United States of America

The pilot, co-pilot, and eight passengers departed on a cross-country flight in the twin-engine airplane. One witness located on the ramp at the airport reported that the airplane sounded underpowered immediately after takeoff “like it was at a reduced power setting.” Another witness stated that the airplane sounded like it did not have sufficient power to takeoff. A third witness described the rotation as “steep,” and other witnesses reported thinking that the airplane was performing aerobatics. Digital video from multiple cameras both on and off the airport showed the airplane roll to its left before reaching a maximum altitude of 100 ft above ground level; it then descended and impacted an airport hangar in an inverted attitude about 17 seconds after takeoff and an explosion immediately followed. After breaching a closed roll-up garage door, the airplane came to rest on its right side outside of the hangar and was immediately involved in a postimpact fire. Sound spectrum analysis of data from the airplane’s cockpit voice recorder (CVR) estimated that the propeller speeds were at takeoff power (1,714 to 1,728 rpm) at liftoff. About 7 seconds later, the propeller speeds diverged, with the left propeller speed decreasing to about 1,688 rpm and the right propeller speed decreasing to 1,707 rpm. Based on the airplane’s estimated calibrated airspeed of about 110 knots and the propeller rpm when the speeds diverged, the estimated thrust in the left engine decreased to near 0 while the right engine continued operating at slightly less than maximum takeoff power. Analysis of available data estimated that, 2 seconds after the propeller speed deviation, the airplane’s sideslip angle was nearly 20°. During the first 5 seconds after the propeller speed deviation, the airplane’s roll rate was about 5° per second to the left; its roll rate then rapidly increased to more than 60° per second before the airplane rolled inverted. Witness marks on the left engine and propeller, the reduction in propeller speed, and the airplane’s roll to the left suggest that the airplane most likely experienced a loss of thrust in the left engine shortly after takeoff. The airplane manufacturer’s engine-out procedure during takeoff instructed that the landing gear should be retracted once a positive rate of climb is established, and the propeller of the inoperative engine should be feathered. Right rudder should also be applied to balance the yawing moment imparted by a thrust reduction in the left engine. Examination of the wreckage found both main landing gear in a position consistent with being extended and the left propeller was unfeathered. The condition of the wreckage precluded determining whether the autofeather system was armed or activated during the accident flight. Thus, the pilot failed to properly configure the airplane once the left engine thrust was reduced. Calculations based on the airplane’s sideslip angle shortly after the propeller speed deviation determined that the thrust asymmetry alone was insufficient to produce the sideslip angle. Based on an evaluation of thrust estimates provided by the propeller manufacturer and performance data provided by the airplane manufacturer, it is likely that the pilot applied left rudder, the opposite input needed to maintain lateral control, before applying right rudder seconds later. However, by then, the airplane’s roll rate was increasing too rapidly, and its altitude was too low to recover. The data support that it would have been possible to maintain directional and lateral control of the airplane after the thrust reduction in the left engine if the pilot had commanded right rudder initially rather than left rudder. The pilot’s confused reaction to the airplane’s performance shortly after takeoff supports the possibility that he was startled by the stall warning that followed the propeller speed divergence, which may have prompted his initial, improper rudder input. In addition, the NTSB’s investigation estimated that rotation occurred before the airplane had attained Vr (rotation speed), which decreased the margin to the minimum controllable airspeed and likely lessened the amount of time available for the pilot to properly react to the reduction in thrust and maintain airplane control. Although the airplane was slightly over its maximum takeoff weight at departure, its rate of climb was near what would be expected at maximum weight in the weather conditions on the day of the accident (even with the extended landing gear adding drag); therefore, the weight exceedance likely was not a factor in the accident. Engine and propeller examinations and functional evaluations of the engine and propeller controls found no condition that would have prevented normal operation; evidence of operation in both engines at impact was found. Absent evidence of an engine malfunction, the investigation considered whether the left engine’s thrust reduction was caused by other means, such as uncommanded throttle movement due to an insufficient friction setting of the airplane’s power lever friction locks. Given the lack of callouts for checklists on the CVR and the pilot’s consistently reported history of not using checklists, it is possible that he did not check or adjust the setting of the power lever friction locks before the accident flight, which led to uncommanded movement of the throttle. Although the co-pilot reportedly had flown with the pilot many times previously and was familiar with the B-300, he was not type rated in the airplane and was not allowed by the pilot to operate the flight controls when passengers were on board. Therefore, the co-pilot may not have checked or adjusted the friction setting before the flight’s departure. Although the investigation considered inadequate friction setting the most likely cause of the thrust reduction in the left engine, other circumstances, such as a malfunction within the throttle control system, could also result in loss of engine thrust. However, heavy fire and impact damage to the throttle control system components, including the power quadrant and cockpit control lever friction components, precluded determining the position of the throttle levers at the time of the loss of thrust or the friction setting during the accident flight. Thus, the reason for the reduction in thrust could not be determined definitively. In addition to a lack of callouts for checklists on the CVR, the pilots did not discuss any emergency procedures. As a result, they did not have a shared understanding of how to respond to the emergency of losing thrust in an engine during takeoff. Although the co-pilot verbally identified the loss of the left engine in response to the pilot’s confused reaction to the airplane’s performance shortly after takeoff, it is likely the co-pilot did not initiate any corrective flight control inputs, possibly due to the pilot’s established practice of being the sole operator of flight controls when passengers were on board. The investigation considered whether fatigue from inadequately treated obstructive sleep apnea contributed to the pilot’s response to the emergency; however, the extent of any fatigue could not be determined from the available evidence. In addition, no evidence indicates that the pilot’s medical conditions or their treatment were factors in the accident. In summary, the available evidence indicates that the pilot improperly responded to the loss of thrust in the left engine by initially commanding a left rudder input and did not retract the landing gear or feather the left propeller, which was not consistent with the airplane manufacturer’s engine out procedure during takeoff. It would have been possible to maintain directional and lateral control of the airplane after the thrust reduction in the left engine if right rudder had been commanded initially rather than left rudder. It is possible that the pilot’s reported habit of not using checklists resulted in his not checking or adjusting the power lever friction locks as specified in the airplane manufacturer’s checklists. However, fire and impact damage precluded determining the position of the power levers or friction setting during the flight.

On June 21, 2019, about 1822 Hawaii-Aleutian standard time, a Beech King Air 65-A90 airplane, N256TA, impacted terrain after takeoff from Dillingham Airfield (HDH), Mokuleia, Hawaii. The pilot and 10 passengers were fatally injured, and the airplane was destroyed. The airplane was owned by N80896 LLC and was operated by Oahu Parachute Center (OPC) LLC under the provisions of Title 14 Code of Federal Regulations (CFR) Part 91 as a local parachute jump (skydiving) flight. Visual meteorological conditions prevailed at the time of the accident. OPC had scheduled five parachute jump flights on the day of the accident and referred to the third through fifth flights of the day as “sunset” flights because they occurred during the late afternoon and early evening. The accident occurred during the fourth flight. The accident pilot was the pilot-in-command (PIC) for each of the OPC flights that departed on the day of the accident. The pilot and 8 of the 10 passengers initially boarded the airplane. These eight passengers comprised three OPC tandem parachute instructors, three passenger parachutists, and two OPC parachutists performing camera operator functions. The pilot began to taxi the airplane from OPC’s location on the airport. According to a witness (an OPC tandem instructor who was not aboard the accident flight), the two other passengers—solo parachutists who had been on the previous skydiving flight and were late additions to the accident flight—“ran out to the airplane and were loaded up at the last minute.” The pilot taxied the airplane to runway 8 about 1820, and the airplane departed about 1822. According to multiple witnesses, after the airplane lifted off, it banked to the left, rolled inverted, and descended to the ground. One witness stated that, before impact, the airplane appeared to be intact and that there were no unusual noises or smoke coming from the airplane. A security camera video showed that the airplane was inverted in a 45° nose-down attitude at the time of impact. The airplane impacted a grass and dirt area about 630 ft northeast of the departure end of the runway, and a postcrash fire ensued. The airplane was not equipped, and was not required to be equipped, with a cockpit voice recorder or a flight data recorder. The accident flight was not detected by radar at the Federal Aviation Administration’s (FAA) Hawaii Control Facility, which was the air traffic control (ATC) facility with jurisdiction of the airspace over HDH. The FAA found no audio communications between the accident airplane and ATC on the day of the accident.

During a cross-country flight, the pilot initiated a descent to his intended destination. During the descent, the pilot informed air traffic control (ATC) that he could not retard power on the right engine. Later in the descent, the pilot decided to shut down the right engine. The pilot communicated his desire to land at the nearest airport to ATC, and ATC provided the pilot with the clearance to divert. Radar data showed the airplane in a steady descent toward the airport. When the airplane was at an altitude of about 2,500 ft mean sea level, the pilot contacted ATC and stated that he was trying to get the airplane under control; radar data showed the airplane in a 360° right turn at the time. The pilot contacted ATC again and stated that he was going to land on a highway. No further transmissions were received from the pilot. After the right turn, the airplane continued in a descent through 1,300 ft mean sea level, at which point radar contact was lost. A witness saw the airplane and stated that the airplane was low and slow but appeared to be in stable flight with both propellers spinning. She did not see any smoke coming from the airplane. She saw the airplane flying northeast to southwest when it suddenly descended nose first into the ground. The airplane impacted a gravel road adjacent to a 100-fttall grain silo about 1 mile from the highway and about 3.3 miles from the airport.

The pilot departed on the cross-country flight with the airplane about 730 lbs over its maximum gross weight. While in cruise altitude at 27,000 ft mean sea level (msl), the pilot reported to air traffic control that he observed weather on his radar along his route and ahead of his position. The areas of weather included thunderstorms with cloud tops up to 43,000 ft msl. The controller acknowledged the weather; however, she did not provide specifics to the pilot, including the size and strength of the area of precipitation or cloud tops information, nor did she solicit or disseminate any pilot reports related to the conditions, as required. The airplane entered an area of heavy to extreme precipitation, likely a thunderstorm updraft, while in instrument meteorological conditions. Tracking information indicated that the airplane climbed about 300 ft, then entered a right, descending spiral and broke up in flight at high altitude. The recovered wreckage was found scattered along a path about 2.6 miles in length. Both wings separated, and most of the empennage was not located. The airplane was likely about 148 lbs over the maximum allowable gross weight at the time of the accident. Examination of the wreckage revealed no evidence of a pre accident malfunction or failure that would have prevented normal operation. The pilot, who owned the airplane, did not possess an instrument rating. The pilot-rated passenger in the right seat was instrument-rated but did not meet resency of experience requirements to act as pilot-in-command. Toxicology testing detected a small amount of ethanol in the pilot’s liver but not in muscle. After absorption, ethanol is uniformly distributed throughout all tissues and body fluids; therefore, the finding in one tissue but not another is most consistent with post-mortem production. Hazardous weather avoidance is ultimately the pilot’s responsibility, and, in this case, the airplane was sufficiently equipped to provide a qualified pilot with the information necessary to navigate hazardous weather; however, the controller’s failure to provide the pilot with adequate and timely weather information as required by Federal Aviation Administration Order JO 7110.65X contributed to the pilot’s inability to safely navigate the hazardous weather along his route of flight, resulting in the penetration of a thunderstorm and the resulting loss of airplane control and inflight breakup.

The pilot was conducting a personal cross-country flight in a turbofan-powered airplane. Shortly after departure, the airplane entered a witness-estimated 90° left bank with the nose parallel to the horizon; as the airplane began to roll out of the turn, the nose remained at or below the horizon before it dropped and the airplane impacted the ground. Flight track data revealed that, shortly after departure, the airplane's ground speed immediately began decreasing from its maximum of 141 knots during takeoff and continued decreasing until the last recorded data point, which showed that the airplane had a ground speed of 100 knots. The surface wind reported about 10 minutes before the accident was from 170° at 9 knots, gusting to 14 knots, which resulted in a 1- to 2-knot tailwind component. Given this information and the airplane's configuration at the time of the accident, the airplane's indicated airspeed (IAS) would have been between about 86 and 93 knots. The airplane's stall speed was calculated to be 100 knots IAS (KIAS) with a bank angle of 45° and 118 KIAS with a bank angle of 60°. Thus, the pilot failed to maintain airspeed or accelerate after departure, which resulted in an aerodynamic stall A pilot who had flown with the accident pilot twice before the accident reported that, during these flights, the pilot had flown at reduced power settings and slower-than-normal airspeeds. During the flight 1 year before the accident, he reached over and pushed the power levers forward himself. He also stated that every time he had flown with the pilot, he was "very behind the airplane." Postaccident examination of the engines revealed no signs of preimpact mechanical failures or malfunctions that would have precluded normal operation, and both engines exhibited circumferential rub marks on all rotating stages, blade tip bending opposite the direction of rotation, and debris ingestion through the gas path, indicating that the engine had power at impact. Further, the right engine full authority digital electronic control (FADEC) nonvolatile memory recorded no faults. (The left engine FADEC could not be downloaded due to damage.) The Airplane Flight Manual stated that the pilot must, in part, advance the throttle lever to the maximum takeoff detent for the FADEC's nonvolatile memory to record a logic trend snapshot 2 seconds after takeoff. The lack of a FADEC logic trend snapshot is consistent with the pilot not fully advancing the throttles during the takeoff and initial climb and is likely why he did not attain or maintain sufficient airspeed. The flight track data, pilot witness account, and airplane damage are consistent with the pilot failing to fully advance the power levers while maneuvering shortly after takeoff, which led to his failure to maintain sufficient airspeed and resulted in the exceedance of the airplane's critical angle of attack and a subsequent aerodynamic stall.

The commercial pilot was conducting his first scheduled commuter flight from the company’s seaplane base to a nearby island seaplane base with one passenger and cargo onboard. According to company pilots, the destination harbor was prone to challenging downdrafts and changing wind conditions due to surrounding terrain. Multiple witnesses at the destination stated that the airplane made a westerly approach, and the wind was from the southeast with light chop on the water. Two witnesses reported the wings rocking left and right before touchdown. One witness stated that a wind gust pushed the tail up before the airplane landed. A different witness reported that the airplane was drifting right during the touchdown, and another witness saw the right (downwind) float submerge under water after touchdown, and the airplane nosed over as it pivoted around the right wingtip, which impacted the water. Flight track and performance data from the cockpit display units revealed that, as the airplane descended on the final approach, the wind changed from a right headwind of 6 knots to a left quartering tailwind of 8 knots before touchdown. The crosswind and tailwind components were within the airplane’s operational limitations. Examination of the airframe, engine, and associated systems revealed no evidence of mechanical malfunctions or failures that would have precluded normal operation or egress. During the final approach and descent, the pilot had various wind information available to him; the sea surface wind waves and signatures, the nearest airport observation winds, the cockpit display calculated wind, and the visual relative ground speed. Had the pilot recognized that the winds had shifted to a quartering tailwind and the airplane’s ground speed was faster than normal, he could have aborted the landing and performed another approach into the wind. Although crosswind landings were practiced during flight training, tailwind landings were not because new pilots were not expected to perform them. Although the crosswind component was well within the airplane’s limits, it is possible that combined with the higher ground speed, the inexperienced pilot was unable to counteract the lateral drift during touchdown in a rapidly shifting wind. The pilot was hired the previous month with 5 hours of seaplane experience, and he completed company-required training and competency checks less than 2 weeks before the accident. According to the chief pilot (CP), company policy was to assign newly hired pilots to tour flights while they gained experience before assigning them to commuter flights later in the season. The previous year, the CP distributed a list of each pilot’s clearances for specific types of flights and destinations; however, an updated list had not been generated for the season at the time of the accident, and the flight coordinators, who were delegated operational control for assigning pilots to flights, and station manager were unaware of the pilot’s assignment limitations. Before the flight, the flight coordinator on duty completed a company flight risk assessment that included numerical values based on flight experience levels. The total risk value for the flight was in the caution area, which required management notification before releasing the flight, due to the pilot’s lack of experience in the accident airplane make and model and with the company, and his unfamiliarity with the geographical area; however, the flight coordinator did not notify management before release because the CP had approved a tour flight with the same risk value earlier in the day. Had the CP been notified, he may not have approved of the pilot's assignment to the accident flight. The pilot's minimal operational experience in seaplane operations likely affected his situational awareness in rapidly changing wind conditions and his ability to compensate adequately for a quartering tailwind at a higher-than-normal ground speed, which resulted in a loss of control during the water landing and a subsequent nose-over.

The commercial pilot was relocating the multiengine airplane following the completion of an extensive avionics upgrade, which also included the installation of new fuel flow transducers. As the pilot neared the destination airport, he reported over the common traffic advisory frequency that he had "an engine out [and] smoke in the cockpit." Witnesses observed and airport surveillance video showed fire emanating from the airplane's right wing. As the airplane turned towards the runway, it entered a rightrolling descent and impacted the ground near the airport's perimeter fence. The right propeller was found feathered. Examination of the right engine revealed evidence of a fire aft of the engine-driven fuel pump. The fuel pump was discolored by the fire. The fire sleeves on both the fuel pump inlet and outlet hoses were burned away. The fuel outlet hose from the fuel pump to the flow transducer was found loose. The reason the hose was loose was not determined. It is likely that pressurized fuel sprayed from the fuel pump outlet hose and was ignited by the hot turbocharger, which resulted in the inflight fire.

On May 13, 2019, about 1221 Alaska daylight time, a float-equipped de Havilland DHC-2 (Beaver) airplane, N952DB, and a float-equipped de Havilland DHC-3 (Otter) airplane, N959PA, collided in midair about 8 miles northeast of Ketchikan, Alaska. The DHC-2 pilot and four passengers sustained fatal injuries. The DHC-3 pilot sustained minor injuries, nine passengers sustained serious injuries, and one passenger sustained fatal injuries. The DHC-2 was destroyed, and the DHC-3 sustained substantial damage. The DHC-2 was registered to and operated by Mountain Air Service LLC, Ketchikan, Alaska, under the provisions of Title 14 Code of Federal Regulations (CFR) Part 135 as an on-demand sightseeing flight. The DHC-3 was registered to Pantechnicon Aviation LTD, Minden, Nevada, and operated by Venture Travel, LLC, dba Taquan Air, Ketchikan, Alaska, under the provisions of Part 135 as an on-demand sightseeing flight. Visual meteorological conditions prevailed in the area at the time of the accident. According to information provided by the operators, both airplanes had been conducting sightseeing flights to the Misty Fjords National Monument area. They were both converging on a scenic waterfall in the Mahoney Lakes area on Revillagigedo Island before returning to the Ketchikan Harbor Seaplane Base (5KE), Ketchikan, Alaska, when the accident occurred. According to recorded avionics data recovered from the DHC-3, it departed from an inlet (Rudyerd Bay) in the Misty Fjords National Monument area about 1203 and followed the inlet westward toward Point Eva and Manzanita Island. At 1209, at an altitude between 1,900 and 2,200 ft, the DHC-3 crossed the Behm Canal then turned to the southwest about 1212 in the vicinity of Lake Grace. Automatic dependent surveillance-broadcast (ADS-B) tracking data for both airplanes, which were provided by the Federal Aviation Administration (FAA), began at 1213:08 for the DHC-3, and at 1213:55 for the DHC-2. At 1217:15, the DHC-3 was about level at 4,000 ft mean sea level (msl) over Carroll Inlet on a track of 225°. The DHC-2 was 4.2 nautical miles (nm) south of the DHC-3, climbing through 2,800 ft, on a track of 255°. The DHC-3 pilot stated that, about this time, he checked his traffic display and “there were two groups of blue triangles, but not on my line. They were to the left of where I was going.” He stated that he did not observe the DHC-2 on his traffic display before the collision. The ADS-B data indicated that, about 1219, the DHC-3 started a descent from 4,000 ft, and the DHC-2 was climbing from 3,175 ft. During the next 1 minute 21 seconds, the DHC-3 continued to descend on a track between 224° and 237°, and the DHC-2 leveled out at 3,350 ft on a track of about 255°. Between 1220:21 and 1221:14, the DHC-3 made a shallow left turn to a track of 210°, then a shallow right turn back to a track of 226°. The airplanes collided at 1221:14 at an altitude of 3,350 ft, 7.4 nm northeast of 5KE. The ADS-B data for both airplanes end about the time of the collision. The DHC-2 was fractured into multiple pieces and impacted the water and terrain northeast of Mahoney Lake. Recorded avionics data for the DHC-3 indicate that at 1221:14, the DHC-3 experienced a brief upset in vertical load factor and soon after entered a right bank, reaching an attitude about 50° right wing down at 1221:19 and 27° nose down at 1221:22. The DHC-3 began descending and completed a 180° turn before impacting George Inlet at 1222:15 along a northeast track.

On May 13, 2019, about 1221 Alaska daylight time, a float-equipped de Havilland DHC-2 (Beaver) airplane, N952DB, and a float-equipped de Havilland DHC-3 (Otter) airplane, N959PA, collided in midair about 8 miles northeast of Ketchikan, Alaska. The DHC-2 pilot and four passengers sustained fatal injuries. The DHC-3 pilot sustained minor injuries, nine passengers sustained serious injuries, and one passenger sustained fatal injuries. The DHC-2 was destroyed, and the DHC-3 sustained substantial damage. The DHC-2 was registered to and operated by Mountain Air Service LLC, Ketchikan, Alaska, under the provisions of Title 14 Code of Federal Regulations (CFR) Part 135 as an on-demand sightseeing flight. The DHC-3 was registered to Pantechnicon Aviation LTD, Minden, Nevada, and operated by Venture Travel, LLC, dba Taquan Air, Ketchikan, Alaska, under the provisions of Part 135 as an on-demand sightseeing flight. Visual meteorological conditions prevailed in the area at the time of the accident. According to information provided by the operators, both airplanes had been conducting sightseeing flights to the Misty Fjords National Monument area. They were both converging on a scenic waterfall in the Mahoney Lakes area on Revillagigedo Island before returning to the Ketchikan Harbor Seaplane Base (5KE), Ketchikan, Alaska, when the accident occurred. According to recorded avionics data recovered from the DHC-3, it departed from an inlet (Rudyerd Bay) in the Misty Fjords National Monument area about 1203 and followed the inlet westward toward Point Eva and Manzanita Island. At 1209, at an altitude between 1,900 and 2,200 ft, the DHC-3 crossed the Behm Canal then turned to the southwest about 1212 in the vicinity of Lake Grace. Automatic dependent surveillance-broadcast (ADS-B) tracking data for both airplanes, which were provided by the Federal Aviation Administration (FAA), began at 1213:08 for the DHC-3, and at 1213:55 for the DHC-2. At 1217:15, the DHC-3 was about level at 4,000 ft mean sea level (msl) over Carroll Inlet on a track of 225°. The DHC-2 was 4.2 nautical miles (nm) south of the DHC-3, climbing through 2,800 ft, on a track of 255°. The DHC-3 pilot stated that, about this time, he checked his traffic display and “there were two groups of blue triangles, but not on my line. They were to the left of where I was going.” He stated that he did not observe the DHC-2 on his traffic display before the collision. The ADS-B data indicated that, about 1219, the DHC-3 started a descent from 4,000 ft, and the DHC-2 was climbing from 3,175 ft. During the next 1 minute 21 seconds, the DHC-3 continued to descend on a track between 224° and 237°, and the DHC-2 leveled out at 3,350 ft on a track of about 255°. Between 1220:21 and 1221:14, the DHC-3 made a shallow left turn to a track of 210°, then a shallow right turn back to a track of 226°. The airplanes collided at 1221:14 at an altitude of 3,350 ft, 7.4 nm northeast of 5KE. The ADS-B data for both airplanes end about the time of the collision. The DHC-2 was fractured into multiple pieces and impacted the water and terrain northeast of Mahoney Lake. Recorded avionics data for the DHC-3 indicate that at 1221:14, the DHC-3 experienced a brief upset in vertical load factor and soon after entered a right bank, reaching an attitude about 50° right wing down at 1221:19 and 27° nose down at 1221:22. The DHC-3 began descending and completed a 180° turn before impacting George Inlet at 1222:15 along a northeast track.

The pilot was performing a personal cross-country flight. While en route to the intended destination, the pilot contacted air traffic control to report that the airplane was having a fuel pump issue and requested to divert to the nearest airport. The pilot stated that the request was only precautionary and did not declare an emergency during the flight; he provided no further information about the fuel pump. As the airplane approached the diversion airport, witnesses observed the airplane flying low and rolling to the left just before impacting terrain, after which a postcrash fire ensued. An examination of the airframe revealed no preimpact mechanical malfunctions or failures that would have precluded normal operation. A postaccident examination and review of recorded data indicated that the left engine was secured and in the feather position, and that the right engine was operating at a high RPM setting. The left engine-driven fuel pump was found fractured. Further examination of the fuel pump revealed fatigue failure of the pressure relief valve. The fatigue failure initiated in upward bending on one side of the valve disk and progressed around both sides of the valve stem. As the cracks grew, the stem separated from the disk on one side and began to tilt in relation to the disk and the valve guide due to the non-symmetric support, which caused the lower end of the stem to rub against the valve guide, creating wear marks. The increasing stem tilt would have impinged against the valve guide, and the valve might have begun to stick in the closed position. If the valve were stuck in the closed position, it would not be able to open, and the outlet fuel pressure could rise above the set point pressure. Because the pump was driven by the engine, there would not be a way for the pilot to shut it off, disconnect it, or bypass it. Instead, the fuel pressure would continue to rise until the valve were to unstick. Thus, the pilot was likely experiencing variable fuel pressure as the valve became stuck and unstuck. Examination of the spring seat and the diaphragm plate, which were in contact with each other in the fuel pump assembly, revealed wear marks on the surface of each component, with one mark on the diaphragm plate and two wear marks on the spring seat. The two wear marks on the spring seat were distinct features separated by material with no wear indications in between. The only way that these wear marks could have occurred were if the spring seat was separated from the diaphragm plate and reinstalled in a different orientation. Thus, it is likely that the pilot had encountered a fuel pump problem before the accident flight and that someone tried to troubleshoot the problem. The last radar data point indicated that the airplane was traveling at a groundspeed of about 98 knots, and had passed north of the airport, traveling to the southwest. The minimum control speed for the airplane with single-engine operation was 88 knots. However, it is likely that if the pilot initiated a left turn back toward the airport, that the right engine torque and the 14 knot wind with gusts to 24 knots would have necessitated a higher speed. Because appropriate control inputs and airspeed were not maintained, the airplane rolled in the direction of the feathered engine (due to the left fuel pump problem), resulting in a loss of control. The pilot's toxicology report was positive for cetirizine, sumatriptan, gabapentin, topiramate, and duloxetine. All of these drugs act in the central nervous system and can be impairing alone or in combination. Although this investigation could not determine the reason(s) for the pilot's use of these drugs, they are commonly used to treat chronic pain syndromes or seizures. It is likely that the pilot was experiencing some impairment because of multiple impairing medications and was unable to successfully respond to the in-flight urgent situation and safely land the airplane.