PSI - Issue 7

A. Rotella et al. / Procedia Structural Integrity 7 (2017) 513–520 Antonio Rotella et al. / Structural Integrity Procedia 00 (2017) 000–000

515

3

the SDAS of the different configurations have been measured (Figure 1c). The measure has been conducted by identifying the dendrites with at least 6 arms (Figure1b). The SDAS is finally measured as the average distance between the secondary arms. An average of 160 measures have been conducted on specimens of the six castings. The finest SDAS (23 µm ± 6 µm) has been obtained for the configuration classed as cavity shrinkage grade 3, the coarser SDAS (36 µm ± 8 µm) has been obtained for the casting classed as sponge shrinkage grade 2.

Reference (grade < 1)

Sponge Shrinkage grade 2 Sponge Shrinkage grade 3

SDAS = 36 µm ± 8 µm

SDAS = 32 µm ± 7 µm

(a)

SDAS = 31 µm ± 8 µm

Cavity Shrinkage grade 2 Cavity Shrinkage grade 3 Cavity Shrinkage grade 4

SDAS = 33 µm ± 7 µm

SDAS = 25 µm ± 6 µm

SDAS = 23 µm ± 6 µm

(b)

(c)

Fig. 1. (a) Microstructure of the A357-T6 Aluminum alloy, (b) schematic representation of the technique used to measure the SDAS, (c) micrographs of the different microstructures obtained for the six castings, specimens are mirror polished with colloidal silica (0.04 µm). Even if a variation of 13 µm between the coarser and the finer SDAS have been detected, a study conducted on the A356-T6 Aluminum alloy showed that a variation over the SDAS of about 10 µm will not influence the fatigue limit of the alloy for a fatigue tensile test at a positive load ratio. 2.2. High cycle fatigue tests All fatigue tests have been conducted using a vibrophore Amsler 10 HFP 420 (electromagnetic resonance machine), the testing machine works with the resonance frequency of the specimen and it is possible to reach a maximum load of 100 kN and a maximum frequency of about 100 Hz. The estimation of the fatigue limit has been conducted by using the step by step method [Lanning (2005)]. The choice of this methodology has been dictated by the need of a fatigue limit for each type of defect. A standard testing using a staircase method is not suitable because each specimen has a natural defect that is not repeatable in size, morphology and position, thus it is necessary to have a methodology that is capable to give an estimation of the fatigue limit for a given number of cycles. The end of the test corresponds to a frequency drop of 1 Hz, which corresponds to a fatigue crack depth of about 3 mm. In order to estimate the fatigue limit, the first step is to test the specimen for a given stress σ , if the specimen not fails and reaches the fixed number of cycles (for the current study N limit = 2 ∙ 10 6 cycles) the fatigue test is started again on the same specimen by increasing the stress of an arbitrary ∆σ. The procedure is repeated until the detection of a frequency drop of 1 Hz, which defines the end of the test (the failure stress is defined as σ n ). Finally the fatigue limit is calculated using the Lanning proposition [Lanning (2005)] that is shown in Equation 1. If the specimen fails at the first step (before reaching the value of N limit ), the fatigue limit is estimated using a Basquin’s law (Equation 1).

∙ ( σ n - σ n-1 ) + σ n-1

N failure N limit

m

σ D =

or

σ D = C ∙ N

(1)

In the context of this study, the value of ∆σ is imposed equal to 10 MPa. The application of the step by step method implies that the material is less, or not, sensitive to load history (no influence on the fatigue limit). For a cast Aluminum alloy (A356-T6), that is quite similar to the one of this study, any effect of the load history has been identified for different kinds of loads [(Iben Houria et al. (2015); Roy et al. (2012)]. The same result has been obtained for others alloys such as Titanium alloys [Bellows et al. (1999); Lanning (2005)] or C35 steel [Billaudeau et al. (2004)].