This month Steve Ayres reflects on where and when a fresh crack may have lethal consequences and whether some other form of warning is essential if carbon monoxide could be present
Steve Ayres
14 April 2021
Researching this month’s column on St Patrick’s Day (17 March) led me to reflect on the two spellings for what is essentially the same word and how, when used in aviation, it is about as far away from the term ‘good craic’ – frequently used in the north and south of the Emerald Isle – as you can get. That, and a recent event at the airstrip, triggered more thoughts on cracks, how they propagate and when they might first be detected. But, it was an accident that was reported in the UK this month that convinced me something more than simply inspecting exhaust systems for cracks was needed to prevent being poisoned.
The pilot and five passengers of a de Havilland Canada DHC-2 Beaver floatplane, boarded the aircraft for a return charter flight from Cottage Point to Rose Bay, Sydney. Shortly after take-off, the aircraft conducted a 270° right turn in Cowan Water and then entered Jerusalem Bay, below the height of the terrain. The aircraft stopped climbing, continued along the bay and then made a very steep right turn. The aircraft’s nose then dropped and it collided with the water. All on board were fatally injured and the aircraft destroyed.
“Higher than normal levels of carboxyhemoglobin likely due to elevated CO levels in cabin”
Some of the circumstances regarding the accident were unexpected, given the nature of the operations and the pilot’s significant level of experience. Specifically, the aircraft entered a known confined area (Jerusalem Bay) below the height of the terrain, with no need to be operating in the bay. The aircraft did not continue to climb despite being in the climb configuration.
The aircraft was capable of turning within the bay, it could have been turned earlier, and there was sufficient distance remaining to land at the position of the steep turn. A steep turn was performed at low-level and at a bank angle in excess of what was required. It was established that pilot control column and rudder inputs were necessary to travel at least half-way through the final steep turn as observed.
However, the propeller was at a ‘lower power condition’. The aircraft likely aerodynamically stalled, with insufficient height to effect a recovery before colliding with the water.
Further, the front seat passenger was regularly taking photographs, but stopped during the turn in Cowan Water, and it was very likely the middle right passenger was unrestrained at impact.
Toxicology results identified that the pilot and passengers had higher than normal levels of carboxyhemoglobin in their blood. This was almost certainly due to elevated levels of carbon monoxide (CO) in the aircraft cabin.
Wreckage examination established that several pre-existing cracks in the exhaust collector ring very likely released exhaust gas into the engine/accessory bay, which then very likely entered the cabin through holes in the main firewall where three bolts were missing from the magneto access panels.
In addition, the examination also found that the in-situ bolts used by the operator’s external maintenance provider to secure the panels were worn, and were a combination of modified AN3-3A bolts and non-specific bolts.
A 27-minute taxi, with the pilot’s door ajar, before the passengers boarded, likely exacerbated the pilot’s elevated carboxyhemoglobin level. As a result, the pilot would have almost certainly experienced effects such as confusion, visual disturbance and disorientation. Consequently, it was likely that this significantly degraded the pilot’s ability to safely operate the aircraft.
The aircraft was fitted with a disposable CO chemical spot detector, which was likely not effective due to sun bleaching. Commonly used in General Aviation, these types of detectors have known limitations and can be unreliable at detecting CO in the cabin. Further, they are passive relying on the pilot regularly monitoring the changing colour of the sensor to detect elevated levels of CO. In contrast, electronic CO detectors are designed to attract the pilot’s attention through auditory and/or visual alerts, so are more likely to be effective.
While inexpensive and readily available, there was no regulatory requirement from the Civil Aviation Safety Authority for piston-engine aircraft to carry a CO detector with an active warning. Similarly, other international investigation agencies have made safety recommendations to aviation regulators to mandate the carriage of active detectors.
However, despite the ongoing threat CO exposure poses to aircraft occupants, these recommendations have not been accepted.
Consequently, the Board has recommended that the Australian Civil Aviation Safety Authority consider mandating the carriage of active warning CO detectors in piston-engine aircraft with a maximum take-off weight less than 5,700kg. In addition, while the aircraft carried a passive CO detector, there was no mechanism for monitoring their serviceability.
On the day of the accident the pilot decided to go flying to stay current as the weather was forecast to be good. He took 20 litres of mogas, in a jerry can, which he stated would have given about 90 minutes endurance. Prior to departure the pilot recalled checking Notams on his portable electronic device and placed it on the passenger’s seat and started the aircraft. He remembers taking off into wind but not what runway he used or whether he used a grass or concrete runway.
“The pilot remembered taking off into wind but not what runway he used…”
His only recollection of the flight was leaving the circuit in a northerly direction for a period of time, but did not go so far as to lose sight of the airfield, before returning to the airfield to fly some visual circuits. A witness, located about one nautical mile north of the airfield, saw the aircraft downwind in the visual circuit and commented that there appeared nothing untoward with the aircraft. The pilot’s next recollection was regaining consciousness at a very low altitude but too late to recover the aircraft before it struck the ground, he then lost consciousness. The pilot’s next recollection was him being tended to by another witness to the accident. Police, RFFS vehicles and ambulances started arriving at the scene 37 minutes after the accident. Due to the limited access to the scene and the pilot’s injuries, an air ambulance also attended.
From the ground marks it was evident the aircraft was heading east towards the airfield at the time of the accident. The first impact marks were made by the right wing tip, which touched the ground three times before the leading edge of the wing hit a small bus parked along the tree-line of the west field. On striking the bus, the wing detached and landed by the bus, while the remainder of the fuselage and left wing glanced off a flat roofed cabin located to the left of, but in line with, the vehicle.
The fuselage bounced off the top right side corner of the building’s roof and continued into the tree-line. The left wing and fuselage sliced through the trees before hitting the east field at a shallow angle. The remaining wing detached and landed to the left of the aircraft’s path. The fuselage continued sliding along the ground before finally stopping in a slightly right nose down attitude pointing towards the airfield.
The propeller’s spinner had large impact dents but there were no radial score marks to indicate the propeller was rotating under power when it hit the ground. It is likely the engine stopped when the propeller sliced through the trees, causing one blade to separate near the root and land in the east field, while the other was bent backwards but remained attached to the hub.
The investigation found evidence of exhaust system gas leakage in the engine bay and pathways by which the gas could have reached the cockpit via the degraded firewall seals and grommets. Although leakage may have been minimal, the effects of CO are cumulative and would have built up over the duration of the flight. The available evidence is consistent with the pilot having suffered CO poisoning and being incapacitated before the accident occurred. Although he reported regaining consciousness, it was not in time to prevent the accident.
Let’s face it metal cracks, composite too, but most of the time experience tells us where to look for possible signs of damage.Airframe design has improved to the point where cracks are rare, mostly predictable and seldom critical, at least initially. Add a bit of thermal cycling, plenty of nasty harmonics and a raft of random stressors and the predictable becomes less so. I am, of course, referring to the engine bay where all these things are routinely in play.
Regular checks mean that most of the time we pick up the big stuff but it has left me wondering if every 50 hours or every 150 is really often enough to check on something as critical as an exhaust leak. Yes, we all inspect exhaust systems as part of our pre-flight routine, but when cowlings remain fitted, that usually limits us to a visual check at arm’s length, or just handling those sections we can see for security that poke out the bottom. And then it would be down to the integrity of the firewall and those leaky bits of rubber, RTV, caulking or heater flap to keep you safe. As the other accidents bear testament, even the smallest of leaks can be lethal.
The Beaver pilot may well have experienced mild CO poisoning before he even took off, simply because he left the cabin door ajar with the engine running while waiting for his pax and during taxi. Ultimately though, the critical source of CO was a cracked pipe and these incidents remind us, cracks can occur at any time and can be difficult to spot when they do.
I had been lucky to watch the Beavers plying the waters around Pittwater, north of Sydney, earlier in the year so I followed this investigation closely. The initial findings only served to deepen the mystery. Rarely had I read of a crash where the engineering, operational and flying documentation was of such a meticulous standard. Especially when associated with what appeared to be a reckless bout of flying indiscipline. Something was out of place.
In the end, though, the only explanation for the pilot’s behaviour was down to three missing AN3-3A bolts from an access panel that let sufficient CO into the cabin from a cracked exhaust. Fixing any potential leak paths through the firewall is probably the last line of defence, but given how fickle cracking of exhaust systems can be, is that something to be relied upon? Regular inspection of the exhaust system is vital but fitting an active CO monitor on all aircraft is surely the only effective last line of defence… and that is what the Aussie investigators said. So, I have just fitted one!
