Integrated Computational Materials Engineering (ICME)-based tools and techniques have been identified as the best path forward for distortion mitigation in thin-plate steel construction at shipyards. ICME tools require temperature-dependent material properties—including specific heat, thermal conductivity, coefficient of thermal expansion, elastic modulus, yield strength, flow stress, and microstructural evolution—to achieve accurate computational results for distortion and residual stress. However, the required temperature-dependent material property databases of US Navy-relevant steels are not available in the literature. Therefore, a comprehensive testing plan for some of the most common marine steels used in the construction of US Naval vessels was completed. This testing plan included DH36, HSLA-65, HSLA-80, HSLA-100, HY-80, and HY-100 steel with a nominal thickness of 4.76 mm (3/16-in.). This report is the second part of a seven-part series detailing the pedigreed steel data. The first six reports will report the material properties for each of the individual steel grades, whereas the final report will compare and contrast the measured steel properties across all six steels. This report will focus specifically on the data associated with HSLA-80 steel.

There are two broad modeling paradigms in scientific applications: forward and inverse. While forward modeling estimates the observations based on known causes, inverse modeling attempts to infer the causes given the observations. Inverse problems are usually more critical as well as difficult in scientific applications as they seek to explore the causes that cannot be directly observed. Inverse problems are used extensively in various scientific fields, such as geophysics, health care and materials science. Exploring the relationships from properties to microstructures is one of the inverse problems in material science. It is challenging to solve the microstructure discovery inverse problem, because it usually needs to learn a one-to-many nonlinear mapping. Given a target property, there are multiple different microstructures that exhibit the target property, and their discovery also requires significant computing time. Further, microstructure discovery becomes even more difficult because the dimension of properties (input) is much lower than that of microstructures (output). In this work, we propose a framework consisting of generative adversarial networks and mixture density networks for inverse modeling of structure–property linkages in materials, i.e., microstructure discovery for a given property. The results demonstrate that compared to baseline methods, the proposed framework can overcome the above-mentioned challenges and discover multiple promising solutions in an efficient manner.

Thermodynamically driven short-range order (SRO) is known to exist in multi-component alloys and can influence their mechanical properties. Atomistic and mesoscale simulations are frequently used to study the effects of SRO, but introducing SRO into the lattice usually requires costly Monte Carlo simulations. Here, we develop a novel method and open-source code to create lattices with specified SRO parameters that can be used as inputs to simulations to study the interaction of SRO with dislocations or other deformation mechanisms. The method, order through informed swapping (OTIS), uses a statistical smart swapping procedure to create lattices with known SRO parameters from an initially random lattice. We illustrate the use of OTIS on ternary and quinary equimolar alloys to create both BCC and FCC lattices with hundreds of thousands of atoms. The computation time and size scaling are discussed. As an example application of the method, the generated SRO lattices are incorporated into a phase-field dislocation dynamics to study dislocation loop growth.

Three new advanced acquisition strategies for lab-based diffraction contrast tomography are presented. They are named Helical Phyllotaxis, Helical Phyllotaxis Raster, and Helical Phyllotaxis HART and enable grain mapping of longer, larger, high-aspect ratio samples. The implementation of these advanced acquisition strategies combines a golden angle rotation with vertical and horizontal translations to perform a seamless data collection that has a uniform sample illumination both angularly and spatially. Reconstruction of the corresponding data is equally seamless, simultaneously using all data to reconstruct the full illuminated volume without the need for registration or stitching of data subsets or sample subvolumes. Examples of performing Helical Phyllotaxis and Helical Phyllotaxis Raster scans on a selection of samples, which have either been mapped previously or come from the same batch of samples, show a substantial reduction in data collection time and/or a significant improvement in grain statistics. The Helical Phyllotaxis HART (high-aspect ratio tomography for plate-like samples) strategy enables investigations of a hitherto inaccessible class of sample geometries comprising industrially relevant materials like rolled metal sheets and electrical steels. The advanced acquisition strategies take lab-based non-destructive 3D grain mapping to the next level of throughput, grain statistics and versatility and hold great promise for integrated computational materials science and engineering applications. While the throughput warrants 4D studies of materials microstructural evolution, a representative sample volume is a prerequisite for successful model predictions of the evolution, and the versatility enables studies of samples or components under more realistic in situ or in operando conditions.

A method for generating high-fidelity, boundary conforming tetrahedral mesh of three-dimensional (3D) polycrystalline microstructures is presented. With growing interest into subgrain scale micromechanics of materials, crystal plasticity finite element (CPFE) models must adapt not only their respective constitutive laws, but also their model geometry to the finer scale, namely the representation of grains and grain boundary junctions. Additionally, with the increasing availability of microstructure datasets obtained via 3D tomography experiments, it is possible to characterize the 3D topology of grains. From these advancements in experiment emerge both an opportunity and challenge for researchers to develop model microstructures, specifically finite element meshes, which best preserve grain topology for the accurate representation of boundary conditions in polycrystalline materials. To accomplish this, an open-source code called XtalMesh was created and is presented here. XtalMesh works by smoothing input voxel microstructure data using a feature-aware Laplacian smoothing algorithm that preserves complex grain topology and leverages state-of-the-art tetrahedralization code fTetWild to generate volume mesh. In this work, the workflow and associated algorithms of XtalMesh are described in detail using a synthetically generated example microstructure. For demonstration, we present a case study involving mesh generation of an experimentally obtained microstructure of nickel-based superalloy Inconel 718 (IN718).

Materials scientists are facing increasingly challenging multi-objective performance requirements to meet the needs of modern systems such as lighter-weight and more fuel-efficient aircraft engines, and higher heat and oxidation-resistant steam turbines. While so-called second wave statistical machine learning techniques are beginning to accelerate the materials development cycle, most materials science applications are data-deprived when compared to the vastness and complexity of the search space of possible solutions. In line with DARPA’s vision of third wave AI approaches, we believe a combination of data-driven statistical machine learning and domain knowledge will be required to achieve a true revolution in materials discovery. To that end, we envision and have begun reducing to practice a system that fuses three forms of knowledge—factual scientific knowledge, physics-based and/or data-driven analytical models, and domain expert knowledge—into a single ‘Compound Knowledge Graph’ in which contextual reasoning and adaptation can be performed to answer increasingly complex questions. We believe this Compound Knowledge Graph-based system can be the nucleus of a collaborative AI assistant that supports stateful natural language back-and-forth dialogs between materials scientists and the AI to accelerate the development and discovery of new materials. This paper details our vision, summarizes our progress to date on a steam turbine blade coating use case, and outlines our thoughts on the key challenges in making this vision a reality.

It is well known that the process parameters chosen in metal additive manufacturing (AM) are directly related to the melt pool dimensions, which can be related to microstructure characteristics, properties, and printability. Thus, the determination of melt pool dimensions and resulting printability, given a set of process parameters, is crucial to understanding the performance of AM parts. Unfortunately, experiments relating process parameters to the melt pool have a high execution and data collection cost. On the other hand, simulations do not suffer this cost penalty but can never be as good as the ground truth experiments. A number of capabilities exist to predict melt pool geometry from process parameters that range from highly accurate codes with long simulation times to less accurate analytical solutions that are nearly instantaneous. This work leverages multifidelity Gaussian process (GP) surrogates to examine how the fusion of information from different fidelities, including experiments, influences the resulting predictive model. Both a multifidelity GP regression and a novel multifidelity GP classification are examined. The multifidelity models are compared to standard GPs and a withheld set of test data. The results show good predictive accuracy for the melt pool width and depth, but suggest multifidelity models may not significantly improve classification of printability.

This study intends to optimize the electromagnetic interference (EMI) shielding effectiveness of multi-walled carbon nanotubes/polyvinyl alcohol (MWCNTs/PVA) composite nanofibers based on multiple nonlinear neuro-regression analysis. The effect of MWCNTs content, thickness, and frequency on the shielding properties of MWCNTs/PVA nanofibers has been investigated in the analysis. The parameters were determined from a literature study focused on the designing and production of MWCNTs/PVA nanofibers as an EMI shielding composite material. Within the scope of the present study, 12 unique candidate models for each response were generated to test the accuracy of the predictions. Determination coefficients and boundedness of models have been checked to acquire the most realistic ones. Nelder-Mead, Simulated Annealing, Differential Evolution, and Random Search algorithms were used to attain the optimum parameters for shielding effectiveness of the composite nanofiber structure by maximizing the absorption while maintaining the minimum reflection. The present paper demonstrates that the proposed methodology can significantly contribute to the experimental studies regarding the materials design by predicting the process reliably.

The quality of powder processed for manufacturing can be certified by hundreds of different variables. Assessing the impact of all these different metrics on the performance of additively manufactured engineered products is an invaluable, but time intensive specification process. In this work, a comprehensive, generalizable, data-driven framework was implemented to select the optimal powder processing and microstructure variables that are required to predict the target property variables. The framework was demonstrated on a high-dimensional dataset collected from selective laser melted, additively manufactured, Inconel 718. One hundred and twenty-nine powder quality variables including particle morphology, rheology, chemical composition, and build composition were assessed for their impact on eight microstructural features and sixteen mechanical properties. The importance of each powder and microstructure variable was determined by using statistical analysis and machine learning models. The trained models predicted target mechanical properties with an R2 value of 0.9 or higher. The results indicate that the desired mechanical properties can be achieved by controlling only a few critical powder properties and without the need for collecting microstructure data. This framework significantly reduces the time and cost of qualifying source materials for production by determining an optimal subset of experiments needed to predict that a given source material will lead to a desired outcome. This general framework can be easily applied to other material systems.

In this paper, phenomenological relationships are presented that permit the prediction of the plastic regime of stress–strain curves using a limited number of parameters. These relationships were obtained from both conventional (wrought + β annealed) and additively manufactured (i.e., “3D printed”) Ti-6Al-4V. Three different methods of additive manufacturing have been exploited to produce the materials, including large-volume electron beam additive manufacturing, large-volume laser hot wire additive manufacturing, and small-volume selective laser melting. The general fundamental expressions are independent not only of the additive manufacturing process, but also of a wide variety of post-deposition heat treatments, however the coefficients are specific to material states. Thus, this work demonstrates that it is possible to predict not only the ultimate tensile strength, but also the full true stress, true strain curves, if certain parameters of the material are known. In general, the prediction of ultimate tensile strength are within 5% of the experimentally measured values across all additive manufacturing variants and subsequent heat treatments. The absolute values of ultimate tensile strength range from ~ 910 MPa to ~ 1170 MPa for the single alloy Ti-6Al-4V. Data representing 113 explicit samples are included in this work.