N14181Piper PA-23-250 · 1971

On June 2, 2020, about 1333 Atlantic standard time, a Piper PA-23-250, N14181, was substantially damaged when it was involved in an accident during takeoff from Fernando Luis Ribas Dominicci Airport (TJIG), San Juan, Puerto Rico. The airline transport pilot and one passenger were fatally injured and a pilot-rated passenger was seriously injured. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 post-maintenance flight test.

The pilot-rated passenger, who occupied the right front seat and also held a mechanic certificate with airframe and powerplant ratings, reported the pilot taxied to runway 9 where an engine run-up was performed and that no discrepancies were noted. The airplane was cleared for takeoff, the pilot applied takeoff power, and the airplane immediately yawed to the left as if the left engine was not producing power. The pilot aborted the takeoff and was able to control the airplane, which, according to the pilot-rated passenger, was “almost sideways.”

The pilot-rated passenger later reported that he attributed the yaw to a loss of power from the left engine but at the time asked the pilot what would have caused the yaw, and the pilot responded that the left landing gear strut had less nitrogen than the right strut. However, according to a partial transcript of the pilot’s communications with air traffic control, he replied “…I think it’s a low tire” when asked if he needed any assistance. The pilot-rated passenger reported that the left engine quit while at idle power as they exited the runway. The pilot was able to restart it with some difficulty then taxied to the ramp.

While at the ramp, the pilot directed the pilot-rated passenger to inspect the landing gear. He informed the pilot that the left strut had more extension than the right, which was opposite the pilot’s expectation. The pilot directed the passenger to get back in the airplane and performed an engine run-up, which included a check of the magnetos and cycling the propellers. The pilot-rated passenger did not detect any discrepancies though he did state that he was wearing a noise-cancelling headset.

The pilot then contacted the air traffic control tower and requested taxi clearance to runway 09. After taxiing to the runway, the airplane was cleared for takeoff. The pilot-rated passenger stated that the airplane again immediately yawed to the left when the pilot applied full power for takeoff. The pilot immediately applied “right rudder, yoke to the right and back pressure like in a crosswind soft field takeoff.” During takeoff, the pilot was "fighting" the airplane as it drifted left of the runway centerline with the left engine not producing full power. The pilot-rated passenger reported the pilot raised the landing gear selector to retract the landing gear but did not recall seeing a green light. The pilot-rated passenger noticed that the airplane was not climbing normally; he estimated the climb rate at 100 feet-per-minute.

Security video from a nearby marina captured the final portion of the flight and depicted the airplane during the upwind leg flying in a near wings-level, slight nose-up attitude. As the flight continued, one of the landing gear was visible. The video depicted the airplane rolling left into an inverted and nose-low attitude. It remained in that position until impacting the water. No parts were seen separating from the airplane.

After impact, the passenger was knocked unconscious then regained consciousness. He initially had difficulty opening the door but was able to exit the airplane after it was fully submerged.

An in-flight video taken by the middle row passenger’s portable electronic device (PED) was provided to the National Transportation Safety Board (NTSB) for review and analysis. A view of the full instrument panel was captured at 03 seconds and frame 21 at which time the airplane was noted to be in a climb pitch attitude, both of the pilot’s hands were on the control yoke, and he was inputting a right roll command. The left throttle, propeller, and mixture controls appeared to be in the full forward position. During that same time and frame, the airspeed indicator appeared to be between 90 and 100 mph. The JPI engine monitor depicted the fuel flow for the left and right engines near zero and top of the indicator bar, respectively. The fuel pressure for the left engine was lower than the fuel pressure for the right engine, and the rpm for the left engine was between 1,000 and 1,500, while the rpm for the right engine was near the top of the green arc.

The pilot also held an mechanic certificate with airframe and powerplant ratings issued September 18, 2008.

The landing gear and flaps were hydraulically controlled by an engine-driven hydraulic pump installed on the left engine. According to the Federal Aviation Administration (FAA)-approved Airplane Flight Manual, in the event of failure of the left engine, the landing gear and flaps should be retracted by utilizing the emergency hydraulic hand pump. The Pilot’s Operating Manual (POM) specified about 50 strokes of the hand pump were required to retract the landing gear.

The published minimum controllable single-engine airspeed and best single engine rate of climb airspeed were 80 and 102 mph, respectively.

The airplane was equipped with a Garmin GTN 650, a Garmin GPSMay 696 and a JPI EDM 960 engine monitor. The devices were submitted to the NTSB Vehicle Recorder Division for read-out. The GTN 650 had no data recording capability and the last recorded flight in the Garmin GPSMap 696 was April 4, 2019.

The JPI engine monitor recorded two sessions on the accident date; the last being associated with the accident flight. The data from the engine monitor indicated that for the accident flight, the left and right engine rpm values were between 1,000 and 1,300, and 600 to 800 rpm, respectively; thus, no data were recorded during takeoff for the accident flight. A review of prior flights revealed the recorded engine rpm for both engines were about 2,300. The last recorded fuel level in the left outboard fuel tank was 9.9 gallons.

The landing gear and flaps were hydraulically controlled by an engine-driven hydraulic pump installed on the left engine. According to the Federal Aviation Administration (FAA)-approved Airplane Flight Manual, in the event of failure of the left engine, the landing gear and flaps should be retracted by utilizing the emergency hydraulic hand pump. The Pilot’s Operating Manual (POM) specified about 50 strokes of the hand pump were required to retract the landing gear.

The published minimum controllable single-engine airspeed and best single engine rate of climb airspeed were 80 and 102 mph, respectively.

The airplane was equipped with a Garmin GTN 650, a Garmin GPSMay 696 and a JPI EDM 960 engine monitor. The devices were submitted to the NTSB Vehicle Recorder Division for read-out. The GTN 650 had no data recording capability and the last recorded flight in the Garmin GPSMap 696 was April 4, 2019.

The JPI engine monitor recorded two sessions on the accident date; the last being associated with the accident flight. The data from the engine monitor indicated that for the accident flight, the left and right engine rpm values were between 1,000 and 1,300, and 600 to 800 rpm, respectively; thus, no data were recorded during takeoff for the accident flight. A review of prior flights revealed the recorded engine rpm for both engines were about 2,300. The last recorded fuel level in the left outboard fuel tank was 9.9 gallons.

The wreckage was recovered the following day. As part of the recovery the fuel tanks were drained; the FAA was not present when the tanks were drained. Examination of the wreckage by a representative of the airframe manufacture with oversight from the FAA revealed extensive impact damage to the airframe and wings; the left wing was separated at the wing root. All landing gears were extended.

A postaccident examination of the flight controls revealed no evidence of preimpact failure or malfunction. The hydraulic flap actuating cylinder piston was extended about 7 inches which corresponded to flaps retracted. The rudder trim jackscrew was extended aft from the drum about 5/8 inch or about 5 threads, which corresponded to a neutral to slight aircraft nose right trim setting. The stabilator jackscrew was extended above the drum about 1/2 inch or about 10 threads, which corresponded to a full aircraft nose up trim setting.

A postaccident examination of the left-wing fuel supply system to include the fuel selector, supply hoses, fuel strainer bowl, fuel tank, and electric fuel pump did not reveal any preimpact failures or malfunctions which would have precluded normal operation.

Examination of the left engine propeller and mixture controls revealed that except for cuts at the left-wing root consistent with recovery, control cable continuity was confirmed from the cockpit to their respective attach points in the engine compartment. Throttle control cable continuity was confirmed from the cockpit to the left-wing root area where cut for recovery, then from there to the ball screw which was fractured; the ball screw connected the throttle cable to the fuel servo throttle arm. Examination of the fractured ball screw by the NTSB Materials Laboratory revealed the fracture features were consistent with fracture from overstress in bending.

During manual rotation of the left propeller, crankshaft, camshaft, and valve train continuity were confirmed to the valves and to the accessory section. Thumb section and compression were felt at each of the top spark plug openings of each cylinder. The magnetos appeared secure on their mounts and undamaged. The magnetos were mechanically rotated and both produced spark at all towers of the distributor cap. Examination of the engine-driven fuel pump revealed during hand actuation, the pump could be heard making noises consistent with actuating. Disassembly examination of the pump revealed the diaphragm appeared to be flexible and no debris was observed within the pump. Examination of the servo fuel injector (fuel servo), the manifold valve, and the fuel injectors revealed the servo inlet screen was free from blockage, the diaphragm of the manifold valve was flexible and intact and three of the injectors appeared free of blockage while three appeared to have some blockage. The fuel supply hoses in the engine compartment contained small quantities of liquid that was consistent with 100 low lead (100LL), with no water detected in any sample. Examination of the propeller governor revealed the screen was clean, the shaft rotated normally and pumped oil. The fuel supply hoses, fuel servo, manifold valve, and fuel injectors were retained for further examination.

Examination of the engine controls of the right engine revealed throttle, propeller, and mixture control cable continuity was confirmed from the cockpit to their respective attach points in the engine compartment. A postaccident examination of the engine did not reveal any preimpact failure or malfunction that would have prevented normal operation of the right engine.

Both propellers remained attached to each engine crankshaft flange and both propeller blades of each propeller remained secured in each propeller hub assembly. There was no evidence of failure or malfunction of either propeller; the left propeller blades were at or near the low pitch position.

According to fueling records, the airplane was fueled with 47.5 gallons of 100LL aviation fuel about 2 hours prior to the accident. According to the fueler, the pilot was present during fueling and both inboard fuel tanks were topped off. After fueling the pilot was observed checking the fuel in the area of each engine nacelle. Another airplane that was fueled from the same fuel truck about 22 minutes earlier flew on multiple flights with no reported fuel related discrepancies.

Examination of the fuel supply hoses revealed no internal obstructions or leaks.

Examination of the left fuel servo at a FAA certified repair station revealed the nylon throttle stop was missing, which allowed the throttle plate to fully close rather than have a 0.006-inch gap. While setting the gap to 0.006 inch between the throttle plate and bore, a gap greater than 0.006 inch was noted at the throttle stop. During initial operational testing, material consistent with sand flowed from the unit. At a test point that required the throttle to be wide open and the mixture control full rich (consistent with takeoff), the unit flowed about 69% of the minimum specified or required value. Subsequent partial disassembly revealed contamination between the mixture and idle valves. Following removal of the contamination, the unit was operationally tested at the same test point and flowed within specification. No attempt was made to identify the contamination. Operational testing of the fuel injector nozzles revealed slight out of tolerance conditions, while operational testing of the manifold valve could not be performed.