Delamination experiments were first performed on unidirectional glass/epoxy
composite specimens to determine a mixed-mode initiation failure criterion
. The tests employed use the double cantilever beam (DCB) specimen for mode
I, end-notched flexure (ENF) for mode II and mixed-mode bending (MMB) for
I/II mixed-mode loading. MMB experiments are analysed according to a method
based on beam theory and an experimental method which uses measured displa
cements. These two analyses give similar results. The initiation failure po
int is defined both by the first non-linearity of the load/displacement cur
ve and by use of the first acoustic signal. These two definitions result in
values very close to each other. A validation of this criterion is then ca
rried out on asymmetric DCB and MMB specimens of the same material. Mixed-m
ode partitioning is performed by three different methods. These are the glo
bal and local methods, which are based on analytical expressions, and the m
odified crack-closure integral (MCCI), which uses a finite-element analysis
. The three failure envelopes obtained are compared with the failure criter
ion determined on symmetric specimens. A very good correlation is observed
between the mixed-mode results from symmetric specimens and the failure env
elopes determined by both the local and finite-element methods of mode part
itioning for asymmetric specimens. On the other hand, the global method doe
s not seem to be well adapted to such tests for this material. (C) 1999 Els
evier Science Ltd. All rights reserved.