Conclusions
The results from the analyses presented here show that the effect of
surface roughness and sub-surface porosity is summative. If either the
roughness or the porosity is neglected, then the development over
multiple load cycles of equivalent plastic strain in the sub-surface
will be under-represented. There is significant coupling between the
porosity and the roughness pits, meaning that two pits that are
reasonably well-spaced can actually work together in combination with
sub-surface porosity to weaken a wedge of material in the sub-surface.
This is also suggestive as a mechanism for stress related corrosion.
Greater levels of porosity in the sub-surface imply greater equivalent
plastic strain penetration into the bulk material; however, for larger
numbers of reasonably homogenously arranged smaller pores the plastic
strain creates a network. This seems to be an emergent phenomenon
arising from the random distribution.
The results presented here for the larger pore examples are not an ideal
fit with the other data generated. This is because of the small numbers
of the pores, and the relatively poorly defined domain over which the
pores could be defined. The problem could be resolved by increasing the
problem size, increasing the number of pores within the model, and
having greater control over the porosity distribution.
Each of the models presented shows some similar features. In each case,
under first loading, the presence of stress raiser features – surface
roughness and porosity – leads to local stresses exceeding the yield
stress of the material, and the formation of localised zones of
equivalent plastic strain. The shape of each of these zones is similar,
but varies in size with the pore size. As the number of load cycles is
increased, the area of these zones of plastic strain remains the same,
or does not grow significantly, but the level of plastic strain
increases within that area. High plastic strain could indicate local
failure, so this is suggestive of a failure initiation mechanism.
Overall, the “PEEQ penetration” – the greatest depth into the
sub-surface where there is plastically strained material – increases
with increasing pore size, and reducing spacing between pores. A more
significant effect is that the combined effect of porosity and surface
roughness is to generate “<” shaped networks of PEEQ into
the material, defined on the surface by nearby deeper furrows in the
surface. Characterization of the PEEQ penetration would depend most
critically on the analysis of the surface roughness data to identify
these potentially damaging furrow spacings.
The data presented here is indicative only, and further modelling work
will be required to describe the trends with more precision. It is
proposed that the “PEEQ penetration” depth be used as a proxy
for the EIDS to be used in the Hartman-Schijve variant of the NASGRO
equation, and life predictions be made on that basis. These computed
lives must then be compared with test data, to test whether this can
provide a conservative basis for life prediction.