A comment on "Discussion on the use of the strain energy release rate for fatigue delamination characterization"

TL;DR

This paper critically examines a recent proposal to use the total work per cycle as a parameter for fatigue delamination characterization, highlighting its equivalence to traditional SERR methods under linear assumptions and discussing its limitations.

Contribution

It clarifies the relationship between the proposed work-based approach and classical SERR methods, emphasizing the non-predictive nature and the need for more fundamental models.

Findings

The work-based approach is equivalent to SERR under linear elastic assumptions.

Paris law dependence emerges only for long cracks, not fundamental physics.

The proposal is observational and not predictive, requiring direct measurement of work input.

Abstract

In a recent very interesting and illuminating proposal, Yao et al. (2014) have discussed the use of the strain energy release rate (SERR) as a parameter to characterize fatigue delamination growth in composite materials. They consider fatigue delamination data strongly affected by R-curve behaviour due to fibres bridging and argue that a better approach is to correlate the crack advance with the total work per cycle measured in the testing machine. This seems to work better than estimating the compliance as a linear fit of experimental curves from Modified Compliance Calibration ASTM standards equations for the SERR in the classical Linear Elastic Fracture Mechanics framework. We show however that if we assume indeed linear behaviour (i.e. LEFM), the approach they introduce is perfectly equivalent to the SERR one, i.e. Paris type of laws. As well known form Barenblatt and Botvina, fatigue crack growth is a weak form of scaling, and it gives Paris classical dependence only when the crack is much longer than any other characteristic sizes. Paris' law is not a fundamental law of physics, is not an energy balance equation like Griffith, and strong size effects due to cohesive zones have been found already in concrete by Bazant. The proposal is very simple, and interesting as it would seem to suggest that a proper scaling with a cohesive model at crack tip could be predicted, although this doesn't seem to have been attempted in the Literature. The main drawback of the present proposal is that it is not predictive, but purely observational, as it requires the actual measurement of work input during the fatigue process.