Page 203 - Failure Analysis Case Studies II

188
was undertaking its 1286th pressurized flight in addition to 255 flights without cabin pressurization.
The Court of Inquiry [l] concurred with the findings of the RAE investigation [2] that the cause of
the accident was sudden cabin failure due to fatigue crack growth followed by the break-up of the
aircraft. The accident which occurred to G-ALYY was attributed to the same cause, as the flight
circumstances were similar, although insufficient wreckage was ever recovered to prove the case.
The root of this rapid failure due to metal fatigue was shown to be high stresses around cut-outs,
such as windows, in the aircraft skin. The aircraft manufacturer, de Havilland, had made estimates
of these stresses averaged over a large area, and ascertained the fatigue life of the aircraft by testing
sections of the cabin and the 22 gauge (0.71 mm) pressure cabin skin was thickened to 20 gauge
(0.91 mm) around the windows. However, the Court of Inquiry reported that the nature of the
sections used meant that they were not representative of a whole aircraft, as bulkheads fitted to
these sections to enable pressurization may have affected the stresses around the cut-outs in the
locality. This enabled the forward cabin section tested by de Havilland to withstand 18,000 cycles
before fatigue failure from a defect in the skin near the corner of a window. In addition, this section
was proof tested to 16.5psi (114kPa), or twice the operating pressure, before the fatigue testing
began, and this may have caused local plastic deformation in the regions of high stress of interest
here [l]. Proof testing of the pressure cabin was undertaken on all Comets during manufacture,
before acceptance by BOAC, and at predetermined times during service, to 11 psi (76 kPa), but
never to 16.5psi (114kPa), and a safety valve to prevent overpressurization of the cabin during
service was set to 8.5 psi (59 kPa).
Cracks were known to be present in the aircraft upon manufacture, and there was an approved
technique for identifying such defects and “locating” them by drilling the end of the crack with a
& in. (1.6mm) drill [l]. In most cases, the crack was seen not to extend beyond the location hole,
and this was assumed to be adequate security against further crack growth. In fact, there was a
“located” crack near the forward port corner of the rear ADF (automatic direction finding) window
(Fig. 3) on Yoke Peter, which did not grow beyond the locating hole until the final failure of the
cabin.
The failure of Yoke Peter was deduced to be a fatigue crack near the starboard rear corner of the
rear ADF window (Fig. 4). This crack emanated from a lOmm diameter bolthole, and propagated
to failure after unexpectedly few pressure cycles of the cabin. This bolthole in such a highly stressed

Fig. 4. Close-up views of the Failure in the skin of Comet G-ALYP. Crown Copyright. Reproduced with the
permission of the Controller of HMSO.