PSI - Issue 5

Miloslav Kepka et al. / Procedia Structural Integrity 5 (2017) 1409–1416 Miloslav Kepka et al. / Structural Integrity Procedia 00 (2017) 000 – 000

1411

3

2. Input data

2.1. Fatigue life curve

The best data sources for fatigue life calculations of particular structural parts are fatigue curves determined from laboratory fatigue tests using statistical evaluation. However, these are not always available (or even possible to obtain). Fatigue characteristics of structural details often need to be established by estimation. Estimates of fatigue curves may be based on empirical formulas or catalogues and standards containing design fatigue curves for various geometric and design configurations of structural details. The source used is this case was British Standard BS 7608, 1993. It relates to the design and assessment of steel structures from the perspective of high-cycle fatigue. It applies to wrought structural steels with minimum yield strength up to 700 MPa. It was suitable for this case because both the rear axle housing and the welded bracket were made of St52 structural steels with a guaranteed strength of 520 MPa. This standard includes a class T fatigue curve. When this curve is used, the fatigue life evaluation must be based on structural hot-spot stress. This procedure is compatible with this case because the actual fatigue cracks initiated in and emanated from the fillet weld toe, i.e. a hot-spot stress location. A schematic view of this situation is shown in Fig. 2 (Hobbacher, 2016).

Fig. 2. Definition of structural hot-spot stress.

In the above standard, the fatigue curve is expressed as follows: log( ) = 12.6606 − 0.2484 − 3 log(∆ ) , (1) where is the number of cycles to fracture, ∆ stands for the stress range and is the number of standard deviations below the mean fatigue life curve. The advantage of this fatigue curve representation is that it enables design curves to be derived for various fatigue failure probabilities (in this case this means the formation of a macroscopic crack). For instance, at d = 0, equation (1) describes a mid-range fatigue curve (failure probability of 50%). Fatigue curves shifted by two standard deviations ( d = 2) below the mean curve, provided that log-normal distribution applies, represent a failure probability of 2.3%. (This means a probability of survival of 97.7%). Fatigue life calculations were carried out using the nCode software. The fatigue curve parameters were therefore converted into the format used by this software. ∆ = 1 ∙ ( ) 1 , (2) where SRI 1 is the intercept at 1 cycle and 1 is the slope.