Page 509 - Biaxial Multiaxial Fatigue and Fracture

Geomehy Yariation and Life Estimates of Biaxial Fatigue Specimens 493

where:

The material cyclic values were taken from Table 3.
Two strain paths were chosen for the life prediction analysis, as is shown in Fig. 5b. The first
path was at a direction e= O', aligned with the slot axis and at a plane that intersected the
surface at y= 45'. The second path was at e= 45' to the slot direction and again at a plane
intersecting the surface at y= 45'.
In Figs 6a and 6b, the life predictions using the subsurface damage model and employing the
two strain paths described above are compared to the experimental results and to a life
prediction carried out using a simpler, hot-spot approach [ll] that searches for the highest
value of the maximum shear strain along the subsurface path. The solid diagonal line in Fig. 6
corresponds to a perfect correlation between experimental and predicted fatigue life, while the
dashed line corresponds to a factor of two on life. Estimated life was obtained by summation of
the subsurface damage up to a radial distance of lmm in Fig. 6a and OSmm in Fig. 6b.
In general, the predicted lives shown in Fig. 6 are within a factor of two on life for all the
tests data, independent of the path or the subsurface distance used. It can also be seen that the
hot-spot approach yields a more conservative life estimate, while the predictions using the
subsurface damage model are non-conservative in all cases. In general, the summation of the
damage of up to a lmm subsurface distance predicted shorter lives compared with the
summation of damage at subsurface distance of up to 0.5mm. It should also be noticed that
although a different strain variation was simulated for the two paths investigated (Figs. 5b and
5c) it does not appear to influence much the predicted lives in Fig. 6. This is further discussed
later.

DISCUSSION

Life prediction analysis
In Fig. 7 the life prediction results using the subsurface model are compared to the
experimental and the hot-spot lives using the same simulated surface maximum shear strain. In
consistency with Fig. 6, it is shown that the hot-spot approach predicts the shortest lives while
the deeper path, up to a lmm subsurface strain, yields the shortest lives. Fig. 7 also
demonstrates that the predicted life using the subsurface damage model is currently sensitive to
the chosen subsurface distance.
In the past, the hot-spot approach assumed that the crack propagation direction coincides
with the maximum shear strain direction, which is at 45' to the slot axis. However, this was not
observed in the testing of the rhombic specimens as is depicted schematically in Fig. 8. It can
be seen that for the specimens subjected to the higher load amplitude levels of 20kN and
17.5kN (Specimens A and B) the crack has initiated and propagated in the e= 45O direction.
However, at the intermediate load level of 15kN (Specimen C), the crack initiates in the e= 0'