Into The Black Void: The View Outside Provides No Information
There comes a time in every pilot’s life when upon rotating the airplane for takeoff, all that can be seen is total darkness. The airplane is accelerating, the pilot’s senses are signaling furiously, the brain insisting a need for prompt, orderly tasks to be done. Yet the view out the windscreen provides no information whatsoever — utterly nothing.
The visual movement of ground and cloud that usually helps to complete the climb feedback loop is just missing. At that point, it’s all too easy to lose orientation and, in a matter of moments, lose control of the airplane. That is what happened to the crew of a Shorts SD3-60 departing the Dutch side of the island of St. Maarten (also known as St. Martin) on Oct. 29, 2014.
The aircraft plunged into the Caribbean Sea less than a mile off the end of Runway 28 at Princess Juliana Airport (TNCM). The two pilots aboard, a 49-year-old captain and his 26-year-old first officer (F/O), were killed, and the airplane wreckage sank 65 ft. to the bottom of the sea. The flight, SKZ 7101, was operated by SkyWay Enterprises (SWE) on an FAR Part 135 cargo flight from Princess Juliana Airport to Luis Munoz Marin International Airport (TJSJ), San Juan, Puerto Rico. The airplane had taken off at night and in rain.
The captain had observed the cargo loading and provided a cargo manifest to the ramp agent. The flight started engines at 1817 (local time), taxied at 1828 and was cleared for takeoff at 1838. The tower ATC clearance was to maintain heading 230 deg. after takeoff until passing 4,000 ft. Airport security video recordings showed normal airplane lighting patterns as the SD3-60 rolled down the runway. At 1839, the tower gave the crew their takeoff time and revised their clearance to maintain heading 230 until passing 3,000 ft.
At 1840, the tower controller observed the airplane descending visually and on radar; he called the flight on tower and guard frequencies, but received no response. After the data block disappeared from radar, the controller immediately notified emergency services, and at 1900, a Coast Guard vessel was dispatched to the scene of the accident. At 2125, the Coast Guard Search and Rescue team notified the tower that airplane debris had been found offshore.
A preliminary playback of the local ATC radar data showed the airplane had reached a Mode C readout altitude of 200 ft. However, data from a handheld GPS device later found in the submerged wreckage showed the airplane had reached a maximum GPS altitude of 433 ft. and 119-kt. groundspeed at 1839:30.
The official weather observation at the time of takeoff showed the winds were from 230 deg. at 10 kt., gusting to 20 kt., varying from 200 to 270 deg. There were light showers and rain, the ceiling was a broken deck at 1,300 ft., and there were towering cumulus clouds in all quadrants.
Cargo loading personnel said when interviewed that there were light to moderate rain showers during loading. One ramp agent stated, “When the airplane was ready to taxi, there was some heavy rain and it was pitch black to the south with clouds. The aircraft taxied to the east; the rain eased a little. The aircraft held to the east waiting for another aircraft to clear and then taxied on the runway where I lost sight of him. The next time I saw the aircraft is when he took off in some heavy rain.” Two other agents confirmed the existing weather at departure and stated, “There was heavy rain that eased off and afterward started heavy again.”
Princess Juliana Airport is located on a strip of land that separates Simpson Bay Lagoon from the mainland, and both runway ends are bounded by water. A low mountain range runs through the center of the island. There is one runway (10/28), and because of prevailing winds from the east, over 90% of takeoffs and landings are on Runway 10. VMC conditions must exist to land on Runway 28 and all night landings must use Runway 10. There is no visual landmass past the shoreline when taking off from Runway 28.
Visibility at the time of takeoff was 4,000 meters, which was less than published takeoff minimums of 300-ft. ceiling and 4,500 meters of visibility for Runway 28. However, SkyWay FAA Operations Specifications allowed a lower takeoff minimum based on the lowest authorized straight-in Category 1 IFR landing minimums.
The wreckage was found 0.8 nm from the far end of Runway 28 about 35 deg. to the left of runway centerline. A Puerto Rico National Police diving team obtained underwater photos and videos of the wreckage. The aircraft had broken up and fragments were scattered about the sea floor. Components of the nose, left and right wings and tail surfaces were found, confirming the location of all four corners of the airplane. The main landing gear hydraulic actuators were in the full up position and the flaps appeared to be fully raised as well. The fragmented wreckage was not consistent with an attempted ditching.
The St. Maarten Civil Aviation Authority conducted an investigation in accordance with ICAO Annex 13. It was assisted by investigators from the U.S. NTSB and FAA, representatives of the U.K.’s Aircraft Accident Investigative Branch and Shorts Brothers plc, and from Canada’s Transportation Safety Board, Transport Canada and Pratt & Whitney Canada (P&WC). The operator, SkyWay Enterprises, declined to participate on investigative groups but cooperated in the investigation.
SkyWay Enterprises Inc. (SWE) was originally certificated as an air carrier in 1979 in Detroit. Operations were moved to Kissimmee, Florida, in 1990, where the company is now based. It was authorized by is operations specifications to serve the U.S., Canada, Mexico, Central America and the Caribbean Sea, including the islands/nations and the Havana FIR. SkyWay’s primary business plan was to provide cargo service within the Caribbean from operating bases in Miami (KMIA) and Puerto Rico (TJSJ) and Rafael Hernandez Airport (TJBQ) in Aguadilla.
SWE obtained the accident airplane from American Eagle Airlines in 2000 and converted the interior to a cargo configuration. At the time of the accident, the airplane had 25,061 flight hours and 32,824 cycles. It was not equipped with a flight data recorder (FDR) or cockpit voice recorder (CVR). SkyWay Enterprises was authorized to operate under Part 135, Part 119.21 (a) (5) On Demand 135 for Cargo only. As such, the FAA airworthiness requirements for passenger aircraft did not apply, and as a result the recorders, GPWS, radar altimeter, one attitude gyro and TCAS were removed. The autopilot was also removed.
The accident captain had been employed by SWE for about four months and had been assigned to the San Juan operation for three weeks. He had previously flown the SD3-60 and the Britten Norman BN-2 Islander in the Caribbean area for other operators. According to his fiance, he had moved to Puerto Rico from Germany to fly. He had an ATP certificate, Airplane Multiengine Land, with type ratings in the SD3, Cessna 500, British Aerospace 3100 Jetstream, Hawker 125 and Learjet — the last three for second in command (SIC) only. He had completed a Part 61.157 type rating check ride at FlightSafety International on June 9, 2014. Based on SWE and FAA records he had 5,318 total flight hours, 361.8 hr. in the SD3 and was current in the airplane, with 50.3 hr. in the previous 30 days and 213.8 hr. in the last 12 months. He had flown from TJSJ to TNCM and back on the two days before the accident and had gotten normal rest at home between trips.
The first officer had been employed for about 13 months. He previously had flown light aircraft for the Civil Air Patrol and had received SWE in-house SIC training. He had a Commercial Pilot certificate, Airplane Single and Multiengine Land, Instrument Airplane, and an SD3 type rating with SIC privileges only. He had 1,040.9 total flight hours, 510.9 hr. in the SD3, and he was current, with 32 hr. in the previous 30 days and 456 hr. in the previous 12 months. He resided at his home in Puerto Rico and had flown the same trips as the captain in the two days before the accident. He had received normal rest in the previous three nights.
The wreckage was recovered from the sea and examined by technical representatives. The fractured ends of both the left and right outer wing boxes showed marked ductile overload failure in downward bending. The damage to the wings was symmetric. X-ray imaging of the cockpit caution and warning light bulbs by the U.K. Royal Navy Lab did not show any filament deformation, indicating there were no systems warnings. GPS data from a handheld battery-operated unit found in the wreckage was downloaded by the NTSB Vehicle Recorder lab, and the unit provided date/time, latitude/longitude, altitude, groundspeed and true course at sequential time intervals. Examination of the flight control pushrods, cables and bell cranks did not reveal any fatigue fractures or mechanical dislocations, and the pushrod failures were ductile, consistent with a high energy impact with the sea.
Both engines were recovered and sent to Pratt & Whitney Canada Service Investigation Facilities in Canada for teardown, examination and lab testing. The damage on both engines was characteristic of propellers striking with a sudden stoppage and with engines producing power at the time of the strike. However, FAA observation of the overall poor engine condition resulted in action to rescind SWE’s authorized 12,000-hr. TBO and to set limitations to 6,000 hr. in accordance with the current P&WC Service Bulletin.
The bulk fuel supply and the supply from the truck that serviced the accident airplane were tested and met quality standards. The airplane carried 10 boxes of cargo for FedEx that weighed 435 lb. Weight and balance documents showed the airplane was well within limits.
Investigators met with and interviewed SWE officials at Kissimmee and FAA officials at the Orlando FSDO, where the principal operations inspector (POI) was based. The government agency officials said it was difficult to oversee Caribbean operations because of travel and budgetary limitations. The POI was not type rated on the SD3-60, had not observed the company’s training or performed any company en route observations, and was not aware of handheld GPS use in the cockpit.
In addition, the company had no formal safety reporting system and the cockpit resource management (CRM) training required by Part 135.330 was said to be “one on one” and most of it was on CD and video. The pilot training manual showed unusual attitude recoveries, wind shear/microburst, and lower than standard takeoff minimums were to be trained by briefing rather than flight training. Although the FAA rated SWE as “low risk,” the former FAA principal maintenance inspector described his relationship with company officials as “difficult.” He said any time he would make a proposal to the company he was challenged with “tell me why” or “show me the regulation.”
Investigators learned the accident crew had experienced a runway overrun in the same airplane at the same airport two days prior to the accident. Although the captain asserted he had experienced hydraulic failure, an inquiry by the director of operations determined that the captain had inadvertently shut off the engines by taking the fuel condition levers too far aft of the detent during landing.
The investigative authority found the cause of the accident was: “the PF [pilot flying] experienced a loss of control while initiating a turn to the required departure heading after takeoff. Flap retraction and its associated acceleration combined to set in motion a somatogravic illusion for the PF. The PF’s reaction to pitch down while initiating a turn led to an extreme unusual attitude and the subsequent crash. PM [pilot monitoring] awareness to the imminent loss of control and any attempt to intervene could not be determined. Crew resource management (CRM) performance was insufficient to avoid the crash.” Contributing factors were “environmental conditions including departure from an unfamiliar runway with loss of visual references [black hole], night and rain with gusting winds.”
The investigation also expressed concern about the company’s safety culture and the extent and effectiveness of FAA oversight. Management did not provide any method to communicate safety issues or a voluntary safety reporting system. By removing safety equipment required for passenger airlines, the company obtained weight reduction and cost savings but exposed crews to fatigue and errors in a high air traffic environment. No flight data monitoring program was possible and without a TCAS, inflight collision risk increased. Manufacturers’ Safety Bulletins were not complied with unless they were FAA mandated. There were no provisions for crew rest facilities during crew wait times.
The report stated “Training of the next generation of young pilots and maintenance engineers is seriously challenged in this environment. It can be a breeding ground for the learning of bad habits. As aviation activity and complexity continues to grow, understanding and managing these challenges and developing a more proactive safety culture encompassing modern SMS concepts will become the imperative for all Part 135 operators.”
The investigation found that the FAA appeared to have difficulty accomplishing its surveillance responsibilities. Rather, its priorities went toward passenger-carrying operations. Risks associated with cargo operations were not given needed resources.
The St. Maarten Civil Aviation Authority made two recommendations to the FAA as a result of the accident. The first was to evaluate the facts, analysis and conclusions of the final report of this loss of control accident and similar cases of CRM breakdown during loss of control. The second was to, within one year, publish a Notice of Proposed Rulemaking (NPRM) to extend the current Part 5 SMS rule to all Part 135 operators.
Night takeoffs into turbulent, overcast conditions can be difficult even for experienced pilots. A somatogravic illusion results from rapid acceleration, causing the pilot to sense the airplane is pitching up more than it really is. Turns immediately after takeoff and accelerations associated with changes in configuration can compound the unusual sensations that can lead a pilot to accept false cues from his inner ear. Two important countermeasures to these sensations are a disciplined instrument crosscheck and another qualified pilot monitoring the actions of the pilot flying and calling out excessive pitch and bank, under- and over-speeds, and failure to reconfigure at the appropriate time. A radar altimeter and a GPWS also would have been extremely useful in this situation if they had been installed.
The F/O had not reported the captain’s misadventure shutting off the fuel levers on landing two days previously, very likely due to differences in age, experience and command authority, and a lack of a reporting system. How likely was he to correct the captain in a demanding flight situation? If the F/O was flying, how much skill and experience did he have to cope with an illusion? Without recorders, we’ll never know for sure.
One thing seems certain, though. When you keep reducing the safety margins, eventually you will have an accident.
Author’s Note: While at the NTSB, I participated in this investigation as the technical advisor for operations to the accredited representative from the U.S.