Too often we see an NTSBfinding that states: “The probable cause(s) of this accident [involving a highly experienced airman] is determined to be the pilot’s decision to initiate and continue flight into known adverse weather conditions, which resulted in spatial disorientation, a loss of airplane control and a subsequent inflight breakup.
Such is the case with a Key Lime Air Fairchild SA227 (Swearingen Metroliner) that broke up in cumulus activity over Camilla, Georgia, on Dec. 5, 2016, killing its pilot, the lone occupant on the FAR Part 135 IFR cargo flight.
The ATP airman was well-respected and highly experienced. He had accumulated 11,133 hr., with some 4,647 hr. in the make and model. He had flown 74 hr. in the previous 90 days and 29 hr. in the previous 30 days. From 2008 to 2016, company records revealed no unsatisfactory competency/proficiency checks.
The pilot’s primary flight assignment was to operate single-pilot cargo flights between Northwest Florida Beaches Airport (KECP) near Panama City and Southwest Georgia Regional Airport (KABY) near Albany. He was the only Key Lime Air pilot based at KECP. The flight was scheduled to depart every weekday, Monday through Friday, at 2130 EST. Upon arrival at KABY, the pilot would typically spend the night at the airport and return to flight duty at 0730 to complete the return flight.
The pilot’s most recent flight duty had ended on Dec. 3, 2016, at 0830. He had been off duty until he checked in with Key Lime Air Dispatch about 1 hr. before the accident flight. What follows comes from the safety board’s investigation into this crash.
About 1 hr. before the scheduled departure time of 2130, the pilot completed a routine check-in call with the Key Lime Air flight follower assigned to cargo flights, who was one of two flight followers working that night at the operator’s headquarters and dispatch office in Englewood, Colorado. The departure time and weather conditions were discussed. The pilot told the cargo flight follower that he was “holding on the ground” for convective activity that was “extreme” and had “tornado activity.” The pilot delayed the departure to continue to evaluate the weather conditions along his route.
At 2140, the customer (UPS) called the operator’s dispatch office, wanting to confirm that the flight (LYM308) would depart, noting if it did not depart soon, the cargo would “not make service.” About 2 min. later, the other flight follower, who was not assigned to the cargo flights, called the pilot, informing him that UPS had called the dispatch office to ask whether the flight was going to depart. According to this non-cargo flight follower, the pilot explained that he would be departing immediately and would try to fly a clear-weather corridor extending northeast toward KABY. The pilot added that if he could not get through the storms to his left, he would make Tallahassee International Airport (KTLH) his alternate. The flight departed about 12 min. later at 2154.
ATC voice communication transcripts revealed that, at 2215, the Jacksonville Center controller who was working the flight advised the pilot of a “ragged line of moderate, heavy and extreme precipitation” along his planned route of flight. The controller also stated, “I don’t show any breaks in the weather.” The controller then cleared the pilot to descend at his discretion from 7,000 ft. MSL to 3,000 ft. MSL.
Subsequently, the controller suggested a route of flight that would have required a diversion to the northeast for 70 nm to avoid the most severe weather. The pilot responded that he had enough fuel for such a diversion but that he would “see what the radar is painting” after completing the descent to 3,000 ft. MSL.
About 1 min., 30 sec. later, at 2218, during the airplane’s descent from 7,000 ft. to 3,000 ft., the controller stated, “I just lost you on radar. I don’t show a transponder; it might have to do with the weather.” Some 40 sec. later, the pilot advised the controller that he intended to deviate to the right of the course, and the controller told the pilot that he could turn left and right as needed. Shortly thereafter, the pilot stated, “we’re going to turn back around to Tallahassee.” The controller cleared the pilot direct to KTLH and instructed him to maintain 3,000 ft.
The pilot responded, “present position direct Tallahassee and we’ll try to maintain 3,000 here.” The air traffic controller then asked, “Do you want to climb back up? I can offer you any altitude.” The pilot responded, “We’ll see if we can get it up to about 3,000.” The air traffic controller then recommended a heading of 180 deg. to “get you clear of the weather quicker,” and the pilot responded with “All right, 180.” There were no further communications from the pilot.
At about 2220, radar data showed the airplane enter a right turn that continued through about 540 deg. before radar contact was lost at 2222:24. Throughout the final turn, the airplane’s reported altitude was near 3,500 ft. A study of the airplane’s radar track revealed that its calibrated airspeed varied between 198 and 130 kt., with estimated bank angles between 40 and 50 deg. The airplane’s calculated load factor for this radar-recorded portion of the flight was about 1.3 G.
The wreckage was located about 3.4 mi. east-southeast of Camilla, Georgia, and was scattered over a large area that included a cotton field and dense forest. The debris field was about 2,640 ft. long and 1,500 ft. wide and oriented on a heading of 049 deg. true. The wings outboard of the engine nacelles, wing extensions, empennage, ailerons and cargo door separated from the airplane during the accident sequence and were located along the debris path leading to the main wreckage. The first components located along the debris field were the outboard sections of both wings, which exhibited damage and paint transfer consistent with contact with the fuselage. Additional components located along the debris path included the empennage and the mid-span portions of both wings. The fuselage was at the end of the debris path beside a residence. The fuselage, cockpit, cabin section, inboard wings and both engines were damaged by the post-crash fire.
The examination revealed that the left outboard wing separated from the airplane just outboard of the nacelle, and the left wingtip extension separated from the outboard wing. The leading edge was separated into several pieces. The left aileron was separated from the wing and recovered in three pieces.
Examination of the right wing revealed the right outboard wing separated from the airplane just outboard of the nacelle and the wing tip extension separated from the outboard wing. All the right outboard wing structure was recovered away from the main wreckage. The wing structure exhibited substantial twisting and crushing damage in a fore-aft direction. The right aileron was separated from the wing and only the inboard portion was recovered.
There was significant fire damage to the fuselage and the forward cargo door frame with some areas consumed by fire. The fuselage was on its right side and displayed lateral crushing damage. The cargo door separated during the accident sequence and was recovered away from the main wreckage. The flanges for the dorsal fin on the top of the fuselage were flattened to the right. The attach fittings for the pitch trim actuator remained installed on the upper fuselage and the actuator rod ends were installed in the fittings. The rod ends were fractured from the trim actuator. The trim actuator was not recovered.
Examination of the horizontal stabilizers revealed they separated from the airplane during the accident sequence. They were recovered away from the main wreckage. The left elevator was intact and installed on the horizontal stabilizer between the center and outboard hinges. The right horizontal stabilizer revealed it was mostly intact but damaged.
The empennage, vertical stabilizer and rudder were recovered in several pieces away from the main wreckage. The fuselage/empennage structure was mostly intact.
All the fracture surfaces examined had a dull, grainy appearance consistent with overstress separation. There was no evidence of pre-existing cracking noted at any of the separation points.
The type and degree of damage observed to the engines and propellers were consistent with both engines being under power and operating at the time of impact. No evidence of pre-existing conditions was found that would have prevented normal operation of either engine.
All the propeller blades from both assemblies were accounted for at the time of the examination. Both propeller assemblies had sudden failure damage as the result of impact. The examination found no evidence of any fatigue failure or pre-impact malfunction.
The Forensics Sciences Div. lf the Georgia Bureau of Investigation performed an autopsy on the pilot. The cause of death was multiple blunt force trauma. The FAA’s Bioaeronautical Sciences Research Laboratory, Oklahoma City, performed forensic toxicology on specimens from the pilot with positive results for dextromethorphan, a cough-suppressant commonly used in over-the-counter preparations.
The Weather
Weather at the accident site was night IFR. Albany — 14 mi. from the accident site — was reporting scattered clouds at 2,600 ft., visibility 8 mi. and wind 14 kt. out of the east in light drizzle.
The National Weather Service (NWS) surface analysis chart for 2200 depicted a low-pressure system over Louisiana located along a frontal wave with a cold front extending southward into the Gulf of Mexico and a stationary front extending eastward along the Gulf Coast into the Florida panhandle and southern Georgia, and then into the Atlantic Ocean. A high-pressure system was located over North Carolina. The stationary front was depicted over the flight route and near the accident site.
The station models in the immediate area surrounding the accident site showed a counterclockwise wind-flow pattern suggesting that a low-pressure area was developing along the stationary front. The surrounding stations also reported moderate to heavy rain and thunderstorms. The station model for Albany reported wind from the north about 10 kt., moderate rain, overcast cloud cover, and a temperature and dew point of 60F. To the south of the front over the Florida panhandle, southerly winds of 10 to 15 kt. were indicated with temperatures and dew points in the 70s.
The NWS national composite radar image for 2220 showed that the accident site was located along the leading edge of a line of convection with reflectivities ranging from 50 to 60 decibels (dBZ) immediately west of the site. The line extended from the Gulf of Mexico immediately west of Panama City, Florida, northeastward to the cities of Albany, Vidalia and Statesboro, Georgia, and then eastward through Savannah and into the Atlantic. Other, more scattered or less organized areas of echoes were located across northern Florida, south and east of Tallahassee, to the west of Jacksonville, and into southeastern Georgia. A narrow corridor clear of echoes extended from Panama City to Tallahassee to Moultrie, Georgia. This corridor was located immediately east of the accident site.
RELATED
The NWS Storm Prediction Center (SPC) graphic convective outlook issued at 2000 depicted where organized thunderstorms were expected to develop and the potential for severe thunderstorms during the period. The chart showed a slight risk of severe thunderstorms over extreme southeast Louisiana and Mississippi, southern Alabama, southwest Georgia and the Florida panhandle, which included the accident site. A marginal risk of thunderstorms surrounded the area and included southern Alabama and Georgia and northern Florida. The slight-risk area implied that an area of organized scattered severe storms was possible, with either short-lived and/or not widespread, isolated intense storms possible. It also implied that one or more tornadoes, reports of intense winds, and 1-in. and possibly 2-in. hail were expected within the designated area.
The automated special observation at KABY at 2221 included wind from 090 deg. at 13 kt.; visibility, 8 sm in rain; clouds, scattered at 2,600 ft. AGL and broken at 12,000 ft. AGL; temperature and dewpoint, 16C; and altimeter, 29.81 in. of mercury.
The automated special observation at KTLH at 2234 included wind from 190 deg. at 16 kt.; visibility, 10 sm in rain; clouds, broken at 800 ft. AGL and overcast at 1,200 ft. AGL; temperature, 24C; dewpoint, 23C; and altimeter, 29.93 in. of mercury.
The closest Weather Surveillance Radar was at NWS Tallahassee, about 50 mi. south of the accident site. Based on the radar scans between 1,960 to 7,000 ft. during the minutes of the flight prior to the accident, the accident flight tracked along the leading edge of a line of heavy-intensity echoes and was operating in light intensity precipitation. The next base reflectivity image for those elevations at 2228:12 showed rapidly developing echoes over the preceding flight track (which ended at 2222:24 in the vicinity of the accident site), indicating heavy-intensity precipitation.
The base reflectivity image at 2234:39 continued to show echoes increasing in intensity over the accident site, indicating heavy to extreme intensity. Several small bowing segments were also indicated to the southwest. The radar records for 2221 and 2228 showed echo tops near 30,000 to 35,000 ft. over the last 4 min. along the flight track and the accident site, with echo tops to 45,000 ft. immediately west of the accident site.
Safety Management
The operator did not have a formal Title 14 CFR Part 5 safety management system (SMS) implemented at the time of the accident nor was it required to have such a program. The operator’s technical programs director reported establishing a “system safety-based program” in 2012. As part of that program, the operator’s company operations manual (COM) required a flight risk assessment tool (FRAT) to be completed before every cargo flight. The FRAT was a worksheet that assigned numerical risk values to a variety of conditions that a flight might experience. The FRAT concluded with a total score that placed a flight in a “Go,” “Consult” or “Permission Needed” category.
According to the COM, a dispatcher or flight follower was required to complete the FRAT, and it was to be preserved for 30 days electronically. During post-accident interviews, the operator’s director of operations (DO) reported that a FRAT had not been completed for the accident flight, and in addition, he reported that there was no record that a FRAT had been completed for flight LYM308 in the past 30 days. The DO stated that the failure to complete FRATs for night cargo flights was a “management oversight” and that the management team was not aware that the FRATs were not being completed for night cargo flights.
During a post-accident interview, the flight follower who was assigned to the cargo operations the night of the accident stated that he did not complete the FRAT for the accident flight. He further stated, “At no point in my initial training or when I started did anyone, or any of my coworkers, or any of my bosses, or anyone in the company, tell me that we were responsible for doing FRATs for any cargo flight at all.”
According to the other flight follower on duty the night of the accident, the dispatch office only completed the FRATs for passenger flights, and, to his understanding, FRATs were not required for cargo flights.
Investigators completed a FRAT for the accident flight using the known risk conditions based upon the available evidence. The resultant score of 19 would have placed the flight in the “Go” category.
According to the COM, the DO “is authorized to exercise operational control in all aspects of Key Lime Air’s operations.” The COM further stated that “the pilot in command (PIC) is authorized to exercise operational control in all areas allowing the safe completion of each flight to which he/she is assigned.” The PIC’s areas of operational control included:
  • The PIC must obtain and check current and forecast weather for the applicable airports. The PIC will do all flight planning to each flight they are assigned.
  • The PIC must select an alternate airport if applicable for the intended flight.
  • The PIC will load the aircraft within its applicable CG limits and weight limitations.
  • The PIC will check to make sure the aircraft is in airworthy condition prior to flight.
According to the COM, a dispatcher had operational control over scheduling of crews and aircraft and for monitoring the progress of flights. A review of the operations specifications and COM found no requirement for flight followers to release cargo flights, as they did not have operational control over flights.
Analysis
Safety Board investigators spent months putting the whole story together. Here’s their analysis of the accident with emphasis on the go/no-go decisions made by the pilot.
Examination of the wreckage indicated that the airplane experienced an inflight breakup at relatively low altitude, consistent with radar data that showed the airplane’s last recorded altitudes to be around 3,500 ft. MSL. The symmetrical nature of the breakup, damage to the outboard wings, and damage to the upper fuselage were all signatures indicative that the left and right wings failed in positive overload almost simultaneously. There was no evidence of pre-existing cracking noted at any of the separation points, nor was there evidence of any mechanical anomalies that would have prevented normal operation.
Review of base reflectivity weather radar data showed that, while the pilot was maneuvering to divert to the alternate airport, the airplane was operating in an area of light precipitation that rapidly intensified to heavy precipitation, as shown by radar scans completed shortly after the accident.
During this time, the flight was likely operating in clouds along the leading edge of the convective line, where the pilot most likely would have encountered updrafts and severe or greater turbulence. The low visibility conditions that existed during the flight, which was conducted at night and in instrument meteorological conditions, coupled with the turbulence the flight likely encountered, were conducive to the development of spatial disorientation. Additionally, the airplane’s maneuvering during the final moments of the flight was consistent with a loss of control due to spatial disorientation. The pilot’s continued flight into known convective weather conditions and his delayed decision to divert the flight directly contributed to the accident.
Although the operator had a system safety-based program, the responsibility for the safe outcome of the flight was left solely to the pilot. Written company policy required completion of a FRAT before each flight by the assigned flight follower; however, a FRAT was not completed for the accident flight. The flight followers responsible for completing the FRATs were not trained to complete them for night cargo flights, and the operator’s management was not aware that the FRATs were not being completed for night cargo flights. Further, if a FRAT had been completed for the accident flight, the resultant score would have allowed the flight to commence into known hazardous weather conditions without any further review. If greater oversight had been provided by the operator, it is possible that the flight may have been canceled or re-routed due to the severity of the convective weather conditions present along the planned route of flight.
Perhaps the bottom line for pilots is that, in the end, it is the PIC’s go/no-go decision. Professionals want to complete the job, especially when the pressure is on. However, getting the job done means getting it done safely. (See “The Spatial Disorientation Trap”sidebar.)