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

516

4

Nevertheless it is important to notice that a small crack can start to grow at the end of a step even if it is not long enough to activate a frequency drop of 1 Hz. In such a case like this, it is necessary to analyze the frequency diagram and correct the estimated fatigue limit. One of the objectives of this study is the definition of the defect size of the pores that caused the failure of the specimen. The analysis has been conducted by observing the fracture surfaces of the specimen, post-mortem, by Scanning Electron Microscopy (SEM). The defect size is estimated as the projected defect surface on the cracking plane. The parameter that has been used to measure the defect size is AREA 1/2 as proposed by Murakami (2002). The defect surface is calculated on the visible contour of the defect and the area is estimated as the surface of a generic polygon [(Luo et al. (2005)]. The validity of this parameter has been confirmed by different studies conducted on different casting alloys [Iben Houria et al. (2015); Mu et al. (2014); Nadot et al. (1997)]. 3. Discussion 3.1. Fatigue test results The two families of defects (cavity and sponge shrinkages) have been tested for a positive load ratio (R=0.1). An S-N diagram has been plotted for the different defects and is shown in Figure 2. The presence of natural defects reduce the fatigue limit of the material. If we consider only the cavity shrinkages (Figure 2a) the diagram shows that for a defect grade that ranges between 2 and 3, the average fatigue limit reduction with respect to the reference material (grade < 1) is of about 51%. The tendency is different for a higher defect grade that corresponds to a cavity shrinkage grade 4, in this case, the fatigue limit reduction is of about 77%.

/ σ a-max

/ σ a-max

σ a (MPa)

σ a (MPa)

0.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4 0.6 0.8 1.0

1E+ 05

1E+ 06

1E+ 07

1E+ 05

1E+ 06

1E+ 07

Number of cycles

Number of cycles

(a)

(b)

Fig. 2. (a) S-N diagram obtained for the specimens classed as CS 2, CS 3 and CS 4 at R=0.1 (frequency of 106 Hz), (b) S-N diagram obtained for the specimens classed as SS 2, SS 3 at R=0.1 (frequency of 106 Hz). For both diagrams σ amp is the stress amplitude, σ amp-max is the value of the maximal stress amplitude used for the stress normalization. The Figure 2b shows the S-N diagram plotted for the specimens containing sponge shrinkages with a grade ranging between 2 and 3. The fatigue limit reduction is of about 49% with respect to the reference material (grade < 1). From a global point of view, this result, shows that the different shrinkage type is not a parameter influencing the fatigue limit. For two shrinkage families, with the same defect grade (grade 2 and 3), the fatigue limit is substantially the same, a higher defect grade (CS 4) has a more detrimental effect. In order to understand the origin of the crack initiation and to estimate the defect size, a fracture surface analysis has been conducted for the different defect families. For the specimens affected by a cavity or a sponge shrinkage, the fatigue cracks always occur on a defect and, in function of the defect position, single or multiple cracks can initiate from a surface defect, from an internal defect or both surface and internal at the same time. Surface crack initiation is the most common case of failure that occurred during the experimental campaign. Figures 3a and 3b show the fracture surface of a specimen classed as cavity shrinkage grade 2 (CS 2). The defect size has been calculated using the method proposed in section 2.2 and is equal to 1619 µm. It is important to notice that the first crack initiation phase of a micro-crack originating from a surface shrinkage is normally related to the Silicon particles debonding in the eutectic phase localized in the convex zones of the pore [Buffière et al. (2001)].