Sao Paulo Flight Crash Raises Questions about ATR 72s

 

On Friday Aug. 9, a regional airline flight carrying 62 passengers and crew bound from Cascavel to Sao Paulo, Brazil crashed in a suburb of Sao Paulo, killing all on board.  The crash was significant for several reasons, and not only because it was the most deadly plane accident of 2024 so far, but because it involved an ATR 72.  The ATR 72 is a turboprop plane made by a consortium of French and Italian firms and intended for regional airline service.  While it is too early to make any definitive statements about the cause of the crash, we can summarize what is known so far.

 

The flight appeared to be going normally until the plane was about eighty miles (130 km) away from the destination airport.  The pilot then appeared to deviate from his intended course, and eyewitnesses on the ground reported that the plane was in a flat spin when it crashed into a residential neighborhood.  Fortunately, there were no casualties on the ground, but all 62 people on board the plane were killed. 

 

Since around 2000, the total global number of fatalities per year in commercial aircraft has been on a mostly steady decline.  As total air miles traveled has increased, this means that the fatality rate per passenger-mile is declining even faster.  2023 was one of the best years to fly safety-wise, as only 102 fatalities were recorded.  Unfortunately, most of that number consists of the 72 people who were killed in the crash of another ATR 72, this time in Nepal.  That crash was due to human error:  during the landing approach, a pilot "feathered" the propellers by mistake when he used the wrong levers in the cockpit.  This turned the propellers at an angle that reduced the engine thrust to zero, and the plane stalled and crashed short of the runway.

 

The cause of the Sao Paulo crash is still uncertain.  Along with the bodies of all 62 victims, the two flight recorders have been recovered, and Brazilian authorities are promising that initial results of the investigation will be available within 30 days.

 

Over its 36-year history of production, the ATR 72 has been involved in accidents resulting in about 500 fatalities.  According to the Wikipedia page on the aircraft, several of these crashes have been attributed to icing conditions.  Icing is a difficulty that modern aircraft are well equipped to deal with, but pilots must be aware of icing conditions in order to take steps to mitigate its adverse effects.

 

For example, the use of autopilots during icing conditions can lead to trouble, as one American Eagle flight was on autopilot, iced up, and crashed in the U. S. on Oct. 31, 1994 after flying a holding pattern at 8,000 feet, killing all 68 people on board, according to an AP report on the Brazilian crash.  In Sao Paulo last Friday, a local meterological agency said there were severe icing conditions at the time of the accident, so it is at least a possibility that icing could have contributed to this crash as well.

 

Turboprop planes use jet-engine-like turbines, not for their jet-propulsion qualities, but to generate torque to turn conventional propellers.  But like any air-breathing turbine, the engines of a turboprop can be compromised or incapacitated if ice builds up on the intake ports.  Ice on control surfaces can prevent the maneuverability needed for proper flight, and ice gathering onto flight instrument sensors can lead to false readings and inappropriate responses from autopilot mechanisms.

 

Modern aircraft have ways of combating all these problems, but not automatically.  De-icing systems use power and can compromise performance in other ways, so they are generally controlled by cockpit switches that the pilot must use.  Preoccupation with other duties or inattention to weather conditions might render these devices useless if the pilots do not notice there is an icing problem until it is too late.

 

And icing can occur in any kind of surface weather.  The fact that a summer thunderstorm on a blazingly-hot day can produce hail is a vivid reminder that the typical height at which the air is at freezing temperature is about 10,000 feet (3,000 meters), which commercial regional flights fly through quite often.  While pilots try to avoid clouds that are likely to produce icing conditions, this is not always possible, which is why anti-icing devices are installed.  But they only work if they are turned on.

 

It is an open question whether the ATR 72 needs attention with regard to its icing problems, but if history is any guide, a combination of icing conditions and poor crew training procedures can lead to a dangerous situation.  As long as robots aren't flying planes yet, we will need highly trained professionals operating them from the cockpit, and those professionals need to pay attention to a lot of possible problems. 

 

I recently found a used copy of Van Sickle's Modern Airmanship, 6th Edition.  It is a kind of encyclopedia of flying, with all the answers to questions a student pilot could ask and many that an experienced professional needs to be reminded of from time to time.  While I always knew that flying was hard, reading this book has brought home to me just how hard it can be, and how many details have to go exactly right for a smooth commercial flight to come off from takeoff to landing. 

 

All the technical training in the world will not keep a careless pilot from making mistakes.  And while pilots are human too, they hold the responsibility for dozens of human lives literally in their hands, just like a surgeon does.  It takes character to live up to that responsibility, and cultivating such character is a big part of pilot training and recertification.  It's possible that the Sao Paulo crash was due entirely to mechanical problems and the crew was blameless.  We will have to await the results of the investigation to find out.  But whatever happened, it's too late for the 62 who died.  It's not too late, however, for the rest of us to learn from whatever mistakes were made, and avoid them next time.

 

Sources:  I referred to the Associated Press articles on the crash at https://apnews.com/article/brazil-plane-crash-sao-paulo-394427322513643f4eaf517e4ac4c85e

and https://apnews.com/article/brazil-airplane-crash-deadly-voepass-sao-paulo-5c5a3371b5e899b745c1cb6e46950230.  I also consulted statistics on plane crashes at https://www.airliners.net/forum/viewtopic.php?t=1490283 and the Wikipedia sites "List of accidents and incidents involving commercial aircraft," "ATR 72," and "Yeti Airlines Flight 691."  

Sriwijaya Air Flight 182 Investigation Focuses on Automatic Throttle

 

On Saturday, January 9 of this year, Sriwijaya Air Flight 182 took off from Soekarno-Hatta International Airport near Jakarta a little after 2 P. M. with 62 passengers and crew on board, headed for Pontianak, Indonesia.  It was raining heavily, and less than five minutes into the flight, shortly after the plane crossed the shore of the Java Sea, flight controllers and radar lost contact with it.  A fishing boat reported hearing a loud explosion in the vicinity of the disappearance, and search parties have since discovered debris spread over a 2-km area (over a mile).  There were no survivors.

 

Although the flight-data recorder was recovered, the cockpit voice recorder (CVR), a separate piece of equipment, broke apart in the crash.  The CVR's memory unit is designed to survive even if the CVR is destroyed, but understandably it is difficult to locate, and has not yet been found.

 

Even without the CVR, investigators with the Indonesian National Transportation Safety Committee have begun to focus their attention on the automatic thrust controller of the Boeing 737 plane, which entered into service in 1994. (This is an earlier version of the 737 than the later MAX series, which was grounded for more than a year over concerns about flight software when two crashes were traced to software malfunctions.) 

 

Manual aircraft throttles are very similar to the gas pedal in a car:  they simply control the rate of fuel supplied to the engines.  Depending on wind conditions, angle of attack, and other factors, a given throttle setting can result in very different amounts of either thrust or speed.  To maintain constant speed, pilots would have to be constantly fiddling with manual throttles.  So modern airliners have autothrottle control systems that can be set either to maintain a constant speed (within safety limits) or a constant thrust, which is convenient for takeoff and landing.  In this connection, many autothrottles have a takeoff/go-around switch that is normally engaged only during takeoff or if the pilot decides to abort a landing.  Pressing the go-around switch provides a controlled amount of added thrust that will enable a smooth transition from an aborted approach to going around for another approach.  But to engage the go-around switch before entering an approach descent is a bad mistake.

 

This is apparently just what happened in another crash involving a Boeing 767 cargo plane heading for Houston's Bush Airport in February of 2019.  Investigators believe that while the plane was a good distance away from the airport, the first officer's wristwatch accidentally hit the "go-around" button on the autothruster, causing a surge in power.  The plane was in the clouds, which obscured all visible horizon references, and a sudden acceleration can feel like an upward tilt to pilots who are not sufficiently trained to ignore such sensations and pay attention to their instruments instead.  Evidently, the first officer, who was flying the plane at the time, decided that the plane was stalling and pushed the control stick forward, sending the aircraft into a dive from which it could not recover.  It crashed into a swampy area east of Houston with the loss of all three crew members. 

 

It's not yet clear what happened to Sriwijaya Flight 182, but the flight data recovered so far show that one of the plane's two engines was producing much more thrust than the other for the final seconds prior to the crash.  News reports indicate that there may have been problems with the autothrottle computer on this particular plane in the weeks before the crash.

 

Unequal thrust on a plane, if not corrected or compensated for, can cause severe rolling and even destabilize its flight and send it into a dive.  Normally, pilots are trained to notice such situations and to deal with them promptly.  But a factor in the Houston crash was inattention:  neither the captain nor the first officer understood what was going on until it was too late to do anything about it. 

 

In the weather conditions under which Sriwijaya Flight 182 took off, the pilots had to rely on their instruments during and after takeoff, and it's possible that they failed to figure out what was happening during the fifteen seconds or so that the aircraft took to fall from 10,000 feet into the Java Sea. 

 

Fifteen seconds is not much time to diagnose one specific problem from a complicated set of instrument readings and take exactly the right evasive action to deal with it.  As aircraft have become more complex, there are more things that can go wrong, and pilots have to be trained to deal with each one of them in the appropriate way.  Especially during takeoff and landing, where margins of error are small, having the correct response to a sudden emergency can mean the difference between life and death.

 

But 99.9% of the time, piloting a modern aircraft is about as exciting as driving to the grocery store—less, if you consider that in driving a car you don't have some friendly ground controller telling every other car exactly where to go so you have plenty of room to drive in.  But during every second of flight, pilots are expected to be in a state of high vigilance with the procedures for dealing with dozens of different kinds of emergencies at their mental fingertips.  While some pilots do meet these standards, as for example Chesley Sullenberger did during the 2009 Flight 1549 ditching in the East River, others do not.  The ones that don't can nevertheless fly without incident for years, because the systems they operate behave themselves.

 

But when something goes wrong, as it evidently did during Flight 182 out of Jakarta, the responses of the pilots are critical.  While we will have to await the full investigation results to be sure, it appears that both maintenance and training procedures for Sriwijaya Air may need to be overhauled.  Better maintenance can prevent mechanical malfunctions such as autothrottle failures from happening, and better training can help pilots more closely approach the ideal of eternal vigilance that is ready to deal with even unlikely emergencies such as autothrottle failures. 

 

Sources:  I referred to an article in Bloomberg News that describes what is currently known of the accident investigation results for Sriwijaya Air Flight 182 at https://www.bloomberg.com/news/articles/2021-01-20/faulty-automatic-throttle-eyed-in-indonesia-jet-crash-probe.  I also referred to Wikipedia articles on the crash as well as that source's articles on autothrottles, Atlas Air Flight 3591, and takeoff/go-around switches. 

Airline Safety: No News Is Good News—Or Is It?

It has been almost four years since the last fatal commercial airline accident in the U. S.:  the crash of Flight 3407 in February 2009, in which pilot and copilot errors combined to send the plane into a house in Clarence Center, N. Y., killing fifty people.  Of course, that could change overnight, but for the moment we can be grateful that the airline safety record looks so good.  However, there’s a fly in this otherwise sweet-smelling ointment:  it turns out that crashes stimulate the Federal Aviation Administration (FAA) more than anything else to improve safety measures such as pilot training and work rules.  And the lack of such stimulus has allowed the FAA to drag out some important pilot-training improvements for over a decade.

In the early years of flying, many accidents were due to mechanical failures, and this remained true at least until the 1980s.  Cargo doors, flammable cargo, and of course bombs (before airport security was beefed up) were responsible for many fatalities.  But with the advent of airport security measures and technical improvements in both airframe construction and restrictions on the types of cargo carried, most of the non-human causes of commercial crashes have been adequately dealt with.  What remains, as the story of Flight 3407 tells us, is the human factor.

Because fatigue seemed to play a big role in the Flight 3407 accident, the FAA began to revise rules on pilot work schedules to prevent the kind of overscheduling that pilot Marvin Renslow and 24-year-old copilot Rebecca Shaw experienced before their fatal accident in 2009.  Shaw had joined the flight after an all-night commute from Seattle, and at the time it was a common thing for pilots to snooze in airport waiting rooms at odd moments rather than checking into a hotel with the accompanying delays.

But only in 2011 were the new work-schedule rules implemented, and then only for planes carrying paying passengers.  Cargo flights are still immune from the new rules, which seems to imply that while we want to protect paying customers, pilots and hardware are expendable.

Regulation is a sparring match between an industry that sees restrictions on how they can use their paid staff in dollars-and-cents terms, and a government agency that is beholden to Congress and the people at large to ensure that airline travel is “safe.”  Of course, “safe” can only be approached, not achieved, and therein lies the difficulty.  The practical outcome is that things slide along, with the FAA taking years to solicit industry input and modify the proposed rules, until an accident prods Congress to come down hard on the agency with a mandate for improved rules that will keep the next horse from getting out of the barn after the present one has escaped.

That is perhaps a cynical view of the process, but it appears to cover the facts.  I have not looked at the proposed new rules, but as I recommended in May 2009 when the National Transportation Safety Board issued its conclusions on the causes of the Flight 3407 crash, we can learn a lot from near-accidents without having to go through the agony of a real one.

A recent news report on the issue of new FAA regulations says that voluntary data-gathering has been emphasized over one-size-fits-all obligatory rules.  And perhaps that is one reason that we’ve had such a long spell without commercial-airline fatalities.  The Air Line Pilots’ Association (ALPA) holds an annual Air Safety Forum, and the September 2012 online issue of the Air Line Pilot Magazine describes topics at the four-day forum such as stall recovery, the new pilot training rules proposed by the FAA, and airport safety issues.  Computerized records of voluntary safety-issue reporting make it easier than ever for pilots to learn from the mistakes of their colleagues.  Under the old cop-and-bad-guy model, airlines were reluctant to publicize pilot errors because of fears that they would get in trouble with the FAA, and consequently, knowledge about errors that could have turned into major disasters stayed in the cockpit.  Under the new atmosphere of collaboration, however, the FAA encourages such sharing of experience, with the result that pilots are more prepared than ever to avoid or deal with dangerous situations that other pilots have encountered.

This is one more example of a general principle: that we often learn more from technical mistakes and errors than we do from uneventful success.  This was something Chesley Sullenberger put into practice on January 15, 2009, when some geese killed all the engines of his Airbus A320 on his descent to New York City.  Sullenberger and his co-pilot successfully ditched the plane in the Hudson River and everyone was safely rescued.  It turned out that Sullenberger had served the NTSB as an accident investigator and was active in safety committees for the ALPA.  A good man constantly improving his safety skills got a big chance to put them into action, and he did.

What Sullenberger did that day, we would hope that the entire industry does all the time:  learn from previous errors, communicate them widely, and use that knowledge to prevent future incidents.  So far, it seems to be working.  It doesn’t make headlines, and it doesn’t stir Congress to action, but the pragmatic engineering criterion “does it work?” seems to be met here.  Let’s hope that the new record of time without a fatal commercial crash keeps getting longer each day.

Sources:  The Associated Press carried Joan Lowy’s article on airline pilot rule making, which was titled “No crashes, so new safety rules stall” in the Austin American-Statesman for Oct. 21, 2012.  I consulted the September 2012 online edition of the Air Line Pilot Magazine at http://www.alpa.org/publications/Air_Line_Pilot_September_2012/Air_Line_Pilot_September_2012.html#06 and the Wikipedia article on Chesley Sullenberger.  My blogs on Flight 3407 appeared on Feb. 16 and May 18, 2009.