Fatigue Crack Paths 2003
transformation between two Cartesian systems can be described in different ways, but
the direction cosines matrix is usually applied. On the other hand, such a matrix
includes nine dependent components, and averaging all the matrix elements is not
effective since the averaged matrix would not satisfy the conditions of transformation
[8]. Moreover, we are not able to select which three independent elements of the
direction cosines matrix should be averaged. The exact algorithm of average procedure
and calculation of Euler angles can be found in [4-5]. The aim of the present paper is to
verify whether the weight function method based on energy parameters can allow us to
determine the expected fatigue fracture plane position.
E X P E R I M E N TDAALT A
Fatigue tests were performed in the high cyclic fatigue regime on the smooth specimen
(Fig. 1) made of 18G2Asteel. Mechanical properties of 18G2Asteel were as follows,
yield stress: Re = 357 MPa, ultimate stress: Rm = 535 MPa, Young’s modulus: E = 210
GPa, Poisson’s ratio ν = 0.3, fatigue limit for fully reversed axial and torsion
respectively: σaf= 204 MPa, and τaf= 170 MPa, exponents of the fatigue curves for
fully reversed axial (σa-Nf) and torsion (τa-Nf) respectively: mσ=8.2 and mτ=8.2.
Figure 1. Specimen geometry.
The specimens were subjected to random bending, torsion and combined bending
with torsion for three values of the cross correlation coefficient (rσ= -0.01, 0.50, 1.00)
between courses of bending Mg(t) and torsional momentMs(t). Histories of loading with
normal probability distribution and narrow frequency band were generated in a
computer in the form of one block. The length of a loading block was equal to 33
minutes and 30 seconds. This block was repeated until to fatigue failure of specimen.
The moments of bending and torsion were recalculated to stress using the elastic model.
Fatigue tests were carried out for different ratio of maximum torsional stress to
maximumbending stress λσ= τmax/σmax (Table 1). As a result of fatigue test, the crack
line for each specimen was established on the basis of photos of specimen surface
(Fig.2). The photos of specimens were made using an optical microscope with
magnification of 60 times connected directly to the computer. Points representing
cracks were approximated to regression line using the least squares methods. The slope
coefficient of regression line was used to calculate the experimental value of angle αexp .
Made with FlippingBook - Online catalogs