Nepalese investigators have been unable to determine why a Dornier 228 lost thrust while climbing out of Kathmandu, leading the turboprop to stall and crash with the loss of all 19 on board.
Although the aircraft hit a bird - a black kite weighing some 700g - while at 70kt on its take-off roll, there was no evidence of ingestion.
The Sita Air aircraft lifted off at 86kt and its landing-gear was retracted. But audio analysis of cockpit-voice recorder signatures indicates that one of its engines ran down to 95% of nominal speed, and then 91% shortly afterwards.
Nepal's governmental air accident commission says the Dornier achieved a nose-up attitude too high to maintain its climb-out speed of 83kt.
It levelled off at a height of 100ft, having decelerated to 77kt, and flew level for 14s. The airspeed continued to bleed away - noticed even by the tower controller - and the aircraft drifted to the left.
The stall warning sounded before the turboprop, entering a left turn, pitched down and struck the ground about 420m southeast of departure runway 20's far end.
Investigators pursued several potential leads with the inquiry. The inquiry discovered loading inconsistencies and calculated that the aircraft was 78kg over the maximum take-off weight for the flight. Trim sheets recorded incorrect positions for the centre of gravity, although the aircraft was not out of trim.
Surveillance video captured a flash in the region of the right-hand engine during the take-off run which appeared to coincide with the sound of a small bang, and a momentary dip in the rotation speed of one engine.
But the inquiry has not been able to tie these circumstances and events conclusively to the accident.
It says the aircraft suffered an "insidious reduction" in power at around 70kt, about the same time as it hit the bird, but that this "seems not to have been noticed" by the pilots because there was no decision to reject the take-off.
As the aircraft lifted off there was a "step change" in the rate of increase in total energy, says the accident commission, and after a further 4s the energy calculations indicate there was "more drag than thrust".
Combined evidence from the Honeywell TPE331 engines and the propellers indicated they were operating at around flight idle power at impact.
"In an accident scenario it is not uncommon for a pilot to pull the power levers back to idle just before ground impact," the inquiry points out.
But it notes that a fuel flow problem "could not be entirely excluded" as a potential source of the power loss.
Extensive damage to fuel-control components prevented further exploration of this possibility. But the inquiry found that the flight-idle fuel flow, which is set while on the ground, had been fixed below the correct level. Investigators issued an interim recommendation in November last year to check that Dornier 228s had correctly-set fuel flows.
"It is possible that there was a problem with one or more engines that caused the power to reduce to a near flight-idle condition," says the inquiry. "A problem which resulted in a lower-than-commanded fuel-flow rate would cause a loss of power."
But the commission admits that it is "unable to determine" the reason for the thrust reduction which led to the 28 September 2012 accident.
The inquiry observes that a fuel flow problem could potentially be responsible for an engine surge - which might, in turn, have been behind the flash and bang during take-off - but the powerplant manufacturer states that the engine type is "very resistant" to surge.
It also believes that the struck bird was "unlikely" to have caused a surge, by disturbing the engine airflow, without being ingested.