Relating polymer matrix composite delamination behavior to constituent properties

Abstract

Developing predictive capabilities of composite material behavior from constituent properties is an important component of accelerating materials insertion. Many models exist that accomplish this objective for a range of material properties, but no such method is available for delamination properties. This thesis explores the issues associated with predicting polymer matrix composite Mode I delamination behavior from constituent properties by examining the topic from a variety of perspectives. Nanoindentation tests of matrix materials in composites and their associated neat polymers analyze the assumption that neat matrix properties are the same as unconstrained in situ properties and therefore may be used in models predicting composite behavior from constituent properties. Quasi-static and fatigue fracture experiments using a graphite/epoxy composite and its neat resin at a variety of temperatures and loading rates allow for an examination of the dominant mechanisms involved in the fracture process, an analysis of the shifts in quasi-static behavior with temperature and potential implications for fatigue predictions, and measurement of values that will act as inputs and verification of a delamination initiation model. A global-local finite element model of a double cantilever beam specimen is used to study the prediction of delamination initiation by examining inelastic matrix deformation at the crack tip. In addition, a fiber bridging fatigue model is created to analyze crack propagation data that effectively separates the bridging and resin crack tip contributions. The final component of the thesis is to tie the various experimental and analytical studies together to create methodologies that may be used in a design or research environment

(cont.) to accelerate materials insertion. An important conclusion from the thesis is that quantitative predictions of composite fracture behavior using unmodified neat matrix properties is not feasible, but the constrained matrix properties may be used to predict composite delamination behavior. It is also shown that following a process of implementing mechanism-based models in conjunction with experimental observations is essential when implementing models that bridge lengthscales.