 The training captain did read the message out loud when he saw it,

27:47 but none of the pilots knew what it meant.

27:51 The relief captain in the back just

27:52 briefly suggested checking whether the reference V speed

27:55 function was turned on, which the training

27:57 captain then did and verified that it was.

28:00 So, that clearly wasn't the reason.

28:03 At this point, if they would have had some extra time available,

28:06 they might have pulled out the manual

28:07 and discovered that takeoff with that message,

28:10 V speed unavailable, was actually prohibited,

28:13 and that could then possibly have led them back

28:16 to their load sheet and helped them to discover their error.

28:19 But, since they didn't, one of the last safety barriers was now crossed,

28:24 and the jaws of disaster had started opening up.

28:29 The training captain now just told his colleagues to proceed.

28:32 So, they just accepted their manually entered speeds

28:35 and went ahead to finish the before start checklist.

28:38 Time to get this aircraft airborne.

28:41 Now, before we go any further here,

28:43 it's important to highlight what some of the potential

28:45 consequences would be of what I've shown you so far.

28:49 Obviously, the first risk with having a too low thrust setting is that the plane

28:53 just won't be able to reach its takeoff speed before the end of the runway.

28:58 And in most cases, that problem is then

29:01 also coupled with a calculated too low rotation speed.

29:06 You see, the rotation speed is designed to be

29:08 high enough so that by the time we reach it,

29:10 the plane naturally wants to become airborne.

29:13 In other words, the lift that the wings can

29:16 produce should be greater than the aircraft's own weight.

29:20 But in order for that to happen, you need air speed.

29:24 So, if you end up rotating too early, well,

29:27 then the plane just won't have enough lift to climb away,

29:31 and that in turn will mean that the nose will come up,

29:34 but the plane will just continue rolling down the runway on its

29:37 rear wheels until it's built up enough speed to actually fly.

29:42 And that's risky for two different reasons.

29:45 Firstly, if the airplane doesn't lift off when the pilot rotates normally, well,

29:50 then the pilot might instinctively continue rotating more or deliberately try

29:55 to pull in order to get the aircraft off the ground,

29:58 which can then cause the tail of the aircraft to strike the runway.

30:02 And now that can lead to structural damage preventing the later

30:06 pressurization of the aircraft or even worse outcomes than that.

30:11 The second issue is that rolling down the runway

30:13 with the nose high up in the air greatly increases the drag.

30:18 And that comes from both the form drag

30:20 due to the greater surface area that is now

30:22 facing the airstream and the induced drag created

30:25 by the high angle of attack of the wings.

30:28 And if that's combined with a tail also striking the ground,

30:32 that obviously also adds drag.

30:35 So, all of this will make it harder

30:37 to accelerate up to the appropriate takeoff speed,

30:40 and if the thrust setting is already insufficient,

30:43 which is what has caused this in the first place,

30:45 well then this could all combine lead to the aircraft

30:49 overrunning the runway and a potentially catastrophic ending.

30:54 Now, you might be thinking,

30:56 "Surely with all of the training that pilots receive,

30:59 there is no way that this could be

31:01 a regular occurrence." And actually, you would be wrong.

31:07 In 2004, for example, an MK Airlines Boeing 747 cargo plane

31:12 crashed after a crew member had accidentally

31:14 populated the takeoff weight field in their OPT

31:17 with the weight from their previous takeoff,

31:20 where they didn't have any cargo on board.

31:24 Then again, in 2009,

31:25 an Emirates Airbus A340 was departing Melbourne in Australia,

31:30 when the pilots accidentally entered a takeoff weight

31:32 that was also exactly 100 tons too low.

31:37 Now, that aircraft then ended up getting airborne beyond the end of the runway,

31:40 striking the localizer antenna array in the process,

31:43 but after that, it luckily returned and landed safely.

31:48 But in yet another incident back in 2015,

31:52 the crew of a Boeing 777 cargo aircraft,

31:54 which was preparing for takeoff from Paris Charles de Gaulle,

31:57 again inadvertently dropped a digit and entered

32:01 a takeoff weight that was 100 tons too low.

32:04 That plane ended up rotating before it had enough speed to do so,

32:08 almost causing a tail strike,

32:10 but it did eventually get airborne just before the end of the runway.

32:15 And the investigation into that incident found that one

32:18 of the pilots had verbalized the incorrect takeoff weight,

32:22 exactly like in this case.

32:25 Now, my point in sharing these stories is to highlight

32:27 that the problem here likely wasn't only the involved crew.

32:31 Something systemic was also at play here.

32:36 Anyway, completely unaware of the danger

32:39 that they had now set up for themselves,

32:41 the pilots of Latam flight 8073 received clearance

32:45 to push back from the gate at time 13:05,

32:47 just 10 minutes after receiving the load sheet.

32:51 Then, at time 13:14, they were cleared to taxi out towards runway 35 left,

32:56 and at time 13:21, the captain reported reaching the holding point.

33:01 Now, the tower initially held them in that position for about 4 minutes or so,

33:05 and then at time 13:25, they were finally cleared for takeoff.

33:10 Moments later, the aircraft lined up with the runway,

33:13 and the captain moved the thrust levers into the takeoff position.

33:16 [music] This caused the 777's two huge GE90 engines to spool up to 92.8% N1,

33:25 which was the limit imposed by the selected assumed temperature of 56°.

33:29 The aircraft now started slowly accelerating down the runway,

33:33 and on a runway that is that long, the pilots didn't initially notice that they

33:38 were gaining speed slower than normal.

33:41 The pilot monitoring called out, "80 knots",

33:43 which was then followed by a "Checked" from the training captain,

33:47 and 35 seconds into the takeoff roll,

33:50 the aircraft's automated voice called out, V1.

33:53 at a speed of 145 knots, which of course we know was about 30 knots too early.

34:00 Now, since nobody had calculated the V1 for the real weight and thrust setting,

34:05 it was at this point actually impossible

34:07 to know where the real decision point actually was.

34:10 So, even if these pilots would have now realized their mistake,

34:14 they wouldn't know if they had enough runway left to reject.

34:20 But, at this point, they were all still blissfully unaware of their mistake.

34:25 So, 3 seconds after that V1 callout,

34:28 the aircraft reached its calculated rotation speed of 149 knots,

34:32 which the pilot monitoring called out,

34:34 and the training captain now started pulling back on his control column,

34:37 like he had done countless times before.

34:41 Only this time, the airplane didn't lift off.

34:46 In fact, at this point, they were nowhere near the speed required

34:49 to generate enough lift for the aircraft to fly.

34:51 So, while the nose did start to rise,

34:54 the main landing gear remained firmly attached to the ground,

34:57 now eating up the remaining runway with an alarming speed.

35:02 This whole thing caught the training captain completely by surprise,

35:06 and for several seconds, he therefore just continued pulling back instinctively

35:09 to get the aircraft up into the air.

35:13 Now recognizing that its tail was about to hit the runway,

35:16 the 777's built-in tail strike protection system

35:19 kicked in and applied a nose down component to the elevator deflection command

35:23 to try and prevent the imminent tail strike.

35:26 But, its efforts were sadly unsuccessful.

35:30 As the plane accelerated through 160 knots,

35:33 the tail slammed hard into the runway, kicking up a dramatic cloud of dust

35:37 and debris as it dragged itself along the asphalt.

35:39 [music] For several seconds, the pilots seemed stunned,

35:43 caught like deer in the headlights in something called the starter effect.

35:48 Well, this monumental tail strike just kept happening.

35:54 Behind them, the tail skid,

35:56 which is designed to absorb some of the forces from a tail strike,

35:58 now gouged a 6 cm deep scar into the runway surface,

36:03 scattering fragments of metal and pavements in its wake.

36:07 In the cockpit, the training captain shouted

36:10 out a curse and exclaimed, "It's not working.

36:13 Something strange.

36:14 It's not working." But, recognizing that the aircraft wasn't responding,

36:19 the relief captain now just shouted from the jump seat,

36:23 "Toga!" urging the pilots to push

36:25 the thrust levers forward into maximum takeoff thrust.

36:29 Still struggling to comprehend what was going

36:31 on, the training captain just replied with a questioning noise.

36:34 So, the relief captain quickly called the same thing again.

36:37 Toga!

36:39 And this time, the training captain

36:41 reacted promptly and firewall both thrust levers.

36:47 This caused the two giant engines to start spooling up further,

36:50 and with that, they began to overcome the drag

36:52 of the high pitch attitude and the tail strike,

36:55 finally forcing the airplane into the air at the speed of 178 knots.

37:00 So, after having dragged its tail for an incredible 723 m,

37:04 flight 8073 finally became airborne about 80% of the way down the runway,

37:09 largely thanks to the relief

37:10 captain's situational awareness and timely intervention.

37:14 Now, I absolutely have to shout out the relief captain's actions here.

37:18 It takes a real pro to not only figure out what to do,

37:21 but also to give clear and unambiguous instructions

37:24 in a situation which included so much confusion.

37:28 Calling toga was basically the only thing he could have done which

37:31 was simple enough to save the situation and not just make it worse.

37:34 So, well done.

37:38 Anyway, as the plane cleared the end of the runway

37:41 and started climbing away into the bright blue sky,

37:44 the trainee captain soon announced "Positive climb."

37:47 And the training captain called for gear up.

37:50 By this point, the flight data recorder had

37:53 also captured a loss of pressure in both

37:55 fire extinguisher bottles for the APU caused

37:57 by the tail strike damage in the back.

37:59 And the tail strike detector had also sensed the impact and was now sending

38:04 a tail strike warning message on the engine

38:06 indication and crew alerting system or ECAM.

38:09 The relief captain noticed that message and called it out,

38:13 and that was then followed by a continuous

38:15 repetitive chime alerting the crew to the damage.

38:19 So, after having climbed to a safe altitude,

38:21 the pilots now pulled out the quick reference handbook

38:23 and followed the abnormal procedure for a tail strike,

38:27 which called for them to avoid pressurizing the airplane

38:30 to avoid any further damage to the hole,

38:32 and then to land at the nearest suitable airport.

38:35 And this meant that the crew just leveled

38:37 off at 6,000 ft to discuss their options.

38:41 With all of the fuel for the transatlantic journey still on board,

38:44 the 777 was now far over its maximum landing weight.

38:48 So, the relief captain suggested dumping that fuel

38:50 or some of it before attempting to land.

38:54 And since there were luckily no signs of any control

38:57 issues or further system failures other than the APU fire extinguishers,

39:01 the training captain now agreed.

39:04 He then also switched over to the left

39:06 seat and let the relief captain take the right

39:09 seat in order to utilize the most experienced

39:11 crew members for the coming approach and landing.

39:13 And all of this showed really good

39:16 failure management and generally nice non-normal handling.

39:20 I should also mention here that the 777 could have

39:23 landed overweight if the emergency would have been more severe,

39:26 but that would have put undue stress

39:29 on the landing gear and brakes and other structures.

39:31 So, since this crew now didn't really know

39:35 how badly damaged the back of their aircraft was,

39:38 reducing the fuel was, in my view, a pretty good idea.

39:42 Air traffic control soon directed them out over a rural area southwest

39:46 of Milan where any side effects of the fuel dump would be minimized.

39:50 And the pilots then circled for about 36 minutes releasing as much

39:54 fuel as they possibly could until the weight was sufficiently reduced.

39:59 After that, they turned inbound for a normal

40:01 approach and landing on runway 35 right,

40:03 which was reasonably uneventful except they also

40:07 then suffered a bird strike on final.

40:10 But at time 14:36, they landed safely back at Milan again.

40:15 Thankfully, none of the 398 passengers and crew were hurt in this incident

40:20 and the aircraft was later ferried back to LATAM Brazil's base in São Paulo,

40:24 where it was repaired and returned to service by February of 2025.

40:30 The investigation by Italy's National Flight Safety Agency,

40:34 ANSV, found that LATAM Airlines procedures contained ambiguities,

40:38 such as a failure to explicitly prohibit verbalizing data during the OPT entry,

40:44 as well as a lack of clarity about

40:46 where the pilots should find the takeoff weight,

40:48 which increased the risk of human error.

40:52 Now, in his statement, the training captain suggested that there should also be

40:56 a final cross-check between the load sheet, the OPT,

40:59 and the CDU to ensure that the data in all three sources agreed,

41:03 but the ANSV pointed out that new

41:06 procedures might not be the best solution here.

41:09 The preflight phase is already jammed full of checklists,

41:13 checks, and cross-checks,

41:14 and with so many dispatch tasks also now being dumped onto flight crews,

41:18 adding even more checks would likely only increase the pressure

41:21 to complete all of the tasks within the expected turnaround time,

41:25 potentially undermining its own benefit.

41:29 Instead, the ANSV wrote that procedures alone doesn't seem to prevent

41:33 flight crews from making performance calculation errors when under pressure.

41:38 So, the best solution would be to develop flight management

41:40 system software that could detect and flag gross errors instead.

41:45 And in fact, that technology already exists.

41:49 In 2025, the European Aviation Safety Agency announced that all

41:53 newly developed large aircraft would be required to have

41:56 a takeoff performance monitoring system capable of identifying and alerting

42:01 the crew to performance errors both before and during the takeoff.

42:07 Existing aircraft types currently in production will also

42:09 have to have this system beginning in 2033,

42:12 but it won't need to be retrofitted onto aircraft that are already in service.

42:18 Airbus right now offers this system as an optional retrofit to older aircraft,

42:22 but most probably still [music] won't have it

42:26 for a quite a number of years from now,

42:28 which personally I think is a real shame.

42:31 In the meantime, LATAM Airlines have also

42:34 updated its pre-flight procedures to remove any ambiguities,

42:37 and they have also improved their flight data monitoring

42:40 program to detect takeoff performance-related events on its [music] fleet.

42:45 But, the lesson that I want to really highlight here is

42:48 that these type of huge errors are probably much more likely than you think,

42:52 and it can happen to absolutely anyone if you're not careful.

42:57 But, at the same time, I also want to point out something else.

43:01 And that is that even after these guys

43:03 underestimated their weight by 100 metric tons,

43:06 there was still enough of a performance margin available to allow

43:09 this giant aircraft to get airborne inside of the available runway.

43:14 Showing just how much safety margins

43:16 that we build into this fantastic [music] industry.

43:20 Aviation remains incredibly safe,

43:22 and my team and I are here to help spread the word

43:25 about these type of lessons to hopefully make things even safer.

43:29 [music] Now, if you want to help us with this mission

43:31 and get the preview all my videos before everyone else,

43:34 well then please scan this QR code or go

43:36 to patreon.com/join/mentorpilot and become a member of my awesome Patreon crew.

43:43 And here, by the way,

43:44 are also some fantastic comments that you have left on last week's video.

43:49 So, please continue to send those in because

43:51 we absolutely love [music] reading them.

43:54 My name is Petter Hörnfeldt, and you're watching Mentor Pilot.

43:57 Have an absolutely fantastic day, and I'll see you next time.

44:00 Bye-bye.