Centrum Naukowo-Produkcyjne-PZL M-18A, 1Z024-14

Pilot and eyewitness descriptions of events immediately prior to the accident suggest that the engine lost a substantial degree of power, but did not completely fail.

It is likely that the engine power loss was gradual. This was evidenced by the pilot's report of the aircraft's increasing nose-down tendency and its extended flight profile. Examination of the engine indicated that it was producing limited power at impact. There was no evidence of any pre-existing condition that may have adversely affected the serviceability of the engine.

A significant amount of water and other contaminants were found in the firewall fuel filter.

Inspection and testing indicated that the water had been in the filter a significant length of time. The water in the carburettor bowl and the fuel pump was clear and the lack of corrosion or other damage inside the carburettor and the pump indicated that the water had only recently been introduced to those components, possibly during the post-accident firefighting operations. No factors, other than water contamination and the deterioration of the fuel filter, were disclosed during the investigation. Consequently, it is considered that the deterioration of the filter most likely contributed to the engine's loss of power.

There was no evidence to indicate that the fuel filter had been removed and inspected during the periodic inspection conducted for the initial issue of an Australian aircraft maintenance release or at any subsequent time prior to the occurrence. The aircraft maintenance schedule elected by the Certificate of Registration holder would not have required a further inspection of the filter until 100 flight hours had been completed after the initial inspection.

During the investigation, it became apparent that there was a general misunderstanding of the intent of certain aspects of the CASA maintenance schedule contained in CAR42B Schedule 5.

Two aircraft were being used to suppress bushfires in the Perth
basin. At approximately 1027 WST, the aircraft were called to a
fire near Maddington, an outer suburb of Perth approximately 6 NM
south-east of Perth airport. The pilot of VH-JTD reported that no
problems were identified during the pre-takeoff checks.

Both aircraft departed at approximately 1035 towards the south.
Air traffic controllers in the tower noted that JTD, the accident
aircraft, flew a shallower departure profile than the other
aircraft. The pilot later reported that the profile resulted from
his operating technique and not from any problem associated with
the aircraft.

Because the bushfire was relatively close to the airport, both
aircraft climbed to only 1,200 ft before descending to conduct the
inspection circuit at about 1,000 ft. While communicating with the
bushfire ground controllers, the pilots positioned their aircraft
in a left circuit pattern to reduce the time during which the
aircraft were climbing out over populated areas. The pilots then
approached the fire from the south-east. The first aircraft dropped
its load of fire retardant without incident.

The pilot of JTD reported that he set up the approach slightly
tighter and steeper than usual due to the proximity of housing.
Just prior to the base turn, he selected full flap and set the
propeller to a fine-pitch setting.

During the base turn, the pilot reported that he needed more
elevator control back pressure than usual to maintain the
aircraft's nose attitude. The pilot initially thought that the
aircraft's trim setting needed adjusting, until the nose dropped
further. The pilot levelled the wings and fully opened the
throttle: this appeared to have little or no effect and he decided
to immediately dump the load of retardant. He could not recall any
engine instrument indications. Because the dump system had been set
up for a partial dump only, he decided to use the manual lever
rather than the electric switch to dump all the fire retardant.
Following the loss of engine power, the pilot turned on the
emergency fuel pump as required in the emergency checklist, but did
not apply carburettor heat.

The pilot reported that the aircraft appeared to travel further
than expected during the descent. He flew the aircraft close to the
stall at approximately 60 kts during the forced landing, using
rudder to control wing drop. Having passed over several streets,
the pilot decided to attempt to land the aircraft on a small
street, parallel and close to the aircraft's flight path.

The aircraft severed a set of powerlines at the northern end of
the street before its right wing struck and broke a power pole. The
aircraft then veered right, glanced off the roof of an occupied
house, and embedded itself into the next house, which was
unoccupied at the time.

Once the aircraft had come to rest, the pilot switched off the
fuel and electrical systems. There was no fire. He then exited the
aircraft and signalled to the other pilot that he was okay. The
pilot was wearing a four-point harness and a flying helmet, both of
which were undamaged. However, the pilot received minor injuries to
his left leg during the impact sequence. The nature of the impact
sequence appeared to have progressively slowed the aircraft,
thereby lessening the peak forces suffered by the pilot.

When the other pilot saw that JTD had crashed, he advised air
traffic services of the accident and gave directions to enable
emergency services to locate the accident site. Emergency services
arrived about 5 minutes later. The pilot switched off the operating
emergency locator transmitter.

Wreckage examination

The aircraft's forward fuselage was substantially damaged and
the main landing gear had collapsed. The wings remained attached to
the fuselage although powerlines had severed 2 m of the left wing.
The cabin was penetrated by two pieces of wood from the house roof
trusses. The engine was torn from its mounts, moving 1 m forward
and rotating approximately 90 degrees to the left.

The on-site examination of the engine revealed no abnormality.
The carburettor was firmly attached and all four butterfly valves
were fully open although their positioning may have resulted from
the impact. The propeller remained attached to the engine with all
blades bent backwards. The general appearance of the blades
indicated that the propeller was rotating at low power at
impact.

Meteorological information

The Bureau of Meteorology reported that there was 2 octas of
cumulo-form cloud at 4,500 ft. The temperature was approximately 35
degrees C, dewpoint was 16.5 degrees C, and there was a relative
humidity of 38% at Perth airport. There was a light wind from the
north-east.

Fuel system

The fuel system included two wing tanks connected to a header
tank. Fuel from the header tank was drawn by the engine-driven fuel
pump and fed through a firewall-mounted filter before entering the
carburettor.

The operator reported that the aircraft departed with a fuel
load of approximately 700 L and that approximately 40 L of fuel was
drained from the left tank soon after the accident. It was
estimated that about another 20 L was spilled during the aircraft
wreckage recovery. The right wing tank contained no fuel when it
was inspected after the accident; however, the fuel would have
drained from the fuel lines that were damaged during the collision
sequence.

A sample of fuel was drained from the left wing tank and
analysed. The analysis confirmed the fuel as Avgas 100/130. Apart
from an extremely low Reid Vapour Pressure, it complied with the
fuel specification requirements. The fuel sample was taken almost
24 hours after the accident. The fuel manufacturer reported that
due to the hot weather in Perth at the time, the lighter fuel
fractions may have evaporated from the ruptured tank, resulting in
the low vapour pressure.

A small amount of clear water was found in the fuel sample
drained from the fuel pump.

Fuel lines to and from the fuel filter, mounted on the firewall,
were severed at the filter. The fuel filter did not have an
external or cockpit warning to indicate when the filter was blocked
and operating in bypass. The aircraft's flight manual noted that
the fuel pressure indication dropped when the filter was blocked,
but the pilot reported that he did not observe such an indication.
A mixture consisting of 50% fuel and 50% black-coloured water was
drained from the filter. Chemical analysis of the water found the
presence of dissolved impurities.

Examination of the filter element found the lower part saturated
with water and heavily contaminated with rust particles to the
point that the lower half of the filter was blocked. On the fuel
inlet side, there was a significant accumulation of large, flat
rust pieces. The lower half of the steel coil inside the element
was severely corroded. Analysis indicated that the large rust
pieces probably originated from a fuel storage tank. Specialist
examination of the filter indicated that the water and subsequent
corrosion had been present for an extended period.

Approximately 250 mL of fuel and a similar quantity of water
were drained from the carburettor bowl. Both samples were free of
sediment. The carburettor was a standard float type, where the
pressure difference between the chamber and venturi regulated the
amount of fuel drawn. The carburettor was opened and inspected. No
sediments, or evidence that water had been present for an extended
period, were found. The examination of the carburettor found no
defect that may have adversely affected its operation.

The aircraft was manufactured in 1993 and operated in the USA
before its purchase by the Australian owner in 1996. The aircraft
Certificate of Registration holder made a valid election to use the
Civil Aviation Safety Authority (CASA) maintenance schedule, CAR42B
Schedule 5, as the aircraft maintenance schedule and the aircraft
had undergone a periodic maintenance inspection prior to the issue
of an Australian maintenance release. A subsequent periodic
maintenance inspection would then be required at the completion of
100 flying hours or 12 calendar months.

An option available to the aircraft Certificate of Registration
holder was to elect to use the aircraft manufacturer's maintenance
schedule, which would require that the fuel filter be inspected
every 50 (+/- 5) flight hours.

On 5 February 1999, CASA informed the Bureau that the aircraft
maintenance schedule, CAR 42B Schedule 5 Part 1, required that a
fuel sample drain of aircraft fuel filters be conducted each day
the aircraft was flown, except when not applicable to the aircraft.
In the case of the PZLM18A aircraft, CASA stated that the engine
firewall fuel filter would not be required to have a fuel sample
drain carried out during the daily inspection.

The Civil Aviation Safety Authority confirmed that the firewall
fuel filter should have been inspected during the initial periodic
inspection for the issue of the aircraft maintenance release.

At the tiine of the accident, the aircraft had accumulated
approximately 53 flying hours since the completion of the periodic
inspection. The investigation could find no evidence that the
firewall fuel filter had been inspected at any time prior to or
during flying operations in Australia.