Issue 35

B. Lian et alii, Frattura ed Integrità Strutturale, 35 (2016) 389-395; DOI: 10.3221/IGF-ESIS.35.44 394 such a way, it is possible to record the crack propagation from early stage at the artificial defect to just before the final fracture phenomenon of the specimen. The crack length measurement results obtained for the experiment using a specimen with an artificial defect (  a = 800 MPa, N f = 4.57 × 10 4 cycles) and using a specimen without artificial defect (  a = 1400 MPa, N f = 2.39 × 10 4 cycles) are depicted in Fig. 10(a) and (b), respectively. Detection of the early stage of the crack propagation is approximately at 10  m in the case of experiment using a specimen with artificial defect, whereas a length of approximately 20  m was necessary for specimen without artificial defect. In the present experiments, the LED stroboscope gives an uneven enlightenment, inducing bright and dark spots along the whole observation field. It is thought that future improvements of the flashing technique would give fatigue crack observations with better clarity. Figure 9 : Series images of fatigue crack initiation and propagation. (a) : N = 7.50 × 10 3 cycles, (b) : N = 3.45 × 10 4 cycles, 2 s = 606  m, (c) : N = 4.50 × 10 4 cycles, 2 s = 3360  m. Figure 10 : Relationship between cycle ratio and crack length. (a) : With artificial defect (  a = 800 MPa, N f = 4.57 × 10 4 cycles), (b) : Without artificial defect (  a = 1400 MPa, N f = 2.39 × 10 4 cycles). C ONCLUSIONS y using the combination of a LED stroboscope, a control system of the stroboscope flash timing, a long working distance microscope and high spec digital camera, an observation system suitable to record the fatigue crack length on a hourglass shape specimen for rotating bending fatigue tests has been developed. This system allows to study fatigue crack initiation and propagation phenomena at a resolution equivalent to the replica method, without any suspension of the fatigue test. B (a) (b) (a) (b) (c)