Ks. Ravichandran et Jm. Larsen, EFFECTS OF CRACK ASPECT RATIO ON THE BEHAVIOR OF A SMALL SURFACE CRACKS IN FATIGUE .2. EXPERIMENTS ON A TITANIUM (TI-8AL) ALLOY, Metallurgical and materials transactions. A, Physical metallurgy andmaterials science, 28(1), 1997, pp. 157-169
The continuous variations in crack shape or aspect ratio, a/c (a is th
e crack depth and c is the half-surface length), of small surface crac
ks, induced by grain boundaries, have been investigated during the fat
igue crack growth of small cracks in a titanium (Ti-8Al) alloy. The si
gnificance of the aspect ratio variations in explaining the ''anomalou
s'' small-crack behavior was evaluated. The aspect ratio data were det
ermined from the measurements of crack compliance, made using a laser
interferometric system, and the measurements of surface crack length (
2c), made using a photomicroscopic system. The variations in aspect ra
tio were found to be large at small crack sizes of the order of a few
grain diameters. The experimental a/c data were compared with the patt
erns of crack aspect ratio variation, obtained from theoretical, simul
ations. The simulated data were generated by assuming alternate crack
propagation at the surface and at the depth, the details of which are
presented in Part I of the study accompanying this article. A good agr
eement was found between the simulated and the experimentally observed
variations. After incorporating the a/c variations in Delta K calcula
tions; the scatter in the growth data of small cracks was significantl
y reduced and was found to be of the same order as in large cracks. Ad
ditionally, it is shown in this study that the conventional methods of
analysis of small-crack data, performed with an assumption of a/c = 1
, can result in significant errors in Delta K calculation and an incre
ased level of scatter in small-crack growth data. Small cracks also we
re found to exhibit low closure levels relative to large cracks. The r
esults of the study strongly indicate that characteristics of small cr
acks, often referred to as anomalous, are due to the assumption of a/c
= 1 in situations of large variations in aspect ratio, the use of con
ventional methods of data analysis, and the lower levels of crack clos
ure found naturally in small cracks.