N196DA

On September 4, 1999, at about 1845 eastern daylight time, a Diamond DA20-A1, N196DA, registered to Sky King Aviation, operating as a 14 CFR Part 91 personal flight, crashed while making a forced landing back to George T. Lewis Airport, Cedar Key, Florida. Visual meteorological conditions prevailed and no flight plan was filed. The airplane sustained substantial damage. The private pilot reported no injuries. The flight was originating at the time of the accident.

The pilot stated he did not conduct any performance data for his departure from George T. Lewis Airport. The departure runway is 2,335 feet long. He started the airplane engine, conducted an engine run-up, taxied out to runway 23, and back taxied down the runway for departure. The takeoff roll was normal. Before reaching rotation speed, the low fuel pressure warning light illuminated about half way down the runway. He checked and verified that the fuel selector valve and electric fuel pump were in the on position. He elected not to abort the takeoff and the airplane became airborne at about 1,800 feet. As soon as the airplane was airborne the engine started to sputter. He made a hard right turn in an attempt to return to the departure runway. Halfway through the turn the engine quit, the stall warning horn activated, due to low airspeed and high angle of bank. He realized he was going to over shoot runway 05. He lowered the nose to regain airspeed, and made a hard left turn towards the runway. The next he remembered was the airplane being on the ground with smoke in the cockpit. He turned the fuel selector valve off, and was assisted out of the airplane by a person who came to render assistance.

A witness stated he observed the airplane on the takeoff roll. The airplane took off and was at about 15 feet when the engine sputtered. It looked like the pilot was going to set the airplane back down on the runway, but he continued the takeoff. He flew out over the water, made a 180-degree turn and over flew the runway. The airplane was observed to make a hard left bank, the left wing collided with the tie down area, the airplane turned sideways, took out a light-box sign, hit another airplane, a tree, and came to a complete stop.

Examination of the airframe and flight controls by the FAA revealed no evidence of a precrash mechanical failure or malfunction. The electrical fuel switch was in the off position. The engine-driven fuel pump was removed from the airplane. Battery power was reconnected and the electrical fuel pump switch was turned on to validate pump operation. The electrical fuel pump did not produce any noise normally heard when the switch is turned on. Power was then turned off and the battery was disconnected.

The FAA inspector asked the pilot if he had conducted a performance takeoff stopping distance calculation as part of his preflight preparation. The pilot stated no. The pilot was asked to calculate the weight and balance of the airplane and takeoff distance required given the following criteria: temperature 74 degrees, dew point 64, wind 310 at 4 knots. The pilot incorrectly calculated the weight and balance and takeoff distance. With continued assistance from the FAA, the pilot calculated the takeoff/stopping decision distance to be 1,100 feet. At this point of the ground roll, the pilot had 1,225 feet of runway remaining. The inspector stated, "It is my opinion that the pilot could have safely aborted the takeoff if he had acted responsibly when the fuel light had illuminated." (For additional information see FAA Inspector statement, an attachment to this report.)

On September 7, 1999, the NTSB investigator-in-charge received a fax from the registered owner of the accident airplane. He stated that another airplane that was used by his flight school had experienced an intermittent engine roughness that would last for a second or more. After numerous inspections by the maintenance facility on the field they could not determine the cause. Inspections included pressure checks on the engine-driven fuel pump and the electric backup fuel pump. A change of all spark plugs and any related electrical items related to ignition, draining of the entire fuel tanks and inspection, replacement of fuel filters, and removal and inspection of the fuel lines. All inspections were found to be without fault.

The registered owner stated further test flights revealed the problem was increasing and the fuel pressure warning light began to illuminate prior to the engine roughness. He suspected an engine-driven fuel pump intermittent failure Rotech of Canada was contacted and a new fuel pump was ordered and installed. The pump lasted about 4 hours and was replaced again. Rotech was contacted and another fuel pump was ordered and replaced. After 10 hours, the pump yielded the same results and was replaced. At this time N196DA began having the same symptoms. A fuel pump was ordered and replaced without any further incident. (For additional information see letter dated September 7, 1999, an attachment to this report.)

A Diamond Aircraft Field Service Representative examined the aircraft on September 13, 1999, after the wreckage had been released by the NTSB to the registered owner on September 6, 1999. The examination was conducted at a salvage yard. The field representative concluded, "Visual inspection: All airframe fuel system components, located below the fuel tank and aft of the spar bridge appeared undamaged from impact of the accident. In addition, all components were properly installed and secured, including the electrical wiring and electrical ground terminal. Operational Check: When power was applied to aircraft's main bus and the "fuel pump" was selected on, the electric boost pump operated normally." (For additional information see Diamond Aircraft Field Service Report an attachment to this report.)

Review of Diamond DA20 KATANA Airplane Flight Manual states on page1-2 Paragraph 1.3 WARNINGS, CAUTIONS, AND NOTES states, "The following definitions apply to warnings, cautions, and notes used in the Flight Manual: WARNING means that the non-observation of the corresponding procedure leads to an immediate or important degradation of the flight safety." Chapter 4 Normal Operating Procedures, paragraph 4.4 Normal Operation Checklist, 4.4.2. Before Starting Engine, 7. Fuel Shut-off Valve, OPEN. Paragraph 4.4.14. Engine Shut-down, does not instruct the pilot to turn the fuel shut-off valve to the off position when the flight is completed.

Title 14 CFR Section 91.7 (b) states, "The pilot in command of a civil aircraft is responsible for determining whether that aircraft is in condition for safe flight. The pilot in command shall discontinue the flight when un-airworthy mechanical, electrical, or structural condition occur."

The flight manual does not address the loss of fuel pressure on the ground; however, it states the following on page 3-4, "III. LOSS OF FUEL PRESSURE

1. Electric Fuel Pump ON, and land at nearest suitable airport

2. If Fuel Pressure Land at nearest suitable airport. Be Warning Light prepared for engine failure and Does not extinguish emergency landing"

The engine fuel pump and electric fuel pump from N196DA was removed from the airplane and forwarded to the NTSB investigator-in-charge (IIC). The IIC forwarded the components to NTSB headquarters and were hand carried to Rotec Canada for further analysis. A Kodiak Research representative obtained the engine assembly from Quality Aircraft Salvage in Groveland, Florida. The engine was shipped to Rotec Research Canada Ltd. for testing and evaluation.

The engine assembly was mounted onto a test stand to determine if the engine had suffered a mechanical failure. The original fuel pump was installed on the engine. Inspection of the fuel pump found no discrepancies and the pump did not rattle internally as originally suspected by the aircraft owner. The engine was pulled through by hand and found to be resistive and sticky. Investigation revealed the starter had failed. The starter was replaced and the engine was cranked over to establish oil pressure. The stator winding, triggers, and coils were impedance tested to confirm serviceability. An engine start was attempted. The engine started immediately and effortlessly. A high audible whine coming from the gearbox was noted. The engine was warmed and taken to mid power (3750 RPM) to perform a magneto check. The magneto check was within limits. The engine was accelerated to full power rpm of 5,800. The engine was run at cruise and various power settings with no hesitation or stoppage.

The engine test stand fuel supply shut-off valve was changed to the off position. After several minutes (about 2.4), the fuel low pressure light illuminated and the engine ran for about 50 more seconds at full power before losing power. The fuel shut-off valve was turned back on and the engine recovered full power immediately. The fuel shut-off valve was activated again and the engine was operated at idle for 4 minutes before the fuel low pressure light illuminated. The fuel supply (shut-off supply valve opened) was restored and the engine recovered immediately. (See NTSB Powerplant Teardown Report of Investigation an attachment to this report.)

A spectrum analysis of the engine was performed to determine vibration amplitudes of the engine. The engine was started and warmed to operating temperature, and then accelerated to 5,000 engine rpm for reading of the spectrum analysis. The result was the engine gearbox was producing overall vibration amplitudes of 21.68 inches-per-second in the 600-12,000 frequency range and overall vibration amplitudes of 29.17 in the 600-60,000 frequency range. These amplitudes are the most severe recorded to date of a 912 engine. A baseline engine will typically produce overall vibration amplitudes of 0.5 inches-per-second.

A visual inspection was conducted on the gearbox. The PTO drive gear had...