Most models have been developed for the macro-crack phase of the crack growth process. However, the crack must first be formed, and must develop to a length that is covered by the model.
As mentioned in Section 
, there are a
number of stages to crack growth. Initially, a structure may
contain no defects. This is called the dormant phase.
During this stage the crack does not exist. The next stage is the
formation of a crack. This is termed nucleation or
initiation. This phase consists of the material gradually
deforming, until such time as the molecules of the material have
become sufficiently dislocated to the extent that a recognisable
defect, or crack has formed.
Then there follows an initial growth phase which is known as
microcrack propagation or short crack propagation.
This is a haphazard growth, is not well explained by the
traditional models described in Chapter , and is
the area of research interest here.
As a crack grows, it enters the macrocrack phase, by which time the various models in the literature become appropriate. The final phase of interest is sometimes referred to as the failure phase, which is the component or specimen giving up its strength. Failure is generally very quick, since the rate of growth of long cracks is exponential. In comparison, microcracks can be present in a structure for a very large proportion of its lifetime. For this reason, they play an important part in determining total lifetime.
There are a number of qualitative differences exhibited by the microcracks in the specimens that were examined, as compared to what happens for macrocracks. The first of these is the rate of growth. Thus the primary aim for this section of research was to model the rate of growth of microcracks within the specimens, and make reliability predictions based upon these models. This aim fundamentally differs from the work of previous authors such as Beretta and Clerici [4], who have concentrated on predicting reliability based upon an overall damage statistic.