Fatigue Life Assessment for Specimens Subjected to Varying Load Levels in a High-Throughput Method for Vibration-Based Fatigue

In vibration-based fatigue testing, a cantilevered rectangular plate is excited in resonance until failure. This technique has recently been modified to clamp smaller gauge specimens into a larger, reusable carrier plate to reduce material costs. This present work modifies the technique even further by inserting multiple clamped specimens into a shared carrier plate, thus fatiguing multiple specimens in parallel. In the time that it takes one specimen to fail, the other specimens each accumulate the same number of cycles as the failed specimen, thus reducing the remaining time for the other specimens to fail. The failed specimen is then removed and either (a) replaced with a fresh specimen, or (b) is left un-replaced. In the case when specimens are replaced, the amplitude of loading is changed to avoid repeating the same data points. When specimens are not replaced, the mode shape of the overall assembly changes in response to the new geometry, consequently changing the strain relationship between specimens. Either way, this means that some specimens accumulate damage at multiple load levels, and thus require a damage accumulation model to assess their net fatigue life.  Based on this approach, a mathematical framework is presented to track the damage accumulation in each of the specimens. Fatigue life is assumed to follow Basquin’s law, while damage accumulation is assumed to follow Miner’s rule.

Using this framework, four scenarios are examined for a carrier plate with six inserts: (1) specimens are removed but not replaced each time that they fail; (2) specimens are replaced until at least one specimen from each location in the assembly has failed at least once, after which they are removed when they fail; (3) the same as case 2, except that the level of excitation is reduced by 5% to reduce repeated datapoints after specimens in each of the first three locations fail; and (4) the same as case 3, except the level of excitation is changed at different intervals. Of the four scenarios, case 4 has the highest throughput, yielding a total of 15 datapoints at fatigue lives spanning three orders of magnitude at a rate 5.49 times faster compared to testing each specimen individually at the same frequencies. As vibration-based methods operate at frequencies on the order of kHz, testing a single specimen is already about 30 times faster compared to measuring the same datapoints in an axial load frame, making the net throughput over 160 times faster by shaking multiple specimens in tandem.