Usable information about the UPS A300F crash at Birmingham, Alabama has been coming in since the NTSB has downloaded the cockpit voice recorder (CVR) and flight data recorder (FDR).
The aircraft was approaching runway 18 in good visibility before dawn, with a slight tailwind (340deg/04kt). Final approach crosses empty fields, there are no runway approach lights and the runway itself only has white edge lighting, so there was potential for the “black hole approach” illusion to develop. The crew was carrying out a non-precision localiser/DME approach because there is no ILS glideslope for 18, but the visual PAPIs (precision approach path indicators) were working and the NTSB has since confirmed that they are correctly set.
The FDR shows the autopilot was engaged all the time until the recording ended. The autothrottle was also engaged and the aircraft was holding 140kt indicated airspeed for the final approach. There is no initial evidence of control or engine malfunction.
With 16s to go to the end of recording, – about 1.4nm to the runway – there was a “sink rate” alert from the enhanced ground proximity warning system. Three seconds later one of the pilots said “runway in sight”, though this may have been a confirmation rather than a first sighting.
At 9s to the end of recording there was the first sound of impact. The aircraft had hit trees about 1.15nm from the runway. Final impact with rising ground took place 0.8nm from the runway threshold. If the aircraft had followed the correct vertical profile over that low hill, the ground clearance is only about 100ft, so the aircraft does not have to be dramatically low to hit it, but the PAPIs would have been showing four reds if it was that low.
The crew were flying their third short leg of the night and had been on duty for about 7.5h. The NTSB is checking whether they had rested properly in the previous 72h.
The fact that the aircraft was too low with autopilot/autothrottle engaged begs the question not only about the efficacy of crew monitoring, but the suitability of the flight management mode they had set, which is unknown at present. Pilots who operate into Birmingham say their airlines require special preparation for non-precision approaches there.
Approach incidents are lining up: 2007 Thomsonfly at Bournemouth, UK; 2009 Turkish Airlines at Amsterdam; 2013 Asiana at San Francisco, and now UPS at Birmingham. The first two were ILS-coupled but the speed decreased unnoticed. In the second two, the crew didn’t have electronic glideslope guidance and both aircraft descended below the approach profile to impact.
The risk of accidents on non-precision approaches has always been more than three times that of precision. Is it time to ban them, or do we train pilots properly for them?