Search

This study introduces the calculation of spatially-resolved chord length distribution (SR-CLD) as an efficient approach for quantifying and visualizing non-uniform microstructures in heterogeneous materials. SR-CLD enables detailed analysis of spatial variation of microstructure constituent sizes in different directions that can be overlooked with traditional descriptions. We present the calculation of SR-CLD using efficient scan-line algorithm that counts pixels in constituents along pixel rows or columns of microstructure images for detailed, high-resolution SR-CLD maps. We demonstrate the application of SR-CLD in two case studies: one on synthetic polycrystalline microstructures with known and intentionally created uniform and gradient spatial distributions of grain size; and one on experimental images of two-phase microstructures of additively manufactured Ti alloys with significant spatially non-uniform distributions of laths of one of the phases. Additionally, we present how SR-CLDs can enable automated, computationally efficient, and robust alignment of large sets of images for merging into accurate composite images of large microstructure areas., Comment: 11 pages, 5 figures

High-cycle fatigue is a critical performance metric of structural alloys for many applications. The high cost, time, and labor involved in experimental fatigue testing call for efficient and accurate computer models of fatigue life. We present FIP-GNN -- a graph neural network for polycrystals that (i) predicts fatigue indicator parameters as grain-level inelastic responses to cyclic loading quantifying the local driving force for crack initiation and (ii) generalizes these predictions to large microstructure volume elements with grain populations well beyond those used in training. These advances can make significant contributions to statistically rigorous and computationally efficient modeling of high-cycle fatigue -- a long-standing challenge in the field.

We present AnisoGNNs -- graph neural networks (GNNs) that generalize predictions of anisotropic properties of polycrystals in arbitrary testing directions without the need in excessive training data. To this end, we develop GNNs with a physics-inspired combination of node attributes and aggregation function. We demonstrate the excellent generalization capabilities of AnisoGNNs in predicting anisotropic elastic and inelastic properties of two alloys., Comment: 25 pages, 7 figures

Microstructure--property relationships are key to effective design of structural materials for advanced applications. Advances in computational methods enabled modeling microstructure-sensitive properties using 3D models (e.g., finite elements) based on microstructure representative volumes. 3D microstructure data required as input to these models are typically obtained from either 3D characterization experiments or digital reconstruction based on statistics from 2D microstructure images. In this work, we present machine learning (ML) approaches to modeling effective properties of heterogeneous materials directly from 2D microstructure sections. To this end, we consider statistical learning models based on spatial correlations and convolutional neural networks as two distinct ML strategies. In both strategies, models are trained on a dataset of synthetically generated 3D microstructures and their properties obtained from micromechanical 3D simulations. Upon training, the models predict properties from 2D microstructure sections. The advantage of the presented models is that they only need 2D sections, whose experimental acquisition is more accessible compared to 3D characterization. Furthermore, the present models do not require digital reconstruction of 3D microstructures.

In this contribution, we present a new Materials Knowledge System framework for microstructure-sensitive predictions of effective stress--strain responses in composite materials. The model is developed for composites with a wide range of combinations of strain hardening laws and topologies of the constituents. The theoretical foundation of the model is inspired by statistical continuum theories, leveraged by mean-field approximation of self-consistent models, and calibrated to data obtained from micromechanical finite element simulations. The model also relies on newly formulated data-driven linkages between micromechanical responses (phase-average strain rates and effective strength) and microstructure as well as strength contrast of the constituents. The paper describes in detail the theoretical development of the model, its implementation into an efficient computational plasticity framework, calibration of the linkages, and demonstration of the model predictions on two-phase composites with isotropic constituents exhibiting linear and power-law strain hardening laws. It is shown that the model reproduces finite element results reasonably well with significant savings of the computational cost., Comment: 39 pages, 5 figures