PSI - Issue 28

Stepan Major et al. / Procedia Structural Integrity 28 (2020) 561–576 Author name / Structural Integrity Procedia 00 (2019) 000–000

567

7

life of all specimens lies within a transition region between the high-cycle and the low-cycle fatigue. The symmetric in-phase sinusoidal bending-torsion loading (both component of frequency f = 29 Hz) was applied to smooth cylindrical specimens made of high-strength low-alloy using the resonance testing machine MZGS-100. The specimens were loaded to the final rupture. Experiments were carried at the room temperature. The differences in the fatigue life are small enough to enable a mutual comparison of fracture morphologies of investigated specimens. The fatigue life of virgin specimens lies within a transition region between the high-cycle and the low-cycle fatigue. Fatigue life of nitrided specimens is an order of magnitude higher than virgin samples at the same load values.

Table 1. Loading condition of virgin specimens. Sample Type of loading σ a [MPa]

τ a [MPa]

L R [-]

N f [cycle] 1229000 1252000 1099100 1700150 4475000 N f [cycle] 2710000 1910000 1830000 3450000 4560000

1 2 3 4 5

Pure bending

620 550 330 140

0

0

200 330 385 390

0.23

Combined bending

0.5

0.73

Pure torsion

0

1

Table 2. Loading condition of nitrided specimens. Sample Type of loading σ a [MPa]

τ a [MPa]

L R [-]

1 2 3 4 5

Pure bending

650 555 343 150

0

0

202 343 414 450

0.27

Combined bending

0.5

0.73

Pure torsion

0

1

3.3. Fracture surface reconstruction Topographical 3D data of fracture surfaces were obtained by means of stereophotogrammetry, see Sherer and Kolednik (2001). Stereoimages were taken using the scanning electron microscope LEO S440 by tilting the specimen by a deviation angle from 7º to 15º in dependence on local surface relief complexity. It can be said, that in the case of nitrided specimens the higher angle was used. Also, quality of reconstruction was in this case lower. Due to frequent imperfections in the reconstruction of highly inclined surfaces (this is case of nitrided specimen), it was necessary to repeat it. Subsequent processing of acquired stereopairs was realized by means of the commercial software MeX. The output digital elevation models consisted of up to 3 � 10 5 non-equidistantly localized points. This set of non equidistantly localized points was processed using Delaunay triangulation, see Sherer and Kolednik (2001). The entire fracture surface was covered with electron microscope images with a magnification of 55x and these images were used for reconstruction of whole fracture surface. Complex images of fracture surfaces are shown in Fig. 4. Each of studied fracture surface was divided by a square grid in the x -y plane in elementary square areas of S j . This uniform square network consists of squares of edge length a 99 = 99 μm. For each of these squares, mean normal vector �⃗ � � was calculated from all normal vectors of triangulation triangles , T i j n  , which lie in the appropriate square. Furthermore, the orientation of certain areas of the fracture surface was studied in more detail, for which the results of stereophotogrammetric reconstruction with higher magnification was used. In this case, the square area of size ��� � ��� mm 2 was chosen on each surface with its center at the distance of 0.8 mm from the fatigue crack initiation site on the specimen surface. Others squares then follow inwardly of the sample. These square sites cover the principal and local direction of the fatigue crack propagation, Fig. 5. These small squares were divided by a square grid, with a standard square edge length a 99 = 99 μm (same as in the previous case) and also a finer mesh with the side length a 05 = 5 μm for one elementary square. The vector �⃗ � � �� and vector �⃗ � � were calculated for each triangle or square intersecting fracture surfaces. Knowledge of these two vectors allows to calculate the deviation angles α and β .