Analysis of the Results of Strain Controlled Fatigue Testing of Reinforced Polypropylene

In Figure 1 , for metallic cyclicly softening materials, one can see that the stress-strain hysteresis loops become very large rapidly, and in a test under load control, failure quickly ensues. The large strains often cause cyclic buckling or cyclic ratcheting; and it is difficult to decide what is the “characteristic” strain for a test of this stress amplitude. The use of strain control in large amplitude tests provides better test stability. Specific details can be found in the references [ 1 ] and [ 2 ]. Briefly, the test process consists of fatigue testing uniaxial metal samples in fully reveresed strain control to failure, and then recording the 1/2 life stress-strain hysteresis loop as a “characteristic” stress for that strain amplitude.

Figure 2 plots these half-life characteristic stress amplitudes for a cyclic softening Boron steel [ 3 ] along with the monotonic tension test. One can see a substantial difference in the stress-strain behavior of the monotonic and cyclic curves. The cyclic stress-strain curve is widely used for fatigue design[ 4 ]. It should be noted that many metals exhibit cyclic hardening behavior during fatigue, but will not be discussed here since the plastic material of interest in this study shows cyclic softening.

Figure 3 plots the change of the hysteresis loop tensile tip stress in the tests for this boron steel. In the plot the stress amplitude of stress-strain hysteresis loops of the axial fully reversed strain controlled tests are plotted at intervals during the fatigue life. Failure is denoted by a downward pointing arrow at the end of each softening line. One can see that after a great deal of initial cyclic softening, much of the life is spent near the value of the half-life stress amplitude. The shortest life test, depicted at the top of the figure with a strain range of 0.0493 fails at 425 reversals, and at 200 reversals, the approximate half life, the stress amplitude was 104ksi. Measurements of stress response in these older tests were recorded by a combination of strip chart and x-y plotter. All of the tests[ 3 ] had less than 1ksi mean stress, with the exception of the runout test with a −5% mean stress. A logrithmic axis for cycles or reversals is traditionally used in this type of plot because a linear scale would de-emphazise the transient softening behavior and it would be difficult to combine long and short life results in one plot.