PSI - Issue 19

P. Cussac et al. / Procedia Structural Integrity 19 (2019) 463–471 P. Cussac / Structural Integrity Procedia 00 (2019) 000 – 000

465

3

Surface imperfections (Fig. 2) were introduced in some of the previously polished samples. For conservatism purposes, they have to be more severe than those that may be encountered in an industrial environment. For this reason, artificially-introduced imperfections have, within the limits allowed by the process, relatively small opening angles and imperfection root radii. Indeed, these parameters can influence fatigue crack initiation and early propagation [Inchekel (1994)]. Moreover, in order to avoid taking advantage of possible residual compressive stresses, which can be generated during the imperfections introduction [Gourdin (2017)], the selected method has to limit as much as possible the generation of this type of stresses. For the present work, a device (called "cracking machine") developed by the Pprime Institute was implemented. Its principle is based on a gradual removal of material, by abrasion, using a rotational zirconia disk associated with a diamond powder lubricant. The use of a disc when machining imperfections induces a slightly curved shape (14 mm radius) at the bottom of the imperfection.

Fig. 2. Example of surface imperfection topography observed by optical microscopy

In order to consider only the mode I cracking configuration, known as being the most unfavorable, all imperfections were introduced at the center of the specimens, perpendicularly to the loading axis. The surface imperfection depths varied from 100 to 350 μm. The electrical potential method, better known under the acronym DCPD was implemented as part of this study to monitor the initiation and the propagation of cracks emanating from surface imperfections. The principle of this method is based on the increase of the electrical resistance in the specimen during crack propagation, generating, according to Ohm's law and for a constant current intensity, an increase of the difference of the electrical potential, denoted V, on both sides of the crack. The measurement is made using 0.1 mm diameter platinum wires welded at a distance of 1 mm on each side of the center of the imperfections. In order to ensure the measurement independent of material properties, V is normalized using the value measured at the maximum of the first cycle, denoted V 0 . This formalism also enables to take into account the plastic deformation occurring at the beginning of the test and which may influence the measurement of the potential [Ljustell (2011)]. Thus, the measurement of V/V 0 during the tests allows to monitor in real time the crack propagation. However, a calibration has to be performed beforehand so as to relate the crack depth to the measured difference of the electrical potential. In that aim, ink markings were made at different stages during the fatigue tests in order to obtain an experimental calibration curve as well as the shape of the crack fronts. The fatigue tests were conducted on an electromechanical machine under air and at ambient temperature. All the strain-controlled tests were performed using an extensometer positioned on the gauge length of the sample. The strain ratio is R ε = -1 and the strain rate is set at 4×10 -3 s -1 . Three levels of total strain were studied: Δε t / 2 = 0.2%, Δε t /2 = 0.3% and Δε t /2 = 0.6%.