Carbon-fiber/SiC-matrix composites are under development for applications in hypersonic vehicles due to their exceptional capabilities at high temperatures. As a subset of these materials, textile-based composites are of particular interest because they offer the possibility of accommodating complex geometries and features in engineering components. Among the numerous obstacles hindering the widespread adoption of these composites, two are addressed in the present work: (i) the incomplete understanding of the influence of textile architecture on thermoelastic properties, damage initiation and failure, and (ii) the lack of robust computational tools for predicting their thermomechanical performance at the appropriate length scales. Accordingly, an experimental study is performed of the thermal and mechanical properties of several prototypical textile C/SiC composites with various fiber architectures. In turn, the experimental results are used to guide the development of computational tools for predicting composite response that explicitly account for fiber architecture.

Textile architecture is found to influence composite response at four length-scales: the panel, the coupon, the tow, and the sub-tow. At the panel scale, distortions to the architecture introduced during weaving or handling of the fabric influence the packing density and the relative rotation of tows. Even when large distortions are intentionally introduced their influence on mechanical response is minimal. At the coupon scale the tow architecture has the largest effects on composite mechanical response. Young's modulus, ultimate tensile strength, and strain to failure are all influenced. Changes in each of these are a function of tow shape, tow anisotropy, and the degree of constraint provided by the matrix. At the tow scale, architecture effects give rise to heterogeneity in measured surface strains under both tensile and thermal loading. Methods for the calibration of tow-scale elastic and thermoelastic properties were developed to enable simulation of these effects with a geometrically-accurate virtual model. Virtual tensile and thermal tests using this model have indicated that interaction between tows has an important influence on local strains. At the sub-tow scale, architecture effects influence the location of matrix cracking. Simulations of the cooling cycle following matrix processing predict that matrix cracks should develop in the matrix above underlying tows due to thermal expansion mismatch between the tows and the matrix. This is consistent with experimental observations. Two methods are presented to extend the virtual tests to explicitly simulate the onset and evolution of these cracks.