FRACTOGRAPHIC ANALYSES OF FATIGUE-CRACK GROWTH IN D16T ALLOY SUBJECTED TO BIAXIAL CYCLIC LOADS AT VARIOUS R-RATIOS

Citation
Aa. Shanyavsky et al., FRACTOGRAPHIC ANALYSES OF FATIGUE-CRACK GROWTH IN D16T ALLOY SUBJECTED TO BIAXIAL CYCLIC LOADS AT VARIOUS R-RATIOS, Fatigue & fracture of engineering materials & structures, 18(11), 1995, pp. 1263-1276
Citations number
19
Categorie Soggetti
Material Science","Engineering, Mechanical
ISSN journal
8756758X
Volume
18
Issue
11
Year of publication
1995
Pages
1263 - 1276
Database
ISI
SICI code
8756-758X(1995)18:11<1263:FAOFGI>2.0.ZU;2-L
Abstract
Fractographic peculiarities of fatigue crack development are studied i n cruciform specimens of D16T aluminium alloy under biaxial tension an d tension-compression In the range of the biaxial load ratios lambda f rom -1 to +1.5, and in the range of R-ratios 0.05 to 0.8, fatigue stri ation formation took place over a crack growth rate near to 4 x 10(-8) m/cycle. The striation spacing and the crack growth rate decrease as the ratios lambda and R increase. The ratio between the increment of c rack growth, da/dN, and the striation spacing, delta, is approximately 1:1 when da/dN is greater than 4 x 10(-8) m/cycle. The relationship b etween the number of cycles from the beginning of a test up to the gro wth rate of 2.14 x 10(-7) m/cycle (N-d), and the crack growth period, N-p, from when the crack initiates up to the instant when that growth rate is reached, was determined for different lambda and R-ratios. The value of N-d increases as the stress ratio, lambda, is increased. Cyc le loading parameters must be taken into account in order to describe the crack growth period when using a unified method involving an equiv alent stress intensity factor K-e = KIF(lambda, R). The values of F(la mbda, R) for the growth rate (F(lambda, R)(v)) and for the striation s pacing (F(lambda, R)(s)) were determined and compared. The fatigue cra ck growth period, N-p(t), applicable to the stage of fatigue striation formation, (predicted by using both of the F(lambda, R)values) is cor related with the experimental data and the error is of the order of 15 %.