Analysis: Well-known hazards seen as likely factors in Asiana crash

SAN FRANCISCO/ SEATTLE Mon Jul 15, 2013 2:32pm EDT

Asiana Airlines Boeing 777 is engulfed in smoke on the tarmac after a crash landing at San Francisco International Airport in California July 6, 2013 in this handout photo provided by passenger Eugene Anthony Rah, released to Reuters on July 8, 2013. REUTERS/Eugene Anthony Rah/Handout via Reuters

Asiana Airlines Boeing 777 is engulfed in smoke on the tarmac after a crash landing at San Francisco International Airport in California July 6, 2013 in this handout photo provided by passenger Eugene Anthony Rah, released to Reuters on July 8, 2013.

Credit: Reuters/Eugene Anthony Rah/Handout via Reuters

SAN FRANCISCO/ SEATTLE (Reuters) - In the seconds before Asiana Airlines Flight 214 slammed into a seawall at San Francisco airport on July 6, pilots realized the plane was flying too low and much too slowly - even though, they told investigators, they had set a control system called an auto-throttle to keep the Boeing 777 at a constant speed.

The pilots belatedly tried to abort the landing, but it was too late. Three Chinese students died in the crash, and at least 14 people remain in hospitals with severe injuries.

The U.S. National Transportation Safety Board, which is investigating the crash, has stressed that it's too early to determine the cause. Still, information from the plane's flight data recorders and cockpit voice recorders released by the NTSB strongly suggest a chain of events similar to what was seen in several other crashes in recent years, aviation experts say.

As in the crash of a Colgan Air jet near Buffalo, the crash of an Air France flight over the Atlantic, and the crash of a Turkish Airlines jet in Amsterdam - which all occurred within six months in 2009 - the Asiana pilots appear to have let the plane's airspeed dip dangerously low and failed to correct it in time. While regulators' reports found multiple causes of the 2009 accidents, including some equipment issues, in each instance they questioned pilots' training and called for more instruction on issues ranging from monitoring of computerized flight systems to dealing with surprises in flight.

The latest crash renews questions about whether increasing reliance on automated flight controls in large passenger jets is eroding pilots' abilities to fly the planes manually and handle unexpected situations.

The Asiana crash also highlights long-standing complaints by the NTSB that the U.S. Federal Aviation Administration, which must implement any recommendations from the safety board, is not moving fast enough to address emerging weaknesses in airline safety. In particular, the NTSB has advocated for better low-airspeed warnings in the cockpit and improved pilot training, but action has been slow.

Airline safety has improved radically in recent decades, and there is broad agreement in the industry that sophisticated cockpit computers have been a major factor in reducing the number of fatal accidents. But it's now clear that computers have also introduced a new set of problems for pilots.

"In the last five years, you'll find the preponderance of accidents have had some element of less-than-optimal flying skill," said David Greenberg, a longtime Delta executive and former chief operating officer at Korean Air. "There was a day when machines were falling out from under humans. Now the humans are riding perfectly good machines into the ground."

In particular, he said, over-reliance on automation was "a global issue right now, not an Asian issue or Korean issue."

Pilots have warned about the loss of manual flying skills ever since Britain's Trident jetliner pioneered automated landings in the 1960s.

The debate has become a major industry-wide issue as later generations of aircraft put ever greater emphasis on automatic systems and risk-averse airlines encourage their pilots to maintain their manual skills in the simulator rather than in flight.

Greg Feith, a consultant and a former senior investigator with the NTSB, said pilot training needs to change.

"Training programs emphasize that the automation can fly better than a human, so a lot of the training is focused around letting the plane do an auto-land or shoot the approach or do a variety of things," Feith said.

"That degrades the pilot's skill because it takes them out of the loop and it breeds complacency."

A Korean government aviation official, who declined to be identified because he was not authorized to speak publicly, said manual flying was once common among Korean pilots, many of whom where former military pilots. But in an effort to improve safety after a 1997 Korean Air crash in Guam, pilots were encouraged to make more use of automated controls.

"Many pilots now get used to automation, and they are given little opportunity for manual flying," he said. "Nowadays, a pilot is required to have management skills rather than flying skills because of automation - it is more important to know the system and understand how to manage it.

"We sometimes make jokes to pilots, saying 'are you a pilot or a manager?'"

In the wake of the Asiana crash, he said, "airlines should overhaul training programs, and beef up manual flying."

An Asiana spokesman said the company could not comment while the investigation is ongoing. In a presentation to the Korean government that was shared with reporters on Monday, Asiana said it would enhance training for pilots seeking to fly new aircraft and take other measures in response to the crash.

The FAA last week approved new rules, demanded by legislators, requiring co-pilots to have as many minimum flying hours as pilots. The measure came in response not to the Asiana crash, but rather to the 2009 Colgan Air crash, where pilots responded incorrectly to a stall warning. A stall, in which a plane loses lift and effectively drops out of the sky, is the most extreme result of a plane going too slowly; a "low and slow" landing can cause an accident even without a stall.

Just days after the Colgan crash, a Turkish Airlines flight stalled after a problem with a radio altimeter was compounded by pilot error in Amsterdam. Four months after that, an Air France flight departing from Brazil crashed after a junior pilot over-reacted to the loss of airspeed readings and inadvertently put the plane into a stall, which the crew then failed to recognize despite numerous voice warnings.

NO ACTION ON SAFETY RECOMMENDATIONS

For close to a decade, the NTSB has been calling for the FAA to study whether it should require airlines and jet manufacturers to install cockpit systems to warn pilots when they are going too slowly, even before they are in immediate danger of stalling. After the Turkish Airlines crash in 2009, Boeing began installing a spoken low-airspeed warning system in 737 airplanes. The 777 is equipped with a low-speed warning tone, but it does not feature a spoken alert. By the time a stall warning known as a "stick shaker" activated on the Asiana flight, the pilots were unable to abort the landing.

In a 2010 safety recommendation published one year after the Colgan crash, the NTSB expressed frustration that the FAA had not moved more quickly to implement low-airspeed warnings: "The NTSB notes that human factors concerns associated with a low-airspeed alert do not require more than 6 years of study for a solution to be implemented."

Scott Maurer, whose daughter Lorin died in the 2009 Colgan accident and who heads a group of families that advocate for more stringent safety standards, blamed the FAA for delaying rule changes because of their cost.

The FAA's delay "gets in the way of doing the right thing and lives have been lost as a result," Maurer said.

Still, Maurer said there has been progress.

"The origins of so many FAA regulations are the result of crashes," Maurer said. "Asiana may be the tipping point, as Colgan 3407 was a tipping point."

In a statement, the FAA said it is still studying possible enhancements for pilot awareness of low speed conditions on large commercial planes.

"That work is continuing and involves considerations for new and existing designs," the FAA statement added.

Senator Charles Schumer of New York has called on the FAA to approve by October another rule passed by Congress that would require pilots to receive more intensive simulator training on stalls.

Robert Mann, an aviation consultant in Port Washington, New York, suggested another possible lesson from the Asiana incident: looking more closely at how pilots are paired on flights. The pilot at the controls of the Asiana flight was attempting his first landing of a Boeing 777 jet in San Francisco and his supervisor was making his first flight as a trainer, and it was the first time the two pilots had flown together, the NTSB said.

"The issue is whether you pair certain crew members with certain characteristics together," Mann said.

Asiana declined to comment on that issue.

Whatever conclusions are eventually reached about the Asiana crash, the lengthy process the FAA undergoes to approve new rules suggests that any change may be slow in coming. Global standards, requiring agreement among FAA counterparts around the world, are likely to take even longer.

(Reporting by Gerry Shih and Alwyn Scott; Additional reporting by Karen Jacobs in Washington, Tim Hepher in Paris and Hyunjoo Kim in Seoul; Editing by Jonathan Weber and Claudia Parsons)

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/
Comments (5)
Tomonthebeach wrote:
It is ironic that aircraft are becoming evermore automated and thereby making pilots evermore marginal while at the same time pilots are decrying the automation as leading to aircraft crashes. It seems doubly ironic that the current failure was apparently pilot error – not using the automation properly. Although recommending more hands-on training might seem reasonable (it does to me), others might conclude that systems should design pilots out of the loop during flight transitions as equipment is not failing – people are.

Jul 15, 2013 4:43pm EDT  --  Report as abuse
PapaDisco wrote:
Yes it likely was Pilot Error, but don’t blame it on the pilot . . .

Without a doubt, more manual flying and practice is needed. Airlines lobbied for and won the right for a pilot flying a new type for the first time to be allowed to make his very first landing in that plane while it’s loaded with passengers. While this is generally safe and does save the airline a lot of money, the flying public would be much better off if we went back to the days of pilots checking out first in the actual aircraft (not the simulator), including a couple of dozen practice landings and manually flown instrument approaches to minimums. Of course this is expensive to do in something as big as a 777, but it is the right way to train.

The whole concept of “Pilot Error” itself is a useless anachronism left over from admiralty law (you know, the whole “captain goes down with the ship” state of mind). In essence, this unique corner of the law intends for anything and everything to be the fault of the captain. While this is an excellent way to ascribe responsibility, it is a terrible way to solve safety issues post accident and has led to increasingly complex automation incrementally removing pilots from the flying process.

Intuitively, that sounds like an improvement, but it is not. What happens is that we have spectacularly complicated flying systems that, when disaster strikes (think AF447), shut themselves off and hand the flying over to a surprised, disoriented pilot. Essentially what our current safety process has done has been to declare the pilot in command to be less capable than the computer in flying the plane, except in emergencies that are so spectacularly complex and disorienting that the computer, without warning, hands the plane back to the pilot and washes its electronic hands of the whole situation.

Perhaps Asiana 214 had a bum autothrottle that neither worked nor alerted the pilot to its failure mode. Regardless, these pilots were distracted by the automation, or counting on it too much, and ignoring the simple (manual) rules of pitch, power and airspeed.

Jul 15, 2013 8:17pm EDT  --  Report as abuse
TommasoT wrote:
I’ll tell of another hazard. It’s lack of handflying experience.

10,000 or 43 hours flying time logged means NOTHING! After take-off Asiana rules that autopilot MUST be engaged at 250 ft = 1min. of handflying. If a manual approach is done, there is 3 mins. of handflying. If it was an auto. approach, then only 30 seconds of handflying. If autoland was done after the auto approach, then 0 handflying was done at the end of the flight. Total handflying for a 12 hour flight was one minute.

A US simulator Instructor, (ex training Capt UAL, ex US Navy, jet fighters) was hired by Asiana. He claimed Korean ex military pilots cannot handfly, and are fast-tracked to Capt. position. Civilian pilots trained in US and Aust. were much better. (No civil flying is allowed in Korea). A Korean National, trained by US military as a transport pilot was superb. This Instructor failed 2-5 pilots every year. The last one he failed was the Route Check Capt!!The Capt later was passed. The Instructor was fired for that. He claims many expat instructors were fired by both KAL and Asiana for insisting on the very Standards that were set by the 2 airlines!

Guam 747 crash. Korean. Handflown approach, ILS was off, (as was the case in SF). The copilot shouted… too low, too low, the ground proximity alarm sounded, BUT there was no response from Capt handflying. They hit high ground, (in cloud). The Copilot F A I L E D to hit full power and take control. ALL died!

One more problem is see. In the ‘old days’ we had round dial speed indicators. Modern ‘glass cockpits’ have digital readouts and verical bars, hardly noticeable in a chaotic situation, such as during the final 500 feet for Asiana.

This reminds me of the Air France Airbus crash out of Brazil. Because the elevator/aeleron control on an Airbus is no longer a big interconnected ‘yoke’ right in front of both pilots, but are two little NOT-interconnected side handles, the pilot not flying the plane failed to notice that the pilot flying the plane had the handle right back, nose-up position, even when the plane HAD stalled, due to incompetence/fear/panic/ etc. With a yoke, the non flying pilot would either have shoved the yolk forward immediately the stall warning sounded, and taken command, or told the other pilot to do so. Innovation not always a good thing, eh?

Jul 15, 2013 9:50pm EDT  --  Report as abuse
This discussion is now closed. We welcome comments on our articles for a limited period after their publication.

Pictures