Crack compliance and dynamic effects during cyclic fatigue testing
The elastic compliance of a cracked specimen increases as the crack gets longer, which can be used to monitor crack growth during a mechanical test. I've recently done some cyclic fatigue tests (at 70Hz) on some compact tension specimens. I have data logs (sampled at 3000Hz) for the instantaneous applied load, and the hydraulic actuator displacement. One way to find the compliance of the sample, and hence the crack length at a given time, is to divide the displacement range by the load range for a cycle or series of cycles. However, even after correcting for the compliance of the sample grips and crosshead, the data is noisy and inconsistent.
When I look closely at the recorded waveforms, I notice the load is out of phase with the displacement, with the load leading the displacement by a small amount. This is the opposite of what I'd expect if the sample's inertia were affecting the test, but could be explained if the sample were behaving viscoelastically - maybe friction on the meeting rough crack faces? If so, is there any recognised way to deal with the problems caused by dynamic behaviour? A way to get a more accurate compliance value that considers these effects, or even get additonal information about the crack closure forces? I could probably have some fun with dynamic mechanical modelling of my samples and the test machine, but I don't want to repeat previous work if anyone's already developed a way of dealing with this. Any ideas?