| Abstract: | This work examines composite failure at multiple scales. The first scale that is examined is the fiber-matrix scale, where fibers and matrix are discretely modeled. A model is developed at this scale which includes randomness in the fiber positions. This randomness is found to signifcantly infuence the stress feld and resulting failure that occurs under thermo mechanical loads as compared to fiber-matrix microstructures with regular arrays of fibers. The fiber-matrix model is utilized to characterize variability and temperature dependence of the composite strength arising from microstructural randomness and the presence of thermally induced stresses. The second scale that is examined is that of a textile unit cell. Failure initiation behavior is examined for a variety of thermo mechanical loadings at this scale, and it is found that failure tends to initiate in a limited number of ways for a wide variety of loadings. A new progressive failure model is then examined for the textile unit cell. This model utilizes cohesive interface elements in the tows, neat matrix pockets, and tow and matrix interfaces to account for crack opening in the textile, as well as a continuum damage model to account for diffuse damage in the tows. Variability and temperature dependence of the transverse tow strength is introduced by specifying varying cohesive strengths in the intra-tow cohesive zones using a Weibull distribution characterized using the random fiber-matrix model. Progressive failure analyses are then performed for the textile unit cell under a variety of thermomechanical loads, and the resulting behaviors are compared to identify characteristic modes of damage development and their effect on the textile response. A continuum damage model for the textile material, which can be applied to engineering structures, is developed based on the characteristic damage modes observed in the textile unit cell analyses. This model tracks the evolution of each characteristic mode of damage based on the structural-scale stress and predicts the degradation in the textile response as a result of this damage. The ability of this model to predict the textile's response under various damage-inducing loads is then compared to the response obtained from textile unit cell progressive failure analysis, and both models are found to be in good agreement for most loadings. The electronic version of this dissertation is accessible from http://hdl.handle.net/1969.1/151792 |
|---|