On March 12, 2019, a professional pilot flying a twin-engine Piper Navajo for a large commercial survey company starved the left engine of fuel, causing it to quit. Declining air traffic control’s offer of vectors to a nearby suitable airport, he continued on the sole remaining working engine toward his original destination (which was also his departure), the Cincinnati Lunken Airport (KLUK) in Ohio. It was 30 miles away. He almost made it.
Before we get into the disturbing details of the crash, let’s note the cause wasn’t fuel exhaustion but, rather, fuel starvation. The former is when there’s no fuel to be burned. The latter occurs when there’s fuel onboard, but for some reason, the pilot can’t or doesn’t get it to the engine, resulting in a loss of power in that engine.
Five miles from the airport, the plane came crashing down into the backyard of a suburban home. What fuel there was in the tanks ignited. The pilot, the only occupant, died in the blazing wreckage. Unpacking the National
Transportation Safety Board’s (NTSB) extensive accident docket reveals several lessons, in things that happened slowly and in things that happened quickly.
The 62-year-old pilot, a father and grandfather, was passionate about flying. However, he had a spotty resume. An old story in The Arizona Republic newspaper reports he was fired from a flight school in 2001 for fighting with another flight instructor. Then, in 2003, he was blamed by the NTSB for crashing a Cessna 172 while simulating engine failure with a student. Rather than stopping at 500 feet AGL, he continued the power-off descent to 20 feet. Unable to recover, they impacted the ground. Both pilots received only minor injuries, but the plane was substantially damaged. The president of the flight school said it was “a classic blunder that many pilots make. They get overconfident and put themselves in a position where they have no options. The facts are, he was too low and too slow.”
That accident didn’t stop him from flying. By 2019, he had over 6,400 total hours, including sitting right seat in a Learjet 35 and 1,350 hours in Piper Navajos. From his home near Phoenix, Arizona, he started flying for an aerial mapping company. They had had a nationwide fleet of 15 Piper Navajos, a workhorse cabin-class piston- twin. The stretched PA-31-350 Chieftain version could fit 10 seats and saw considerable service in decades past for small feeder airlines, corporations and freight outfits. This one, N400JM, built-in 1981, was equipped with two counter-rotating 350-horsepower Lycoming TIO-540-J2B engines. It had just three seats, as most of the floor space was occupied by a specialized aerial imaging platform. In 2017, to better handle long missions, and per a Supplemental Type Certificate (STC), nacelle fuel tanks were added aft of each engine.
Takeoff from KLUK was at 10:51 a.m. on a bright clear day. The pilot flew several closely spaced parallel survey tracks near Cincinnati, Ohio, at 9,500 feet before going north for data gathering near Dayton, Ohio. No flight plan needed, flying VFR taking photos along carefully prescribed GPS tracks for over four hours. Unexciting. Maybe even monotonous. At 3:02 p.m., N400JM called Columbus Approach requesting a progressive routing back to Lunken airport due to a fuel problem. “Say again?” replied ATC.
The “fuel problem” was clearly evident onboard because the left engine had coughed to a stop. The pilot repeated that N400JM wanted direct Lunken because of a fuel problem. Columbus Approach gave him a direct heading, then advised that another airport, Dayton-Wright Brothers Airport (KMGY), was 8 miles away. The pilot said he had KMGY in sight but rejected the offer, adding Lunken was “only 30 miles away.” Columbus Approach acknowledged the pilot’s intentions and offered to declare an emergency. The pilot demurred, pressing on.
NTSB analysis shows the pilot did not feather the left prop. Feathering greatly reduces drag and is the standard procedure after an engine shutdown on a twin-engine airplane. This failure increased drag on the left side, causing control and performance issues. The right engine still had fuel and continued running. Now talking to Cincinnati Approach, the pilot reported he did not require any assistance, there was “a fuel issue,” but it “should be okay.” ATC continued to point out other possible landing airports, but the pilot declined them all.
Cincinnati Approach handed the Navajo over to Lunken Tower when he was about 12 miles out. The pilot checked in, advising he “had a fuel problem and was hoping to make the field.” On an 8-mile final for Runway 21L, the controller cleared N400JM to land. The pilot reported he was unsure if he would make it to the airport. It was now 3:13 p.m., just 11 minutes after the first sign of trouble. Radar contact was lost on a 5-mile final. Unable to see or hear N400JB, ATC activated the crash phone, notifying emergency services.
Descending and slowing on a long straight-in final, the pilot seemed to change plans, giving up on the airport, turning to the right, lining up with the fairway of a golf course. The radar groundspeed went from 140 knots to 98 to 82. Multiple witnesses described unusual engine sounds. They saw the low-flying plane appear crooked, left wing down,“like a stunt in an airshow.”At3:16p.m., the plane rolled left and nose-dived into a neighborhood from a few hundred feet up. It impacted a tree and came to a stop in the backyard of a house. Within three more minutes, it was consumed by flames, a fire fed by the remaining onboard fuel.
After the accident, the NTSB found the low-altitude loss of control to be the defining event. It determined probable cause as “fuel starvation to the left engine and the resulting loss of engine power to that engine, and a loss of airplane control due to the pilot’s failure to maintain the minimum controllable airspeed.” It cited FAA guidance for multiengine aircraft that props should be feathered before rotation stops, warning “the net result of a windmilling propeller is the aircraft total drag exceeds the power available, thus the aircraft is no longer able to sustain level flight.”
The whole time the pilot was heading to Lunken, he was losing altitude and airspeed. Get too slow in any airplane, and you hurt glide performance, and slower still it may stall. But planes with engines on the wings have an additional threat — Vmc roll. With one engine at low or no thrust and the other one producing power, there are several forces acting to yaw and/or roll the plane. We counter these asymmetric forces with rudder and ailerons. However, get too slow, and there isn’t enough control authority. Eventually, the plane will suddenly roll over into the “dead” engine. Vmc with a feathered prop in the PA-31-350 is listed in the POH as 76 knots indicated airspeed. The last groundspeed recorded was 82 knots.
But why did the engine run out of fuel in the first place? The Board found a long history of problems and confusion, starting with the design of the after-market Nayak auxiliary fuel tanks. These are an FAA-approved STCmodification.A27-gallon tank is added in the nacelle behind each engine, using space that was originally a wing cargo locker. That’s a lot of gas. But there’s no fuel gauge for these tanks.
Piper Navajos were built with two fuel cells in each wing, a 56-gallon main inboard tank and a 40-gallon outboard tank. The engine can be selectively fed from either one or crossfed from the other wing’s fuel tanks. Each tank has a fuel quantity sensing unit, and there are two fuel gauges, left and right, that are switchable between inboard and outboard.
The modification piggybacks the added nacelle tank into the main inboard tank via a new fuel line, activated by selectable electric fuel pumps. Adjacent to the switches, there’s a placard: “DO NOT TRANSFER UNTIL MAIN TANK IS BELOW HALF FULL.” The POH notes it takes approximately 55 minutes to transfer all the fuel out of a nacelle tank. There is no fuel quantity sensing in the nacelle tank. There are no pressure or fuel flow sensors in this transfer line. A fact of this design is that it requires careful procedural steps and detailed fuel gauge/time logging to understand your total usable fuel. There are several worrying failure modes. What if there were a leak in the nacelle tank? What if a pump failed? How would you really know without sensors or indicators?
The NTSB found that the company gave no training and had no standardized procedures for fuel management. Different pilots had different techniques. The Safety Board alleged “company pilots had been instructed not to talk to investigators,“ but a few did. And what they said was eye-opening.
One pilot reported, “Sometimes you get on the ground and say, ‘Wow, I just flew six hours.’ Sometimes you get down and wonder how much fuel you do have left?!’” Another told the NTSB he quit the company the day before, as the boss had “just pushed him to his limit.” A third said he “always had a bad feeling…and he always felt they pushed their pilots to fly…and planes were not being maintained appropriately.” Pilots told the NTSB of at least three nacelle transfer pump failures in the company’s fleet within the last few months. Additionally, pilots were allegedly routinely pushed to do their own maintenance on aircraft.
A month before the accident, N400JM allegedly landed early after oil loss forced an in-flight engine shutdown. Asked about this and other incidents, the company CEO told NTSB investigators he was not aware of any maintenance issues or concerns with N400JM, and nobody had reported any problems at all with N400JM.
The last pilot to fly N400JM before the accident pilot flew it had a different tale, telling investigators it had a fuel leak. He had been flying the big Piper for a couple of weeks before a tornado hit his home, forcing the crew change. He shared a photo of blue 100LL fuel stains on a hangar floor after being left overnight. There was a plan to fill just the inboard or outboard tanks overnight to troubleshoot where the leak was coming from. He informed the company of the issue and heard of a plan to ferry the plane to the company base for maintenance.
According to family members, the accident pilot was aware of fuel issues with N400JM. After the crash, the pilot’s sister told a local TV station, “he had concerns…about the plane acting up.” A week before the crash, he told his brother-in-law, “there is a gas leak in this plane, and it’s bothering my sinuses.” The exact nature of any leak or pump failure is now impossible to determine. But forensic analysis shows there was, indeed, a problem.
At the crash site, no evidence of fuel was found in the left-wing inbound or outbound tanks. The NTSB states that in the left engine, “[essentially] no fuel was found during the examination and removal of fuel system components.” However, they do believe the left nacelle tank had fuel in it. “Given the extent of the fire damage to this area of the wreckage, and the witness report that the post-impact fire originated in this area, it is likely [the nacelle] tank contained fuel.” The nacelle fuel pump was too badly damaged to determine its pre-crash condition. The fuel system on the right side seemed nominal: “Fuel was found during the examination of the right engine fuel lines, injector lines and fuel pump.”
The final report notes the crossfeed valve was found closed but doesn’t address why the pilot didn’t open it to supply the left engine with fuel from the right side. The answer to this, and other questions, may have died with the pilot. There are so many links in the long accident chain. Ernest Hemingway succinctly described this kind of timeline in his 1926 novel The Sun Also Rises. Asked how do you go bankrupt, one character replied, “Gradually and then suddenly.”
Before the engine quit, there were years of pilot concerns about safe fuel management with the nacelle tanks. The company had months of fuel leaks, pump failures and alleged lack of maintenance action. After the engine quit, there were 14 minutes where the prop wasn’t feathered, and the fuel crossfeed wasn’t opened. There were 14 minutes of flying past suitable landing airports. And then, in a few seconds, a Vmc roll from 500 feet to the ground ended the chain. PP