N273DBCharles J Norris ARION LIGHTNING · 2016

On July 10, 2017, about 0745 mountain standard time, an experimental light sport Arion Lightning airplane, N273DB, was substantially damaged in a forced off-airport landing near New River, Arizona. The private pilot was uninjured. The airplane was registered to and operated by the pilot as a Title 14 Code of Federal Regulations Part 91 personal flight. Visual meteorological conditions prevailed and no flight plan was filed. The local flight originated from Glendale Municipal Airport (GEU) Glendale, Arizona at 0730.

The pilot was also the builder of the airplane and reported that he had changed the propeller pitch and right wing incidence on July 7, and the purpose of the flight was to assess those changes. The pilot fueled the airplane to a total quantity of about 30 gallons. The engine started normally, and after the engine temperatures reached appropriate values, the pilot conducted an engine runup before beginning his taxi out. That runup was normal. The pilot then taxied to runway 1, where he conducted a second runup, which again was normal. While awaiting takeoff clearance, the pilot noticed that the cylinder head temperature (CHT) on the No. 3 cylinder was higher than both normal and the other five cylinders, but still well below its maximum limit. When the airplane began its takeoff roll, the No. 3 cylinder CHT began decreasing, and aligned with the other CHT values during the climbout.

The pilot reported that he conducted the climbout with an engine speed of 2,660 rpm, and leveled off temporarily at 2,500 ft. Shortly thereafter, the pilot increased the rpm to 2,800 for a climb to a higher altitude. He noticed that the engine "did not feel as if it were operating smoothly," and switched the fuel selector from the left tank to the right tank. As the airplane was passing through 5,980 ft, he felt a loss of power, and the rpm decreased to 1,450. An instrument scan indicated that the engine was still running, but manipulation of the throttle did not result in any rpm changes. The pilot leveled off and began a right turn back to the southwest, towards Deer Valley Airport (DVT) Deer Valley, Arizona. He unsuccessfully attempted to contact DVT air traffic control tower, changed his transponder code to 7700, and began searching for a suitable landing site. He also slowed to the airplane's best glide speed, and verified that the ignition switch was set to the 'BOTH' position. The pilot initially set up for a landing on a plateau, and during his turn from base to final, the engine rpm decreased to about 1,000.

The airplane was going too fast to land on the plateau, so the pilot selected a new landing site just beyond the plateau. He added full flaps, turned off the fuel, and landed on undulating desert terrain. The landing gear collapsed, the propeller struck the ground, and the airplane slid to a stop. The pilot shut off all the switches, except for the emergency locator transmitter (ELT), which had activated. He exited the airplane, called his wife and then the owner of the local facility that he used to build the airplane, in order to begin the airplane recovery process. A local resident drove up to provide assistance. He and the pilot collected some debris, and then the pilot called the DVT manager to notify him of the accident, who then contacted the Federal Aviation Administration (FAA) and National Transportation Safety Board (NTSB). Several first responder ground vehicles, and then a Phoenix Police helicopter arrived, and during the discussions, the pilot was asked to shut off the ELT, which he did. The airplane was recovered later that day to a secure facility for future examination. About 25 gallons of fuel were obtained from the airplane fuel tanks during the recovery process.

The pilot held private and sport pilot certificates, with an instrument airplane rating. He had approximately 1,201 total hours of flight experience, including approximately 60 hours in the accident airplane make and model. His most recent flight review was completed in November 2016. At the time of the accident the pilot was operating the airplane under the medical provisions of his sport pilot certificate.

The airplane was a 2-place low-wing monoplane constructed primarily of composite materials. Each wing was equipped with an integral fuel tank, each with a pilot-reported capacity of 20 gallons. According to pilot, he had purchased the airframe as a kit, and completed assembly of the airplane, including engine installation, at a dedicated build center near Phoenix in 2016. The pilot reported that the airplane and engine had accumulated about 63 hours total time (TT) in service at the time of the accident. The pilot reported that he had about 60 hours TT in the airplane.

According to pilot, he had purchased a brand new, pre-assembled Jabiru model 3300A series engine. The engine was a 6 cylinder, horizontally opposed, normally aspirated, air-cooled engine, with a rated output of 120 hp. The pilot reported that in January 2017, when the airplane had a TT of about 33 hours, due to a rough-running engine, he replaced the engine-driven fuel pump that was found to be internally damaged. The damage was the result of the fuel pump actuation pushrod being about 2mm too long. The engine was purchased from Jabiru USA, with the fuel pump installed. Although the pilot queried Jabiru USA as to why the original pushrod was too long, he never received an explanation.

The 0753 automated weather observation at DVT, located about 17 miles south of the accident site, included variable winds at 3 knots, visibility 10 miles, clear skies, temperature 32o C, dew point 14o C, and an altimeter setting of 29.91 inches of mercury.

The airplane was a 2-place low-wing monoplane constructed primarily of composite materials. Each wing was equipped with an integral fuel tank, each with a pilot-reported capacity of 20 gallons. According to pilot, he had purchased the airframe as a kit, and completed assembly of the airplane, including engine installation, at a dedicated build center near Phoenix in 2016. The pilot reported that the airplane and engine had accumulated about 63 hours total time (TT) in service at the time of the accident. The pilot reported that he had about 60 hours TT in the airplane.

According to pilot, he had purchased a brand new, pre-assembled Jabiru model 3300A series engine. The engine was a 6 cylinder, horizontally opposed, normally aspirated, air-cooled engine, with a rated output of 120 hp. The pilot reported that in January 2017, when the airplane had a TT of about 33 hours, due to a rough-running engine, he replaced the engine-driven fuel pump that was found to be internally damaged. The damage was the result of the fuel pump actuation pushrod being about 2mm too long. The engine was purchased from Jabiru USA, with the fuel pump installed. Although the pilot queried Jabiru USA as to why the original pushrod was too long, he never received an explanation.

The airplane came to rest upright in moderately flat area of desert terrain, about 17 miles north of DVT. Both wings remained attached, and sustained minor damage, primarily at their tips. The engine remained attached to its mount, and the mount to the firewall, but several of the engine mount tubes were bent, buckled, or fractured. The 2-blade composite propeller remained attached to the engine, and both blades were fracture-separated from the hub. The left fixed main landing gear remained intact, but the right strut was partially displaced, and wheel and tire assembly was liberated from the strut. The nose gear assembly and mount was crumpled, and folded aft and up. There was no indication of any fire or thermal damage to the airframe or engine.

The FAA inspector who responded to the accident site reported that the two fuel tanks were not breached, and confirmed the presence of significant fuel in each tank. While at the accident site, the left and right seat bottoms were removed to access some fuel lines and the two transparent, in-line fuel filters (one per tank). Each tank was plumbed to an in-line filter in the fuselage, and both filters were observed to contain fuel. The inspector reported that the fuel in the tanks and filters was clean. The carburetor bowl was removed and contained very little (estimated ½ oz) fuel.

The airframe and engine were examined in detail at the recovery facility. No evidence of any pre-impact mechanical failures was observed. Refer to the public docket for detailed documentation of the examination results.

Fuel System Information

The airplane fuel system was examined in detail, and several differences between the manufacturer's configuration guidance and the actual airplane were observed. No lines, fittings, or components of the fuel system exhibited any indications of fuel leakage, but the system was not able to be tested for integrity.

According to the Jabiru Installation Manual (IM), "The fuel tank must be fitted with an outlet strainer of between 8 and 16 mesh per inch, with a minimum total mesh area of 5 cm2. Ensure the fuel tank is properly vented." The pilot reported that a mesh fuel screen was installed in each tank outlet. The FAA inspector did not observe any fuel screens in the fuel tanks.

The Jabiru IM stated that "A Fuel filter capable of preventing the passage of particles larger than 0.1mm (100um) must be installed between the fuel tank outlet and the fuel pump. The filter must be present in the system for the fuel flow test. The size of the filter should give consideration to allow adequate flow with a used filter." As cited above, fuel filters were found installed in the specified locations. However, no manufacturer identification or part numbers were observed on the installed filters, and therefore their filtration limits were not able to be determined.

The Jabiru IM stated that "Fuel lines are nominally 6mm [0.236 inches] bore." The inside diameters (IDs) of all fuel lines were not measured, but the IDs of the gascolator fittings were 5/32" (0.156 inches).

An Electronics International FT-60 fuel flow transmitter was affixed to an engine mount tube, forward of the firewall. According to the Jabiru IM, "Where a Fuel Flow Meter is to be installed to the aircraft, Jabiru Aircraft recommend that the flow transducer is not installed on the engine side of the firewall. Most transducers are made of either plastic or light aluminium and are not fire resistant. Regulations and common sense both require that every part of the fuel system on the engine side of the firewall must be fire resistant." All fuel lines forward of the firewall were shrouded in fire-resistant sheathing.

The fuel flow transmitter was observed to allow liquid to pass through it, demonstrating that it was not completely blocked. However, due to lack of equipment, the flow capability was not measured or otherwise quantified.

Fuel Pump Information

The engine driven pump (EDP) was a rocker/diaphragm type, driven by a pushrod that was actuated by a cam on the crankshaft. The EDP bore no part number, but based information from the Jabiru Overhaul Manual, this was a "Type 1" pump. The EDP was disconnected and removed from the engine. There was a paper gasket and a nylon/plastic spacer between the pump and the engine case; the gasket abutted the EDP, and the spacer abutted the engine case.

The pump rocker was undamaged, well-lubricated, and free to pivot. The pushrod was free to translate and rotate in its bore, was well lubricated, straight, with no visible scoring or damage. The crankshaft cam for the EDP pushrod was well lubricated, with no visible scoring or damage.

The EDP was sent to the FAA Flight Standards District Office in Nashville Tennessee, with the intent of testing it at Jabiru USA on a known-good exemplar Jabiru 3300 engine. However, before such an engine became available, Jabiru USA ceased operations. Arion Aircraft took over most of Jabiru USA's previous business activities, including equipment and some personnel. In October 2018 an exemplar engine became available for testing of the accident components; the engine was already installed on a similar Arion airplane.

The accident airplane EDP and its pushrod were installed on the engine. The engine started, and operated normally throughout all rpm regimes, with a total run time of about 12 minutes. No abnormalities were noted during this testing.

Carburetor Information

The Jabiru IM stated that a Bing constant-depression type 90/94 carburetor was to be used. The accident carburetor was a Bing model, and bore an engraved "41559" or "41SS9," plus a stamped "94/40/148."

The Jabiru IM also referred to the carburetor as an "altitude compensating" model which "uses bowl float level and two main air circuits – the idle and the needle/main – to control the mixture. Both circuits use jets to meter the rate at which fuel is allowed to flow." The jets are small brass inserts with precisely controlled orifices, and the jets can be changed to adjust engine mixture. Mixture during engine operation is controlled by the position of a custom-profiled, multiple-diameter needle in the jet orifice, which changes the effective size of the jet orifice as a function of needle diameter. The profile of the needle controls the mixture at a given throttle setting, and the needle translates in the jet in response to the air circuit pressure differentials that result from throttle position and ambient air pressure. At low power settings, the jet is nearly completely blocked by the needle. Increased throttle settings move the needle to allow more fuel to flow through the jet.

According to the IM, the "needle… [has] been optimized for use with a propeller, which puts a very non-linear load on the engine; to double the RPM of a propeller a lot more than double the power has to be applied." The IM also stated that the mixture is affected by several variables, including propeller size and propeller loading, which is a function of blade pitch. Jabiru recommends that "whenever a new engine installation is being developed that the engine be fitted with EGT probes and the [carburetor] tuning checked."

The IM stated that "This carburettor has a minimum delivery pressure of 5 kPa (0.75 Psi) and a maximum pressure of 20 kPa (3 psi). To confirm that the fuel system is capable of delivering this pressure a fuel flow test must be performed." Lack of appropriate test equipment prevented the determination of the accident airplane fuel system delivery pressure capability.

In place of the Bing-specified "Balance Tube," the carburetor was equipped with an aftermarket system was known as a "HACman" (for high altitude compensation-manual). The HACman control was a Vernier-type bleed adjustment intended to manually regulate the pressure in the carburetor float bowl, and was advertised to provide better performance and reduced fuel burn. The HACman consisted of four sections of flexible plastic lines connected to the carburetor and other locations, and a cockpit control device labeled "Mixture." Only two of the eight line ends were secured to their attach points with external clamping mechanisms. The IM did not contain any information regarding the HACman system, and therefore did not specify whether such lines were to be secured with clamps. Lack of clamping can permit leaks, but the system integrity was not able to be checked for leaks due to lack of proper equipment.

With the exception of one HACman line end ("port") which appeared to be questionably placed, all lines were found to be properly routed. The Jabiru IM recommended a location that was in "slow-moving" air, but the actual port was installed adjacent to the filter exit, in a region that would possibly be relatively fast-moving airflow. However, a Jabiru representative who examined the setup stated that port location placement was acceptable. The pilot did not modulate the HACman on the accident flight.

The Jabiru North America representative stated that Jabiru was aware of these HACman devices, and they reportedly smoothed engine operation when operating at altitudes above 6,000 to 8,000 ft, where the carburetor tends to run slightly rich. The representative also reported that at lower altitudes, improper setting (in the lean direction) could result in a fuel/air mixture that would not support engine operation.

Similar to the EDP testing, the accident airplane carburetor and several lines (excluding the HACman) were then installed on the exemplar Jabiru engine on the airplane.

The engine was started and allowed to stabilize. It was run approximately 5 minutes at varying speeds up to 2,200 rpm. The engine was shut down to reposition the airplane, and then restarted. It was then run approximately 7 minutes at varying speeds up to 2,650 rpm with the airplane stopped, and reached 2,680 rpm once the brakes were released and the airplane allowed to roll. There were no major abnormalities noted during this portion of the testing.

It was noted that at the high power setting, the engine did run more roughly with the accident carburetor installed. There was a slight decrease in maximum rpm and EGT compared to the performance using the test engine's original carburetor. The Arion representative reported that this was likely due to the fact that the accident carburetor is improperly jetted for the application on the exemplar airplane/engine.

The FAA and Arion representatives reported that it was their opinion that at all times the engine (and thus EDP and carburetor) performance was well with limits and specifications.

Late Jabiru- and Pilot-Provided Information

Post accident email queries of the pilot by the NTSB, that continued through 2018, resulted in additional information regarding the airplane history, and the pilot's experiences prior to and during the accident flight. These communications were also provided to a Jabiru representative, in order to solicit their opinions as to possible reasons for the power loss.

One question raised by the Jabiru representative was regarding the pilot's use of carburetor heat and the engine electric boost pump just after the power loss. In late August 2018, the pilot clarified that due to his relatively low altitude, he focused on finding a suitable off-airport landing site, and did not attempt any engine restart. He did not activate the boost pump, and he did not apply carburetor heat. Temperature and dew point values for the approximate time and location of the power loss indicated that the relative humidity was approximately 35 percent. When the intersection of the two temperature values was located on a chart that depicted carburetor ice envelopes, the point was in the envelope labeled "Icing (glide and cruise power)."

In his August 2018 response, the pilot also stated that "in retrospect it might have been an incomplete turn [of the fuel selector valve handle in response to slight engine roughness] to the right fuel tank that caused the fuel starvation. The proper procedure should have been to engage the electric fuel pump and switch back to the left fuel tank." Mis-positioning of the fuel selector valve can result in insufficient fuel flow for the commanded power setting, and can result in significant or complete power loss. Neither the post-power loss nor post-accident position of the fuel selector valve control was documented by the pilot or first responders.