The characterization and understanding of damage initiation and evolution in SiC/SiC ceramic matrix composites is critical to improve modeling efforts and design guidelines for these high impact materials. In this dissertation work, microscale damage accumulation was investigated in two systems of SiC/SiC minicomposites of different architectures. A combination of mechanical testing with acoustic emission (AE) inside a scanning electron microscope was used to map surface damage at high spatial resolution and compare it to bulk damage accumulation. This combination of in situ damage monitoring enabled the characterization of the evolution of matrix crack density and crack opening displacements and enabled accurate mapping of local damage events to their generated AE. It was found that AE is a strong predictor of transverse matrix crack density, and also likely captures small contributions from secondary mechanisms such as debonding and sliding. The sensitivity of AE was examined using a modeling framework that leveraged simplifications of specimen geometry and assumptions of damage progression to compare point-by-point AE accumulation with estimations of damage accumulation in unidirectional CMCs. This framework resulted in an AE-based estimate of fiber failure evolution that agreed with micromechanics-based predictions when the AE contributions of secondary mechanisms were included. Finally, in situ matrix crack opening displacements (CODs) were characterized, and variations within specimens and across minicomposite systems were related to key microstructural and damage accumulation differences. Experimentally obtained CODs were compared to COD predictions from the literature, and it was found that the accuracy of the model predictions principally depended on consideration of the (i) progressive nature of fiber failure prior to specimen failure; and (ii) variations between the initial crack opening rate (upon matrix crack formation) and the subsequent crack opening rate (upon continued axial loading).

There is great interest in magnesium and magnesium alloys for use as structural materials due to their high strength-to-weight ratio compared to aluminum and steel. However, the relationship between the macroscopic failure behavior of magnesium and the micro- and mesoscopic deformation mechanisms in magnesium are relatively poorly understood. Basal slip is ubiquitous in magnesium deformation, but the factors that determine the secondary deformation mechanisms, non-basal slip and twinning, are less well studied. For this reason, the aim of this study was to identify the relationship between twin occurrence and grain size. Towards this goal, a combination of electron backscatter diffraction (EBSD), scanning electron microscope digital image correlation (SEM-DIC) were used to capture full-field strain maps of a ~1mm x 1mm field of view on a pure magnesium dogbone sample under monotonic compression. It was found that the likelihood that a grain exhibits twinning increases as the size of the grain increases. K-means clustering, filtered cross-correlation, k-means clustering, and an AlexNet convolutional neural network (CNN) were also used to attempt to automate the twin identification process. Though the results of this work exhibited the same trend (increased twin likelihood with increased grain size), the precision, sensitivity, and accuracy of these methods was analyzed and it was found that more work needs to be done before significant conclusions can be drawn from their results.

SiC/SiC composites are refractory, damage tolerant, high-strength materials that are ideal for use in turbine engines. Although these properties allow engineers to design aircraft with better fuel efficiency, the safety-critical nature of their application space necessitates a detailed understanding of how damage initiates and progresses. Recent advances in machine learning (ML) and related statistical tools have created novel characterization pathways for understanding damage accumulation, however their widespread adoption is limited due to a lack of interpretability. This work lays the foundation for trustworthy ML for interpreting acoustic emissions (AE) that are produced when SiC/SiC composites sustain damage. A modified signal representation scheme was combined with unsupervised clustering to form the spectral framework. This represents the first AE-ML approach that can distinguish between fiber break and matrix crack signals in SiC/SiC minicomposites. The spectral framework was then benchmarked against 4 state-of-the-art AE-ML frameworks and shown to achieve superior performance. Community guidelines for standardized benchmarking were proposed to promote greater transparency. Finally, an in situ x-ray computed tomography experiment demonstrated that matrix crack signals overlap with early fiber breaks, and prevents the use of unsupervised clustering for realistic composite geometries. An autoencoder framework was created to overcome this limitation and used to demonstrate that the frequency distribution of fiber break signals is compact. These findings establish a pathway for real-time health monitoring.

Additive manufacturing (AM) of metallic components has become a full-fledged manufacturing technique as the space and aerospace industries require reduced weight, increased design and property control, and reduced lead times. The advantages of additive manufacturing directly correlate to challenges in characterization, property prediction, and component reliability, as printed parts display remarkable variability compared to traditionally qualified components. Regarding characterization, additively manufactured components represent a departure from traditional assumptions such as isotropic (same in all directions) material properties. Particularly for properties such as crystallographic texture, the variable solidification conditions of AM require consideration of arbitrary anisotropy and symmetry on the bulk scale.This work focuses on advancing the quantification of the properties of additively manufactured components with the bulk ultrasonic technique, resonant ultrasound spectroscopy (RUS). Critically, this technique is non-destructive, low-cost, and can be performed in a matter of minutes. However, existing frameworks to characterize properties such as the elastic constants from ultrasonic data only account for either isotropic or single crystalline compounds, with assumptions made to simplify the elastic symmetry of polycrystalline textures. To account for arbitrarily textured microstructures, a novel framework is developed to quantify texture directly from the resonant frequencies, with a novel cobalt-nickel-base superalloy (SB-CoNi-10C) specimen used to demonstrate the effectiveness of the technique. The determination of elastic constants from the resonant frequencies requires inverse problem solving to iteratively calculate the resonant frequencies and compare them to those measured in the laboratory. Given the complexity of the solution space and a need for robust parameter estimates on the independently determined texture, a CPU-parallelizable Bayesian inference technique, Sequential Monte Carlo (SMC) is employed to reduce computational costs to under 24h on a 10-core system. Looking toward crystal scale property prediction, the single crystal elastic constants are critical to understand the behavior of novel AM alloys. Therefore, a framework is developed to determine the single crystal elastic constants from the resonant frequencies of textured AM specimens, provided their texture is known by EBSD or neutron diffraction measurements. Agreement with reported single crystal constants measured on grown single crystal specimens is demonstrated, circumventing the need to grow single crystals. This framework provides an end-to-end determination of the single crystal elastic constants that quantifies error directly from the measured resonant frequency measurements. The determination of single crystal elastic constants from AM specimens has been extended to the lab scale, with 2 mm x 2 mm EBSD scans demonstrated as sufficient to inform the texture within a specimen, in contrast to the prior need for neutron diffraction or large-scale EBSD data. The incorporation of texture coefficient variability from an EBSD measurement in the model enables the single crystal elastic constants to be determined much more accurately. The single crystal elastic constants are determined with EBSD and RUS data of additively manufactured SB-CoNi-10C, Inconel 625, and Ti-6Al-4V, demonstrating agreement with previously reported literature values for each material. Finally, RUS is extended to arbitrary geometry specimens of AM Inconel 625, demonstrating the capability of RUS to quantify microstructure variability by leveraging finite-element forward models of the resonant frequencies and correlating to experimental frequencies. Insightful examples are provided to demonstrate the limitations and constraints of determining the single crystal elastic constants and texture coefficients from resonant frequencies.