The Crash That Killed An Airline - The Scandal Behind Adam Air Flight 574

On the afternoon of New Year's Day 2007, an Indonesian Boeing 737 lifted off from the city of Surabaya for what should have been a routine domestic flight across the Makassar Strait. Ahead of the aircraft, towering tropical thunderstorms were beginning to spread, forcing aircraft throughout the region to start taking vectors to avoid them. But inside the cockpit of Adam Air Flight 574, the pilots were focusing on something else entirely. Their navigation system no longer agreed on where the airplane actually was and what this would then eventually lead to would change the course of aviation in Southeast Asia forever. Before I start this video, I have something to ask from you. My team and I have been nominated for a Digital Creator Award in the Educational category and it is based on your votes. So please use the link in the description or in the comments to this video and go in and vote for me if you think I deserve it. Thank you so much. Now back to the video. Now in order to tell you the story of Flight 574, we first have to take a step back and discuss aviation in Indonesia as a whole. You see, in the early 2000s it was experiencing one of the largest aviation booms anywhere in the world. Indonesia is spread across more than 17 000 islands, so the country has always depended heavily on air travel. But for decades prior, flying had remained so expensive that many Indonesians simply couldn't afford it. But the late 90s then brought in deregulation, cheap tickets and a wave of aggressive new low-cost airlines that suddenly made flying accessible to the wider masses. This caused passenger numbers to explode almost overnight and in just a few years Indonesia's aviation industry then expanded at a pace that would have been difficult for even the most mature regulators to safely oversee, let alone in a developing nation. And among the fastest growing of these new airlines was Adam Air. It had been founded in 2002 by the Speaker of Indonesia's House of Representatives, backed by a wealthy private investor, who eventually installed her 26 year old son Adam as CEO. The new airline soon expanded with astonishing speed, building a fleet of Boeing 737s and rapidly opening routes across the Indonesian archipelago, so on paper it looked like a success story. Flights were cheap, airplanes were full and the airline and its peers were making air travel more accessible than it ever had been before. But behind the scenes this fast pace of growth was creating enormous pressure. They were flying all the aircraft harder and further than ever before, while maintenance personnel were in short supply. Spare parts were always difficult to obtain and regulators were struggling to effectively keep their eyes on an industry that was now really booming. Now none of this necessarily meant that an accident was inevitable. In fact thousands of flights across Indonesia operated safely every single day. But it did create an environment where problems who might once have grounded an airplane, delayed a flight or triggered serious scrutiny would now instead slowly become normalized. And that brings us to Juanda International Airport in Surabaya on the afternoon of January the 1st 2007. During the day the airport had been moving through the usual chaos following the new year holiday rush. Flights were operating at high frequency and the airlines throughout the country were trying to keep their airplanes moving through tightly packed schedules as efficiently as possible to get all of their passengers home. This meant that Adam Air flight 574 was only one of dozens of domestic departures leaving Surabaya that afternoon. And to the passengers boarding their aging Boeing 737 waiting at the gate, there was no real reason to think that this particular flight would stand out in any way. Families settled into their seats, flight attendants completed their checks and eventually 102 people including 2 pilots, 4 cabin crew and 96 passengers were ready for their short trip northeast towards Manado on the island of Sulawesi. Up front everything was looking normal. In the left seat sat the captain, a veteran 737 pilot with 13,356 hours of total flight experience of which 3,356 hours had been flown on this type. He had only joined Adam Air a few months earlier after switching from another Indonesian carrier and he was therefore experienced in Indonesian aviation where he had flown his entire career. And he was generally considered proficient by his colleagues and had no negative marks in his record. To the right of him then sat the first officer who had 4,200 hours of total experience with 998 of them on the 737-400. He had recently returned from several days off and there was no indication that either pilot was fatigued, medically impaired or otherwise unfit to fly. But in hindsight, one of the most important facts about this flight crew was actually the training that they hadn't received. You see, at the time of the accident, Adam Air didn't provide dedicated simulator training for Upset Prevention and Recovery Techniques, or UPRT, which covers things like high altitude loss of control, unusual attitudes and spatial disorientation scenarios. Accidents throughout the 1990s and early 2000s had by this time revealed how quickly modern airliners could become unrecoverable once crews lost their situational awareness, so specific training programs had been created. But in 2007, that training was still quite unevenly applied throughout the global aviation industry, especially in fast-growing developing aviation markets like Indonesia. On top of this, the airplane that they were now about to operate had accumulated quite a history of technical problems. It was an 18-year-old Boeing 737-400, registered as Papa Kilo, Kilo Kilo, Whiskey, and it had at this point already passed through multiple operators before eventually ending up in Adam Air's fleet. Aircraft like this one had become extremely common throughout the booming Southeast Asian low-cost market. Since older Boeing 737s could be leased relatively cheaply, were widely supported, pilots were easy to They allowed airlines like Adam Air to expand routes quickly, without the enormous capital expenditure of having to purchase brand new aircraft. But those advantages also came with consequences. Older aircraft require increasingly careful maintenance, and they sometimes use older parts that might be difficult to source. So airlines operating aggressive schedules often found themselves balancing operational demands against the practical realities of keeping these aging airplanes airworthy. One of the specific systems that had repeatedly caused problems aboard this aircraft was its inertial reference system, or IRS. Now, we humans are prone to quick fixes whenever we can, but that's not how airline maintenance is supposed to work. It's supposed to be more proactive than reactive, and that should go for you as well when it comes to your digital security. Did you know that 2025 was a record year for data breaches? Well, most of those attacks targeted bank and security details, but even if you're not the victim of a direct theft, if your data would leak out in one of those breaches, chances are that it will then end up on data broker marketplaces, available for anyone to purchase for just a few dollars. But today's Incogni is designed exactly to keep your data private, even if that would happen. Once you're signed up with them, Incogni works by continuously scanning the internet for your information, and when they find it, and trust me, they will, they will then send out formal takedown requests to whatever broker sites that are selling it, something that they then have to legally comply with. But Incogni doesn't just stop there. They will also keep monitoring those same sites even after it's taken down, to make sure that it doesn't reappear, and also send you a report every seven days, so that you can see how much progress they've made. And if you check out mine, you can see that they've been really busy. Also, if you would come across your own information somewhere that it shouldn't be, you can then alert Incogni directly through their custom removals feature. Just submit the URL where you saw your details into your Incogni dashboard, and they will do the rest. So, to start protecting your online footprint today, and get 60% off the annual plan, just scan this QR code, or go to incogni.com slash pilot, and use the code pilot at checkout. Thank you Incogni for sponsoring this video. Now, let's get back to those IRS issues aboard this aircraft. In a modern airliner, the IRS acts as one of the aircraft's core sources of navigation and attitude information. Using highly sensitive gyroscopes and accelerometers, this system helps determine the airplane's position, direction, movement, pitch, roll, and heading, without relying on external radio navigation signals or older type gyros. Now, since it was so critical, the 737 carried separate left and right inertial reference units for redundancy, feeding information to the captain's and first officer's displays, the flight management computers, and portions of the autopilot system itself. Now, there was also a separate standby instrument that could be used in case of a complete IRS failure, but under normal circumstances, these two main units quietly agreed with each other in the background, and pilots rarely thought much about them at all. But, if they would begin to disagree, well then the situation could become confusing very quickly, if it wasn't dealt with properly. Especially at high altitude in poor weather and over open water, where crews had few external visual references available. And for the months leading up to this flight, this inertial reference system had been acting up. Maintenance records showed more than 150 recurring IRS-related defects during the three months leading up to January 1st, and even more concerning, similar IRS problems were also appearing throughout other parts of Adam Air's Boeing 737 fleet. This had led pilots to repeatedly write up navigation disagreements, fault the instrument indications, autopilot disconnects, flight director problems, and the other strange anomalies linked directly or indirectly to this system, which was obviously not very good. But since none of these problems individually were seen as a big deal, they were mostly treated operationally as just irritating and manageable technical nuisances, and had slowly become normalized. And that was also because compatible replacement IRS units were hard to come by, and were reportedly on month-long backorder. So replacing the units was impractical, while at the same time, identifying and eliminating the root causes of the often complex IRS issues was beyond the ability of Adam Air's maintenance department. So, forced to make do with what they had, Adam Air technicians often responded to these faults by just cleaning electrical contacts, resetting circuit breakers, swapping units between aircraft, or temporarily reinstalling equipment that had already been removed previously for troubleshooting. That would clear fault codes temporarily, but were absolutely not fixing the underlying issues. Now, the correct solution when a defect can't be fixed or deferred in accordance with the minimum equipment list would be to ground the aircraft. But in an industry where airlines were competing aggressively on scheduled frequency and aircraft utilization, grounding aircraft was seen as potentially ruinous. So the planes were instead kept in the air however they could be, even if that meant cutting some corners. Now here I should point out as well that this wasn't unique to Adam Air. Across Indonesia's booming airline industry, aircraft were at this point flying harder than ever before and all maintenance systems were struggling to keep up. And the issue wasn't that Adam Air's maintenance personnel didn't know about the faults or try to deal with them, but instead that these faults kept coming back over and over again, without ever triggering the kind of deeper engineering investigation that it should have done. And that is an important distinction here. Now, part of the reason for this was certainly structural. Like many rapidly expanding low-cost carriers, Adam Air outsourced much of its maintenance, which meant that the airline itself had limited capability and competence, beyond routine line maintenance and day-to-day operational support. But unfortunately, there were already signs that these navigational issues were becoming more serious than just routine maintenance right ups. Only months before this story took place, another Adam Air Boeing 737 experienced severe navigational difficulties while flying through an area with limited radar coverage. In that case, the crew had become uncertain of their position and were only able to recover their situational awareness after descending and unexpectedly spotting the island of Sumbawa below them. They eventually landed safely, but mostly because they were lucky and happened to exit from the clouds in exactly the correct place at exactly the right time. And perhaps most troubling of all here, the regulator wasn't effectively identifying any of these worrying trends. The Indonesian Directorate General of Civil Aviation was responsible for ensuring that airlines maintained effective systems for tracking and resolving repetitive defects, but investigators would later find significant shortcomings in the oversight process itself. So by the time Flight 574 pushed back from the gate in Surabaya that afternoon, Papa Kilo Kilo Kilo Whiskey was therefore still operating despite months of reoccurring IRS-related defects. Now whether the crew was aware of the full extent of that history is unclear, but there is no indication that they had any issues with accepting it for flights. Anyway, at 13.59 local time, Flight 574 took off from Surabaya with the captain as pilot flying and the first officer as pilot monitoring, and it started climbing northeast towards its cruising altitude. Everything initially seemed completely normal, but ahead of them the weather over the Makassar Strait was quickly beginning to sour. Large areas of convective activity had developed along portions of their route, creating huge towering thunderstorms with turbulence, heavy rain, icing conditions, lightning as well as strong upper level winds. Tropical convection in this part of the world can become particularly dangerous because storm cells often develop rapidly into massive vertical systems that crews need to carefully circumnavigate, and unlike the isolated thunderstorms that we often see in drier climates, tropical storm systems can form sprawling clusters of embedded cells where dangerous turbulence and hail might be hidden inside broader cloud layers. So as this flight now continued north-eastward, the pilots, like many others operating in that region that afternoon, began weaving around these areas of worsening weather. And it was during this phase of the flight that the first navigation discrepancies started appearing. Initially the problem seemed subtle. The captains and first officers displays slowly began showing different wind information as well as slightly different positional data. But then the discrepancies gradually grew larger until the two pilots were effectively looking at two different versions of reality. The captains nav display showed the aircraft tracking normally towards Makassar, while the first officers showed it significantly north of its intended route, since the autopilot was linked to the captains, as it would turn out later, faulty IRS. The two pilots now therefore started discussing the problem and occasionally were also joking about it, which kind of showed that at least initially they didn't treat the situation as especially alarming. Now whilst that doesn't necessarily mean that they were completely unconcerned, it does point to navigational system discrepancies probably not being entirely new to them. But outside of the cockpit, air traffic controllers monitoring the aircraft's radar track were now beginning to notice that Adam Air Flight 574 wasn't following its assigned airway. Instead the airplane was drifting steadily north over the Makassar Strait and the coordinates being reported by the crew when queried no longer seemed to match the aircraft's actual position. Eventually one controller even looked at his radar display and exclaimed in frustration And inside of the cockpit the tone had now also started to shift. The jokes became less frequent and growing concern slowly gave way to confusion as the crew now tried to determine which navigation system was actually correct, where the airplane was, and how they should respond to it. And as they were doing that, the weather around them continued to worsen. Increasingly their attention was now becoming focused not on flying the airplane but on troubleshooting the navigation problem unfolding in front of them. And the captain, who remember was pilot flying, started dividing more and more of his attention to the issue. By this point the disagreement between the aircraft's navigational system had grown even bigger, until eventually the captain remarked that the two flight management systems differed Now in a situation like this, the correct thing to do is to first calmly divide the tasks so that the pilot flying concentrates on flying and the pilot monitoring is left to troubleshoot the issue. If there are actual fault messages, there will be associated non-normal checklists to follow. But if that's not the case, the priority will become to verify which system, if any, is correct, which can normally easily be done by cross-checking the position with air traffic control. Now if it would turn out that both IRS systems are faulty, well then we can revert to conventional navigation and just fly on radio bearings from ground-based navigational aids and radar vectors instead, so that's really not an issue. But that will all include careful crew coordination and communication both within the cockpit and with air traffic control. Now unfortunately one limitation that would later face the investigators was that the cockpit voice recorder only retained the final 30 minutes of cockpit audio from this flight. So by the time that recording began, Flight 574 was already well into its cruise segment and the crew had been clearly discussing the navigation problem for some time. As a result, we don't really know exactly when the pilots first noticed the IRS disagreement, but troubleshooting actions they may have already attempted, or how concerned they initially were when the malfunction first appeared. But we can infer that they most likely didn't attempt any of those cross-checking methods that I just mentioned, because their conversations suggested that they hadn't come to any kind of conclusion about which IRS was correct. Now one factor that might have contributed to the crew's difficulties here was yet another gap in the training that they had received. According to the investigators, Adam Aron didn't provide any simulator training involving IRS malfunctions. And that was despite the fact that IRS-related defects had become a recurring issue throughout the airline's Boeing 737 fleet. Which is just another clue that the company's safety management system wasn't working as it should. In any case, at time 1443, the captain asked the first officer to call air traffic control and have them verify the aircraft's position. But the first officer then never really did that, and the captain didn't immediately follow it up. Then, at approximately time 1447, the crew began consulting the quick reference handbook, the QRH, to try and figure out how to respond. The captain instructed the first officer to find the procedure relating to IRS faults, but almost immediately they then ran into a problem. You see, that checklist was written with the assumption that the pilots had observed an IRS fault warning light. But that light wasn't there. It's not a fault, the captain observed, meaning that there was no fault light. And the first officer agreed. So now, because their situation didn't match the exact scenario described by the checklist, the crew was left unsure whether the corrective actions were even applicable to them. For several minutes, the cockpit conversation then circled around that question, with the pilots comparing indications, discussing the behavior of the IRS systems, and referencing the QRH repeatedly as they tried to understand what exactly had gone wrong. At this point, the captain also asked the first officer two more times to request their position from air traffic control, and this time he actually did it. The controller came back and told them their position in terms of a radial and distance from the Makassar VUR radio beacon, but frustratingly, neither pilot then used that information to actually verify which IRS was correct. And investigators were never able to conclude why they didn't. Then, at approximately time 1454, air traffic control cleared flight 574 direct to the Deola waypoint, northeast of Makassar, which under normal circumstances would have been a completely routine clearance. But now, with this navigation disagreement, that task suddenly became much trickier to do. In order to proceed directly to Deola, the crew first needed to know where the aircraft actually was, because like many modern waypoints, Deola was just a set of coordinates, not a physical radio navigational aid. And that meant that the way that they would normally navigate towards it was by using their IRS. Now today's aircraft also have GPS receivers, which will update the IRS's if they start to drift, but that wasn't available on all classic aircraft, so now this situation started to become a real issue. After this point, the frustration captured on the cockpit voice recorder became increasingly apparent, as the pilots continued comparing indications and attempted to determine which system that they could actually trust. In many ways, the cockpit atmosphere now appears to have become increasingly chaotic, with both pilots trying to diagnose the problem at the same time as the aircraft continued flying on autopilot through increasingly deteriorating weather. And as they did, their outside visual references likely now had started to disappear. Eventually, the crew decided to attempt the troubleshooting actions described in the quick reference handbook anyways, and that checklist now called for the crew to cut off navigational data from the faulty inertial reference units by moving that unit's mode selector switch from NAV to ATTITUDE mode. Under normal circumstances, the IRS continuously provides both navigation and attitude information to the navigation systems and cockpit instruments on its respective side, and switching to ATTITUDE mode strips away the navigation functionality and allows the affected unit to reboot in a configuration where it will only give attitude information and some other basic functions, but then hopefully without the fault. Now to accomplish this reboot, the gyroscopes in the IRU needs to undergo a realignment process in order to establish the neutral baseline against which changes in aircraft attitude are measured. And switching an IRS into ATTITUDE mode also automatically disconnects the autopilot, no matter which autopilot is engaged at the time. After that, the attitude and heading indicators on the affected side will temporarily go blank until the alignment process is complete, which would normally take around 30 seconds or so. But during those 30 seconds, the checklist instructed the pilots to maintain a straight and level attitude at a constant airspeed in order to ensure that the gyroscopes established that neutral baseline correctly. And the cockpit voice recorder recordings clearly shows that at this point the crew still had no idea on which IRS was actually faulty. We know that, because at one point the first officer questioned whether the left side system was the problem, before the captain directed him towards the right side system instead. And maybe most importantly here, there was no indication that this crew actually stopped to discuss what this coming procedure was going to require from them, which is a crucial step before executing virtually any non-normal procedure in a cockpit. So now here there was no meaningful discussion on how the aircraft should be flown during the alignment process, who would be responsible for maintaining the required flight path, whether the weather maybe wasn't very suitable at that point, or what systems might be affected once the selector was moved. And that meant that the anticipated autopilot disconnect that would result from switching the IRS into attitude mode wasn't mentioned here either. Now we pilots are trained specifically to manage abnormal situations without becoming overwhelmed by them, but modern crew resource management depends heavily on effective communication between the crew as well as on delegation of tasks. That's important because the human attention span is quite limited, so if both pilots would become deeply focused on troubleshooting a specific problem, well then they can begin unintentionally sacrificing awareness somewhere else. So like I mentioned before, one pilot should always fly the aircraft and the other deal with the malfunction, whilst both should still communicate each step with each other to make sure that nothing is missed. Aviate, navigate and communicate in other words. But on board this flight, the cockpit voice recorder now increasingly indicated that both pilots were mentally starting to focus on solving this IRS problem, while fewer and fewer mental resources remained dedicated to basic flight monitoring. Anyway, Flight 574 was at this point already operating in an environment full of threats. The aircraft was surrounded by convective activity, they were dealing with conflicting navigational information, and they were carrying a defect that the crew just didn't fully understand. In that kind of a situation, a TEM-oriented mindset would have encouraged the pilot to slow down rather than to speed up. So before moving that IRS mode selector over to attitude, they should have ideally discussed what system might be affected, what indications they expected to see afterwards, what role the autopilot was currently playing and how they would divide responsibilities as the situation became even more complicated. But none of that really happened. Instead, the captain now instructed the first officer to switch the right inertial reference system into attitude mode in an attempt to reset the number 2 IRU, which was actually the one functioning correctly. And this then would immediately remove the attitude information from the first officer's display. Without questioning the captain, the first officer now just complied, which meant that the autopilot immediately disconnected together with the Boeing 737's distinctive cavalry charge warning. But then, almost immediately after that, someone just reached over and cancelled that warning. And after that, nothing. Nobody announced that they had control of the aircraft or called out that the autopilot had disconnected, nor did the pilots discuss who would manage to transition from troubleshooting back to manually flying the airplane. Now, the captain had been the pilot flying up until that point, and the first officer was dealing with the checklist, so that division probably felt quite obvious. But as we will soon see, it clearly wasn't. You see, both pilots now stayed laser focused on the navigational problem, which had now become even more confusing. The first officer's instruments had partially disappeared, navigation failure flags were appearing on his displays, and the crew was still trying to determine which IRS was faulty, and whether the procedure they had just performed was having the intended effect. And that fixation now caused them to miss the sudden appearance of yet another problem, one that the autopilot had been masking for much of the flight up until this point. You see, this aircraft reportedly had a tendency to roll slightly to the right in cruise due to a slight out of trim condition in its ailerons. Now, on its own, this wasn't really a big issue. Aircraft often developed small handling quirks over their years of service, and the autopilot systems routinely compensate for them automatically. But this also meant that the autopilot had been applying continuous corrective five degrees of left aileron input throughout the flight in order to maintain wings level, and as long as it stayed engaged, it had worked beautifully. But as soon as the autopilot now disconnected, this Boeing 737 began rolling slowly slowly over to the right. At first, that motion was subtle, around one degree of bank per second, and in stable 1G flight, a shallow bank developing over several seconds can be surprisingly difficult to notice, especially when the pilots are distracted and focus on other things than their instruments. And what made this even worse was the fact that aircraft was now inside of clouds, with no way for the pilots to visually perceive the horizon, in turbulence, and with the first officer's attitude indication turned off. Unfortunately, the human body itself is extremely bad at detecting slow changes in attitude without visual cues, and that's because the inner ear senses acceleration, not stable motion, which means that a gradual turn at a constant 1G rate will feel the same as straight and level flight. This is actually one of the reasons why spatial disorientation has remained such a huge threat throughout aviation history. Once we pilots stop actively cross-checking our instruments, our body will quickly begin to lie to us, and the only way to counteract that is by diligently get back to scanning our instruments. So, whilst this roll was now starting, the two pilots just continued focusing on the IRS problem. And in fact, the first officer now tried to input heading information into the system before the alignment process had even completed, which kind of suggests that he might not have fully understood the importance of stabilizing the aircraft and allowing the system to realign properly. And as he was doing that, the aircraft just kept quietly rolling. 5° became 10°, then 10° became 20°. Now, a couple of times as this was happening, at least one of the pilots must have noticed this because someone briefly turned the control wheel left, but only long enough to temporarily stop the right roll, not to actually reverse it. So, over the next few seconds, the aircraft's bank angle continued increasing, with the nose of the aircraft also soon starting to drop, and with the deepening turn and acceleration that followed with that, the forces acting on the aircraft were now also starting to increase. Then, as the roll surpassed 35°, the GPWS warning system announced and that warning activates when the aircraft exceeds the safe bank parameters, which under normal circumstances should trigger an immediate recovery response from the crew. But here, instead, the way that these pilots responded was fragmented and inconsistent, and bafflingly, the captain's response to this warning was to say, don't turn it, this is our heading, even though he was supposed to be pilot flying. Here, it is important to point out that this situation was still very much both recoverable and survivable. A Boeing 727 can withstand substantial bank angles and unusual attitudes if pilots recognize the situation quickly and apply the correct recovery techniques. But that depends on prioritizing the correct control inputs in the right order. But now, instead of first decisively bringing the roll back to level, the captain instead pulled back on the yoke whilst the aircraft remained in a steep right bank, which just increased the g-load and turn rate. Flight recorder data then showed that he also initially actually did input a little bit of left roll, which was the correct thing to do given the aircraft's developing right bank angle, but then he almost immediately afterwards reversed that input and began commanding additional right roll instead, which of course now just made the situation worse. That is actually one of the strongest indications that this captain had already become spatially disoriented. This would have meant that his brain likely told him that he was flying straight and level when he was actually banking over 35 degrees, so when he started rolling back to the left, it would have felt like he was rolling into a left turn instead of levelling out. And the cockpit voice recorder now also captured another telling clue. Because rather than calling out the bank angle, identifying the upset or coordinating a recovery, the captain instead started shouting, put it back on nav again, put it back on nav again, put it back on nav again, four times. You see, the crew had now performed an action, the situation had immediately become worse, and the captain's instinct was therefore to undo the action that appeared to have triggered the problem. So instead of flying his aircraft, his attention was still fixated on that IRS issue. But that instruction also revealed a misunderstanding about the system itself. You see, once an inertial reference unit has become switched from nav into attitude mode in flight, it cannot simply be returned to nav mode, since it has then lost its initial realignment. So that option just didn't exist. And it is also possible here that the captain didn't trust his own instruments since he didn't fully understand the IRS issue, and therefore trusted his inner ear more than he normally would. In any case, as all of this was happening, the aircraft continued rolling further and further to the right. And that increasing bank angle soon caused the nose to drop even more, which in turn caused the aircraft to accelerate. And faced with a rapidly descending airplane, the captain then responded instinctively again by pulling back more on the control column. Flight recorder data from this point onwards shows increasingly erratic control wheel movements as he attempted to regain control, swinging the wheel back and forth while the bank angle just kept increasing. In this case, the correct thing to do would have been to roll in the shortest direction towards wings level, which might actually be continuing the roll over rather than returning, then carefully reduce thrust and extend speed brakes in order to control the speed and gently start pulling back once the bank is close to wings level. But without having received that specific training, when an aircraft enters an unusual attitude like this, the natural human instinct can often be to pull instead. Because the ground is below you, so pulling back feels obvious. But an airplane that is in a steep bank angle doesn't behave the way that intuition suggests. Much of its lift vector is no longer acting vertically instead, it is being directed sideways into the turn itself. So by pulling harder, whilst the wings remain steeply banked, you won't stop the descent instead, it just tightens the spiral and increases the aerodynamic loads acting on the aircraft. Which was now exactly what was happening. The engines, still producing thrust, also continued to accelerate the aircraft, so together with the increasing dive, the airspeed now increased dramatically. The Mach number, which is speed shown as a percentage of the speed of sound, now climbed higher and higher, while aerodynamic loads across the entire airframe continued pulling on it. Only 1 minute and 22 seconds after the autopilot had disconnected safely up at 35,000 feet, the first officer now frantically shouted out, pull up, pull up, pull up, as the dive peaked at an absolutely unbelievable negative 53,000 feet per minute, and the Mach number now shot past 0.9, approaching supersonic. The aerodynamic forces acting across the wings, tail and flight surfaces were now far beyond what the aircraft was certified for, but consumed with terror, the pilots aggressively continued pulling back on their controls even more, which ultimately sealed their fate. Once airspeed begins increasing beyond its designed limits in a dive, recovery becomes progressively more difficult, even if the crew would eventually recognize the correct escape path. The flight recorder data showed the descent actually starting to decrease as the plane began to pull up, but at 3.5 G's and at a speed of Mach 0.926, the airplane finally just broke apart. At an altitude of around 9,000 feet, the cockpit voice recorder captured two loud thumps as both horizontal stabilizers were likely ripped from the tail, and the G-forces now abruptly reversed from plus 3.5 G's to negative 2.8. Then, both flight recorders stopped working as the rest of the aircraft separated into multiple large pieces, all of which impacted the water just a few seconds later, killing all 102 people on board. In the weeks that followed, the scale of the disaster became painfully clear. At first, Indonesian authorities weren't even certain where the aircraft had gone down. Floating debris and human remains were recovered across a huge area of the sea, but the main wreckage remained missing, buried somewhere thousands of meters below the surface. And almost immediately, the recovery effort itself became entangled in the same financial and institutional dysfunction that had helped create the conditions for this accident in the first place. You see, the Indonesian government, Adam Air, insurance companies and private contractors all became locked in disputes over who would actually pay for the underwater search operation. Specialized deep ocean recovery ships were enormously expensive to operate, and progress stalled whilst arguments continued over responsibility and cost. This meant that months passed until Adam Air begrudgingly agreed to pay for one week of black boxes, but they then refused to pay for any recovery of the wreckage or victims. The US Navy survey vessel Mary Sears was eventually called in to scan the seabed of the Makassar Strait with sophisticated sonar equipment, and just a few days into their search, they discovered the wreckage of the Boeing 737 at a depth of 2000 meters. Both flight recorders were then found and pulled from the crushing depths by remotely operated submersibles. And when investigators finally listened to the cockpit recordings and analyzed the flight data, what had happened soon became clear. This disaster had been ultimately caused by an airline system riddled with shortcuts, weak oversight, falsified records and a corporate culture that normalized unsafe operations. The investigation also exposed deeper problems within Indonesia's aviation oversight system itself, where regulators had failed to adequately supervise the rapidly expanding low-cost carriers. But this crash ultimately became a turning point. It triggered intense international scrutiny of Indonesian aviation safety, leading to major reforms, stricter oversight requirements and temporarily bans on all Indonesian airlines from operating into Europe and the United States. Audemars itself never recovered from this scandal. And a following series of additional serious incidents, mounting financial pressure and collapsing public confidence, eventually pushed the airline into bankruptcy the following year. Which in my view was a more than fitting end. Now if you want to learn how we pilots safely operate the Boeing 737 today, complete with explanations of all little buttons, switches and procedures, then check out me and Ben's Virtual 737 course. It was initially made for flight simmers who want to really transform the realism of their hobby, but the explanations, checklists and video tutorials can be used by anyone who is curious. So please go to virtual737course.com and use the code MENTORPILOT for a 20% discount right now. My name is Peter Hörnfeld and you're watching MENTORPILOT. Have an absolutely fantastic day and I'll see you next time.