Statuts:	Enquête Officielle
Date:	17 MAR 1982
Heure:	08:10
Type/Sous-type:	Airbus A300B4-203
Compagnie:	Air France
Immatriculation:	F-BVGK
Numéro de série:	070
Année de Fabrication:	1979-02-22 (3 years )
Heures de vol:	9053
Cycles:	3376
Moteurs:	2 General Electric CF6-50C2
Equipage:	victimes: 0 / à bord: 13
Passagers:	victimes: 0 / à bord: 111
Total:	victimes: 0 / à bord: 124
Dégats de l'appareil:	Perte Totale
Conséquences:	Written off (damaged beyond repair)
Lieu de l'accident:	San'a International Airport (SAH) (Yémen)
Phase de vol:	Au décollage (TOF)
Nature:	Transport de Passagers Intern.
Aéroport de départ:	San'a International Airport (SAH/ODSN), Yémen
Aéroport de destination:	Cairo International Airport (CAI/HECA), Egypte
Numéro de vol:	125

Détails:
While travelling at 95 knots during take-off, an explosion was felt, followed by shimmy and vibrations. The crew, thinking a tyre had exploded, aborted the take-off. The explosion was in fact caused by an uncontained failure of the stage 1 HP compressor disk of the no. 2 engine. Debris punctured the fuel tank, causing a fire on the right hand side of the aircraft. The propagation of a low cycle fatigue crack on one of the embossments of the disk rim caused the disk fracture.

CAUSES: "The accident directly resulted from the uncontained explosion of a first stage high pressure turbine disk of the right hand engine. The propagation of a low cycle fatigue crack on one of the embossments of the disk rim was at the origin of the disk fracture.
This cracks, which existed before, had not been detected during the inspection conducted in the operator's workshops, according to the method defined by the manufacturer and in accordance with the requirements of the airworthiness authority. The investigations showed that it was necessary, not only to reduce the intervals between the inspections of these disks and conduct two independent sequential inspections, but also to modify the design and make sure they were progressively replaced by the new type. It seems obvious that the limit life validation methods used at the time of the accident, were still insufficient, as far as the probability of non-detection of cracks was concerned."

Sources: