Computer-Related Incidents with Commercial Aircraft

British Mediterranean Airways, A320, false navigation data on non-precision approach. Addis Ababa, 31 March 2003

31 March 2003

Synopsis The A320 aircraft was making an approach into Addis Ababa airport during stormy weather, with a thunderstorm off to the left. The aircraft performed one missed approach due to fluctuating navigation data from the VOR navigation beacon. After querying whether the VOR was operational and receiving an affirmative reply, they performed another approach. This approach was also abandoned close-in due to signal problems. As the crew started the go-around, the Enhanced Ground Proximity Warning System (EGPWS) announced "Too Low Terrain. The aircraft was determined to have missed high terrain only by some 56 ft.

The incident aircraft was not equipped with GPS, but only with standard ground-based navaid receivers. The EGPWS, which has a "moving map"-like display, likewise obtained its signal from the receivers for the ground-based navaids. The VOR at Addis Ababa was waterlogged, due to the weather and also that it was inappropriately sealed against water intrusion. Although ICAO rules require that VORs are installed with a self-check, which declares the signal to be invalid to receivers when it is out of tolerance, this VOR was determined to be sending "valid" signals that were up to 30 degrees off. The aircraft, tracking the VOR, was up to 3 nautical miles off course. The navigation receivers, as did the EGPWS "moving map", showed the aircraft to be on-course.

I had predicted that EGPWS might not turn out to be the perfect terrain-avoiding device in my Inside Risks column Risk of Technological Remedy, published in the Communications of the ACM 40(11):160, November 1997. I speculated in Air Safety Week that the pilots might have found the false-reading EGPWS display to be "confirming" of the false inbound course. I was assured in correspondence with the Safety Manager of British Mediterranean Airways, Robin Berry, that this was not the case, and indeed the data on the flight confirm Robin's assertion: go-around because of faulty data had commenced at the time that the low terrain warning occurred. Robin was actually on the accident flight.

The U.K. Civil Aviation Authority (CAA) issued Bulletin 16/2003 on receipt of the airline's notification about the incident. The most substantial account of the incident is to be found in Robin's slides, which I hope to publish here with his permission. Air Safety Week published an article False Navigation Data Leads to Near-Crash (archived version) on 15 December, 2003. David Evans, editor of Air Safety Week, wrote a commentary on 1 February 2004 about the incident, entitled Safety: Perils of a Lone Positioning Source.