A Pilot’s Perceptions

If there's a frequently recurring theme when the NTSB writes the probable causes for aviation accidents, it's that the pilot was responsible for the outcome: he or she failed to maintain airplane control or didn't engage in proper decision-making or, perhaps, mismanaged fuel. Often, the conclusions are based on incomplete evidence because the accident involved fatalities and the pilot, having been one of them, isn't around to tell his or her side of the story and explain the pre-crash thinking and actions. That wasn't the case in the June 3, 2017, accident involving a Piper PA-23-250 Aztec that crashed shortly after takeoff from San Juan, Puerto Rico. Although one passenger on the Part 135 on-demand air taxi flight was killed, the pilot and two other passengers survived. And, although the pilot described taking actions that to some would be heroic, the NTSB's probable cause makes him anything but the hero of the piece.

The airplane was operated by Air America, Inc., which flies throughout the Caribbean. The airplane received its airworthiness certificate on January 15, 1975, and was configured with six forward-facing seats: two in the front; two in the middle; and two in the back. Power came from two Lycoming IO-540 engines rated at 250 horsepower each and turning Hartzell controllable pitch full-feathering props. The most recent annual inspection was completed on November 16, 2016, at an airframe total time of just over 9,087 hours and just over 695 hours since major overhaul on the engines. Investigators calculated that the airplane had flown 95 hours between the annual and the accident.

Investigators used a written statement submitted by the pilot and interviews with the pilot and survivors to help reconstruct the sequence of events.

The plane was based at the Luis Muñoz Marín International Airport in San Juan, and the flight was to be to Benjamin Rivera Noriega Airport at Isla de Culebra in Puerto Rico. Although visual meteorological conditions prevailed, an IFR flight plan had been filed.

The pilot reported that he had gotten to work between 7:30 and 7:45 a.m. on the day of the accident, a Saturday, after waking up at about 6 a.m. He had been at home the day before and went to sleep at about 9:30 p.m. He said that at about 8:20 a.m., he departed in a company Britten-Norman BN-2A-21 Islander and flew to Virgin Gorda. He left at around 9:25 and flew back to San Juan. "After I cleared customs and finished the post flight and secured the airplane, I went to the terminal and had lunch," he wrote. After lunch, the pilot went to the FBO and did the paperwork for the afternoon flight to Isla de Culebra. The flight would use the call sign Airmek 303.

The pilot reported that he completed a preflight inspection and, as the passengers were coming out to the plane, did a final walk-around and checked the fuel in the tanks. He said that when the baggage had been loaded and the passengers seated, he gave a safety/emergency briefing and then decided to start the engines and contact air traffic control (ATC) for his clearance.

According to the FAA, the pilot was cleared to taxi for an intersection departure from Runway 8 at Taxiway S5. Runway 8 was 10,500 feet long, and the intersection was at about the halfway point.
The pilot said that he performed the run-up "!item by item on the checklist, and everything was working on acceptable parameters."

At 2:14:09 p.m., the pilot radioed the tower local controller and reported ready for departure. The controller advised that an Embraer 190 had departed about 2 minutes and 30 seconds earlier, and he still had 30 seconds left in the required hold for wake turbulence. The pilot said he was waiving the delay, and the controller then cleared the flight for takeoff. In an interview with NTSB investigators, the pilot said at 85 knots "!I rotate, the airplane started climbing. So I put the gear up."

At 2:16:33, the tower controller radioed the pilot to "!contact San Juan departure, safe flight." Two seconds later, the pilot radioed, "Unable, airmek three zero three going back to!" The controller asked the pilot to "!say again, sir." The pilot responded, "Ah, we are trying hard, we lost power on an engine, we're trying to go back airmek three zero three."

At 2:16:51, the controller asked "!what engine, sir." The pilot replied, "we lose an engine airmek zero three." This was the last transmission from the airplane.

The pilot reported to investigators that when the airplane was about 100 feet above the runway and the landing gear was coming up, he felt that it wasn't "flying as usual and was slow on speed and didn't want to climb." He said he checked the magnetos, fuel pumps, throttles, rpm levers, mixtures and engine instruments, and at first they looked good, but then he looked at the left tachometer and "!the RPMs on the left side, they're starting to decrease. So at that point, the airplane is starting to yaw to the left side, and at that point I just start flying to the left, just trying to see if I can make it back to the airport."

The airplane was captured on video by several airport security cameras. One camera in particular provided images of the airplane after takeoff as it flew over the departure end of the runway, followed runway heading and subsequently turned left. Analysis of the video showed what looked similar to puffs of white smoke briefly trailing from the airplane. An NTSB investigator said they likely were puffs of clouds that were not related to the airplane. When I enlarged frames from the video until individual pixels in the images became highly visible, I could not determine exactly what the puffs were. However, the color of the pixels associated with the puffs was a brighter shade of white than the pixels associated with what clearly were small clouds off in the distance.

The airport's weather observation showed a few clouds at 2,400 feet, scattered clouds at 4,000 feet and 7,000 feet, wind from 070 degrees at 17 knots and visibility 10 miles.

According to the transcript of the pilot's interview with the NTSB, the pilot stated that "!the airplane's just going down [at] a pretty fast rate. So I'm at the shoreline next to the airport, and I remember seeing all the, like, buildings and places that people are eating, and I saw the people, and I thought to myself, ’If I crash over there, I'm going to kill more people.' So!I look forward and I see a reef over here and there's another reef over here, and I saw boats, saw people on the beach. So I thought to myself, ’If I try to make it beyond the reef, I might still kill people. If I hit the reef, I will kill myself.' And the only option that I had at that time was just trying to get a path of water, just water."

Two FAA inspectors interviewed the front-seat passenger, who said although he was not a pilot, he was an aerospace engineer and had spent time in a Boeing 757 simulator. He told a different story than did the pilot. He said the pilot did not do a safety briefing and it was he who told the other passengers to fasten their seatbelts. He said the pilot did not conduct a run-up using a checklist and, once airborne, he mentally questioned the flying ability of the pilot as the airplane was yawing left, banking left and descending to impact. He said although he was tempted to do so, he did not touch the airplane's controls. The passenger said the engine noise was steady during the flight, and he did not notice any smoke or smell fuel.

The pilot painted a picture of carefully selecting a touchdown point. He described noticing a dark area of water, which to him indicated that the water was deeper. "So at that point, with no altitude, no airspeed to play with, I just bank it to the right, and just when it was about to hit the ground, I mean the water, I just get the yoke back and just try to hold it to get a smooth, a smoother landing on the water."

The crash took place at about 2:17 p.m. Atlantic Standard Time. Both wings separated during impact. There was a fire, but the NTSB didn't provide many details. It's possible that it was a result of fuel spilling after impact, but a photograph of the wreckage shows what appears to be a burned area along the fuselage from the last window on the airplane's left side running across the top to the vertical stabilizer. Investigators determined that the airplane's four main tanks were full at takeoff, with 144 gallons of 100 LL.

The pilot told investigators, "What I remember is not just really an explosion. I feel like!heat all over my body, and I saw a really bright light. And when I got conscious again, what I remember is just!floating on the water with the fire. So I got burned, and I dive again into the water. It's like I reached out [a] second time, and I got burned one more time. So without getting any chance to get air, I started diving, start swimming away from the fire."

The pilot continued, "I can hear the fire burning the airplane, and I start looking for my passengers because I had two girls and an adult. So I see the guy, and I see the little girl. When I find out that somebody's missing, I try to get back to the airplane. So I start swimming, but halfway going to the airplane, I wasn't able to get any---I didn't have enough air. I didn't have enough force to reach the airplane." The second girl was found still in the airplane with her seatbelt holding her in place. Investigators tested the quick release buckle on the seatbelt and found that it worked as designed.

The NTSB reported that the pilot held a commercial certificate with ratings for single-engine and multi-engine, was a flight instructor, and was rated for instruments. His first-class FAA medical certificate was current, and he had about 1,200 hours total time, with 200 in multi-engine airplanes and 20 in hours in PA-23 aircraft.

The Safety Board noted that he received disapproval notices on his first attempts to obtain his private pilot and instrument tickets, and it took three tries to pass his flight instructor check ride.

Investigators calculated that the airplane's takeoff weight was 4,850 pounds, 350 pounds below its maximum gross takeoff weight of 5,200 pounds. They noted that the takeoff weight calculated by the pilot was 4,335 pounds. They said the aircraft's center of gravity was near the forward limit. Using the air temperature of 88 degrees that existed at the time of the accident and an altitude of sea level, they figured the airplane should have had a single-engine best-rate-of-climb speed of 102 knots (120 mph) and been able to climb at 240 feet per minute on one engine.

Did the pilot follow procedures for an engine power loss or total failure? There aren't details in his written statement nor the transcript of his NTSB interview from June 3, 2017. Photos taken when the wreckage was still underwater showed the right throttle lever full forward with the mixture and prop levers midrange. The left throttle lever was halfway back while the left mixture and prop levers were forward. Because the wings broke off and various cables were severed, investigators couldn't be sure that's actually where the levers had been before impact. The left propeller blades were found in the normal operating range and not feathered.

The Safety Board's report noted that the airspeed indicator in front of the pilot had an outer ring of larger numbers indicating mph, with knots indicated on an inner ring in smaller numbers. They suggested that the pilot may have lifted off at 85 mph instead of 85 knots.

The report also pointed out that the airplane did not have counter-rotating engines that would have helped offset the left yaw you'd expect to encounter at takeoff power and low airspeeds. Remember your first lesson when the instructor told you to add right rudder to keep the plane climbing straight?

In his interview with the NTSB, the pilot referred to looking at the left tachometer and noticing the needle was reading lower than the right tachometer. The Safety Board's report suggested that this was not a reliable indicator of a problem with the left engine, since it was driving a constant speed propeller. The NTSB said reduced manifold pressure would have indicated engine power loss. Post-crash examination did not identify any mechanical problems with either engine.

The Safety Board's analysis suggested that regardless of whether the pilot attempted to climb the airplane at 85 mph or 85 knots, either airspeed was significantly less than the 120 mph, or 102 knots, best-rate-of-climb speed. The Safety Board said it was likely the airplane climbed while in ground effect, but once out of it, could not maintain a climb and began to descend back into ground effect. What I saw in the video, however, looked very much as if it was doing just fine out of ground effect. The Safety Board said that rather than lower the nose, correct for the left yaw and continue ahead into the 15 knot headwind to increase airspeed, the pilot allowed the left yaw to continue to intensify, and the critical angle of attack was quickly exceeded.

The NTSB determined that the probable cause of the accident was the pilot's failure to maintain adequate airspeed, properly correct for left yaw, and his exceedance of the airplane's critical angle of attack during initial climb, which resulted in an aerodynamic stall and subsequent uncontrolled descent.

An omission from what the Safety Board has done, in my opinion, is their discounting the pilot's perception of a problem, his actions and his belief that he remained in control without helping us understand why he was so at odds with what the investigation subsequently concluded. Of course, had the pilot been a fatality, this wouldn't be an issue.

Peter Katz is editor and publisher of NTSB Reporter, an independent monthly update on aircraft accident investigations and other news concerning the National Transportation Safety Board. To subscribe, visit ntsbreporter.us or write to: NTSB Reporter, Subscription Dept., P.O. Box 831, White Plains, NY 10602-0831.