The relationships between the properties of materials and their microstructures have been a central topic in materials science. The microstructure-property mapping and numerical failure prediction are critical for integrated computational material engineering (ICME). However, the bottleneck of ICME is the lack of a clear understanding of the failure mechanism as well as an efficient computational framework. To resolve these issues, research is performed on developing novel physics-based and data-driven numerical methods to reveal the failure mechanism of materials in microstructure-sensitive applications. First, to explore the damage mechanism of microstructure-sensitive materials in general loading cases, a nonlocal lattice particle model (LPM) is adopted because of its intrinsic ability to handle the discontinuity. However, the original form of LPM is unsuitable for simulating nonlinear behavior involving tensor calculation. Therefore, a damage-augmented LPM (DLPM) is proposed by introducing the concept of interchangeability and bond/particle-based damage criteria. The proposed DLPM successfully handles the damage accumulation behavior in general material systems under static and fatigue loading cases. Then, the study is focused on developing an efficient physics-based data-driven computational framework. A data-driven model is proposed to improve the efficiency of a finite element analysis neural network (FEA-Net). The proposed model, i.e., MFEA-Net, aims to learn a more powerful smoother in a multigrid context. The learned smoothers have good generalization properties, and the resulted MFEA-Net has linear computational complexity. The framework has been applied to efficiently predict the thermal and elastic behavior of composites with various microstructural fields. Finally, the fatigue behavior of additively manufactured (AM) Ti64 alloy is analyzed both experimentally and numerically. The fatigue experiments show the fatigue life is related with the contour process parameters, which can result in different pore defects, surface roughness, and grain structures. The fractography and grain structures are closely observed using scanning electron microscope. The sample geometry and defect/crack morphology are characterized through micro computed tomography (CT). After processing the pixel-level CT data, the fatigue crack initiation and growth behavior are numerically simulated using MFEA-Net and DLPM. The experiments and simulation results provided valuable insights in fatigue mechanism of AM Ti64 alloy.

Despite its relevance for law enforcement applications, the design of soft armor has mainly been based on a trial-and-error approach. A combined experimental and finite element analysis framework is used to build a predictive numerical model for the analysis and hence, design of soft armor. The material models for major components of the soft armor certification system—bullet, shoot pack, straps, and clay backing, are first constructed using laboratory tests and publicly available data. Next, three metrics, namely, back face signature (BFS), number of penetrated shoot pack layers, and mushrooming of the bullet, are established to gauge the model’s accuracy with respect to the laboratory ballistic test data. Finally, optimized material model parameters are obtained by calibrating a coarser model. The final accuracy test of the developed framework is carried out using laboratory ballistic test data involving multiple shots on the shoot pack. Subsequently, the impacts of incorporating stitching into the final model were examined and compared. The results indicate that reliable predictive data can be obtained using the developed process and can likely be extended for use in modeling other impact simulations.

Composites are replacing conventional materials in aerospace applications due to their light weight, non-corrosiveness, and high specific strength. This thesis aims to characterize the input data for IM7-8552 unidirectional composite to support MAT213, an orthotropic elasto-plastic damage material model and MAT_186, a mixed mode cohesive zone model used to model delamination. MAT_213 in conjunction with MAT_186 can be used to predict the behavior of composite under crush and impact loads including delamination. MAT_213 requires twelve sets of stress-strain curves, direction-dependent material constants, and flow rule coefficients as input. All the necessary inputs are obtained through the post-processing of a total of twelve distinct quasi-static and room temperature (QS-RT) experiments. MAT_186 is driven by a set of Mode I and Mode II fracture parameters and traction separation laws, a constitutive law that derives the relationship between stresses and relative displacements at integration points of cohesive elements. Obtaining cohesive law parameters experimentally is a tedious process as it requires close monitoring of the crack length during the test, which is a difficult task to achieve with accuracy even after using sophisticated equipment such as Digital Image Correlation (DIC). In this thesis, a numerical inverse analysis method to precisely predict these parameters by using finite element analysis with cohesive zone modeling and response surface methodology (RSM) is proposed. Three steps comprise RSM. The process in Step 1 involves calculating the root mean square error between the finite element and experimental load-displacement curves to produce the response surface. In step 2, the response surface is fitted with a second-order polynomial using the Levenberg-Marquardt algorithm. In step 3, an optimization problem is solved by minimizing the fitted function to find the optimum cohesive zone parameters. Finally, the obtained input for MAT_213 and MAT_186 material models is validated by performing a quasi-isotropic tension test simulation.

Aromatic polymers, with benzene-like rings in their main chains, include materials such as polyurea, an amorphous elastomer capable of dissipating large amounts of energy under dynamic loading, which makes it a promising coating for defensive systems. Although computational research exists that investigates the atomic-level response of polyurea and other amorphous aromatic polymers to extreme conditions, there is little experimental work to validate these models 1) at the atomic-scale and 2) under high pressures characteristic of extreme dynamic loading. Understanding structure-property relationships at the atomic-level is important for polymers, considering many of them undergo pressure and temperature-induced structural transformations, which must be understood to formulate accurate predictive models. This work aims to gain a deeper understanding of the high-pressure structural response of aromatic polymers at the atomic-level, with emphasis into the mechanisms associated with high-pressure transformations. Hence, atomic-level structural data at high pressures was obtained in situ via multiangle energy dispersive X-ray diffraction (EDXD) experiments at the Advanced Photon Source (APS) for polyurea and another amorphous aromatic polymer, polysulfone, chosen as a reference due to its relatively simple structure. Pressures up to 6 GPa were applied using a Paris Edinburgh (PE) hydraulic press at room temperature. Select polyurea samples were also heated to 277 °C at 6 GPa. The resulting structure factors and pair distribution functions, along with molecular dynamics simulations of polyurea provided by collaborators, suggest that the structures of both polymers are stable up to 6 GPa, aside from reductions in free-volume between polymer backbones. As higher pressures (≲ 32 GPa) were applied using diamond anvils in combination with the PE press, indications of structural transformations were observed in both polymers that appear similar in nature to the sp2-sp3 hybridization in compressed carbon. The transformation occurs gradually up to at least ~ 26 GPa in PSF, while it does not progress past ~ 15 GPa in polyurea. The changes are largely reversible, especially in polysulfone, consistent with pressure-driven, reversible graphite-diamond transformations in the absence of applied temperature. These results constitute some of the first in situ observations of the mechanisms that drive pressure-induced structural transformations in aromatic polymers.

An approach for modeling resistance spot welding of thin-gauge, dissimilar metal sheets with high electrical conductivity is presented in this work. In this scenario, the electrical and thermal contact resistances play a dominant role in heat generation and temperature evolution within the workpieces; these interactions ultimately control the weld geometry. Existing models are limited in modeling these interactions, especially for dissimilar and thin-gauge metal sheets, and at higher temperatures when the multiphysics becomes increasingly interdependent. The approach presented here uses resistivity measurements, combined with thermal modeling and known bulk resistance relationships to infer the relationship between electrical contact resistance and temperature for each of the different material interfaces in the welding process. Corresponding thermal contact resistance models are developed using the Wiedemann-Franz law combined with a scaling factor to account for nonmetallic behavior. Experimental and simulation voltage histories and final weld diameter were used to validate this model for a Cu/Al/Cu and a Cu/Al/Cu/Al/Cu stack-ups. This model was then used to study the effect of Ni-P coating on resistance spot welding of Cu and Al sheets in terms of weld formation, mechanical deformation, and contact resistance. Contact resistance and current density distribution are highly dependent on contact pressure and temperature distribution at the Cu/Al interface in the presence of alumina. The Ni-P coating helps evolve a partially-bonded donut shaped weld into a fully-bonded hourglass-shaped weld by decreasing the dependence of contact resistance and current density distribution on contact pressure and temperature distribution at the Cu/Al interface. This work also provides an approach to minimize distortion due to offset-rolling in thin aluminum sheets by optimizing the stiffening feature geometry. The distortion is minimized using particle swarm optimization. The objective function is a function of distortion and smallest radius of curvature in the geometry. Doubling the minimum allowable radius of curvature nearly doubles the reduction in distortion from the stadium shape for a quarter model. Reduction in distortion in the quarter model extends to the full-scale model with the best design performing 5.3% and 27% better than the corresponding nominal design for a quarter and full-scale model, respectively.

Layer-wise extrusion of soft-solid like cement pastes and mortars is commonly used in 3D printing of concrete. Rheological and mechanical characterization of the printable binder for on-demand flow and subsequent structuration is a critical challenge. This research is an effort to understand the mechanics of cementitious binders as soft solids in the fresh state, towards establishing material-process relationships to enhance print quality. This study introduces 3D printable binders developed based on rotational and capillary rheology test parameters, and establish the direct influence of packing coefficients, geometric ratio, slip velocities, and critical print velocities on the extrudate quality. The ratio of packing fraction to the square of average particle diameter (0.01-0.02), and equivalent microstructural index (5-20) were suitable for printing, and were directly related to the cohesion and extrusional yield stress of the material. In fact, steady state pressure for printing (30-40 kPa) is proportional to the extrusional yield stress, and increases with the geometric ratio (0-60) and print velocity (5-50 mm/s). Higher print velocities results in higher wall shear stresses and was exponentially related to the slip layer thickness (estimated between 1-5μ), while the addition of superplasticizers improve the slip layer thickness and the extrudate flow. However, the steady state pressure and printer capacity limits the maximum print velocity while the deadzone length limits the minimum velocity allowable (critical velocity regime) for printing. The evolution of buildability with time for the fresh state mortars was characterized with digital image correlation using compressive strain and strain rate in printed layers. The fresh state characteristics (interlayer and interfilamentous) and process parameters (layer height and fiber dimensions) influence the hardened mechanical properties. A lower layer height generally improves the mechanical properties and slight addition of fiber (up to 0.3% by volume) results in a 15-30% increase in the mechanical properties. 3D scanning and point-cloud analysis was also used to assess the geometric tolerance of a print based on mean error distances, print accuracy index, and layer-wise percent overlap. The research output will contribute to a synergistic material-process design and development of test methods for printability in the context of 3D printing of concrete.

Investigation into research literature was conducted in order to understand the impacts of traditional concrete construction and explore recent advancements in 3D printing technologies and methodologies. The research project focuses on the relationship between computer modeling, testing, and verification to reduce concrete usage in flexural elements. The project features small-scale and large-scale printing applications modelled by finite element analysis software and printed for laboratory testing. The laboratory testing included mortar cylinder testing, digital image correlation (DIC), and four pointbending tests. Results demonstrated comparable performance between casted, printed solid, and printed optimized flexural elements. Results additionally mimicked finite element models regarding failure regions.

A framework to obtain the failure surface of a unidirectional composite which can be used as an input for Generalized Tabulated Failure Criterion in MAT_213 – an orthotropic elasto-plastic material model implemented in LS-DYNA, a commercial finite element program, is discussed in this research. A finite element model consisting of the fiber and the matrix is generated using the Virtual Testing Software System (VTSS) developed at Arizona State University (ASU). The framework is illustrated using the T800-F3900 unidirectional composite material manufactured by Toray Composites. The T800S fiber is modeled using MAT_213. The F3900 matrix phase is modeled using MAT_187-SAMP1. The response of the virtual tests in 1-direction tension, 1-direction compression, 2-direction tension, 2-direction compression and 2-1 plane shear are verified against the results obtained from experiments performed under quasi-static and room temperature conditions (QS-RT). Finally, a roadmap to generate the failure surface using virtual test is proposed.