Last Saturday, Asiana 214, a 777 flying from Seoul, South Korea clipped the sea wall near the beginning of runway 28R at San Francisco International. The main landing gear was torn off, and the rear section of the aircraft hit the runway with enough force to remove the entire tail assembly from the fuselage and rip open the back wall of the cabin, ejecting several passengers from the aircraft. The plane then skidded on its belly for about 2500 feet, coming to rest in the grass on the side of the runway. Given the severity of the impact and the intensity of the fire that quickly consumed most of the cabin, it is remarkable that only two people were killed – a testament to the heroic efforts of the cabin crew, who managed to evacuate about 300 people, some of them severely injured, in only two or three minutes. (And a good reminder that flight attendants are actually very well trained in emergency procedures and are not just there to serve you food.)
How did a state-of-the-art aircraft operated by a highly trained crew crash at a major airport in near-perfect weather conditions? While we’ll have to wait for the full National Transportation Safety Board (NTSB) report for the full story, there are some thing that we know at this point.
At major airports like San Francisco International (SFO), many approaches are flown with guidance from an instrument landing system (ILS). The ILS emits a radio beam from near the touchdown point of the runway, directed upwards along the optimal approach path for a landing aircraft. All modern airplanes have sensors which detect the horizontal and vertical deviation from the center of the ILS beam – and thus the deviation from this optimal path. With this information, the pilot – or the autopilot – is able to fly the entire approach using only the instruments, without any visual cues from outside. This is how aircraft land in clouds or poor visibility.
While automated and guided approaches are common in airliners, all pilots are trained from day one to fly and land visually, using only basic instruments showing airspeed, altitude, and vertical speed, as well as their view out of the front window. On Saturday, the ILS at SFO was out of service and so Asiana 214 was directed to fly a visual approach over the bay to runway 28R. The sky was clear, the winds were calm, and the visibility was greater than 10 miles. The approach should have been easy.
The pilots descended, reduced speed, and configured the aircraft for landing normally and without any problems until the last 30 seconds or so of the flight. The target landing speed for the 777 was about 137 knots – and it is essential that the speed not deviate more than a few knots from this target. Too fast, and you’ll damage the landing gear or overshoot the runway; too slow, and you’re at risk of stalling the wings. An early analysis of the flight data recorder shows that Asiana 214 reached a minimum speed of 103 knots, and crashed less than 2 seconds later. At that speed, the aircraft was just marginally above a stall.
For most of the flight, the engines were controlled by a part of the auto-flight system called the auto-throttle, which adjusts engine power in order to maintain a commanded speed. The pilots likely had the auto-throttle engaged during the approach as well – this is normal, as it reduces pilot workload by removing the need to constantly adjust and monitor engine power.
However, the flight data recorder also shows that at the time of the crash, the auto-throttle was not engaged, but merely “armed” – meaning it was not controlling the plane’s airspeed. It is possible that the pilots thought the auto-throttle was functioning, either due to poor communication with one another, poor situational awareness, or a misunderstanding of the auto-flight system – or some combination of the three.
But one way or another, the pilots failed to monitor the basic flight parameters during the approach. They let the airspeed get too low and the decent rate get too high, and they didn’t notice until it was too late. There are a lot of potential factors in this accident, but the three major ones seem to be the pilots’ understanding of the aircraft’s automated systems, the pilots’ degree of familiarity and practice flying the aircraft manually, and finally – and potentially most importantly – the communication procedures inside the cockpit. There are multiple pilots so that if one makes an error – like letting the plane get too slow – the other will notice and point it out. That didn’t happen here. We need to find out why.
Now, this is the first fatal airplane crash in the US in over four and a half years – a pretty good record. The aviation community is very good at learning from its mistakes, and no doubt the investigation into Asiana 214 will produce concrete recommendations for changes in aircraft design and/or pilot training. Disturbing as this crash may be, it is an exceedingly rare event – just remember that flying in an airplane is still significantly safer than walking.
Photo courtesy of the National Transportation Safety Board/FLICKR