Micromechanically derived lowest threshold stress intensity factor for macroscopic stress-corrosion cracking in unidirectional GFRP composites

Abstract

Observation by scanning electron microscopy of a typical fracture surface in glass fiber-reinforced polymer (GFRP) composites shows that the fracture surface of each fiber is characterized by a mirror region surrounded by a hackle region. This indicates a two-stage fracture process: (1) a ‘‘slow’’ (meaning time-dependent) fracture of a portion of the glass fiber is followed by (2) a ‘‘fast’’ unstable fracture across the remainder of the glass fiber. On the premise that the time-dependent failure of GFRP composites in acid environments is controlled by the initiation and slow growth of a crack from a pre-existing inherent surface flaw in a glass fiber, a micromechanical model of stress-corrosion crack growth (sometimes called the Sekine–Miyanaga–Beaumont model) was constructed [1–3]. Using this model, an equation was derived for the macroscopic crack growth rate as a function of the apparent crack tip stress intensity factor for the mode I. If tougher and more ductile matrices are used, there exists a lowest threshold value of the stress intensity factor. By assuming that the polymeric fibrils of matrix or ligaments stretched between the crack surfaces behave according to an ideal cohesive force model, i.e., the Dugdale model [4], the lowest threshold stress intensity factor was given in an explicit form [3]. However, the examination in the case of a more general cohesive force model, i.e., the Barenblatt model [5], has been left until now, although this examination will give a physically definite description of the lowest threshold stress intensity factor. We begin this examination with a brief review of our previous article [3]. In bulk glass, the stable crack growth rate due to stresscorrosion cracking da/dt can be given by