HISTORY OF FLIGHTOn December 8, 2023, about 1329 Pacific standard time, an experimental amateur-built Series 5 Kitfox, N66180, was destroyed when it was involved in an accident near Eloy, Arizona. The pilot was fatally injured. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 personal flight. The pilot had flown from his home base of Ryan Field (RYN), Tucson, Arizona, earlier in the day to have lunch at the Eloy Municipal Airport (E60) restaurant. The pilot told a friend, who flew alongside him on the inbound flight in another airplane, that the flight was uneventful. The accident flight was to be the return leg to RYN. ADS-B data revealed that after takeoff from runway 20, the airplane climbed to about 2,450 ft mean sea level (900 ft above ground level) before making a left turn to the south. About that time, a witness just below the airplane’s flight path heard an engine “sputter” and then go quiet. He was not concerned with the lack of engine noise, because he assumed they were practicing engine-out procedures into the local field across from his house. He looked up and watched as white pieces fell from the airplane. The airplane continued to maintain a level flight attitude and he looked away. When he looked up again, he could see the airplane was rapidly descending in a nose-down attitude. It then struck the ground in the field across from his house and burst into flames. The airplane was not trailing smoke or vapors at any time before impact. AIRCRAFT INFORMATIONConstruction of the airplane was completed in 2004. At that time, it was equipped with a six-cylinder Corvair automobile engine and a composite propeller manufactured by Warp Drive Propellers. In 2015, the engine was replaced with a four-cylinder 125 horsepower Lycoming O-290-G engine; at that time, according to the airplane’s maintenance logbooks, it underwent 5 hours of Phase II flight testing. The pilot, who was also a licensed FAA Airframe and Powerplant mechanic, purchased the airplane in December 2019; by that time it had accrued about 200 hours of flight time. Over the next two years, the airplane accumulated a further 200 hours and underwent a series of engine changes. By September 28, 2022, the pilot had reinstalled the original O-290-G engine, which had now been overhauled and upgraded to 160 horsepower through the installation of cylinder assemblies designed for the Lycoming O-320 engine. A month before the accident, during the airplane’s most recent condition inspection, the pilot replaced the propeller (a Catto Propellers fixed-pitch 76-inch two-blade model), with a three-blade, composite, ground-adjustable propeller manufactured by NR Prop. There was no evidence in the logbooks to show that the airplane had gone through a flight test after the installation of the upgraded engine and new propeller. Correspondence between the pilot and representatives from NR Prop indicated that the two-blade propeller configuration had been tested with Yamaha engines, which require a PSRU, up to 170 horsepower; however, NR Prop had no data for installation on a Lycoming engine. The pilot purchased the three-blade adjustable pitch SR-118 assembly, with a diameter of 2,024 mm, in addition to a two-blade hub that was also compatible with the blades. The SR-118-2000 is advertised on the NR Prop website for installation on engines between 90 and 150 horsepower. According to online forum correspondence, the pilot stated that when the propellor arrived it did not include installation instructions and he did not know the mounting hardware torque specifications. Review of subsequent correspondence between the pilot and the manufacturer indicated that he then received torque values for the propeller hub, but there was no evidence he received formal installation or ongoing maintenance instructions. A friend of the pilot stated that after installation, the engine would not reach its rated rpm. The pilot therefore decided to switch to the two-blade configuration, using the two-blade hub and two of the three blades. AIRPORT INFORMATIONConstruction of the airplane was completed in 2004. At that time, it was equipped with a six-cylinder Corvair automobile engine and a composite propeller manufactured by Warp Drive Propellers. In 2015, the engine was replaced with a four-cylinder 125 horsepower Lycoming O-290-G engine; at that time, according to the airplane’s maintenance logbooks, it underwent 5 hours of Phase II flight testing. The pilot, who was also a licensed FAA Airframe and Powerplant mechanic, purchased the airplane in December 2019; by that time it had accrued about 200 hours of flight time. Over the next two years, the airplane accumulated a further 200 hours and underwent a series of engine changes. By September 28, 2022, the pilot had reinstalled the original O-290-G engine, which had now been overhauled and upgraded to 160 horsepower through the installation of cylinder assemblies designed for the Lycoming O-320 engine. A month before the accident, during the airplane’s most recent condition inspection, the pilot replaced the propeller (a Catto Propellers fixed-pitch 76-inch two-blade model), with a three-blade, composite, ground-adjustable propeller manufactured by NR Prop. There was no evidence in the logbooks to show that the airplane had gone through a flight test after the installation of the upgraded engine and new propeller. Correspondence between the pilot and representatives from NR Prop indicated that the two-blade propeller configuration had been tested with Yamaha engines, which require a PSRU, up to 170 horsepower; however, NR Prop had no data for installation on a Lycoming engine. The pilot purchased the three-blade adjustable pitch SR-118 assembly, with a diameter of 2,024 mm, in addition to a two-blade hub that was also compatible with the blades. The SR-118-2000 is advertised on the NR Prop website for installation on engines between 90 and 150 horsepower. According to online forum correspondence, the pilot stated that when the propellor arrived it did not include installation instructions and he did not know the mounting hardware torque specifications. Review of subsequent correspondence between the pilot and the manufacturer indicated that he then received torque values for the propeller hub, but there was no evidence he received formal installation or ongoing maintenance instructions. A friend of the pilot stated that after installation, the engine would not reach its rated rpm. The pilot therefore decided to switch to the two-blade configuration, using the two-blade hub and two of the three blades. WRECKAGE AND IMPACT INFORMATIONThe airplane came to rest in a dirt field 1.5 miles southwest of the departure end of runway 20. The entire structure was consumed by fire with only steel airframe and burnt aluminum, composite, and flight instrument remnants remaining. The first identified point of impact was an almost-complete impression of the airplane’s forward profile in the dirt, which included the main landing gear strut, wheels, wing leading edges, and lift struts. The engine had separated from the firewall and was in the center of the impression. About 20 feet beyond the engine, both wings had come to rest in line with the impact point. The cabin and tail structure were crushed aft, such that the rudder pedals were comingled with the remnants of the empennage. A single intact propeller blade (figure 1) was located 1 mile southwest of the departure runway, in the general vicinity of the flight path. A section of exhaust pipe, along with various items of cabin contents, an iPad, and clear plexiglass fragments were distributed another 1/2 mile closer to the main wreckage site. All remaining primary airframe structure, flight control surfaces, and engine components, along with the thermally damaged second propeller blade, which had fractured from the propeller hub on impact, were accounted for in the main wreckage. The propeller hub remained attached to the engine crankshaft, and the root of the ground-impacted blade remained clamped within the hub. Examination of the engine did not reveal any catastrophic failures or anomalies that would have precluded normal operation. Figure 1- ADS-B flight path with debris field, and propeller blade as-found. TESTS AND RESEARCHAn accredited representative from the National Transportation Investigation Bureau of Ukraine, which was the state of the propeller manufacturer, was assigned to assist with the investigation. The propeller blade and hub assembly, along with the unused third blade, were sent to the NTSB Materials Laboratory for examination. The manufacturer did not have construction or layup documents available, so the third blade was used as a reference. Propeller Design The assembly consisted of two composite propeller blades and an aluminum two-piece hub, clamped together by bolts. Blade retention was achieved through friction between the outer cylindrical surface of the blade and an inner clamping surface on the hub through inboard and outboard grip areas. A collar in the middle of the blade retention area slotted into a groove on the hub to locate the blade spanwise within the hub (figure 2). Figure 2 - Root of the exemplar propeller blade Exemplar Blade The root sections of the exemplar blade were examined by X-ray computed tomography (X-ray CT), revealing that it was constructed of composite fiber polymer matrix layers built around an aluminum tube that was flush against the root of the blade. The innermost layer consisted of unidirectional glass fiber wound circumferentially around the tube. The next layer consisted of unidirectional glass fibers oriented along the axial direction of the blade. The next layer comprised four plies of plain-woven glass fiber fabric with warp and weft at +/- 45°. The next layer was comprised of multiple plies of carbon fiber fabric, and the final outermost layer consisted of a +/- 45° glass fiber fabric ply. Within the collar feature, the fiber layers bowed radially outward, partially filling and forming the collar. The rest of the collar consisted of resin-rich wedge-shaped regions of epoxy along the upper and lower edges. In some regions, the excess volume between the bowed layers was filled by resin pockets or voids, and in other regions it was filled with unidirectional circumferential glass fiber (figure 3). The outermost fiber ply in this collar was severed by the collar machining process. Figure 3 - X-ray CT of the exemplar blade root and collar. Separated Blade The collar of the separated blade was still retained inside the groove in the hub (figure 6). The blade exhibited fracture and wear features at its root, and there were axial (spanwise) rub marks on the grip section at the base of the blade root (figure 4) and on the clamping faces of the hub. The outer resin and glass fiber layers were worn away on either side of the built-up unidirectional axial fiber layer close to the blade trailing edge; on the opposite side of the blade, an elliptical wear region had exposed the +/- 45° glass fiber layers in the collar region. Loose fiber regions were visible on the exposed fiber surfaces at the blade root, with an appearance consistent with incomplete or absent resin encapsulation (resin starvation). Figure 4 – Separated blade root with collar detached X-ray CT scans revealed that the aluminum backing tube was about 10 mm longer than the exemplar blade. It was slightly concave at the location of the collar and not positioned against the base of the blade root, but instead was recessed from the base of the blade by approximately 4 mm (figure 5). The blade exhibited multiple voids, several of which were visible from the outside where the collar had separated. The voids ranged in size, with some as big as 18 mm long and 8 mm wide (figure 5). Additional void regions were visible on other scan images. See the X-ray CT scan report for additional details. Figure 5 - X-ray CT scan image of separated blade root near leading edge location showing multiple exposed surface and sub-surface voids in area of clamping collar. Detached Blade Root The retained blade root appeared to have shifted outward (figure 6), and there was a dirt/dust mark on the hub clamping surface approximately 4.4 mm inboard of the root end of the blade (figure 7). The collar was still within the groove and had slid out of position toward the base of the blade root. Resin and fiber material had extruded into the gap between the hub halves, and dirt/dust marks appeared to follow the edges of some of the remaining extruded material (figures 6,7). Figure 6 – Hub with separated blade collar still in place (right), and mark on hub showing were retained blade root migrated outward (left). Figure 7 – Blade root with shifted collar, dirt mark, and extruded resin/fibers at the hub join. The inboard grip region on the blade root had two regions that exhibited different surface textures. The inboard portion, extending approximately up to the lower edge of the displaced collar feature, had a flat and comparatively smooth appearance. The outer portion, extending up to the original edge of the collar, had a comparatively textured, raised, and dimpled appearance (figure 8). Figure 8 – Blade root with shifted collar, and different surface textures. A section view of the blade root revealed that along the spanwise outboard end of the collar, the resin-rich wedge had separated from the fiber plies. The outermost three or four plies had fractured near the spanwise outboard edge of the collar and had displaced spanwise inboard toward the root of the blade. Additionally, the plies near the spanwise inboard edge exhibited buckling and peel separation (figure 9) and the outer four or five plies within the inboard grip area had sheared inboard relative to the underlying plies. Figure 9 - Top Image: Overview image of axial cross section. Lower image: Higher magnification image of axial displacement of collar feature. Examination by scanning electron microscope (SEM) revealed multiple areas of resin-starved dry fiber features (figure 10), along with regions filled with fibers but with gaps between the fibers. Figure 10 – SEM image of dry fiber regions