We got into an off-topic discussion in the comments section of a previous post recently about the latest in the Gol/ExcelAire midair collision in Brazil, and I thought it was worth a closer look. Fortunately, I was able to turn to a pilot friend of mine for an excellent piece about the accident. I know it’s lengthy, but I highly recommend reading the whole way through.
With the release of investigative reports by the Brazilian Air Force and U.S. NTSB, as well as the publication of an article in January’s Vanity Fair magazine by William Langewiesche, there has been renewed interest in the September 2006 collision of a GOL Airlines 737 and an ExcelAire Legacy business jet. The tragic accident claimed the lives of all 154 passengers and crew aboard the 737; the Legacy jet landed safely at a military airport in the Amazon. The on-line version of the Vanity Fair article provides links to the entire recordings of the both aircraft’s cockpit voice recorders (CVRs), which Langewiesche, himself a private pilot, uses to analyze the accident.
As an airline pilot whose every word is recorded at work, I am deeply troubled by Vanity Fair’s decision to publish the entire actual audio recordings of both CVRs. The sole purpose of the CVR is to aid in the investigation of an accident, not to serve as entertainment or to help sell magazines. The audio recordings are protected from release by U.S. and international laws, with only the portions relevant to the accident transcribed in official NTSB reports. It is therefore unconscionable that the entire 30 minutes of the Gol recording and all 2 hours of the Legacy recording now appear on Vanity Fair’s website. What possible value other than entertainment is provided by having subscribers listen to the 737 pilots chat about their friends and family for 30 minutes before their deaths? It is within this context that pilot opposition to cockpit video cameras must be understood.
Releasing the CVR recordings to Langewiesche also allows him to investigate and dramatize the accident, despite the fact that his private pilot certificate is completely inadequate to qualify him as an accident investigator. He makes a number of small technical errors, including a statement that autopilot use at 37,000’ is required by law (autopilots should be used in Reduced Vertical Separation Minimum (RVSM) airspace but may be disengaged any time), and that transponder failures are routine (failures are common in Cessnas but not in modern aircraft operating above 30,000’). But more troubling is his dramatization of the events in the Legacy cockpit and his portrayal of the thoughts and character of the Legacy pilots, none of which is warranted by the recordings themselves. This is important in light of the fact that far more people will read the Vanity Fair article than will ever come in contact with the official NTSB or Brazilian Air Force Reports.
I therefore hesitated to provide my own thoughts based on the recordings and article, but ultimately thought that the perspective of someone who has flown often in South America could fill in some of the holes in the article and answer some of the questions it raises. For example, the article correctly concludes that this accident could have only happened with the extremely accurate navigation in use on modern aircraft, and is an example of the unintended and unexpected consequences of improvements in technology. But the article didn’t mention the procedural changes that had already begun before this accident to address the problem and are now in wide spread use – namely SLOP (strategic lateral offset procedure).
SLOP is now in use by airlines in all oceanic and non-radar environments, and also by many crews over South America, where radar coverage can be spotty. Essentially, SLOP is a 1-mile offset to the right. The Legacy pilots seemed largely clueless about international procedures, so it’s unlikely they would have been SLOP-ing, but if the GOL airplane had been doing it they would have missed the Legacy by a mile. Unfortunately, no one was using SLOP over South America in 2006.
Still, the majority of the responsibility for the accident clearly rests with air traffic control (ATC). The interactions between the Legacy aircraft and ATC are very representative of my experience dealing with controllers in South America, an experience that is far inferior to the service provided by ATC in the rest of the developed world. It’s not just a language problem, although that certainly plays a factor. But Chinese and Russian controllers are often just as difficult to understand and provide far better service. Language issues did not prevent the Brazilian controller from issuing a clearance to 36,000’ at Brasilia, from realizing that the Legacy’s transponder had stopped working, or for allowing a system to be in use that showed a filed altitude on a radar display instead of the actual altitude. Regardless of language, the controller should have realized that the two airplanes were in conflict, and could have at least moved the GOL airplane.
Flying over Brazil is much like flying over an ocean, although in many ways it’s more dangerous. Communication is difficult, as shown by the inability of the Legacy crew to be able to communicate with an ATC facility on the emergency frequency (121.5) following the collision. Radar coverage is inconsistent, and in fact the Legacy flew out of radar contact for 50 minutes without having the benefit of non-radar separation standards or procedures. At least over the ocean, altitude separation for the entire oceanic portion would have been provided prior to entering the airspace.
Much has been made of the fact that the Legacy pilots should have known that they were at the wrong altitude; Langewiesche equates it to “drive-on-the-right highway rules”. The reality in international flying is unfortunately a little more complex. While there are direction-of-flight altitude rules under radar contact, these rules largely don’t exist over oceanic airspace. There are flights everyday flying eastbound to Europe at 36,000’ and 37,000’, as well as westbound to Asia at 38,000’ and 39,000’. I’m almost embarrassed to admit that I have to stop and think which altitude is appropriate for an eastbound flight across the United States; it is not instinctive. I seriously doubt that I would have noticed the aircraft was at the wrong altitude following a slight left turn at Brasilia and a course somewhat to the west of north. Without a new clearance from ATC, we would have likely stayed at 37,000’ as well.
Assuming the transponder was turned off (as opposed to failing) the level of warning for the Traffic Alert and Collisions Avoidance System (TCAS) being off is fairly typical across aircraft. Turning the transponder off is normal system function and is done on every flight after landing. Warnings are generally not given for normal system functions – i.e, turning a Pack off will also generally give a white status message without an aural warning. On the 777, turning the transponder off would result in a white TCAS-Off message on the upper Engine Indicating and Crew Alerting System (EICAS) screen. It seems the Legacy warnings were even less apparent – a message on the Flight Management System (FMS) and on the Primary Flight Display (PFD) are really easy to overlook.
Having said that, it’s probably very appropriate to upgrade the TCAS-Off message in flight to a higher-level warning. On a Boeing, it should most likely be at the Master Caution level, resulting in an aural warning and illumination of lights in front of both pilots if the TCAS fails or is turned off. Had this occurred on the Legacy aircraft, it is extremely unlikely the accident ever would have happened.
Despite the majority of the responsibility lying with Brazilian ATC, the Legacy crew cannot escape without criticism. It’s very apparent that they were in over their heads on a flight from Brazil to the United States. Not to be too condescending, but this was a regional jet crew who really had no business operating in this environment. The problem is that they didn’t even realize it. I have a lot of sympathy for the feelings of confusion over language and procedures internationally – I’ve felt it myself a lot in South America and Asia. The difference is that at my airline the other 2 or 3 crewmembers are usually very experienced in the area, or we at least compensate by paying a lot more attention. These guys were treating this like a flight from Chicago to Dallas, and weren’t doing any of the things we normally do internationally (plotting, monitoring position, etc.). After the impact, it’s apparent they really didn’t even know where they were.
Their cockpit resource management after the impact is poor. That’s another area where I see major differences between an airline crew and a couple of corporate pilots. They didn’t do any of the things we would normally do following an emergency – one pilot will normally fly and handle all of the radio communications with ATC, while the other pilot will work on the problem and utilize any other resources.
The captain was essentially useless. The copilot rightly took over to a certain extent, but then he tried to do everything and didn’t manage the cockpit well. They also spent far too much time trying to establish communication with ATC and not nearly enough time flying the aircraft or utilizing any of the other resources available to them.
If I had been the captain (in the perfect world of hindsight and sitting at a computer), I would have had the First Officer fly and talk to ATC/relay via Polar on 121.5. I would have first evaluated the condition of the airplane and solicited help from the back. With the airplane intact and still flying, it may not have been necessary to initiate an emergency descent until a plan was in place. I would have used the SATCOM (they had a satellite phone they apparently never even used) to get Dispatch on the line to assist with determining the nearest suitable airport and with frequencies for ATC. They wouldn’t have had our Dispatch capabilities, but their Operations should have been at least able to help.
With all of that more efficiently accomplished, they should have had time for a bit more of an investigation into how the aircraft was flying. I would have incrementally extended the flaps and attempted to gauge some controllability; they would have been better served by slowing for landing. Langewiesche incorrectly states that they couldn’t slow down; the truth is that they never attempted to. They seemed to really have had no idea what they were doing or why they were doing it, and I can’t say I’m very impressed with how they handled the situation following the impact.
None of that performance changes the fact that Brazilian ATC put them on a collision course with another aircraft, or failed to realize their error despite numerous opportunities. The Legacy pilots’ performance also doesn’t rise to the level of criminal negligence; the fact that these pilots face trial in Brazil is deplorable. With the reliability of TCAS and the introduction of SLOP, it’s not an accident likely to ever be repeated, but then neither is any airline accident that occurs in the modern era. I can only hope that the accident will spur some much needed improvements in Brazilian Air Traffic Control, and that next time the CVR recordings will stay where they belong – in the hands of the investigators.