Aeroperu Flight 606

"Shit! What the hell have these assholes done!?"


You know what I really hate? Air travel. Everyone tells me "oh, you know, statistically speaking, airplanes are the safest way to travel, so don't worry!" Intellectually, I am perfectly aware of this. People fly every single minute of every single day without incident. I personally have flown several times, and of all those times, I've only ever had two experiences that led me to be alarmed, both through the same carrier: AirTran, formerly known as ValuJet. In the first instance, we were informed after the plane had already taken off that one of the engines was not functioning, but not to worry, the DC-9 was designed to fly using just two of its three engines. I ordered a small single-serving bottle of vodka the instant I was able to do so. The next occasion was not far from our final destination. The plane started to stall and began to vertically drop. The pilot(s) put the aircraft into a noticeably nose-down position to regain speed in quick order. The woman sitting next to me (whom I did not know) reached over and grabbed my hand tighter than a vise grip. It was the most terrifying moment of my life. The descent was bumpy and turbulent. Our flight attendant -- a pretty young woman named Jenny -- was a consummate professional, though I could tell that she was rattled from the experience by the way her eyes bulged and she ran her hand through her hair when she smiled and asked me if I had anything I wanted to throw away. As I exited the plane, the pilot was standing there with a retail smile on his face to see everyone off. My wife had not yet arrived to the airport after I ran off the plane like I had stolen something, so I had to sit and wait. That was fine because I felt like smoking four packs of cigarettes. As I waited for my ride, I saw the pilot walk out of the terminal with his luggage. A car pulled up and a woman got out. She went to hug him and he limply patted her on the back. She asked him "what's wrong?" He replied "worst fucking landing I've ever done. I don't know what these numb nuts are doing anymore." I still don't know what that means and I'm not sure I want to.

Some people are afraid of heights. I'm not. Enclosed spaces don't bother me either. I hate the feeling of not being in control of my own mode of transportation. Basically all passenger planes today are flown almost exclusively through the use of an autopilot system. In theory, this should comfort people like me. A perfect machine incapable of mathematical error is in control of my fate. All real people have to do is take off, land, and serve me my food and drink. But then the question arises: what data do these machines use? Is there any possible way in which they can be compromised? As the crew and passengers of Aeroperu flight 603 were to discover shortly after 1:00 in the morning on the 2nd of October, 1996, the answer was "yes."

The Flight

Aeroperu flight 603 was a flight originating in Miami, Florida, bound for Santiago, Chile. It had stopovers in both Ecuador and Peru. Flight 603 was originally intended to be flown by a Boeing 757 but due to mechanical issues, a 727 had to be substituted until the second layover in Lima, Peru, at which point the 757 was deemed ready for service. Flight 603 was commanded by Captain Eric Schreiber and First Officer David Fernandez. The takeoff for the final leg of the flight at 12:42 AM started out all right; things were so routine and normal that Schreiber joked "such precision, and we're not even Swiss." Within seconds of achieving flight, however, the 757 began behaving very strangely. Almost all planes in service today utitlize conspicuously audible warning systems to alert their pilots to any issues with the aircraft. Right after the plane took off, the plane's windshear warning sounded. This is not terribly alarming on its own, but it indicates a significant difference between the direction the plane is travelling and the direction of the wind blowing against its hull. More worrying, however, was the fact that one of the plane's altimeters (which measures altitude) and one of its air speed indicators (which shows how fast the plane is moving) both read "0." Although the plane was obviously flying and ascending, the readings showed the plane was still parked on the ground.

Now in daylight hours, this wouldn't necessarily be as confusing. It would be pretty clear to the pilots what was going on because they would be able to see what was going on outside. In the middle of the night, though, was a different story. After lifting off from the airport, there wouldn't be any visual indicators to speak of. Flying at night is generally done through reading the instruments and turning the autopilot on. The 757 has three computer systems that determine and display airspeed and altitude independently. It requires identical readings from two of the three systems to engage the autopilot. While the pilot's console showed the plane was sitting on the runway, the co-pilot's console indicated that the plane was actually rising with speed. The third system displayed similar information but at a different level and rate. With all three systems indicating mutually contradictory information, it would be impossible to activate the autopilot.

Within seconds, the plane's rudder ratio warning came on. The rudder ratio system activates when the computer detects that the plane is travelling above a certain speed. Once this speed is reached, it is unsafe to steer the plane with its rudder beyond a certain angle, so it restricts the range of motion available to the pilots. Considering the pilot's air speed indicator was still at 0, the activation of the rudder ratio system made little sense. Suddenly, Schreiber's altimeter and air speed indicator began working. As his readings surged to catch up with those of his co- pilot, the plane issued a mach speed trim warning, which generally means the plane is travelling too fast and that its stabilizer could be damaged by abrupt inputs. The air speed indicators were still displaying roughly 250 knots, though, which is not particularly fast. Since there were identical inputs, however, Schreiber turned on the autopilot.

Four minutes after taking off and with no idea how reliable their instruments were, Schreiber wisely had Fernandez declare an emergency to Lima air traffic control. Seconds later, the autopilot disconnected since the air speed indicators again began to display different readings. They decided it would be impossible to continue the flight and that their best course of action would be to return to the Lima airport. By this time, though, the plane was flying over the Pacific Ocean. Their only external visual cue at this point would have been the moon, which for obvious reasons was not especially helpful for two pilots who had no idea what their plane was telling them or why. Their air speed indicators again displayed zero, but their altimeters showed they were climbing at a fast rate, which is physically impossible.

Increasingly frustrated, Schreiber and Fernandez suspected maintenance workers at the Lima airport had either tampered with the aircraft or not fixed whatever issue originally kept the plane on the ground, including the quote that leads this essay. Schreiber would continually reactivate the autopilot and it would continually disengage shortly afterward. Whenever the autopilot automatically disengages in most planes, a master caution alarm loudly blares. This, coupled with the repeated rudder ratio and mach speed alarms, made for an extremely stressful cockpit environment. Schreiber and Fernandez now had to rely on their air traffic controller's radar to tell them their speed, their altitude, their heading, and their distance from the airport. The controller was able to give them accurate readings for all of these except their altitude; he was receiving this piece of information from the plane's obviously dysfunctional transponder, although nobody involved in the ordeal was aware of that at the time.

At this point, Schreiber and Fernandez decided to focus on preparing the plane for an emergency landing back at Lima. Their controller gave them the information they would need to land on runway number 15 when they were approximately 40 miles away from the airport. They were now travelling at a rate of just under 300 knots. Their air speed indicators, however, now displayed a rate of 350 knots. This triggered another audible warning, this time the overspeed alarm. This means exactly what it sounds like; when the plane begins to travel faster than its recommended speed, structural damage can occur. This is a cue to slow the plane down, which is exactly what they did. Once they did this, the stick shaker alarm activated. The stick shaker is a warning system that violently shakes the controls of the plane to get the pilots' attention. It means that the plane is now about to stall and fall out of the sky unless and until the pilots pick up more speed. Both the stall warning and the overspeed warning were active at the same time, which is contradictory. Any semblance of calm that remained disintegrated.

In a panic, Fernandez requested that Lima send another plane to their position to guide them back to the airport. Schreiber attempted to overrule him, but the air traffic controller informed them that a Boeing 707 was already being prepared to take off and meet them. The pilots then began arguing about whether the plane was going too fast or too slow. The air speed indicators now displayed a speed of 395 knots, which should have been enough to physically damage the plane. The air traffic controller informed them that they were showing a speed of 220 knots, very close to the stall speed. They had to quickly increase that rate if they had any hope of making it back to Lima alive. As they began to try to recover from the stall, the plane's ground proximity warning system activated. The GPWS is designed to tell a plane's crew they are flying dangerously close to the ground. It does this by loudly blaring "TOO LOW TERRAIN" over and over again. If this is not corrected and the plane continues to descend, the GPWS will sound two tones followed by a voice insistently saying "PULL UP!" This was even more confusing because the Lima air traffic controller had just informed them that they were close to 10,000 feet in the air.

This put the pilots in an awkward position. They were already close to stalling the plane, but pulling up to increase their altitude would slow the plane down even more. On the other hand, though, putting the nose of the plane in a downward position to regain speed could cause them to crash into the ocean. In attempt to give the pilots some good news, the air traffic controller informed them that the 707 being sent to rescue them would be taking off in 15 minutes or less. Eventually they are able to pull the plane out of the stall and pull up far enough to disengage the terrain alarm. Schreiber and Fernandez began to focus on preparing the plane to land. Within minutes, however, the GPWS comes back on. They dismiss the alarm as incorrect and continue their landing preparations. As they begin to turn the plane to make its return to Lima, the horrible truth is finally revealed: the left wing hits the surface of the Pacific Ocean. Their displayed flight level of 9,700 feet was off by about 9,650 feet at that point. Schreiber and Fernandez frantically scrambled to pull the plane up away from the water, but the damage was irreversible. They rose approximately 200 feet, at which point the plane turned upside down and sank into the ocean. The passengers and crew of Aeroperu flight 603 -- aside from the pilots -- had no idea what was happening. The cockpit voice recorder shows that neither Schreiber nor Fernandez addressed the cabin at all during the flight. There were no survivors among the 61 passengers and 9 crew members. Chillingly, it was later determined that although some people were killed instantly upon impacting the water, the vast majority of the plane's occupants were alive when it began to sink and as a result drowned in the largely intact plane. From its takeoff to its crash landing in the Pacific Ocean, the final leg of Aeroperu flight 603 lasted just 31 minutes. The last sound captured by the cockpit voice recorder was the GPWS futilely declaring "WHOOP WHOOP PU--" before the plane is heard falling into the water.

Investigation and Aftermath

Peru's Commission of Accident Investigations is responsible for investigating plane crashes in its territory. Because the CAI did not have much experience dealing with accidents involving Boeing 757s, the American National Transportation Safety Board volunteered its help, which the Peruvians willingly accepted. Peru's lead investigator was a man named Guido Fernandez, the uncle of First Officer David Fernandez. While at first concerned that Fernandez would be unable to dispassionately investigate his own nephew's death, the NTSB team would later go on to be impressed by his professionalism in dealing with the matter.

Because the plane was essentially complete, it was very easy to locate both of its black boxes: the cockpit voice recorder and the flight data recorder. While the CVR's function is to capture all sounds emanating from the cockpit, the FDR captures all of the control inputs and information reported to the plane's computers. Based on the analysis of the FDR and extrapolating the relevant control inputs into a simulator, the investigators were able to determine that immediately following the plane's takeoff, it proceeded to descend at an almost constant rate for the duration of the flight. While the windshear, overspeed, rudder ratio, and mach trim warnings were all discarded as erroneous, both the GPWS alarms and the stall warnings were found to be correct.

The investigation of the crash was complete within two months. While this might seem like an alarmingly short period of time, there was a good reaon for it. The crash of Aeroperu flight 603 was eerily similar to the crash of Birgenair flight 301, which occurred only 8 months earlier off the coast of the Dominican Republic. Flight 301 was also a 757. The pilots of 301 were also forced to endure contradictory overspeed and stall warnings as well as simultaneous rudder ratio and mach trim alarms. Both the altimeters and the air speed indicators were useless. Birgenair flight 301 also crashed into the ocean resulting in the deaths of all aboard. It was later determined that Birgenair flight 301 suffered a catastrophic failure of its pitot-static system. The pitot-static system uses forward-facing tubes on the side of the aircraft and sensor ports underneath or on the side of it to compare the rate of the air coming into the tubes to the still (or static) air to determine a plane's speed. The investigation into the Birgenair crash discovered that the plane had been left unattended on a runway for three weeks without the customary pitot tube covers in place. The pitot tubes became clogged, likely by wasps building nests inside of them.

It did not take long for the investigators of the Aeroperu crash to confirm that that plane had also been brought down by a failure of its pitot-static system. While it could never be proven 100% that wasps were responsible for blocking Birgenair 301's pitot tubes, the cause of 603's issue was immediately obvious. Given the relatively intact nature of the downed jet, investigators were able to clearly see duct tape covering its static sensor ports. The duct tape had been put there by an Aeroperu maintenance worker during a routine cleaning of the plane who had forgotten to remove it prior to the plane taking off. Boeing had specifically said in the past that duct tape should not be used to cover the static ports of its planes because the grey duct tape would likely blend in with the color of the plane and be easily missed during a pre-flight inspection, which is exactly what happened in this instance. Bright pink or yellow tape is recommended for this very reason. Schreiber had been unable to see the duct tape on the static ports in the dead of night as he was doing his inspection and cleared the plane for takeoff.

Also at issue was the fact that after the Birgenair crash, the NTSB had instructed Boeing to issue notices to its 757 operators as to how best to handle simultaneous rudder ratio and mach speed trim warnings which were almost certainly indicative of a failure in the pitot-static system. For whatever reason, Boeing had not made particular haste in getting this information out to its clients. While pilots like Eric Schreiber and David Fernandez would surely have been aware of the Birgenair crash, they could not be reasonably expected to know all the details of how to handle the exact same situation properly.

At the same time, the CAI investigators did ultimately assign partial responsibility to Schreiber and Fernandez. Schreiber was taken to task for failing to notice the duct tape and for not paying attention to the second and final terrain warning, which is governed by a system entirely separate from the pitot-static system and therefore not liable to the same types of failures and inaccurate readings. This is by design to ensure that the GPWS alarms are heeded regardless of indicated flight levels and speeds. Similarly, Fernandez should have been more insistent in getting Schreiber to respond to the GPWS alarms. The crash would ultimately be categorized as a CFIT -- controlled flight into terrain -- complicated by multiple instrument failures. CFIT implies that an otherwise airworthy plane crashed under the control of its pilots. In theory, Aeroperu 603 could have continued flying until it ran out of fuel as long as the directions of the GPWS alarms were followed. It would have been annoying and frustrating and confusing with the multitude of alarms that would have been constantly sounding, but it would have been possible.

Lawsuits inevitably followed when the results of the investigation were made public. Both Aeroperu and Boeing were sued by surviving family members of the crash victims: Aeroperu for improper maintenance and training and Boeing...well, Boeing because they had money and Aeroperu didn't. The case against Boeing was not particularly strong, resting mainly on the premise that they should have been more actively involved in saying that duct tape should not be used to cover static ports during cleaning and that they should have worked harder in the aftermath of Birgenair 301 to inform their customers of the issues surrounding pitot-static failures. Legal actions were brought in both Peru and the United States. Additionally, Boeing and Aeroperu reached an out of court settlement with other victims' families who did not wish to pursue lengthy trials. The case in Peru resulted in a $29 million judgment against Aeroperu, forcing it into bankruptcy. Boeing settled out of court in the United States, and while the details of the settlement were not made public, it is said that it exceeded $1 million per victim.

The final court case involved Eleuterio Chacaliaza. He was the Aeroperu maintenance worker who initially put the duct tape on the static sensors and then failed to remove them. Chacaliaza was arrested and charged with 70 counts of negligent homicide. Guido Fernandez has publicly spoken out against Chacaliaza's arrest, trial, and subsequent imprisonment. While he was, of course, literally responsible for putting the duct tape on the plane, Fernandez argued that Chacaliaza's low level of education and lack of supervision (the maintenance manager who was supposed to be on duty that night of the flight called in sick) made it impossible to truly consider him legally liable. While it is unknown whether or not the lead Peruvian investigator's words had any bearing on his case, Chacaliaza was released after two years and given a suspended sentence.

The destruction of Aeroperu flight 603 is, to me, one of the most horrifying plane crash scenarios. I can't imagine what the pilots must have felt throughout the flight; nevermind the passengers who had no clue what was going on right up until the plane hit the water. The most alarming thing about the crash, however, is the minimal impact on air safety that it has had. There's really no way to prevent someone else from putting the incorrect color of tape on a static port or the pilot missing it during an inspection other than to say "don't do that." Likewise, it's very easy to conceptualize the idea that "all these mutually contradictory warnings should be ignored" while you're sitting at a desk, but I highly doubt it's that easy in the air when lives are at stake. In 2009, Air France flight 447 -- an Airbus A330 flying from Rio de Janeiro to Paris overnight -- crashed into the Atlantic Ocean, killing all 228 passengers and crew on board. It took almost 2 years to find the plane's wreckage at the bottom of the ocean floor as well as its intact black boxes.

The mystery of 447 was solved when it was revealed that when the plane entered a storm, the precipitation froze on the plane's pitot tubes and began giving incorrect air speed indications. The relatively inexperienced co-pilot who was at the controls subsequently pitched the nose of the plane up to compensate and repeatedly ignored the various stall warnings the plane sent out. Eventually, the stall warnings stopped because the plane's computer saw no action was being taken to correct the problem and assumed the readings it received were faulty. The captain -- who had been attempting to sleep at this point during the long flight -- returned to the cockpit and asked "what the hell are you doing?" At this point, the plane did stall and began a descent that would last about 3 minutes. This happened at 38,000 feet, meaning that there was plenty of time to recover from the stall. Shortly before impacting the ocean, the co-pilot let out that he had been pitching the nose up the whole time, at which point the captain immediately told him to pitch the nose down to get out of the stall. The GPWS sounded at this point, though, meaning that the plane had lost too much air speed to recover. The co-pilot said "fuck, we're going to crash," which it did. Thirteen years after the crash of Aeroperu 603, with fewer problems and an even more sophisticated plane, over 200 people lost their lives due to an improper response by a pilot to a computer error. The skies may be friendly, but the people flying them are, in the end, people.

Log in or register to write something here or to contact authors.