Summary and Conclusions
The fractographic and fracture mechanics evidence assembled in the second investigation allows a revised failure scenario to be proposed, in which the accident is still primarily due to poor maintenance, occurring over a significant period of time. However, in contrast to the conclusions of the first investigation, the accident is now proposed to have been 'sudden and unexpected', in the sense that it could not be foreseen by typical operating personnel, who would not expect fatigue cracks to exist in the shear bolt. Such defects would not have been detected or, indeed, looked for during routine maintenance. The underlying cause of the accident is hence proposed to be the plastic deformation of the yoke holes which contain the 11 mm diameter shear bolts. Over a number of years, these holes have become hourglass shaped during the application of high loads to the yoke and associated plastic deformation of the shear bolts. The bolts are replaceable items during routine maintenance, but little attention would be paid to re-sizing the holes in the yoke.
This plastic deformation in the holes allows deflection of the bolt to occur in the yoke, and changes the applied stress state on the bolts from double shear to a mixture of shear and, over the central region of the bolt, surface bending stress. The direct cause of the accident is then initiation and growth of fatigue cracks under service loading, to a size which was critical under the stress applied at the time of the accident.
The applet calculations have indicated that whilst the fracture stress of the bolt remains fairly high in the presence of the likely fatigue cracks, the applied stress is equally high for small deflections of the bolt. The proposed scenario is reasonable and accounts for the observed deformation of the towbar at the tractor coupling side.
This case study provides an interesting demonstration of the advantages that simple solid mechanics and fracture mechanics calculations, in combination with fractography in the SEM, can bring to a failure analysis. Without such evidence the case has to argued on a much more subjective basis. This theme, of eliminating the subjectivity present in many failure analyses, will occur repeatedly in these case studies.
Nonetheless, it is very instructive to note the compounding of errors which led to this failure. If the aircraft braking system had been operative, the person in the cockpit would have brought the aircraft to a standstill before any damage to it had occurred. The only damage would then have been to the towbar, a very inexpensive item to repair.
Perhaps the most useful point to take away from this case study is the relevance of detailed fault tree analysis is assessing the consequences of engineering failures, and in identifying the 'true' factors which must be controlled to avoid failure consistently. In the present case, one could propose a greater level of training for the personnel, but possibly the most failsafe system would be one that immobilised the aircraft unless the brake system had sufficient pressure and reserves to be safely moved.
The concepts of diagrammatic support for analysis of system failure and operator error are illustrated in the linked paper
by the University of Glasgow Accident Analysis Group
. See also the page dealing with Hazard Analysis Methodologies
on the US Department of Labor OSHA website.