NTSB Factual Report

HISTORY OF FLIGHT

On April 18, 1994, at 1314 hours Hawaii standard time, a Hughes 369D, N1103N, lost power while maneuvering and collided with rocky terrain below the Manawaupuna Falls, located about 7 miles northeast of Hanapepe, Kauai, Hawaii. The helicopter was being operated by Smoky Mountain Helicopters, Inc., dba Inter-Island Helicopters, Hanapepe, Hawaii, as a sightseeing flight under 14 CFR Part 91. The helicopter was destroyed. The certificated commercial pilot and three passengers were seriously injured. One passenger was fatally injured. The local flight originated at the Port Allen Airport, Hanapepe, Hawaii at 1135 hours. Visual meteorological conditions prevailed.

The helicopter was maneuvering about 150 feet above the surface of a man-made pond below the waterfalls when the engine lost power. The pond was in a valley formed by the Hanapepe River. The surrounding terrain was mountainous covered with thick vegetation.

The pilot stated he noticed a drop in "N2" with an immediate descent. The pilot attempted to fly out of the descent and increased the collective. The pilot stated he then transmitted a Mayday call and "informed the passengers a crash was eminent." The pilot flared the helicopter at treetop level and applied initial cushioning pitch approximately 5 feet above the ground. The pilot stated he did not observe any warning or caution lights during the accident.

The helicopter landed hard in the rocky riverbed about 300 feet downstream from the dam of the man-made pond. All the survivors reported losing consciousness and could not give clear accounts of the circumstances during postaccident interviews.

PILOT INFORMATION

The pilot holds a Commercial Pilot Certificate with a Rotorcraft- Helicopter rating. According to the operator's accident report, the pilot's total aeronautical experience consists of about 3,100 hours, all of which were accrued in helicopters.

AIRCRAFT INFORMATION

The helicopter, a Hughes 369D, had accumulated a total time in service of 7,154.3 hours. Examination of the maintenance records revealed that the most recent 100-hour inspection was accomplished on April 14, 1994, 14.3 hours before the accident.

An Allison 250-C20B turboshaft engine, serial number CAE-834267, was removed from the airframe by the operator on January 1, 1994, at 7,322.5 hours total engine time, and later installed in the airframe by the operator on January 28, 1994. The operator's maintenance records indicated about 227.8 hours were flown on the engine after the installation. The engine had accrued a total time in service of 7,550.3 hours at the time of the accident.

According to a statement from the operator, he performs acompressor wash on a daily basis in accordance with theinstructions contained in Allison Service Letter CSL-1135. The procedure was timed at 31 seconds, and 2.5 quarts of water are sprayed into the engine during the process.

WRECKAGE AND IMPACT INFORMATION

The initial impact point was about 10 feet upstream from the helicopter's point of rest. The landing gear was broken and portions were found upstream with aluminum transferred to submerged rocks.

The belly was crushed upward. The sheet metal was torn from the left side under the pilot's seat aft to the cabin. The cabin floor was displaced upward about 3 inches. The rear seat passenger seatbelt center attach point was torn loose from the aft cabin bulkhead.

The pilot's seat pan was bent downward. The collective torque tube under the pilot's seat was deformed in the same manner contouring the seat pan. The throttle linkage that runs through the collective torque tube was bent corresponding to the seat pan and found in the off position.

The engine turbine section was broken from the accessory gearbox and the fuel control unit was damaged. The engine fuel filter housing was found broken and the filter was found lying on the ground.

The helicopter's fuel cell was found shredded from underneath. No fuel was found in the cells. There was an aroma of jet fuel in the accident area and fuel was noted floating on the water in stagnant pools. One passenger complained of minor skin injuries due to exposure to fuel after the accident. The right rear passenger's seatbelt was found to have jet fuel on it.

The tail boom was severed from the airframe and found about 20 feet downstream in trees on the west bank of the stream. One of the helicopters main rotor blades was found separated and came to rest near the tail boom.

The tail rotor gear box was found attached to the tail boom with about 10 inches of the tail rotor blades broken off. Both outboard segments of the tail rotor blades were found in the accident area. One blade was found about 75 feet downstream of the fuselage. The other blade was found about 100 feet away from the fuselage in vegetation on the east bank of the stream.

Flight control and drive train continuity was established. According to the helicopter airframe manufacturer, the breaks in the drive train and flight controls were consistent with damage generated from impact with the ground.

TESTS AND RESEARCH

Engine Examination

The helicopter's engine was examined at the operator's facilities in Hanapepe, Hawaii, on April 21, 1994. There was no evidence or damage found in the engine compressor or turbine section that indicated there was a mechanical failure or malfunction with the rotating or stationary parts.

Fuel Control Inlet Filter

The fuel control inlet filter was tested at Allied Signal Controls and Accessories Division, Burbank, California, on June 1, 1994. The inlet filter was visually inspected and placed in a test fuel control. Metered amounts of fuel were introduced to the fuel control and differential pressures were measured across the control filter. The differential pressures were also measured from randomly selected fuel control inlet filters. The results of the test are attached to this report.

Fuel Spray Nozzle Assembly Flow Tests

The fuel spray nozzle assembly (P/N 6890917, serial No. AG83474) from the accident helicopter was tested on a flow bench at Helipower Services, Santa Paula, California, on June 15, 1994. The nozzle fuel strainer appeared to be collapsed when examined through the fuel inlet end. The nozzle assembly did not meet the flow limits published in the Allison 250-C20 overhaul manual. The tests measure fuel flow in pounds per hour at various specified inlet pressures (PSI). The following table depicts the test results.

Inlet Minimum Maximum Accident Pressure Flow Limit Flow limit Nozzle Flow

60 PSI 23 LB/HR 30 LB/HR 20 LB/HR 125 PSI 50 LB/HR 70 LB/HR 38 LB/HR 200 PSI 138 LB/HR 178 LB/HR 52 LB/HR 400 PSI 260 LB/HR 305 LB/HR 80 LB/HR

After the tests, the fuel nozzle was disassembled and the internal 229 micron fuel strainer was found contaminated and collapsed.

Fuel Filter Information/Research

The helicopter's airframe fuel filter, fuel control unit filter, and fuel nozzle filters are constructed of various micron mesh values. The progression of the filtration runs from fine to course with the 5 micron airframe filter being the finer of the three and the 229 micron fuel nozzle filter being the courser. The fuel control filter is rated at 64 micron. The fuel nozzle filter is final stage of fuel filtration before the fuel enters the combuster section of the engine and is ignited.

The engine manufacturer has established a time limit on the fuel nozzle of 2,500 hours, after which the nozzle must be overhauled. The fuel nozzle screen is not inspected by operators as part of any recurring inspection program approved by the engine manufacturer.

Fuel Filter Examination

The three fuel filters were submitted to Fowler, Inc., Gardena, California, for analysis. According to Fowler, Inc., the airframe filter "was relatively clean and only a few random particles were found. The fuel control filter and fuel nozzle filter were contaminated. The quantity of contamination in the fuel nozzle filter was more prevalent than in the fuel control filter.

The fuel nozzle filter was found completely covered with a brownish material. The brown materials were analyzed using energy dispersive spectroscopy (EDS). The EDS identified the materials comprised of sodium, oxygen, and carbon with small amounts of sulfur, chlorine, and calcium.

The fuel control filter unit was also examined. The EDS trace identified materials comprised of sodium with smaller amounts of aluminum, silicon, sulfur, and calcium. According to Fowler, Inc., the quantity of contaminants on the fuel control screen were much less than that found on the fuel nozzle filter.

Fuel Sample Tests

There was no fuel retained from the accident helicopter. The helicopter's fuel system was damaged to the extent any fuel on board at the time of the accident would have drained through the damaged areas. Two fuel samples were taken on April 19, 1994, from the truck that serviced helicopter before the accident. An additional sample was taken from the truck on August 12, 1994, and submitted for testing.

On April 20, 1994, a fuel sample from the truck was tested by BHP Petroleum Americas Refining, Inc. According to the test results, the sample met all routine delivery specifications. However, tests for sodium are not part of the routine.

Additional samples were submitted to Inspectorate, Carson, California, for analysis. The samples were tested specifically for sodium content using Atomic Absorption Spectrophotomer, (A.A.S.). During the test, a random "on grade" Jet A fuel sample and a salt-saturated Jet A fuel sample were introduced for comparison.

The analysis of the truck fuel sample [Sample B], taken one day after the accident in April, resulted in a .09 parts per million (PPM) quantification. Analysis of the truck sample [Sample A], taken in August after the accident, resulted in a .06 PPM quantification.

The random "on grade" Jet A fuel sample was analyzed three times. The first sample [Sample C] was analyzed as received and produced a .05 PPM sodium concentration. The sample was then mixed with salt crystals taken from the Salt Pond Park located adjacent to the Port Allen Airport in Hanapepe, Hawaii. The salted "on grade" Jet A fuel sample [Sample C(I)] was analyzed as received and produced a .04 PPM sodium concentration. The salted "on grade" Jet A fuel sample was then filter through a .8 micron millipore filter and tested for sodium content. The filtered salted "on grade" Jet A fuel sample [Sample C(ii)] resulted in a 0.04 PPM sodium concentration.

The Sample C(ii) was then mixed with concentrated salt water and left to settle. After settling, the Jet A was analyzed for sodium content and produced a 0.74 PPM concentration.

According to the Inspectorate Regional Lab Manager, the sodium concentration of the truck fuel sample [Sample B] taken one day after the accident in April would be considered minute.

Allison Compressor Wash Procedures

The engine manufacturer publishes procedures to remove all forms of contamination using a daily water rinse in Commercial Service Letter (CSL-1135, Revision 4, dtd. June 30, 1992). The procedure specifically lists "salt water air" as an example of contamination to which the procedure applies. CSL-1135 identifies the Hawaiian Islands as an area of severe corrosion.

The procedure specifies clean water should be sprayed at a rate of 1 quart in 9 to 11 seconds into the compressor inlet.

The manufacturer indicates the water injection should start about 3 seconds before energizing the starter motor to provide a full water flow rate to the compressor. The starter should be energized for 10 seconds without exceeding 10 percent engine N1 rpm while spraying the water. The water spray should continue during engine coast down after the starter is released.

If during the starter energize the engine N1 rpm starts to exceed 10 percent, the starter should be released and water spray continued. The N1 rpm should be allowed to reduce to 5 percent, and then the starter should be energized again to obtain a full 10 seconds of engine rotation while water is sprayed into the compressor.

The rinse procedure may be repeated if exposure to excessive salt has occurred.

CSL-1135 indicates the engine should be operated for 5 minutes within 15 minutes of the water rinse to purge and evaporate all residual water. The commercial service letter does not address when the 15 minute period begins, when the last water drains out of the combuster, or at the end of starter energizing.

Compressor Wash Water Drain Test

A compressor wash drain test was conducted by the Safety Board. An Allison 250-C20B combustion camber was installed in a fixture tilted about 45 degrees to simulate the installation in the helicopter airframe. Three quarts of water were poured against the side of the combuster. The water was directed away from the fuel nozzle. The flow rate was approximately 1 quart in 10 seconds. The rate at which the water drained from the combuster through the combuster drain valve was timed with the sweep second hand of a wristwatch.

The first quart of water immersed the fuel nozzle. After the 3 quarts were added it took 30.5 minutes for the water to drain. The fuel nozzle was immersed in the wash water about 25 minutes.

During the test, water was noted dripping from the fuel nozzle inlet. Disassembly of the nozzle revealed the filter screen was wet and water had seeped past the check valve in the fuel nozzle.

It should be noted the fuel nozzle was checked for proper assembly before the tests.

The test was performed again, but this time the fuel drain check valve was removed. Again, the nozzle was immersed by the first quart of water, but all of the wash water drained in less than 1 minute. Removal of the drain valve is not mandated by the Allison compressor wash procedures. However, it was reported to the Safety Board that many operators removed the drain plug to accomplish wash procedure.

ADDITIONAL INFORMATION

Wreckage Release

The wreckage was released to the helicopter owner on April 22, 1994, with parts retained for further examination. The parts were released by the Safety Board on August 17, 1994.

ALLISON RESPONSE TO INTIAL DRAFT

Allison Engine Company responded to the NTSB Factual Report Draft Narrative on February 24, 1995. A copy of the response is included in this report.

NTSB Probable Cause

INADEQUATE TURBINE ENGINE COMPRESSOR CLEANING PROCEDURES BASED ON INFORMATION IN THE MANUFACTURER'S SERVICE LETTER, AND LOSS OF ENGINE POWER DUE TO BLOCKAGE OF THE FUEL NOZZLE STRAINER WITH FOREIGN MATERIAL (SODIUM). FACTORS RELATED TO THE ACCIDENT WERE: THE LACK OF A SPECIFIED SERVICE REQUIRMENT FOR INSPECTION OF THE FUEL STRAINER, AND TERRAIN CONDITIONS IN THE EMERGENCY LANDING AREA.