The theories of strength of materials, elasticity, and plasticity, lose much of their power when the structure of a metal can no longer be regarded as a homogeneous medium of unvarying physical properties. An example of this exists in the ductile-to-brittle transition that occurs in plain carbon steel and other metals.
Adequate ductility is an important engineering consideration, because it allows the material to redistribute localised stresses. This factor enables design for static situations to proceed in many cases, on the basis of average stresses. However, this method is fallacious when applied to materials which are "notch brittle sensitive". Localised stresses continue to build up when there is no localised yielding, a crack finally appears at one or more points, and propagates rapidly over the whole section without warning and showing no obvious deformation.
This fracture takes the form of the commonly recognised brittle fracture.
It is not to be thought that high localised stresses are the sole cause of brittle fracture. Rather, a stress concentration is only one of the many aggravating factors which include low temperatures, dynamic loading, and myriad structural (microscopic and macroscopic) features of the material, such as composition, grain size, impurities etc. Thus, to avoid brittle fractures, the engineer must apply new criteria to the design problem. The simple mechanical tests such as tensile and bend tests must be viewed with suspicion when regarded as an indication of the merits of the metal if there isany doubt as to its behaviour under adverse conditions.
While it is realised that a new test must be employed, the form of this test has not been firmly determined. The principal area of agreement is that it must be some form of notched bar test, but the choice of a specific specimen or interpretation of derived data is subject to much dispute. At present, Australian practice rather favours the use of the Vee-notch Charpy Impact Test, but English practice is largely based on the cantilevered "Izod" impact test. This discussion is not meant to detract from the merits of such excellent and expensive tests as the Robertson, Kahn, S.O.L, or B.W.R.A. tests among others, but to show the lack of agreement and the obvious dangers in attempts to correlate or infer values obtained using different test methods.
Accordingly, after the celebrated collapse of Kings Bridge, Melbourne, the University of Queensland began a program of investigation into the problem of brittle fracture in which the author participated. The approach considered was an extension of previous work by Hendrickson, Wood, and Clark et alia (Ref. 1, 2 etc.). It was hoped that, by following this approach, the susceptibility of Australian Standard Al steel to embrittlement could be determined, tested, and a design procedure developed to enable consideration of these factors and reduction of the brittle failure danger at the design stage.
This thesis discusses some of the factors involved in brittle fracture and the progress made by previous workers. It describes the work undertaken by the author and considers some methods of reducing the possibility of brittle fracture. While the author has not presented a fait accompli it is his hope that such preparatory work may some day give future investigators a fuller understanding of the principles involved in brittle failure.