Crystal plasticity simulations with representative volume element of as-build laser powder bed fusion materials.

Additive manufacturing of as-build metal materials with laser powder bed fusion typically leads to the formations of various chemical phases and their corresponding microstructure types. Such microstructures have very complex shape and size anisotropic distributions due to the history of the laser heat gradients and scanning patterns. With higher complexity compared to the post-heat-treated materials, the synthetic volume reconstruction of as-build materials for accurate modelling of their mechanical properties is a serious challenge. Here, we present an example of complete workflow pipeline for such nontrivial task. It takes into account the statistical distributions of microstructures: object sizes for each phase, several shape parameters for each microstructure type, and their morphological and crystallographic orientations. In principle, each step in the pipeline, including the parameters in the crystal plasticity model, can be fine-tuned to achieve suitable correspondence between experimental and synthetic microstructures as well as between experimental stress-strain curves and simulated results. To our best knowledge, this work represents an example of the most challenging synthetic volume reconstruction for as-build additive manufacturing materials to date.

Artificial intelligence inspired design of non-isothermal aging for γ-γ' two-phase, Ni-Al alloys.

In this paper, a state-of-the-art Artificial Intelligence (AI) technique is used for a precipitation hardening of Ni-based alloy to predict more flexible non-isothermal aging (NIA) and to examine the possible routes for the enhancement in strength that may be practically achieved. Additionally, AI is used to integrate with Materials Integration by Network Technology, which is a computational workflow utilized to model the microstructure evolution and evaluate the 0.2% proof stress for isothermal aging and NIA. As a result, it is possible to find enhanced 0.2% proof stress for NIA for a fixed time of 10 min compared to the isothermal aging benchmark. The entire search space for aging scheduling was ~ 3 billion. Out of 1620 NIA schedules, we succeeded in designing the 110 NIA schedules that outperformed the isothermal aging benchmark. Interestingly, it is found that early-stage high-temperature aging for a shorter time increases the γ' precipitate size up to the critical size and later aging at lower temperature increases the γ' fraction with no anomalous change in γ' size. Therefore, employing this essence from AI, we designed an optimum aging route in which we attained an outperformed 0.2% proof stress to AI-designed NIA routes.

Generation of highly realistic microstructural images of alloys from limited data with a style-based generative adversarial network.

In materials science, the amount of observational data is often limited by operating protocols that require a high level of expertise, often machine-dependent, developed for a time-consuming integration of valuable data. Scanning electron microscopy (SEM) is one of those methodologies of characterisation for which the number of observations of a given material is limited to just a few images. In the present study, we present the possibility to artificially inflate the size of SEM image datasets from a limited ([Formula: see text] of images) to a virtually unbounded number thanks to a generative adversarial network (GAN). For this purpose, we use one of the latest developments in GAN architectures and training methodologies, the StyleGAN2 with adaptive discriminator augmentation (ADA), to generate a diversity of high-quality SEM images of [Formula: see text] pixels. Overall, coarse and fine microstructural details are successfully reproduced when training a StyleGAN2 with ADA from scratch on at most 3000 SEM images, and interpolations between microstructures are performed without significant modifications to the training protocol when applied to natural images.

A universal Bayesian inference framework for complicated creep constitutive equations.

Evaluating the creep deformation process of heat-resistant steels is important for improving the energy efficiency of power plants by increasing the operating temperature. There is an analysis framework that estimates the rupture time of this process by regressing the strain-time relationship of the creep process using a regression model called the creep constitutive equation. Because many creep constitutive equations have been proposed, it is important to construct a framework to determine which one is best for the creep processes of different steel types at various temperatures and stresses. A Bayesian model selection framework is one of the best frameworks for evaluating the constitutive equations. In previous studies, approximate-expression methods such as the Laplace approximation were used to develop the Bayesian model selection frameworks for creep. Such frameworks are not applicable to creep constitutive equations or data that violate the assumption of the approximation. In this study, we propose a universal Bayesian model selection framework for creep that is applicable to the evaluation of various types of creep constitutive equations. Using the replica exchange Monte Carlo method, we develop a Bayesian model selection framework for creep without an approximate-expression method. To assess the effectiveness of the proposed framework, we applied it to the evaluation of a creep constitutive equation called the Kimura model, which is difficult to evaluate by existing frameworks. Through a model evaluation using the creep measurement data of Grade 91 steel, we confirmed that our proposed framework gives a more reasonable evaluation of the Kimura model than existing frameworks. Investigating the posterior distribution obtained by the proposed framework, we also found a model candidate that could improve the Kimura model.

Data-based selection of creep constitutive models for high-Cr heat-resistant steel.

There are two types of creep constitutive equation, one with a steady-state term (steady-state type) and the other with no steady-state term (non-steady-state type). We applied the Bayesian inference framework in order to examine which type is supported by experimental creep curves for a Grade 91 (Gr.91) steel. The Bayesian free energy was significantly lower for the steady-state type under all the test conditions in the ranges of 50-90 MPa at 923 K, 90-160 MPa at 873 K and 170-240 MPa at 823 K, leading to the conclusion that the posterior probability was virtually 1.0. These findings mean that the experimental data supported the steady-state-type equation. The dependence of the evaluated steady-state creep rate on the applied stress indicates that there is a transition in the mechanism governing creep deformation around 120 MPa.

Automatic steel labeling on certain microstructural constituents with image processing and machine learning tools.

It is demonstrated that optical microscopy images of steel materials could be effectively categorized into classes on preset ferrite/pearlite-, ferrite/pearlite/bainite-, and bainite/martensite-type microstructures with image pre-processing and statistical analysis including the machine learning techniques. Though several popular classifiers were able to get the reasonable class-labeling accuracy, the random forest was virtually the best choice in terms of overall performance and usability. The present categorizing classifier could assist in choosing the appropriate pattern recognition method from our library for various steel microstructures, which we have recently reported. That is, the combination of the categorizing and pattern-recognizing methods provides a total solution for automatic quantification of a wide range of steel microstructures.

Pattern recognition with machine learning on optical microscopy images of typical metallurgical microstructures.

For advanced materials characterization, a novel and extremely effective approach of pattern recognition in optical microscopic images of steels is demonstrated. It is based on fast Random Forest statistical algorithm of machine learning for reliable and automated segmentation of typical steel microstructures. Their percentage and location areas excellently agreed between machine learning and manual examination results. The accurate microstructure pattern recognition/segmentation technique in combination with other suitable mathematical methods of image processing and analysis can help to handle the large volumes of image data in a short time for quality control and for the quest of new steels with desirable properties.