NTSB Factual Report

HISTORY OF FLIGHT

On August 10, 2010, about 0215 eastern daylight time, a Piper PA-32R-301, N220ST, operated by Coastal Aviation, was substantially damaged when it impacted trees while making a forced landing following a loss of engine power during cruise flight near Washington, Georgia. The certificated commercial pilot was fatally injured. Visual meteorological conditions (VMC) prevailed for the flight that departed Madison County Executive Airport (MDQ), Huntsville, Alabama, destined for Mt. Pleasant Regional Airport (LRO), Mount Pleasant, South Carolina. An instrument flight rules (IFR) flight plan was filed for the positioning flight conducted under 14 Code of Federal Regulations (CFR) Part 91.

On the morning of August 9, 2010 after receiving a call from a freight forwarding company, the operator's general manager (who was also the accident pilot) flew the accident airplane to Charleston International Airport (CHS), Charleston, South Carolina to conduct a 14 CFR Part 135 air taxi flight. After loading approximately 400 pounds of freight, at approximately 1207 the pilot departed for MDQ. After arriving at 1414, and off-loading the freight, the pilot attempted to depart; but, the airplane would not start. After a mechanic replaced the starter, the pilot departed MDQ for LRO the following morning at 0048.

According to air traffic control (ATC) information provided by the Federal Aviation Administration (FAA), during cruise flight at 7,000 feet above mean sea level, the pilot reported a loss of engine power. He was then radar vectored by ATC towards the closest airport which was Washington-Wilkes County Airport (IIY), Washington, Georgia. ATC continued to monitor the airplanes flight path as it descended. Approximately 3 nautical miles southeast of IIY, radar contact was lost as the airplane descended below the radar coverage area.

PERSONNEL INFORMATION

According to FAA records, the pilot held a commercial pilot certificate with multiple ratings including airplane single-engine land, and instrument airplane. His most recent FAA second-class medical certificate was issued on June 30, 2010. On that date, he reported that he had accrued 2,600 total hours of flight experience.

The pilot had at one time been listed in the operator's FAA approved Part 135 operations specifications when the operator had been operating Part 135 with multiple pilots. According to the FAA, at the time of the accident however, the operator was only authorized to operate its Part 135 flights with one pilot. The pilot of the accident airplane was not that authorized individual.

AIRCRAFT INFORMATION

The accident aircraft was a low wing, unpressurized, six seat, single-engine monoplane of conventional stressed skin construction. It was powered by a 300 horsepower, Lycoming IO-540-K1G5 fuel injected engine, equipped with a Hartzell HC-13YR-1, controllable pitch, 3- bladed propeller.

According to FAA and airplane maintenance records, the accident airplane was manufactured in 2004. An engine overhaul had been completed on July 22, 2010. At the time of accident, the engine had accrued 18 hours since the overhaul had been completed.

The airplane’s most recent annual inspection was completed on July 30, 2010, and at the time of the inspection, the airplane had accrued 1959.5 total hours of operation.

METEOROLOGICAL INFORMATION

The recorded weather at Greene County Airport (3J7), Greensboro, Georgia, approximately 19 nautical miles southwest of the accident site, at 0220, included: wind 100 degrees at 3 knots, visibility 10 miles, sky clear, temperature 26 degrees C, dew point 21 degrees C, and an altimeter setting of 29.99 inches of mercury.

WRECKAGE AND IMPACT INFORMATION

The wreckage was located in a heavily wooded area approximately 1.6 miles southeast of IIY.

All major components of the airplane were located at the accident site. After impacting trees, the airplane came to rest inverted on a magnetic heading of 350 degrees. A 100-foot long debris field, oriented on a magnetic heading of 302 degrees was present. Multiple portions of broken tree limbs were spread throughout the area. Evidence of a flash fire was present, and the majority of the cabin area had been consumed by a post crash fire.

Both the landing gear and the flaps were in the retracted position. Both fuel caps were found to be closed and locked. The fuel selector valve was in the right tank position and the valve contained approximately a teaspoon of fuel. The mixture control was in the idle cutoff position.

Both wings were fragmented into multiple pieces, exhibited fracturing and various degrees of crush damage. The wing flaps and ailerons exhibited multiple breaks and separations and were spread throughout the debris field.

The empennage was separated from the remains of the aft fuselage. The rudder panel remained attached to its fittings on the vertical stabilizer and the left and right portions of the stabilator exhibited fracturing and crush damage.

The pitch trim jackscrew was found in the full nose-up position, however; no preimpact failures or disconnects of the primary flight control system were discovered and control continuity was established from the ailerons, stabilator, and rudder to the broken ends of the control cables, which exhibited evidence of tensile overload.

Examination of the 3-bladed constant speed propeller revealed that there was no evidence that it had been rotating when the accident occurred. One blade was bent back approximately 90 degrees at the mid span position, one blade was straight, and the third blade exhibited slight aft bending toward the tip. Further examination also revealed that the propeller governor was in the high pitch position.

Examination of the engine revealed that there was no evidence of any blockage of the intake system or exhaust system, the crankshaft could be rotated, and continuity of the valve train, and crankshaft were confirmed. All of the spark plugs electrodes were intact and light gray in color, and thumb compression was noted on all six cylinders. All six cylinders were also examined internally with a lighted borescope, and no anomalies were observed. Both magnetos were fire damaged, but exhibited no internal or external evidence of any preimpact malfunction. The oil suction and oil pressure screens were absent of debris, and oil was noted in the rocker boxes.

External examination of the fuel injection system revealed no indication of any preimpact failure or malfunction of the fuel nozzles or fuel flow divider. However, during internal examination of the fuel servo, it was discovered that the stud which attached to the hub in the diaphragm assembly was fractured.

MEDICAL AND PATHOLOGICAL INFORMATION

An autopsy was performed on the pilot by the Georgia Bureau of Investigation's Division of Forensic Sciences. The cause of death was smoke, soot, and super-heated gas inhalation and multiple blunt force injuries.

Toxicological testing of the pilot was conducted at the FAA Bioaeronautical Sciences Research Laboratory, Oklahoma City, Oklahoma. The specimens were negative for carbon monoxide, cyanide, basic, acidic, and neutral drugs.

TESTS AND RESEARCH

Fuel Servo

According to airplane maintenance records during the engine overhaul which was completed on July 22, 2010, the fuel servo was shipped to AVStar Fuel Systems Inc. where the unit was overhauled using AVStar components in accordance with FAA approved data from the manufacturer's Component Maintenance Manual.

According to the engine manufacturer, the fuel injection system was a mass-air flow type fuel injection system. The fuel servo as part of that system would sense the amount of air moving past the throttle by use of a venturi. A diaphragm in the fuel servo would meter the amount of fuel supplied to the flow divider by comparing the air pressure from the venturi. The flow divider would then divide the metered fuel among the individual cylinder fuel nozzles.

Examination of fuel servo technical specifications revealed that the diaphragm utilized the hub stud to mechanically communicate its position to the system and that if the hub stud became disconnected; the fuel servo would be unable to properly meter fuel which could result in a loss of power.

Materials Laboratory Examination

Examination of the fractured stud by the NTSB's Materials Laboratory revealed that hub stud was fractured where it was threaded into the hub. The hub retained two aluminum plates that had originally clamped to the diaphragm in the fuel injector servo, and the hub stud retained a spring holder on its outer end fixed by two nuts. Some discoloration of the components was apparent, consistent with the effects of the post-crash fire.

Examination of the mating fracture surfaces of the hub stud revealed that the fracture surfaces had a nondescript appearance consistent with fatigue cracking, and the fractures displayed diametrically opposed ratchet marks indicating that the fatigue cracking originated at multiple locations consistent with reversed bending loads. There was no indication of any intergranular fracture at the origin areas of the fatigue cracking around the perimeter of the hub stud. The fatigue cracking propagated across an estimated 98 percent of the surface, leaving a region approximately 0.01 inch by 0.002 inch in size characterized by ductile dimple features which indicated overstress separation.

Exemplars

At the request of the Investigator-In-Charge, the fuel servo manufacturer provided an exemplar diaphragm assembly (hub stud threaded into hub and brazed) from the same lot as the assembly from the accident. The manufacturer also provided three exemplar hubs and three exemplar hub studs for comparison. The exemplar hubs were from the same lot as the hub in the accident, but the exemplar hub studs came from a different lot than the hub stud in the accident.

A fuel injector servo from another accident (ERA11LA015) was also disassembled to examine the diaphragm assembly. The hub stud and the hub in that assembly were intact, as was the brazed joint, and no blockage or other damage to the fuel servo was observed. This diaphragm assembly was also used as an exemplar for comparison with the parts from the accident airplane.

Materials

The material used for both the hub and the hub stud was identified as American Iron and Steel Institute (AISI) 416 or the equivalent Unified Number System (UNS) S41600 which is a martensitic stainless steel with added sulfur to improve machinability. The chemical composition of the hub stud was explored using X-ray energy dispersion spectroscopy (EDS) in a scanning electron microscope (SEM). Two spectra were done; one from the center of the fracture surface and one after cutting a section from the opposite end of the stud to test its hardness The EDS spectra were generally consistent with the specification, with major peaks for iron and smaller peaks for chromium, silicon, sulfur and manganese. The EDS spectrum from the fracture surface showed peaks for zinc not present in the spectrum from the cut end, and a higher peak for oxygen.

Assembly and Brazing

During assembly, the hub stud would be threaded into the hub, and then the assembly would be locked in place by subsequently brazing the two components using American Materials Society (AMS) 4765 silver braze, with the brazing process carried out in a hydrogen furnace.

AMS 4765 has a composition of 56 percent silver, 42 percent copper and 2 percent nickel, with a solidus temperature of 771 degrees F and a liquidus temperature of 893 degrees F. Records indicated that the braze material was supplied in powder form for this process, and the braze suppliers certification document indicated that the braze was suitable for atmosphere furnace brazing, “but used where zinc fumes in the furnace are not permissible.”

An EDS spectrum of a sample of the braze material that had remained adhered to the hub stud revealed that the spectrum was consistent with the braze composition, with a major peak for silver and a large peak for copper, superimposed on the type 416 steel spectrum; a small peak for nickel was also detected. This spectrum also showed the presence of zinc.

Geometry

In general, the parts conformed to the drawings; however, there was a discrepancy between the drawing for the hub stud as compared to the part from the accident and the exemplars provided by the manufacturer. On the hub end of the hub stud, the threads terminated in a groove perpendicular to the axis of the stud. This groove had a depth and profile similar to the #0-80 UNF – 3A thread profile called out for the stud threads. This groove did not appear on the part drawing. The hub stud involved in the accident fractured at approximately the middle of this groove on a plane nearly perpendicular to the axis of the hub stud, at the position of the minimum diameter. All three of the exemplar hub studs had this groove, as did the part from the accident and the exemplar assembly from the same lot as the accident. The hub-end threads on the hub stud from accident ERA11LA015 terminated without a groove.

A second discrepancy was also observed when the hubs were sectioned. The part drawing for the hub called out the threads for insertion of the hub stud to be #0-80 UNF – 3B, which indicated a specific internal thread profile and tolerance. Sectioning of the hubs showed that the thread-cutting operation left remnants of the hub material at the crown of each thread, so that the thread profile did not conform to the specification.

Braze Joint

The drawing for the assembly of the hub stud and hub indicated that there should be a fillet braze all around at the corner between the hub stud and the surface of the hub perpendicular to the hub stud. On the drawing of the assembly, there was a note on the callout for the braze stating that the fillet formed by the braze material was not to exceed 0.050 inch over the hub and hub stud. The assembly drawing also indicated that the brazing process should follow AMS 26644, which states that “Examination of all visible joint edges shall show presence of brazing filler metal for 100% of each joint.”

The braze process documented by the manufacturer's vendor indicated that joining was accomplished in a belt furnace at 1900 degrees F, in a hydrogen atmosphere, with a post-braze heat treatment of the assembly performed in a belt furnace at 1500 degrees F, also in a hydrogen atmosphere. The braze joints were to be inspected between the joining and post-braze heat treatment steps, and final hardness was to be checked after the final heat treatment step. The quality of the joint was to be assured by application of a counterclockwise torque of 22 to 24 inch-ounces to ensure that the hub stud would not unscrew from the hub. Examination of the brazing revealed however, that the braze material did not extend up to the surface of the hub, which indicated that the braze material did not fill the gap and contact the shoulder at the end of the hub stud. The braze material also did not have a visible fillet at the corner between the hub stud and the hub.

Examination of an exemplar assembly from the same lot as the accident assembly, also revealed that its braze fillet was only well defined on one side of the hub stud, and there was a visible gap on the other side of the hub stud. Additionally, a well defined fillet was not present.

ADDITIONAL INFORMATION

In order to increase safety the parties to the investigation took the following actions:

AVStar Fuel Systems

AVStar Fuel Systems issued Mandatory Service Bulletin AFS-SB6 requiring affected fuel servos that were either overhauled by AVStar or contained an affected AVStar fuel diaphragm assembly installed by another facility to be removed from the aircraft and returned to AVStar for immediate inspection and repair.

Lycoming Engines

Lycoming issued Mandatory Service Bulletin No. 596 which contained a table of affected Lycoming engines and a reprint of AVStar Fuel Systems Mandatory Service Bulletin AFS-SB6.

Piper Aircraft

Piper Aircraft issued Vendor Service Publication VSP-211 which contained a reprint of Lycoming's Mandatory Service Bulletin No.

NTSB Probable Cause

The manufacturer's inadequate quality control and improper manufacture of the fuel servo diaphragm assembly, which resulted in fatigue cracking of the hub stud and subsequent loss of engine power due to fuel starvation.