N6948LDavid B Lunn VANS RV8 · 2023

On December 16, 2023, about 1150 Pacific standard time, an experimental amateur-built Vans RV-8 airplane, N6948L, was substantially damaged when it was involved in an accident near Corona, California. The pilot and passenger were fatally injured. The airplane was operated as a Title 14 Code of Federal Regulations Part 91 personal flight.

According to the owner of a paint shop at the Corona Municipal Airport (AJO), Corona, California, the pilot had flown the airplane to AJO on November 13, 2023, to have it painted. The accident flight was the first flight after it was painted. The paint shop required the pilot to disassemble and reassemble the airplane before and after painting because it was an experimental amateur-built airplane. To address fuel leaks, the pilot had drained and repaired the left fuel tank before flying the airplane to AJO.

Witnesses reported hearing the airplane during the takeoff and remarked that the engine sounded like it was “missing” and producing partial power.

Airport security camera footage revealed that, at 1149:23, the airplane was midway down runway 7 (passing by the fuel tanks) with all of its landing gear on the surface. The airplane continued down the runway and rotated about 4 seconds later. Video analysis determined that the airplane’s speed was 71 kts at the time of rotation, and a sound spectrum analysis revealed that the engine rpm was about 2,500 at that time. The airplane then climbed slightly, just passing over a berm at the airport’s perimeter, reaching a maximum altitude of about 95 ft above the runway. The estimated ground speed was 58 kts. Thereafter, the airplane descended and made a left turn, disappearing from the security camera’s field of view.

Figure 1. Flight path with composite image (inset) showing takeoff.

A review of the maintenance logbooks revealed no evidence that the airplane had undergone any maintenance after receiving a Special Airworthiness Certificate on May 11, 2023. The pilot purchased the airplane kit in April 2022. The tachometer was consumed by fire.

Invoices revealed that the most recent sealant that the pilot purchased was two-part sealant Flamemaster CS-3204 B2 and Pro Seal Fuel Tank Sealant PS 890 (part A & B). An invoice dated January 31, 2023, revealed that the pilot purchased a quart of Bill Hirsch alcohol-resistant gas tank sealer. The sealer manufacturer stated that if the tank is not dry or has foreign matter, it may prohibit the sealer from adhering and then lead to it coming loose, eventually entering the fuel system and blocking the fuel pump.

The pilot/builder had deviated slightly from the original RV-8 fuel plan design and used several automotive parts in its construction. The left fuel tank was likely empty and the right tank likely contained about 10 gallons, but the quantities could not be verified because of the severe fire damage. The fuel source at AJO indicated that the pilot did not purchase fuel at the airport.

A review of the maintenance logbooks revealed no evidence that the airplane had undergone any maintenance after receiving a Special Airworthiness Certificate on May 11, 2023. The pilot purchased the airplane kit in April 2022. The tachometer was consumed by fire.

Invoices revealed that the most recent sealant that the pilot purchased was two-part sealant Flamemaster CS-3204 B2 and Pro Seal Fuel Tank Sealant PS 890 (part A & B). An invoice dated January 31, 2023, revealed that the pilot purchased a quart of Bill Hirsch alcohol-resistant gas tank sealer. The sealer manufacturer stated that if the tank is not dry or has foreign matter, it may prohibit the sealer from adhering and then lead to it coming loose, eventually entering the fuel system and blocking the fuel pump.

The pilot/builder had deviated slightly from the original RV-8 fuel plan design and used several automotive parts in its construction. The left fuel tank was likely empty and the right tank likely contained about 10 gallons, but the quantities could not be verified because of the severe fire damage. The fuel source at AJO indicated that the pilot did not purchase fuel at the airport.

The accident site was located in soft dirt about 3,650 ft from the departure end of runway 7. The airplane came to rest upright under a tree, and the surrounding 5–10 ft of terrain was burned.

The wreckage was consumed by a postimpact fire, and a majority of the fuselage, wings, and skin panels were charred, with areas of ash and resolidified puddled aluminum. The fuselage was on a heading of about 55° (see figure 2). The entire airplane was located at the accident site, except portions of the wheel pants and outboard left wing, which were found in the debris path.

Figure 2. Main wreckage.

The first identified points of contact were on the west side of the flat, paved road at the west end of the debris field. The first impact was a divot in the concrete wall at the edge of the road. There were paint transfers and a pile of paint chips adjacent to the divot, consistent with the left wing tip (see figure 3).

Figure 3. Debris field before main wreckage.

About 85 inches south of the divot was a black smudge on the road’s surface, which continued toward the wreckage and made a left track. The debris path continued with grooves and indentations on the road making a parallel right track. About 12 ft from the road’s western edge were two 18-inch gouges in the concrete that were oriented perpendicular to and between the two parallel tracks. These were consistent with propeller slashes.

The dimensions and orientation of the witness marks were consistent with the left wing impacting the road first, followed by the left wheel and then the right wheel.

Postaccident examination of the airframe established partial control continuity. The control cables were continuous to molten bellcranks and portions of the airframe. The cockpit was completely thermally consumed. The canopy appeared to be in the unlocked and open position. The main landing gear remained with the airframe and was folded aft and located under the wings. The left and right rotors on the wheels were shaved down with flat areas, consistent with sliding against the pavement during the accident sequence.

Postaccident examination of the engine revealed no evidence of any preimpact mechanical malfunctions or failures that would have precluded normal operation. The engine remained attached to the airframe via the engine mount and sustained significant thermal damage from the postimpact fire. The crankshaft rotated freely by hand, in both directions, using the attached propeller. Thumb compression was observed on all four cylinders. The complete valvetrain operated properly. Normal lift action was observed at each rocker assembly, and clean, uncontaminated oil was present in all four rocker box areas. Mechanical continuity was confirmed throughout the rotating and reciprocating group, valvetrain, and accessory section during hand rotation of the crankshaft.

The combustion chambers of each cylinder were examined via the spark plug holes using a lighted borescope. The chambers and bottom spark plug electrodes were mechanically undamaged, with no evidence of foreign object ingestion or detonation. All valves were intact and undamaged, with no observed valve-to-piston contact. The gas path and combustion signatures at the spark plugs, combustion chambers, and exhaust components exhibited coloration consistent with normal operation. No oil residue was observed in the exhaust gas path, and the exhaust system was free of obstructions.

Fuel System Examination

The fuel system (see figure 4) was examined, and components were sent to the NTSB Materials Laboratory for analysis.

Figure 4. Fuel system schematic.

The fuel system sustained substantial thermal damage and the fuel tanks were consumed. The fuel lines from the tanks to the selector were not directly routed and instead were looped before attaching to the valve’s inlet fittings. The fuel line from the selector valve’s outlet to the fuel filter was kinked at the bottom of a loop. The investigation was unable to determine if the crushing of the fuel line was a result of the impact.

Trace amounts of liquid exited from the fuel selector valve when manipulated; when subjected to water-detecting paste, the liquid tested positive for water, consistent with fire suppression efforts. Disassembly of the selector revealed white material on the inner surface of the exposed straight run end of the T-shaped fitting. A glossy, variegated tan and bubbled material was adhered to the inner surface of the branch of the T-shaped fitting.

From the fuel selector valve, fuel was plumbed to a 40- micron fuel filter; the outlet fitting was loose and the entire fitting assembly could be moved in and out of the filter body. The fuel line continued to the electric (boost) pump, which was mounted on the floor, aft of the firewall. The pump and connected return line system remained affixed in the brackets. The pump housing was covered in black soot deposits. Disassembly of the unit revealed a solid white, slightly glossy material between two curved plate magnets inside the pump. A distinct groove extended along the length of the material with an overall appearance consistent with molded and melted plastic.

From the electric pump, the fuel line was routed to a 90° firewall elbow to a 10-micron cylindrical pleated filter. The exterior surfaces of these components (except the pleated filter element) were covered by black soot deposits. The firewall elbow fittings contained a tan and bubbled material on the inner surface of the fittings. The interior surface of the filter housing contained a hard, shiny, off-white and bubbled material. A similar off-white material was visible between filter pleats.

The fuel line continued to the engine-driven fuel pump. A hard, glossy, white material was observed within the fuel chambers of the pump (see figure 5). Disassembly of the pump revealed this material was adhered inside both chambers. To assess flow between the chambers, water was introduced into the inlet chamber and slowly dripped through the check valve into the opposite chamber under gravity. When water was added to the opposite chamber, it slowly dripped back into the inlet chamber with no appreciable flow observed to the outlet chamber. When compressed air was applied to the outlet chamber, bubbles were observed emanating from around the outlet check valve in the opposite-side chamber, consistent with small open pathways through the valves, with more significant flow observed on the inlet valve side.

Figure 5. Mechanical fuel pump.

The fuel line was routed forward under cylinder Nos. 4 and 2, where an elbow connected to a fuel flow transducer. The transducer was not mounted, and the surrounding fuel lines were unsecured. The transducer and its fittings exhibited light thermal discoloration, with the red and blue anodized coloring remaining visible. The outer surfaces were covered in black soot deposits. Disassembly of the transducer revealed a white material, similar in appearance to that found in the engine-driven fuel pump, adhered to the internal surfaces and pooled within the inlet and outlet fittings. The internal black wheel component was partially embedded in the white material and could not be rotated by hand (see figure 6).

Figure 6. Transducer.

From the transducer, the fuel was routed to the fuel injection servo. A blue elastomeric O-ring was present on the filter side of the fitting, and a smaller black elastomeric O-ring was present around the filter at the base of the fitting. A hard, white material, similar in appearance to that observed in the engine-driven fuel pump and the fuel flow transducer, was found filling and protruding from the screen filter element. The interior of the fitting was obstructed by the same white material.

The fuel distributor (spider) was disassembled, revealing that the diaphragm seal was present; there was no liquid in the body.

The NTSB Materials Laboratory analyzed specimens extracted from various fuel system components to determine their chemical composition. The samples showed the presence of multiple organic/polymeric chemistries. Material from the fuel selector matched polyimide or polyacrylic polymers, while the electric fuel pump sample was consistent with a nylon-based material. Samples from the firewall elbow fitting, 10-micron filter screen and housing, engine-driven fuel pump, fuel flow transducer, and fuel injector servo filter were consistent with polyvinyl chloride or polyvinyl acetate-based polymers or co-polymers. The sample from the 10-micron filter screen displayed a composite spectral profile containing elements of polyimide, polyacrylic, polyvinyl chloride, and polyvinyl acetate.

Thermal testing of a white material removed from the engine-driven fuel pump showed that exposure to a 100°C temperature resulted in the specimen softening and become tacky, resembling a melted marshmallow. Continued heating resulted in the material darkening while remaining pliable. Upon cooling, it returned to a hardened state, a thermal behavior characteristic of thermoplastic polymers and not consistent with thermoset or chemically cured materials.

Comparative analysis was conducted against materials reportedly used in the airplane build. These included two polysulfide-based sealants, a one-part alcohol-resistant tank sealer (Bill Hirsch), and two Axalta exterior coatings (epoxy primer and polyurethane topcoat). The fuel system deposits were inconsistent with polysulfide, epoxy, or polyurethane compositions. The Bill Hirsch alcohol-resistant gas tank sealer, which contains VMCH resin (CAS# 9005-09-8), displayed chemical compatibility with the deposits found in the firewall elbow, 10-micron filter screen and housing, engine-driven fuel pump, fuel flow transducer, and fuel injector servo filter.