It’s become widely accepted in aviation that nearly every accident can be traced not to one single cause, but rather a chain of threats and errors that all contribute to the final outcome.
On October 10, 2009, N108GF—a Piaggio P.180 Avanti twin turboprop—was destroyed during an attempted emergency landing on the Greenland ice cap. It’s difficult to imagine a more illustrative case of how accumulated errors during flight stack up to ultimately paint the pilot into a corner from which recovery is impossible.
Even at long-range power settings, authorities calculated that the Piaggio would have only 40 minutes of reserve fuel after a diversion to the alternate at Kangerlussuaq. But if the pilot flew the approach into Narsarsuaq correctly, he wouldn’t have had to divert.
In flight planning, errors in preparedness and execution—as well as fundamental knowledge of the accident aircraft’s performance and systems—were made that, cumulatively, overwhelmed the pilot. Because these errors often are made (albeit in isolation) by pilots conducting a North Atlantic crossing for the first time, a careful digestion of this accident is extremely valuable for turbine pilots interested in performing a transatlantic flight.
The aircraft operator was located in Kuwait, using two P.180 aircraft in a charter operation. The accident flight was operated under Part 91 as a ferry flight from Kuwait to Texas, where the aircraft was scheduled for maintenance. Because the aircraft did not have approvals for flight in reduced vertical separation minimum (RVSM) or minimum navigation performance specifications (MNPS) airspace, which begin at FL290, a flight plan was filed at FL280 for the leg between Keflavik, Iceland (BIKF), and the next planned fuel stop in Narsarsuaq, Greenland (BGBW).
In the flight planning, the first glimmer of trouble appears. The Danish Accident Investigation Board (AIB) calculated that had the flight been flown at long range cruise (LRC) power, the optimal power setting for maximizing fuel reserves, flying at FL280 to Narsarsuaq, then diversion to the filed alternate of Kangerlussuaq (BGSF, required by Narsarsuaq’s forecast weather) would leave only 40 minutes of fuel reserves. This falls short of the 45-minute reserve legally required under FAR 91.
Threat number one, and error number one, occurred when the pilot accepted this flight as planned.
The fuel situation quickly worsened. Once airborne, the pilot was assigned FL200. At FL200 and LRC power, the aircraft would now have only 18 minutes of fuel reserves. However, the pilot never set LRC power, instead flying the leg to Narsarsuaq at “recommended cruise” power—a higher and less efficient setting. At this power setting, the pilot would have an 11-minute reserve at FL270 and, prophetically, exactly zero reserve at FL200.
Threat number two, and errors two and three, were in both continuing the flight at a lower altitude rather than turning back, and in not setting power to optimize fuel efficiency with reserves so tight.
Many pilots are not conversant in the subtleties of cruise power setting, or in utilizing the aircraft’s published performance data to optimize cruise performance (see “Mentor Matters: The Narsarsuaq Trap,” July 2014 Turbine Pilot). Pilots inexperienced in North Atlantic crossings also are caught unaware by the separation requirements where aircraft can be held significantly lower than filed for hundreds of miles.
This could have ended at Narsarsuaq with a successful approach and landing, and the zero fuel reserve situation would forever remain merely an uncomfortable memory of the pilot. Indeed, one wonders how many flights conducted in remote areas are completed quietly with inadequate fuel reserves, simply because a diversion to the alternate so rarely occurs. However, on the accident flight, the pivotal moment occurred with the attempted NDB-DME approach to Runway 7—Narsarsuaq’s only approach.
With the METAR reporting visibility of 6 miles in light rain and overcast ceilings at 3,000 feet, and with the NDB approach providing for a minimum descent altitude of 1,800 feet, completion of the approach and landing should not have been in question. However, the pilot began the inbound approach leg at 6,000 feet, much higher than the minimum segment altitude of 3,600 feet. The extra altitude, combined with a 15-knot tailwind on final, meant that the aircraft was still in the clouds when it crossed the missed approach point, and the pilot appropriately went missed.
The PIC reported to Narsarsuaq aerodrome flight information service (AFIS) that “he could not get into this place.” Even after being informed that the ceiling was 3,000 feet, he made the critical decision to divert to Kangerlussuaq, nearly 400 nm away, and stated he had just enough fuel to get there. At this point the pilot’s margins were razor thin—the AIB calculated that diversion from Narsarsuaq to Kangerlussuaq at FL280 required 721 pounds of fuel, and that actual fuel on board at the time of the missed approach was 780 pounds.
When he switched over to Sondrestrom Flight Information Center (FIC), the PIC requested direct Kangerlussuaq and FL280 or FL300. As Sondrestrom only controls the airspace at or below FL195, it cleared the flight to FL190, and stated it would contact Gander to coordinate permission for higher. At this point the PIC recognized the critical fuel situation and stated that if he couldn’t fly higher than FL190 he wouldn’t make Kangerlussuaq. Yet when asked if he wanted to declare an emergency to get higher, the PIC declined to do so.
Error number four, not immediately declaring an emergency and climbing to the altitude that would give the aircraft the best fuel efficiency en route to Kangerlussuaq.
It is inexplicable that the pilot knew flight at FL190 would result in engine flameout before Kangerlussuaq, yet accepted FL190 and initially leveled off there. Further, he stated the aircraft was picking up “a lot” of ice at FL190 before he finally took proper emergency authority and initiated a climb without clearance to FL320.
This error likely reflects another common knowledge gap—understanding how crippling the separation standards over the North Atlantic are to ATC’s ability to accommodate rerouted traffic or climbs above the low 20s. In the case of the accident aircraft, Gander was unable to offer clearance to FL300 because of a flight of F–16s at FL270 that were later estimated to be between 50 and 100 nm from the Piaggio. In this case, climbing to FL300 would pose no danger of separation loss, but couldn’t be offered by ATC to a non-emergency aircraft.
Even then, the pilot should have barely been able to reach Kangerlussuaq, once proper action of climbing to FL320 was taken; leveling at FL320 he reported predicted landing fuel at Kangerlussuaq of 61 pounds. The pilot was again operating at normal cruise, rather than LRC, and presumably would have had greater reserve had the diversion been performed at LRC.
Next, an even safer course of action was suggested by ATC but rejected by the PIC: diversion to Nuuk (BGGH). Nuuk features a 3,117-foot runway and lies along the route between Narsarsuaq and Kangerlussuaq; it was more than 100 nm closer than Kangerlussuaq. A LOC DME approach serves Runway 23, which was favored by the winds at the time of accident. With ceilings of 600 feet and an MDA of 350 feet, the pilot would have been able to successfully complete an approach to Nuuk, and the aircraft should have been able to stop on the available runway.
Unfortunately, the operator had chosen a flight management system (FMS) navigation database option that only included airports with runway lengths of 4,000 feet or greater. There’s a valuable lesson to be learned here, particularly on flights to remote areas—be able to find and navigate to any airport at which you could potentially land. I frequently find pilots have set up their nearest airport filter to an arbitrary number that jibes with the low end of their normal operating practice; e.g., 4,000 feet.
Similarly, for a trip across remote areas such as Greenland, even something as seemingly straightforward as the runway surface filter in some flight decks requires consideration. Many light jets have an AFM restriction that landings are only allowed on paved surfaces, so many owners filter out unpaved airports from the nearest airport return. But consider that lying just before the halfway point and about 150 nm offset from the great circle line between Narsarsuaq and BIKF is Kulusuk (BGKK), a 3,934-foot gravel runway.
Yes, it may be gravel with only an NBD approach that has 800-foot minimums. But given that there’s a small village nearby, it’s vastly preferable to the ice cap or ocean if an emergency landing must be made. If a pilot has a hard-surface-only filter applied in the avionics, he or she will not be presented with Kulusuk as a diversion possibility.
The pilot’s final error was the one that doomed the flight to an off-airport landing: He elected to shut down the left engine and start a slow, continuous descent to BGSF in a misguided attempt to reduce fuel consumption. By doing so, he reduced the aircraft’s per-mile efficiency enough that reaching Kangerlussuaq was no longer possible. This displayed a fundamental lack of knowledge of both the aircraft’s performance (flying on two engines would have been more efficient), as well as basic high-altitude aerodynamics (staying high as long as possible, followed by an aggressive descent at idle, will use less fuel than a continuous descent from farther out).
Realizing that he no longer has enough fuel to make Kangerlussuaq, the pilot asked ATC for vectors to the ocean, planning on ditching. But by this point the ocean was even farther away than Kangerlussuaq—a statement of the pilot’s situational awareness in the final minutes of the flight—and a forced landing was the only course left.
Approximately 60 nm from Kanger-lussuaq, with less than 100 pounds of fuel remaining on board, N108GF made an approach and landing on the ice cap with the left engine stopped and the right engine producing power. After contact with the terrain, the aircraft separated into several pieces, with the cockpit and cabin coming away from the wings and tail.
The PIC survived the landing with only minor injuries and a search and rescue helicopter was able to quickly locate the wreckage and pick up the PIC, despite lowering weather on the ice cap and the fact that the PIC had switched off the transponder and ELT before landing. There was no arctic survival equipment on board the aircraft; the pilot had only his own clothing to protect him from the below-freezing conditions on the ice cap.
Realizing that all of the errors made by the pilot likely occur without consequence every day, it’s remarkable how tolerant of mistakes aviation actually can be.
Related Articles
