Air Canada Flight 759 Landing Accident at SFO
Technology Does Not Replace Human Awareness
There is little doubt that the use of big data to establish preventative solutions, but as recently noted, reliance on the numbers is not enough. Like the well documented phenomenon of automation removing/diminishing cockpit awareness, aviation’s dependency on numbers should not supplant human vigilance.
While it would be premature to ascribe either of the points to be mentioned as a “probable” cause for the Air Canada 759 accident, reports about the loss of the CVR and the omission of an enhancement to the ADSE-X equipment are valuable reminders that all aviation professionals cannot assume that all of the technology and data systems are replacements for human awareness.
As Forbes’ Christine Negroni reported, the relevant segment of the Cockpit Voice Recorder for the flight was taped over as no one took the time to preserve the recording. The specific aircraft continued in service for such a long time that the closed loop recording captured subsequent cockpit conversations and thus over recorded the AC 759 pilots’ dialogue on approach to SFO. The NTSB noted in an update:
The incident airplane’s cockpit voice recorder had been overwritten, so NTSB investigators did not have that data.
The author of the article asks whether Air Canada should have reported the “incident” to the Board earlier. It appears that the FAA was the first to identify the A-320’s proximity to the taxiway.
According to one report, the NTSB believes that the absence of this tape will not hinder its investigation
Whoever may be responsible, it is clear that multiple persons had knowledge of the breach of the separation between the AC A-320 and the planes on the taxiway and the first notice that this flight segment was an incident was too late. Proper vigilance by any one of those responsible individuals would have, should have, captured this useful data. One could suppose that the failure to act expeditiously was cause by expectations that the machines will, as always, preserve everything.
Lesson #1: Attitude Adjustment
As reported by Alan Levin of Bloomberg, the NTSB in 2011 recommended a software upgrade to ground radar systems that would warn when a plane is landing in the wrong place. But the FAA dismissed the recommendation, declining to even study whether it was feasible, according to government records.
In an announcement issued since the San Francisco near-collision, the FAA says it has begun over the past year doing what the safety board recommended and testing could begin in a few months.
The 2009 Delta Air Lines Inc. incident at Atlanta involved the pilots landing on a taxiway in Atlanta. The fact that no one was injured may have lessened the FAA’s focus on this risk. The NTSB found that pilot fatigue was a contributing factor and that existing radar systems at major airports could be enhances to warn controllers if a landing plane was headed to a taxiway instead of a runway.
Captain Randolph Babbitt, the FAA’s Administrator in 2011, reported that the agency’s experts were of the opinion “that the ASDE-X system wasn’t up to the task,” according to NTSB records of correspondence in the case. The ASDE-X’s are installed to warn controllers when there’s a risk of a plane colliding with another aircraft or vehicle on a runway.
Given the low incidence of these incidents, the need to introduce these systems appears to have, under the SMS risk management matrix, a HIGHLY SEVERE consequence, but a LOW PROBABILITY.
The severity of an aircraft landing on other aircraft was not a factor in the 2008 incident, but by expanding the scope of the risk to the SFO-type situation, a low risk priority may have been increased?
A 2015 Alaska Air taxiway landing incident, this one involving an Alaska Air plane in Seattle in 2015, evidentially elevated the probability classification. The FAA escalated the development to add this feature to ASDE-X.
The good news is that technology has evolved in the intervening years— Saab Sensis Corp., a division of Sweden-based Saab AB, confirmed that its ASDE-X could be programmed to detect a potential errant landing as far as 0.75 mile. Too bad that this enhancement was not available to Administrator Babbitt.
Lesson #2: In setting probabilities for severe consequence cases, the risk assessment might be more generous than in other cells of the SMS matrix.