PSI - Issue 19

Fabian Becker et al. / Procedia Structural Integrity 19 (2019) 645–654 F. Becker et al. / St uctural Integrity P o edi 00 (2019) 000– 00

648 4

Fig. 3. Experimental setup of the three point bending experiment with clamping

epoxy resin. The force signal calibration was done in the same configuration as in the clamping group with a 150 kN universal tensile testing machine of Zwick, Ulm, Germany. The signal was processed with a VLink 200 amplifier of Lord MicroStrain, Williston, USA. To distinguish the two signals, the bolt loads are denoted with ”green” and ”black” in the following, according to the color of the plastic cover of the strain gauges.

2.2. Experimental set-up of the reference testing

In a first step, the reference for the fatigue life was tested with only the upper half of the clamping group acting on the specimen. All tests were run displacement controlled, since the failure in fiber parallel direction is assumed to be dominated by the strain to failure of the fiber (Talreja et al. (2012)). In a static preloading step, the bending modulus of a specimen was determined by approaching two displacement values and noting the reaction force values at the loading pin of the testing rig. The preloading displacements were set low enough so that the linear bending theory was still valid. Nominal bending stresses could be calculated from the reaction force values and nominal bending strains could be calculated from the displacement values along with the geometry of the specimen and the testing setup. The bending modulus was then determined from the ratio of di ff erences between the respective values. The desired nominal bending strain was calculated with respect to the geometric nonlinear problem by the iteration algorithm of Knickrehm (1999). The three point bending support was designed with roller bearings to avoid additional heating at the supports. The support span was 160 mm. All tests were performed with a frequency of f = 10 Hz, a sinusoidal wave form and a strain ratio of R = 0 . 1 related to the strain on the tension side of the specimen. All tests were run under ambient conditions with monitored but uncontrolled temperature. During the fatigue testing the dynamic sti ff ness was recorded, defined as di ff erence between maximum and minimum force divided by the di ff erence of the corresponding displacement values within a cycle (Eq. 1):

F max − F min u max − u min

k dyn =

(1)

In a postprocessing step, the dynamic sti ff ness was normalized to the value of the 1000th cycle. A residual sti ff ness based criterion was used to determine the fatigue life.