1950 Australian National Airways Douglas DC-4 crash

Investigation

Three investigators from the Department of Civil Aviation began work at the crash scene the day after the accident. They found the Amana had crashed in a heavily timbered area on the Inkpen family property Berry Brow, on the easterly track between Perth airport and Kalgoorlie, at a point where the elevation was about 1100 ft above sea level. The aircraft struck the tops of tall gum trees while descending at an angle of about 15°. Its speed at impact was estimated at 250 mph. It crashed through large trees, breaking them off, before impacting the ground violently and gouging a long, wide furrow. The left wing was torn away from the fuselage and then the aircraft broke up and burst into flames. Only the rear fuselage with the fin and rudder were not affected by fire. The wreckage trail was about 280 yd long and 35 yd wide. At the time of impact the Amana's left wing was lower than its right, suggesting it may have been turning left. It was heading north, not east towards Cunderdin. Investigators speculated that the crew may have been turning with the intention of returning to Perth airport; or they may have been preparing for a crash-landing in a large clear area to the north of the crash site.

Possibly as a result of rough-running of one or more of its engines, the Amana was observed flying over Perth's outer-eastern suburbs at an unusually low altitude. No witness report was received from anyone along the next 16 nmi of the Amana's track from Perth's outer suburbs to within 5 nmi of the crash site. In the minute before it crashed, eight witnesses heard a large aeroplane in distress and reported unusual engine noise, including engine noise ceasing on at least one occasion, followed by the sudden return of loud engine noise. This suggested that, on at least one occasion, none of the engines were producing power, followed by a resumption of power on some of the engines. The investigation team concluded that the Amana failed to reach its assigned altitude of 9,000 feet, and that it experienced intermittent engine problems of such severity that all engine power was lost on at least one occasion. Without power and with only one of its propellers feathered, a Douglas DC-4 loses altitude at a great rate, possibly as fast as 100 feet per second (6,000 feet per minute).

Engines and propellers numbers 1 to 3 suffered substantial damage in the crash, but engine and propeller number 4 suffered much less damage. The investigators determined that at the time of impact, propellers 1, 2 and 3 were turning normally and their engines were producing power but propeller number 4 was feathered and its engine was not operating. There was also some evidence that action was taken by the crew to unfeather propeller number 4 in the moments before impact. None of the engines contained evidence of any internal failure prior to impact. All the magnetos were tested and the results indicated normal ignition was available to all engines up to the time of impact.

Engine number 4 suffered only minor, external damage. It was dismantled by the investigation team in an attempt to determine why it might have been shut down by the crew. A corrosion product was found in the passages of the fuel flow meter on engine number 4. Western Australia's Deputy Mineralogist identified it as magnesium hydroxide. This is formed by reaction of magnesium and water, suggesting the fuel passages had been filled with water in the months between the crash and the detailed examination of the engine. Charles Gibbs, an engine specialist employed by the Department of Civil Aviation, estimated at least 45 cubic centimetres of water must have been involved. Rain falling on the crash site before engine number 4 was removed could not account for this much water in the fuel passages. Gibbs first examined the fuel system of engine number 4 and discovered the corrosion about two months after the accident. He conducted a test on an identical flow meter and found that after he left water in the fuel flow passages for approximately 8 weeks a similar amount of corrosion product developed. This suggested the rough running heard by witnesses on the ground may have been caused by water in the fuel reaching engine number 4. The steel rotor in the fuel pump of engine number 1 was slightly corroded but the fuel systems of engines 2 and 3 showed no evidence of corrosion. Investigators formed the opinion that the rough running heard by witnesses on the ground, and the crew's decision to shut down engine number 4 and feather its propeller, may have been related to water in the fuel reaching that engine. The intermittent loss of power on all engines in the final minutes of the flight may indicate that all engines were receiving fuel contaminated with water.

The only abnormality found in all four engines was the vapour vent float in the fuel strainer chamber of the carburettors. The floats had been crushed by extreme fuel pressure. Inquiries were made to the engine manufacturer and other civil aviation authorities but none had prior experience of vapour vent floats collapsing. Tests on carburettors were also carried out in Australia by the Aeronautical Research Laboratories but without finding any suitable explanation. Whether the floats were crushed in flight or in the crash could not be determined, but even if it had occurred in flight it would not have affected operation of the engines.

The earliest reports from the crash site speculated that the Amana was already on fire when it struck the tops of trees because those trees, and pieces of the aircraft's left wing torn off in the impact with them, showed signs of scorching. Several eyewitnesses reported seeing flames in the sky before the aircraft struck the ground. Department of Civil Aviation investigators discounted this speculation because only one of the Amana's push-button engine fire extinguishers had activated and this had most likely occurred during the crash or the fire.

Australian National Airways (ANA) ground staff in Sydney checked the Amana's fuel tanks for the presence of water prior to its first departure on 26 June. They found none. The Amana was subsequently re-fuelled in Melbourne and Adelaide but no check of the fuel tanks was made on these occasions. After being re-fuelled in Perth immediately prior to the fatal flight, the fuel filters in all 4 engines and the fuel drain serving the cross-feed pipe in the wing centre-section were all checked for the presence of water. The fuel tanks themselves were not checked, partly because, on the night of 26 June, the ground staff were "pressed for time" because one despatch engineer was absent due to illness.

ANA was of the opinion that if a small amount of water entered a fuel tank during refuelling it would only reach the drain cocks when the aircraft was in level flight so it could not be detected immediately after re-fuelling. For 15 years ANA had operated in the knowledge that the only satisfactory time to check fuel tanks for the presence of water was prior to the first flight of the day, after the aircraft had been stationary overnight. Throughout this time ANA checked fuel tanks for the presence of water prior to the first flight of the day.

Prior to its final flight, the Amana received 1756 USgal of fuel from a tanker operated by the Vacuum Oil Company. The tanker had been checked for the presence of water in the morning and again at 6:30 pm, about 2 hours prior to refuelling the Amana. It had also supplied fuel to 3 de Havilland Dove aircraft, none of which suffered any engine problems or were found to have water in the fuel.

The Department of Civil Aviation performed tests on parts of the DC-4 fuel system. Tests on the engine fuel system showed that when the engine boost pump was operating, a vortex formed in the engine fuel tank. If a small amount of water was present, this vortex held the water in suspension and prevented it from entering the engine. The tests also showed that when the boost pump was turned off, the vortex dissipated and any water would soon find its way into the engine. Investigators believed this might explain why all engines were operating normally during the takeoff but at least one engine began to run roughly around the time the engine boost pumps would be turned off.