Browsing by Author "Fan, Shaoxun"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Accelerate microstructure evolution simulation using graph neural networks with adaptive spatiotemporal resolution
    (IOP Publishing, 2024) Fan, Shaoxun; Hitt, Andrew L.; Tang, Ming; Sadigh, Babak; Zhou, Fei
    Surrogate models driven by sizeable datasets and scientific machine-learning methods have emerged as an attractive microstructure simulation tool with the potential to deliver predictive microstructure evolution dynamics with huge savings in computational costs. Taking 2D and 3D grain growth simulations as an example, we present a completely overhauled computational framework based on graph neural networks with not only excellent agreement to both the ground truth phase-field methods and theoretical predictions, but enhanced accuracy and efficiency compared to previous works based on convolutional neural networks. These improvements can be attributed to the graph representation, both improved predictive power and a more flexible data structure amenable to adaptive mesh refinement. As the simulated microstructures coarsen, our method can adaptively adopt remeshed grids and larger timesteps to achieve further speedup. The data-to-model pipeline with training procedures together with the source codes are provided.
  • Thumbnail Image
    Item
    Self-supervised learning and prediction of microstructure evolution with convolutional recurrent neural networks
    (Elsevier, 2021) Yang, Kaiqi; Cao, Yifan; Zhang, Youtian; Fan, Shaoxun; Tang, Ming; Aberg, Daniel; Sadigh, Babak; Zhou, Fei
    Microstructural evolution is a key aspect of understanding and exploiting the processing-structure-property relationship of materials. Modeling microstructure evolution usually relies on coarse-grained simulations with evolution principles described by partial differential equations (PDEs). Here we demonstrate that convolutional recurrent neural networks can learn the underlying physical rules and replace PDE-based simulations in the prediction of microstructure phenomena. Neural nets are trained by self-supervised learning with image sequences from simulations of several common processes, including plane-wave propagation, grain growth, spinodal decomposition, and dendritic crystal growth. The trained networks can accurately predict both short-term local dynamics and long-term statistical properties of microstructures assessed herein and are capable of extrapolating beyond the training datasets in spatiotemporal domains and configurational and parametric spaces. Such a data-driven approach offers significant advantages over PDE-based simulations in time-stepping efficiency and offers a useful alternative, especially when the material parameters or governing PDEs are not well determined.
  • Thumbnail Image
    ItemEmbargo
    Towards More Powerful Materials Microstructure Simulation: Extension to Soft Matter Systems and Acceleration by Machine Learning
    (2024-07-16) Fan, Shaoxun; Tang, Ming
    This thesis explores two significant advancements in modern materials science: simulation techniques and the application of machine learning. Simulations enable scientists to not only predict the properties of new materials but also gain insights into the behavior and underlying mechanisms of existing ones. Machine learning can further enhance materials research in numerous ways. In this thesis, we first present our phase-field simulation of micelles and emulsion, followed by the application of deep learning models to accelerate simulations and analyze simulation results. A comprehensive understanding of the morphological transformations of micelles and emulsions can benefit various fields. However, direct and especially in-situ observation of these systems faces certain limitations. In the first topic of this thesis, we investigate the formation and transformation mechanisms of micelles and emulsions at mesoscale using phase-field simulation. The two-stage growth of micelle cluster during fission was discovered. In the first stage, the volume of micelles grows exponentially while in the second stage, the cluster growth is limited by long-range transport of surfactant monomers. A quantitative model explaining the observed cluster growth kinetics was then established. The phase-field model was extended to water-oil-surfactant ternary system to enable the study of emulsion's morphological transition. Surrogate models driven by sizeable datasets and scientific machine-learning methods have emerged as an attractive microstructure simulation tool with the potential to deliver predictive microstructure evolution dynamics with huge savings in computational costs. In the second topic of the thesis, we present a computational framework based on graph neural networks with not only excellent agreement to both the ground truth phase-field methods and theoretical predictions, but enhanced accuracy and efficiency compared to previous works. As the simulated microstructures coarsen, our method can adaptively adopt remeshed grids and larger timesteps to achieve further speedup. In the third topic, we study the potential of recurrent convolutional neural network models in regressing parameters from spatial and temporal features in image sequences generated by simulations, showing machine learning's capability in automatically extracting information that are critical for understanding behaviors of various materials. It provides a novel data-driven approach for materials research and analysis.