Mixed numerical-experimental identification of non-local characteristics of random-fibre-reinforced composites

In the present study, the relationship between the initial microstructure, the specific damage mechanisms and the length parameter which arises in hig her-order continuum models, is investigated for three materials with a grad ient-enhanced damage model. For this purpose, a series of experiments has b een carried out on polypropylene-based composite materials with different g lass-fibre lengths or microstructures. The process zone close to the crack tip has been investigated with a digital image correlation technique, in wh ich images of the deformed process zone are digitized and stored. A correla tion analysis between the successive images permits quantification of the d isplacements in the process zone. An inverse mixed numerical-experimental a pproach then permits the determination of the length parameter and the gove rning parameters for damage development. The influence of the microstructur al characteristics of each material and the failure mechanisms on the param eters has been scrutinized. Conclusions are drawn with respect to the physi cal nature of some non-local aspects, which relate the transforming microst ructure to the macroscopically observed behaviour. (C) 1999 Elsevier Scienc e Ltd. All rights reserved.