A twin with a history of fuel problems tricks yet another generation of pilots
A modification to install fuel panel warning lights may need to be installed on all Convair 580s, not just those of Air Tahoma, which had a Convair go down last year.
At 12:47 a.m. on Aug. 13, 2004, Air Tahoma Flight 185, a Convair 580, reported engine trouble just before crashing onto a golf course short of runway 36R at Cincinnati/Northern Kentucky International Airport. The twin-engine turboprop hit a tree and broke in half. The cargo flight was arriving from Memphis, Tenn., and was carrying out a visual approach. At the time the National Transportation Safety Board (NTSB) said that the Air Tahoma flight had appeared completely routine until its final minutes, and would not speculate as to what had brought the plane down into a Florence golf course just 1.2 miles short of its intended runway. But investigators did say the crew was heard on the cockpit voice recorder talking about low power from first, and then both, of the plane’s engines. The last radar contact with the plane, more than a mile south of the airport, showed it was moving at what was known to be a Convair 580’s normal landing approach speed.
The co-pilot was killed in the crash. His body was finally found buried in the gnarled wreckage of the tail section and it was removed around noon the following day. The pilot was able to walk away and was treated at University Hospital for his injuries. It was the first fatal aircraft accident at the busy Cincinnati airport in 21 years. Local emergency rescue personnel were on the scene shortly after Flight 185 went down. Boone County Coroner Doug Stith, who was notified of the crash at 1:38 a.m., said the co-pilot was pronounced dead at the scene. “There was an enormous amount of jet fuel around,” Stith said. “It’s a wonder it didn’t ignite. The fact that it didn’t catch fire is probably what saved the pilot’s life.” Little did he know that by the time the aircraft hit, the engines had cooled off and the electrics were long dead.
Without Warning
The plane had received clearance to land at 12:44 a.m., with no sign of trouble. Three minutes later, at 12:47 a.m., the pilots indicated they were having engine problems. The tower responded by asking if the pilots wanted emergency trucks to respond. At a press conference, NTSB Member Carol Carmody said the pilots told the tower “negative,” saying that they were “tracking direct” (for the runway). At that point the plane’s altitude was 2,200 feet and its speed was 150 knots, or about 173 mph. The final radar contact was less than three minutes later, with the air speed down to 110 knots or 127 mph and the altitude at 1,200 feet. The typical approach speed for a Convair 580 is 130 knots or 150 mph. 110 knots or 127 mph is the touchdown speed. A preliminary review of the cockpit voice recorder (CVR) indicated that the pilots had animated conversations about a minor control problem, fuel management, and “low power in first one, and then both of the engines.” NTSB Member Carol Carmody said that, significantly, both the voice and data recorders quit operating just before impact.
Relevant Accident History of the CV580
The Convair 580 is typically used to haul freight, although the plane can hold up to 56 passengers. Convair 580s have been involved in 24 accidents or crashes since 1963 involving 196 fatalities. The last fatal crash of a CV580 was in October 2003 in New Zealand. The Air Freight NZ flight was carrying mail from Christchurch to Palmerston North on a stormy night. People reported engine noises, a bang and objects falling from the sky. The two crewmembers were killed.
As is often the case, the clues to the Achilles heel that had brought down Flight 185 was only to be found in this aircraft type’s accident history. There turned out to be ample precedents, starting with a United Airlines (Convair) incident in the early sixties. In that incident, the captain had been trying to transfer fuel from one tank to another, in contravention of instructions in the airplane flight manual (AFM). Both engines quit due to fuel starvation, and the aircraft was dead-sticked into a carrot patch just north of the San Fernando Valley in California, with little damage. It was flown out four days later by a UAL management crew.
On Jan. 20, 1989, an Aspen Airways plane landed on a dirt road, collapsing the nosewheel (and was written off) after “inadvertent deactivation of fuel to the left engine” following a shutdown of the right engine. On Dec. 6, 2001, a CV580 freighter of Trans Air Link (TAL) was written off following a night ditching off Sunny Isles, Fla. The engines had failed after the captain “began transferring fuel from the left tank to the right engine.” On Feb. 22, 2005 a TAM CV580 crashed just after takeoff at Trinidad Airport, Bolivia, when both engines died – injuring 28. This fuelish thread has been continued by the Air Tahoma accident.
About 33 minutes before the accident, the Air Tahoma captain announced: “Just gonna balance out the fuel here.” Copilot (flying): “OK.” During descent for their visual approach, the maximum airspeed clackers sound and the two pilots laugh openly about that. There’s no evidence of any descent checklist. Seven minutes before the accident they report “field in sight,” and are cleared for a visual approach to 36R with an air traffic control (ATC) request to “keep their speed up.” A minute later (still doing 250kts indicated), ATC again enjoins them to keep their speed up but the copilot is unhappy with the plane’s handling. A minute further on and the copilot gives voice again to his rising concern: “Man, I’m telling you. What is wrong with this plane? It is really funny. I got something all messed up here.” Sound of landing gear warning follows (assume here that throttles have been reduced to idle in order to slow and extend gear, thus triggering the undercarriage safety horn). Copilot: “Can you feel it? It’s, like, swinging back and forth.” Captain: “Yeah, we’ve got an imbalance on this (expletive) crossfeed that I left open.” Copilot: “Oh, that’s what it is?”
Turboprop engines will run on a “not too lean” and aerated mixture of jet fuel, particularly if they’ve got continuous ignition – so what the copilot felt was possibly a pulsing of variable torque as an engine increasingly sucked entrained air, at least on the port side. Landing with a considerable/gross lateral fuel imbalance becomes more of a handling issue as speed reduces and ailerons/rudder become less effective, and it can be a real issue (wing drop) in the landing flare. The captain knows this so he goes for a last minute “save” – although he must realize that the fuel will take time to reach anywhere near balanced. Expletives follow until the copilot says (evidently acquiescing in reference to the captain’s hand motion of reaching back and up to flick the fuel valves on the overhead panel): “We’re gonna flame-out if we don’t, aren’t we?” The captain responds: “I got the crossfeed open, just keep the power on.”
A sound is heard similar to decreasing engine RPM. Copilot: “(Expletive) we’re losing power.” Captain: “On number … OK.” Copilot (flying): “We’ve lost both of them, did we?” Captain: “Nope.”
The CVR recording ends (simultaneous with FDR power loss) as both generators drop off line.
The captain later told investigators that he did not recall whether he had turned on the crossfeed valves, which move fuel from tanks on different sides of the aircraft. However, investigators found no fuel residue on the left side of the accident scene and found that the fuel controls’ positions indicated that the pilot had been draining fuel from the left side to the right side. Such a procedure is not normally done, according to information included in the report, but may have been necessary to balance the plane properly for handling on landing – a key concern since the flight was fully loaded with cargo headed for DHL’s sorting hub.
Air Tahoma president Noel Rude said the airline had already started installing fuel panel warning lights to indicate when a cross-feed operation was in progress.
The Crossfeeding Frenzy
It may be simply that the crossfeed manifold is vulnerable to taking gulps of air at lower fuel states when receiver-tank pumps are switched off (and tank-valves opened) to allow another tank’s fuel-pump pressures to push fuel in. This might be due to wing dihedral. It may also be due to the two engines being mounted high up on the low wing (in comparison to the C130’s underslung and the Electra’s level engine-to-wing relationship). If the engine’s higher than the tank (as in the MD-11’s tail engine, say) there’s no way that you can gravity-feed, even though it may be only required for a short period while another tank is filling the cross-feed manifold with fuel under pump-pressure. Engine- mounted or driven boost-pumps would only suck air if the positive head of low pressure fuel isn’t already there from a tank supply line. Whatever the reason, the Convair AFM says (and probably for good reason) not to do it near the ground.
Forgetting about an underway crossfeed set up in cruise and thereby creating a gross imbalance is a sure-fire way of later sucking yourself into a panic-stricken switch- flicking of fuel-switches at lower fuel-states during an approach. The CV580 used to be a CV440 Metropolitan with reciprocating engines – in its earlier life. Piston engines don’t flame-out when you rob them momentarily of fuel – but then again they’re not ingesting the same volumes at the higher flow-rates of a turboprop. Maybe the turboprop conversion personnel put insufficient thought into that aspect (the lower-slung CV440 wing’s tanks and dihedral) when they dreamt up the turboprop conversion STC (Supplemental Type Certificate).
Here’s a further thought. Looking up and back over your shoulder at an overhead fuel panel full of valves, switches and indicators is somewhat akin to trying to shave the back of your neck using two mirrors. The gymnastics just increases the chance of getting it wrong (and also of forgetting about any interim set-up – as in “outta sight = outta mind”). And once both engines have flamed out, where will the electrics come from that are needed to move the fuel-valves, light the indicators and run the pumps? And is there a battery- based hot ignition relay for attempting relight? Perhaps the final report will address that detail.
One final thought might be that if you ever needed “crash”-proof of the value of a pre-descent checklist, this one might be it – as long as they mount those new fuel-panel warning lights down where they will be seen and if their critical status is written into a descent checklist. But will those new fuel-panel warning lights be going into all CV580s – or just Air Tahoma’s? It appears that it will probably be a local modification only and not springing from the mod squad (certificate holder Kelowna). If so, then this deadly gotcha can maintain its enduring status throughout the far-flung CV580 fleet as a “lesson never learnt.”