Accident Overview
On July 25, 2011, at 1417 eastern daylight time, a WSK PZL Mielec M18A airplane, registration N92640, collided with terrain near Anderson, Indiana, during an aerial application flight following a loss of engine power. The airline transport rated pilot was not injured. The airplane sustained substantial damage. The flight was operated by Forest Air Services under 14 Code of Federal Regulations Part 137 without a flight plan. The local flight originated from Marion, Indiana, about 1335.
Pilot Actions
The pilot stated he was beginning to mark his second field when the accident occurred. As he started the second marking pass, he experienced what he thought was an engine failure. He reported the engine was still running but not producing power. He turned away from a nearby town and a corn field, locating a bean field for landing. The pilot pushed the power lever forward, but the engine power did not increase; instead, there was an increase in engine temperature. He noted the engine was idling fast and forced the airplane onto the ground about 150 to 200 yards before a tree line. The pilot performed a ground loop prior to contacting the trees.
Propeller Inspection
The airplane was equipped with a Hartzell 5-bladed, hydraulically operated, constant speed propeller, model number HC-B5MP-5BL, hub serial number EVA3036. The blades showed varying degrees of impact damage. Impact marks between the link arms and the bottom of the blade counterweights indicated the blades were not in either the feather or reverse position at impact. A postaccident inspection of the propeller did not reveal any failure or malfunction that would have precluded normal operation.
Engine and Component Testing
The engine, a Honeywell model TPE331-11U-612G, serial number P44089C, was last installed on the accident airplane on October 28, 2009. It was shipped to Honeywell for a teardown inspection under Federal Aviation Administration supervision. The teardown inspection did not reveal any failure or malfunction that would have precluded normal operation. The fuel control, fuel pump, propeller governor, fuel nozzles, and fuel manifold were removed for further testing. All components tested normally except the propeller governor, which showed the minimum stop adjustment was out of limits. The governor was delivering sufficient oil volume and pressure.
Propeller Governor Examination
The propeller governor, Woodward part number 897410-16, serial number 1758842, was examined at Woodward. A bench test revealed the maximum speed was 44 rpm above the nominal limit at the maximum control lever position of 112 degrees, and the minimum speed was 410 rpm below the nominal limit. The maximum control lever position, which should have been 84 +/- 5 degrees, was set at 112 degrees. The maximum speed to minimum speed control lever position was set at 30 degrees, whereas it should have been 10.2 degrees. A non-Woodward lockwire was present on the control lever, indicating adjustments had been made to the control lever position.
The governor was x-rayed, and the x-ray appeared to show a fracture of the speeder spring, part number 1522-666. Two of the spring coils were in contact with each other near the spring seat. Disassembly revealed the speeder spring was fractured at a point 1-1/2 to 1-3/4 coils from the end of the spring.
Speeder Spring Failure
A metallurgical examination of the speeder spring determined the fracture was a result of a fatigue crack that propagated through the fracture surface. Contact marks were visible between the coils. The depth and wear of marks on the contacting surfaces of the coil adjacent to the fracture indicated the spring continued to function after the fracture was initiated. The surface finish of the spring adjacent to the fracture origins contained pitted areas. Maintenance records show the propeller governor was overhauled on May 19, 2009. Speeder springs do not contain serial numbers, so the history of the fractured spring could not be determined.
Further testing on an exemplar spring was performed by Woodward to determine the change in load rating at a condition where wear was occurring before and after the fracture. With an applied load of 8 pounds, the spring length was 1.1842 inches before a simulated cut and 1.1657 inches after; with a 16-pound load, the lengths were 1.0707 inches and 1.0642 inches, respectively. It was calculated that, during the time the crack was present but before complete failure, the resulting governor speed output would have been 244 rpm (6.5 percent) lower than with a non-cracked spring, with an additional 2.7 percent reduction in speed.