Olney, TX, USA
N8517Q
AIR TRACTOR AT502
The pilot was conducting a cross-country business flight of the newly manufactured airplane when during initial climb the airplane’s thrust began to oscillate every 1-2 seconds between what the pilot perceived as idle and full engine power. He was unable to resolve the malfunction and a forced landing was made in a rough pasture. The airplane sustained substantial damage to the fuselage, wings, and empennage during the forced landing. The investigation determined that a combination of the propeller design and the airplane’s nose-high angle while on the ground likely resulted in trapped air in the propeller’s external piston cavity. Air is present in the piston cavity of new or overhauled propellers before installation, is introduced during maintenance when a propeller is separated from an engine, or replaces the oil that gradually drains from the external piston cavity while the engine is not operating. Trapped air in the external piston cavity can act as an unintended spring in the system and confuse the interaction between the propeller and governor, which can result in an undamped oscillation event. While the airplane is operated on the ground, it is unlikely that all trapped air is purged from the external piston cavity when the propeller is cycled between maximum speed and feather. The pilot stated that he cycled the propeller 3 times before takeoff. The airplane manufacturer’s test pilots typically cycle the propeller 4 to 5 times before takeoff in a newly-manufactured airplane. Although the airplane flight manual states to feather the propeller twice before takeoff, it does not address propeller surge behavior, how to avoid it, or how to respond when it occurs during flight. The airplane manufacturer’s chief test pilot stated they do not change the engine power lever in the event of a propeller surge; they slightly retard the propeller speed lever. Implementing a small, less than ¼ inch, aft movement of the propeller speed lever changes the propeller speed 10-20 RPM and stops the propeller surges. However, this slight reduction of propeller speed as a resolution to a propeller surge is not discussed in the airplane flight manual. The airplane manufacturer stated that although their flight test department has used a slight aft change of the propeller control lever to resolve propeller surge events, they have not conducted a formal evaluation to determine how or why this resolves the propeller surge event or if it could be implemented in an effective way that also ensures pilots do not inadvertently exceed the engine’s maximum torque limitation.
HISTORY OF FLIGHTOn August 29, 2019, about 1100 central daylight time, an Air Tractor AT-502A airplane, N8517Q, was substantially damaged when it was involved in an accident near Olney, Texas. The pilot sustained minor injuries. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 business flight. According to maintenance documentation, the airplane had accumulated about 1 hour before it received its airworthiness certificate on August 26, 2019. The pilot reported that the purpose of the flight was to fly the airplane from the Air Tractor factory at Olney Municipal Airport (ONY), Olney, Texas, to his company’s base-of-operations at Lane Airpark (T54), Rosenberg, Texas. He reviewed paperwork and the airplane logbooks for about an hour before he accepted airplane from the airplane manufacturer. The pilot did not observe any anomalies during his preflight inspection. The airplane was topped-off with fuel (234 gallons) before he arrived, and the fuel samples observed during the preflight inspection were free of water and particulate. The pilot stated that the airplane was equipped with a ferry tank system; however, the ferry tank (hopper) did not contain any fuel. The pilot reported that the engine started normally and that he did not observe any anomalies on the engine gauges. He completed a brake check and cycled the propeller 3 times while he taxied to runway 17. The engine startup, taxi, and takeoff were conducted with the fuel selector valve positioned to use fuel from the main tank. Before takeoff, the pilot locked the tailwheel and increased engine power to 1,500 foot-pounds (ft-lbs) of torque as he held the brakes. He then increased engine power to 2,200-2,300 ft-lbs of torque during the takeoff roll and completed a normal liftoff at 95 to 100 knots. After liftoff, he maintained a higher-than-normal climb angle to quickly climb to a safe altitude above the runway; however, about 200 ft above ground level and halfway down the 5,101-ft-long runway, the engine began to surge every 1 to 2 seconds between what he perceived as idle and full power. He described the sensation as alternating between being pushed back into the seat and being thrown forward until the seat belt restrained any further motion. The pilot stated there was insufficient runway remaining to ensure a safe landing, so he entered a left turn to land on runway 31. The pilot reported that during the turn he saw an amber annunciator panel light illuminated, which he believed to be the fuel filter warning light. The airplane was too low and slow to complete the turn to runway 31 without entering an aerodynamic stall, so the pilot leveled the wings and made a forced landing in a rough pasture southeast of runway 31. The airplane landed hard on the left main landing gear before it collided with a tree stump, and then swerved to the left and down into a ditch. After the accident, the pilot moved the fuel condition lever to the cutoff position and evacuated the airplane through the left door. The airplane manufacturer’s chief test pilot witnessed the airplane’s engine start, taxi, and takeoff. He heard the airplane’s propeller surging during the initial climb after liftoff. The propeller surge continued as the airplane climbed to about 200 ft agl where it began to decelerate, descend, and increase pitch attitude. The airplane disappeared behind an airport hangar while in a gradual descent. When the airplane did not reemerge from behind the hangar, he immediately drove his truck to the accident site to render assistance. After ensuring that the pilot was safe, another individual climbed into the airplane and turned-off the master switch, confirmed that the fuel condition lever was in the fuel cutoff position (the pilot had already accomplished this task before he evacuated the airplane), and removed the fire extinguisher from airplane. WRECKAGE AND IMPACT INFORMATIONThe accident site was in a pasture about 850 ft southeast of runway 31 at ONY. The distribution and orientation of the wreckage debris and ground impact marks were consistent with the airplane impacting the ground on a northeast heading. The airplane’s left wing impacted a tree and the airplane crossed through a drainage ditch before it came to a stop facing west and in an upright, left wing down attitude. The outboard portion of the left wing separated when it impacted the tree. The left aileron separated from the wing and was found on the ground near the main wreckage. The left flap remained attached to the wing attachment points. The right wing remained attached to the fuselage, with the impact related damage to the wingtip. The right flap and aileron remained attached to the wing. The aft fuselage with empennage separated during the impact sequence. The horizontal and vertical stabilizers, elevator, and rudder exhibited various impact related damage. All airframe structural components and flight control surfaces were identified at the accident site. Flight control continuity for the right aileron was established from the cockpit stick to the control surface. Flight control continuity for the left aileron, elevator, and rudder could not be established due to impact related damage; however, all observed separations were consistent with overstress. The flap actuator was fully retracted, and the flap drooping system remained intact with impact related damage. Examination of the airframe fuel system did not reveal any preimpact anomalies that would have resulted in a clogged fuel filter amber warning light that the pilot reportedly observed during the flight. The airframe fuel filter housing was full of fuel with some small particles, but the filter element was free of substantial blockage. The fuel selector valve was in the main tank position and the fuel condition lever was in the fuel cutoff position. The engine remained attached to the airframe mount ring. The engine control cables and pushrods were continuous but were deformed or bent, which prevented their movement using the cockpit controls. All engine modules were bent and/or fractured. The reduction gear box was fractured circumferentially through 360°. The 1st stage compressor and power turbine blades were intact and undamaged; however, the blades were locked against the deformed outer shroud and prevented rotation of the 1st stage compressor and power turbine. Bench testing and/or disassembly of the fuel control unit, constant speed unit (propeller governor), high-pressure fuel pump, airframe fuel hoses, fuel-to-oil heat exchanger, and low-pressure electrical and mechanical fuel pumps did not reveal any evidence of a preimpact mechanical failure. The propeller separated from the engine propeller shaft during impact. The propeller exhibited cylinder and piston impact marks that were consistent with the propeller operating in the normal operating blade angle range at initial impact. There was no evidence that the propeller was feathered or in the beta/reverse range at impact. Propeller blade damage including camber side chordwise/rotational abrasion, aft bending, bending opposite direction of rotation, leading edge gouging with material deformation toward low pitch, and progressive compound bending/twisting toward low pitch were all consistent with a shallow aircraft impact angle with the propeller rotating at a low power setting. Additionally, there were no conditions identified that would result in beta system binding on the newly manufactured propeller with only 1 hour of operation and no prolonged exposure to moisture and/or products commonly used in agricultural operations. Postaccident examination and/or testing of the engine and its accessories, engine controls, propeller, and airframe fuel system did not reveal any mechanical failures that would have precluded normal operation during the flight. TESTS AND RESEARCHPropeller Surge The maximum acceleration and deceleration of the engine’s gas generator is about 4 to 5 seconds and 2 to 3 seconds, respectively, and the average full engine output cycle is about 7 seconds. The pilot stated that the engine thrust fluctuated every 1 to 2 seconds during the flight, which is not consistent with an engine power fluctuation. The propeller blade angles change at a quicker rate, and manufacturer testing after the accident revealed that the propeller had an average cycling time of about 1.2 seconds, which is consistent with the pilot’s recollection of the thrust oscillation event during the flight. The speed of the propeller is limited to either a pilot-selected lower speed or the maximum set speed by propeller governor of 100%. In the event of a failure of the propeller governor, the overspeed governor serves as an emergency safety device and limits the maximum speed of the propeller to about 106% of the maximum propeller governor speed. The known propeller surge events do not exceed the maximum 100% set limit of the propeller governor and, thus, the overspeed governor did not contribute to the thrust oscillation condition. An interview with the airplane manufacturer’s chief test pilot revealed that the pilots in the flight test department are aware of a condition with some of their Air Tractor airplane models that is commonly referred to as propeller surge. Most of their surge events occurred on airplane models equipped with a smaller Pratt & Whitney PT6 engine (PT6-15AG, -27AG, or -34AG). The chief test pilot estimated that about 1-in-15 of these airplanes had a propeller surge event during their initial flight after manufacture. He stated that there was no clear pattern between the airplanes that demonstrate propeller surge and those that do not; however, the factory test pilots have observed a higher frequency of propeller surges in airplanes that have not been operated for a month or longer. The chief test pilot stated that his predecessor told him not to change the engine power lever in the event of a propeller surge, but to slightly retard the propeller speed lever instead. According to the chief test pilot, implementing a small, less than ¼ inch, aft movement of the propeller speed lever changes the propeller speed by 10 to 20 rpm and stops the propeller surge. The chief test pilot offered two possible explanations for why a small reduction in propeller speed would stop a propeller surge: the cycling process introduces hot oil into the propeller, freeing the internal propeller mechanisms and allowing it to respond quicker to changes in oil control pressure; or the small reduction of propeller speed purges any trapped air from the propeller cavity. The chief test pilot further stated that all company pilots have been trained to cycle the propeller multiple times (4-5 times) before takeoff in a newly manufactured airplane. Full propeller cycles during ground operations are made by increasing the propeller to maximum speed and then use the propeller control lever to select feather; however, the propeller is not cycled through the reverse-thrust (beta) position during this cycling process. The investigation did not reveal any mechanical anomalies with the engine or propeller, and a review of the propeller design and its installation on the Air Tractor AT-502A was completed. The review determined that because the airplane has a nose-high pitch attitude while on the ground, air could be trapped in the propeller’s external piston cavity. Air is present in the piston cavity of new or overhauled propellers before installation, is introduced during maintenance when a propeller is separated from an engine, or replaces the oil that gradually drains from the external piston cavity while the engine is not operating. Trapped air in the external piston cavity can act as an unintended spring in the system and confuse the interaction between the propeller and governor, which can result in an undamped oscillation event. While the airplane is operated on the ground in a nose-high pitch attitude, it is unlikely that all trapped air is purged from the external piston cavity when the propeller is cycled between maximum speed and feather; the propeller must be cycled into the reverse-thrust (beta) position to displace the greatest amount of oil/air from the external piston cavity. Additionally, horizontal installations of the engine/propeller on airplanes with tricycle landing gear do not have the propeller surge events that have been reported with the Air Tractor airplanes with a conventional landing gear and a nose-high angle while on the ground. The airplane flight manual states to feather the propeller twice before takeoff; however, it does not address propeller surge behavior, how to avoid it, or how to respond when it occurs during flight. The airplane manufacturer stated that although their flight test department has used a slight aft change of the propeller control lever to resolve propeller surge events, they have not conducted a formal evaluation to determine how or why this resolves the propeller surge event or if it could be implemented in an effective way that also ensures pilots do not inadvertently exceed the engine’s maximum torque limitation.
A loss of thrust control due to a propeller surge event, which resulted in a forced landing shortly after takeoff.
Source: NTSB Aviation Accident Database
Aviation Accidents App
In-Depth Access to Aviation Accident Reports
