Application of three-dimensional (3D) woven composites is growing as an alternative to the use of ply-based composite materials. However, the design, analysis, modeling, and optimization of these materials is more challenging due to their complex and inherently multiscale geometries. Herein, a multiscale modeling procedure, based on efficient, semi-analytical micromechanical theories rather than the traditional finite element approach, is presented and applied to a 3D woven carbon–epoxy composite. A crack-band progressive damage model was employed for the matrix constituent to capture the globally observed nonlinear response. Realistic microstructural dimensions and tow-fiber volume fractions were determined from detailed X-ray computed tomography (CT) and scanning electron microscopy data. Pre-existing binder-tow disbonds and weft-tow waviness, observed in X-ray CT scans of the composite, were also included in the model. The results were compared with experimental data for the in-plane tensile and shear behavior of the composite. The tensile predictions exhibited good correlations with the test data. While the model was able to capture the less brittle nature of the in-plane shear response, quantitative measures were underpredicted to some degree.

The polymer spray-on foam insulation used on NASA’s Space Shuttle external fuel tank is analyzed via the high-fidelity generalized method of cells micromechanical model. This model has been enhanced to include internal pore pressure, which is applied as a boundary condition on the internal faces of the foam pores. The pore pressure arises due to both ideal gas expansion during a temperature change as well as outgassing of species from the foam polymer material. Material creep and elastic stiffening are also incorporated via appropriate constitutive models. Due to the lack of reliable properties for the in situ foam polymer material, these parameters are backed out from foam thermomechanical test data. Parametric studies of the effects of key variables (both property-related and microstructural) are presented as is a comparison of model predictions for the thermal expansion behavior of the foam with experimental data.

Enhanced finite elements are elements with an embedded analytical solution that can capture detailed local fields, enabling more efficient mesh independent finite element analysis. In earlier research, this method was applied to adhesively bonded joints. The adherends were modeled as composite Euler–Bernoulli beams, and the adhesive layer was modeled as a bed of linear shear and normal springs. The field equations were derived using the principle of minimum potential energy, and the resulting solutions for the displacement fields were used to generate shape functions and a stiffness matrix for a single bonded joint finite element. In this study, the capability to model large rotations and nonlinear adhesive constitutive behavior is developed, and progressive failure of the adhesive is modeled by remeshing the joint as the adhesive fails. The results obtained using this enhanced joint element are compared with experimental results.

The behavior and response of ceramic matrix composites (CMCs), in particular silicon carbide fiber reinforced silicon carbide matrix (SiC/SiC), is affected by many factors such as variation of fiber volume fraction, residual stresses resulting from processing of the composites at high temperature, random microstructures, and the presence of matrix flaws (e.g., voids, pores, cracks etc.) as well as general material nonlinearity and heterogeneity that occurs randomly in a composite. Residual stresses arising from the phase change of constituents are evaluated in this paper and it is shown that they do influence composite strength and need to be properly accounted for. Additionally, the microstructures (location of fiber centers, coating thickness etc.) of advanced CMCs are usually disordered (or random) and fiber diameter and strength typically have a distribution. They rarely resemble the ordered fiber packing (square, rectangular, or hexagonal) that is generally assumed in micromechanics-based models with periodic boundary conditions for computational expediency. These issues raise the question of how should one model such systems effectively? Can an ordered hexagonal packed repeating unit cell (RUC) accurately represent the random microstructure behavior? How many fibers need to be included to enable accurate representation? Clearly, the number of fibers within an RUC must be limited to insure a balance between accuracy and efficiency. NASA’s in-house micromechanics-based code MAC/GMC provides a framework to analyze such RUCs for the overall composite behavior and the FEAMAC computer code provides linkage of MAC/GMC to the commercial FEA code, ABAQUS. The appropriate level of discretization of the RUC as well as the analysis method employed, i.e., Generalized Method of Cells (GMC) or High Fidelity Generalized Method of Cells (HFGMC), is investigated in this paper in the context of a unidirectional as well as a cross-ply laminated CMC. Results including effective composite properties, proportional limit stress (an important design parameter) and fatigue are shown utilizing both GMC as well as HFGMC. Finally, a few multiscale analyses are performed on smooth bar test coupons as well as test coupons with features such as open-hole and double notches using FEAMAC. Best practices and guidance are provided to take these phenomena into account and keep a proper balance between fidelity (accuracy) and efficiency. Following these guidelines can account for important physics of the problem and provide significant advantages when performing large multiscale composite structural analyses. Finally, to demonstrate the multiscale analysis framework, a CMC gas turbine engine vane structure is analyzed involving a progressive damage model.