PSI - Issue 7

I. Milošević et al. / Procedia Structural Integrity 7 (2017) 327 –334 I. Milosˇevic´ / Structural Integrity Procedia 00 (2017) 000–000

329

3

Fatigue testing

The experiments were carried out on servohydraulic testing rigs, which are operating at testing frequencies up to 30 Hz. Specimens D 7 . 5 were tested on the Sinus machine (Sinus 100.25, Instron GmbH, ± 100 kN). Specimens D 4 were tested on the Type 4 ( ± 50 kN) own build test rig. These test rigs were designed for tests at room temperature T = 20 ◦ C. Experiments at elevated temperatures at T = 350 ◦ C, D 7 . 5 were carried out on a Instron ± 100 kN testing machine. This setup was prepared for high temperature fatigue tests by cooled clamping tools. The temperature was induced by induction heating and monitored by thermocouples to ensure proper specimen temperature. A stress ratio of R = − 1 was applied to all fatigue tests.

Fractrography

Fractured surfaces within the TA were analysed by SEM because of crack origin identification. A EVO MA15 (Carl Zeiss SMT) was used for the analyses, which was also equipped with a EDS (Energy Dispersive X-Ray Spectroscopy). The EDS enabled the determination of inclusion types and their chemical composition.

3. Fatigue life

Recent questions lead to a closer inspection of fatigue data used for general lifetime assessments. Conventional stress based lifetime models were based on fatigue data, which were provided by fatigue tests. Specimens were loaded with a constant stress amplitude until failure occured. Thus a stress / cycles to failure (S / N) relationship was established. After a statistical evaluation (standardised ASTM E739-91 by ASTM International (1998), arcsin √ P transforma tion by Dengel (1975)) S / N curves were determined. A S / N curve represents a certain survival probability (Figure 2, I, 50 %). After the inflexion (turn-o ff ) point a horizontal course was assumed. S / N curves were characterised by parameters k (slope), S aD (fatigue strength or constant amplitude limit) and N D (turn-o ff point). The presented curve I (Figure 2) is a suitable proposal for bcc materials like structural steels. A di ff erent behaviour was examined for aluminium cast alloys (III, by Eichlseder and Leitner (2002)) and high strength steels (actual material, I + II, by Marines (2003)).

Fig. 2. Example sketch of di ff erent approaches for fatigue modelling > 10 7 cycles.(50 % survival probability). I: Wo¨hler (1866); I + II: Marines (2003); III: Eichlseder and Leitner (2002)

Fatigue life is influenced by di ff erent additional e ff ects like notched geometries and elevated temperatures. The e ff ect of notches, for example, on the life estimation was discussed by Eichlseder (1989, 2002) and Milosˇevic´ et al. (2016) by various examples, where parameters of the S / N curve (k, S aD , N D ) were modfied. According to Eichlseder (2002) a second slope k 2 (III, Figure 2) was invented for aluminium cast alloys (nonferrous materials) instead of applying a horizontal course. Defect based approaches like √ area by Murakami (2002) were used to calculate the according fatigue strength. In the actual study Murakami’s approach was used. Due to the fact that all specimens failed within the TA (10 5 to