PSI - Issue 5

Tomasz Tomaszewski et al. / Procedia Structural Integrity 5 (2017) 840–847 Tomasz Tomaszewski et al. / StructuralIntegrity Procedia 00 (2017) 000 – 000

845

6

Clear differences in the fatigue strength for axial loads and bending loads were observed. Analysis of the position of fatigue characteristics σ a -N was performed using the statistic test of parallelism of slopes. The test was performed for σ a -N characteristics for a standard specimen. It was demonstrated that the characteristics for axial load are parallel, despite minor differences in the value of a coefficient. In the case of characteristics for bending load, a hypothesis was rejected that assumed the equality of a coefficient. The characteristics are not parallel. Material 1.4301 features a definite sensitivity to changing the cross-section size. This is confirmed by the high value of K HC coefficient, which is dependent on the level of stress amplitude. It is clearly noticeable in the case of four-point bending (Fig. 5). The mean value for axial stress is at a level of 1.086, and equal to 1.275 for bending stress. The mean values were determined for extreme fatigue life.

Fig. 5. K HC coefficient for axial and bending load

4. Analytical approach

The differences in fatigue properties obtained are determined by the model based on highly stressed volume. For the specimens tested, V 90% volume is graphically represented on Fig. 6, and the determined values of V 90% are presented in Table 7.

Table 7. Highly stressed volume parameters.

w 1

t

V 90%

Type of geometry Standard specimen

7

4

8.4

Minispecimen

2.5

1.4

0.35

Fig. 6. Highly stressed volume for tested specimens

The highly stressed volume model was applied for estimating fatigue life for the cross-section area of the standard specimen. The calculations were performed basing on fatigue life σ a -N obtained for the minispecimen. The analyses were performed for the results obtained from four-point bending. Based on the (2) equation, fatigue life was calculated of an object that corresponds to the sectional area of the standard specimen. The regression line obtained based on experimental tests of the standard specimen was deemed as the reference characteristics. Evaluation of the matching of the estimated characteristics based on the model to the experimental data was performed by calculating the standard error of the estimate and the coefficient of residual variation. In the previous works of the authors it was confirmed experimentally that the values of cross-sectional area coefficient determined from static and fatigue tests are similar ( K S ≈ K HC ) (Tomaszewski et al. (2016)). Detailed