Get extra lift from AOPA. Start your free membership trial today! Click here

Frozen fuel

Why some turbines require icing inhibitors

First observations about the March 22, 2009, crash of a Pilatus PC–12/45 at the Bert Mooney Airport in Butte, Montana, centered on the number of fatalities. There were 14 aboard, and all died. The airplane had 10 seats; the flight plan listed nine. The crash occurred on short final to Runway 33, less than a half-mile from the runway threshold.
Turbine Accident Review
Zoomed image
Illustration by Alex Williamson

The pilot was a 65-year-old, 8,800-hour ATP with more than 1,700 hours in type. The rest of the victims included three couples and their seven children, aged 1 to 9.

Witnesses who saw the Pilatus on short final reported that it appeared to be too high, and that it then began turning in a steepening bank to the left at about 300 feet agl. The bank angle increased, the nose dropped, and the airplane crashed into a cemetery and burst into flames. The wreckage was extensively fragmented and largely consumed by fire, but investigators determined the fire damage was most severe along the left
side. Furthermore, the yaw damper and aileron trim—both jackscrew-driven systems not susceptible to displacement by impact forces—were at their extreme settings for nose left and right wing down.

Turbine Accident ReviewThe extent of the destruction initially made it seem unlikely that the cause would ever be determined, but the National Transportation Safety Board’s aviation investigators are resourceful. A sub-board of the airplane’s central advisory and warning system (CAWS), itself too badly damaged to be read by its manufacturer, yielded viable flash memory chips that had stored a series of event and caution codes spanning some 480 flights and ending moments before the accident. The chips helped to unlock some of the secrets behind the crash. Most of the clues pertained to the airplane’s fuel system, and the way it balances fuel loads.

The PC–12/45’s fuel system is designed to automatically balance the fuel load between each wing tank. It does this by means of fuel boost pumps installed in each wing’s fuel collector tank. If one tank’s fuel load exceeds the other by more than 70 pounds (about 10.5 gallons), the heavier tank’s boost pump automatically runs until the imbalance is corrected. Each pump activation is annunciated, and logged by the CAWS. Fuel boost pumps also will come on automatically if fuel pressure drops below 2.0 psi. The boost pumps can be turned on manually if necessary.

If the imbalance increases beyond 270 pounds, the fuel pump shuts down to protect against fuel exhaustion in the event of a malfunction of the quantity sensors, or a fuel leak.

The Limitations section of the PC–12/45’s Airplane Flight Manual (AFM) defines 178 pounds (about 26.7 gallons) as the maximum permissible fuel imbalance, and requires landing as soon as practical if it cannot be remedied.

The Pilatus had departed its base at the Redlands, California, Municipal Airport at 7:42 a.m., bound for the Nut Tree Airport in Vacaville, California. The airplane had been fueled the previous day with 222 gallons of Jet-A. The flight to Vacaville took two hours and 18 minutes. The fuel boost pumps in both tanks began cycling about one and a half hours into the flight. About 15 minutes later, the left pump came on continuously and the right pump stopped. The airplane landed at Vacaville about 30 minutes later, where 128 gallons of Jet-A were loaded from the self-serve pump station, and four adults and five children boarded the airplane. The Vacaville self-serve Jet-A was not premixed with an icing inhibitor, such as Prist, and there was no evidence that the pilot added any himself. A fuel receipt from Redlands confirmed the line manager’s recollection that none was supplied there, either. The PC–12’s AFM specifically requires that an icing inhibitor be used on all flights in subfreezing temperatures, noting that ice accumulation in the filters can obstruct fuel flow to the point of starving the engine.

During the next 13-minute positioning leg to the Oroville Municipal Airport, the boost pumps did not activate. Four more passengers—two adults and two children—boarded at Oroville, and at 11:10 a.m. the airplane departed on an IFR flight plan to Bozeman, Montana, where the families planned a ski vacation.

One of the airplane’s co-owners later told investigators that they never worried about weight and balance; “there were just not enough seat belts.”One of the airplane’s co-owners later told investigators that they never worried about weight and balance; “there were just not enough seat belts.” Their solution was to put some of the children “on the floor to sleep.” In fact, the Pilatus was 572 pounds overweight on that takeoff, although its center of gravity was within limits.

Flying into Montana in March at FL 250, subfreezing temperatures could be anticipated. Data recovered from the engine information system indicated an outside air temperature of minus 40 degrees Celsius in cruise during the accident flight.

Twenty-two minutes after takeoff, the right boost pump ran for three minutes and 45 seconds, a typical duration for load-balancing. Both pumps remained off for the 48 minutes that followed.

One hour and 13 minutes into the flight, both boost pumps began to cycle in response to persistent indications of low fuel pressure. Five minutes later, the automatic balancing system switched the left pump to continuous operation. It continued to run for the next 34 minutes, 20 minutes beyond the time the imbalance surpassed 270 pounds. Meanwhile, the right pump resumed cycling. The combination suggests that although the pilot had switched on the left pump manually, its output was no longer sufficient to maintain fuel pressure. This implied that the fuel in the left tank was freezing into a slushy mixture that couldn’t pass through to the engine.

Worse, the higher output pressure from the right tank would have suppressed the weaker flow from the left side, even as the engine returned surplus fuel to both tanks. One hour and 21 minutes into the flight, the left tank actually began refilling, while the right was drained at double the normal rate. By the time the pilot requested a diversion to Butte, NTSB calculations showed that the fuel imbalance had reached 732 pounds (about 110 gallons), more than four times the action limit set in the AFM—and the pilot still hadn’t chosen the nearest suitable airport, even though the aileron forces needed to keep the wings level must have been increasingly noticeable. Three other airports were closer than Butte, which at this point was more than 100 nautical miles ahead. Challis, Idaho’s airport was a mere five minutes away, and almost exactly on the current course.

Twelve miles from Butte, the R FUEL LOW annunciator illuminated, indicating that fewer than 133 pounds (about 20 gallons) of usable fuel remained in that tank. The pilot activated the right boost pump. By the time of the last radar hit two miles southwest of the airport and 3,600 feet above the runway threshold, the right tank held less than 10 minutes of usable fuel and the imbalance was more than 1,300 pounds—requiring more than 50 percent aileron deflection with full right-wing-down trim to keep the wings level.

The NTSB interpreted the position of the yaw damper as suggesting the pilot tried a forward slip after coming in high, but couldn’t lose enough altitude. With terrain allowing no room to make a right turn in that direction, the only alternative to a landing overrun was to bank toward the heavy wing—at low altitude, at a low airspeed, and with the airplane uncoordinated. An uncontrollable roll and stall ensued, with fatal consequences.

The instructor who had given the pilot his most recent recurrent training described his competence in the airplane as “very high” and his professional judgment as “superb.”

But in its findings, however, the NTSB cited the pilot’s “(1) ... failure to ensure that a fuel system icing inhibitor was added to the fuel before the flights on the day of the accident; (2)...failure to take appropriate remedial actions after a low fuel pressure state (resulting from icing within the fuel system) and a lateral fuel imbalance developed, including diverting to a suitable airport before the fuel imbalance became extreme; and (3) a loss of control while...maneuvering the left-wing-heavy airplane near the approach end of the runway” as probable causes of the accident.

The details of the accident suggest the pilot had come to view the dictates of the AFM as mere suggestions.

They weren’t.

ASI Staff
David Jack Kenny
David Jack Kenny is a freelance aviation writer.

Related Articles