13th International Conference on Fracture June 16–21, 2013, Beijing, China -3- Figure 2. Experimental set up [2&3] In the present fretting cracking investigation the displacement amplitude was monitored in order to maintain partial slip conditions, keeping constant the tangential force amplitude during the whole test duration. A cylinder/plane configuration was applied. The fretting pad consists of a Ti-6Al-4V alloy which displays the following elastic properties: 119500 MPa for the elastic modulus and 0.287 for the Poisson’s coefficient. The cylinder’s radius was fixed at 80 mm and the applied normal force at P=436.5 N/mm which induces a maximum hertzian constant pressure p0=300 MPa and a hertzian constant radius aH=0.92 mm. The lateral width of the cylinder pad was about 8 mm which corresponds to plain strain conditions along the median axis of the fretting scar. Previous tests were done to determine the friction coefficient at the sliding transition µt=0.85. [4] 3. Experimental results Our objective was to identify the evolution of the crack length as a function of the fretting cycle to establish the "plateau" value related to the crack arrest condition. The tangential force was kept constant for all the experiment with Q*=200 N/mm. This tangential loading was previously adjusted to guarantee significant fretting cracking for elastic configurations (i.e. the stress is inferior to the yield stress R0.2). Figure 3 shows a typical crack path observed for the studied configuration after 2.106 cycles. As usually observed the incipient crack propagation a near 45° orientation compared to the normal of the surface until about 30 µm in depth. Below 200 µm, the crack propagation direction is normal to the plan. Between 30 and 200 µm, the crack path is considered to be linear with a 10° angle with respect to the normal of the surface. A schematic description of the crack path is given in Figure 3. The major part of the crack expertise leads to similar evolutions, so that the overall crack path description can be provided by linear segments (Fig. 3), and by making the average, we kept the first angle at 47.5°, the second at 15.5 and the last at 0° for the numerical model exposed hereafter.
RkJQdWJsaXNoZXIy MjM0NDE=