Abstract:The development of novel materials in recent years has been accelerated greatly by the use of computational modelling techniques aimed at elucidating the complex physics controlling microstructure formation in materials, the properties of which control material function. One such technique is the phase field method, a field theoretic approach that couples various thermophysical fields to microscopic order parameter fields that track the phases of microstructure. Phase field models are framed as multiple, non-linear, partial differential equations, which are extremely challenging to compute efficiently. Recent years have seen an explosion of computational algorithms aimed at enhancing the efficiency of phase field simulations. One such technique, adaptive mesh refinement (AMR), dynamically adapts numerical meshes to be highly refined around steep spatial gradients of the PDE fields and coarser where the fields are smooth. This reduces the number of computations per time step significantly, thus reducing the total time of computation. What AMR doesn't do is allow for adaptive time stepping. This work combines AMR with a neural network algorithm that uses a U-Net with a Convolutional Long-Short Term Memory (CLSTM) base to accelerate phase field simulations. Our neural network algorithm is described in detail and tested in on simulations of directional solidification of a dilute binary alloy, a paradigm that is highly practical for its relevance to the solidification of alloys.

What problem does this paper attempt to address?

### Problems Addressed by the Paper This paper aims to address the computational efficiency issues in multi-scale material simulations. Specifically, the authors propose an algorithm named "LeapFrog," which combines Adaptive Mesh Refinement (AMR) technology with a machine learning method based on Convolutional Long Short-Term Memory networks (CLSTM) to accelerate Phase Field (PF) simulations. ### Background and Challenges 1. **Complexity of Phase Field Simulations**: - The phase field method describes the evolution of material microstructures by coupling multiple nonlinear partial differential equations, which describe the changes in thermophysical and microstructural order parameter fields. - Due to the need to handle high-resolution arrays across multiple length scales, phase field simulations are extremely time-consuming at each time step. 2. **Limitations of Existing Solutions**: - Adaptive Mesh Refinement (AMR) technology significantly reduces the computational load per time step by dynamically adjusting the resolution of the numerical grid, thereby shortening the total computation time. - However, AMR cannot achieve adaptive time stepping, which limits its ability to further improve computational efficiency. - Although hardware advancements (such as the integration of GPUs) have significantly improved computational performance, most experiment-related simulations still require weeks to complete. ### Solution 1. **LeapFrog Algorithm**: - Combines AMR and CLSTM neural networks to achieve adaptive time stepping by predicting phase field evolution through machine learning. - Uses a U-Net architecture CLSTM network to effectively predict changes in phase and concentration fields over multiple time steps. 2. **Implementation Details**: - **Data Generation**: Uses AMR to generate initial phase field simulation data. - **Neural Network Training**: Trains the CLSTM network using the generated data to predict future states of the phase and concentration fields. - **Alternating Execution**: During the simulation, alternates between direct phase field simulation and neural network prediction, forming a "LeapFrog" pattern. After each neural network prediction, the results are fed back into the phase field code for correction and reintroduction of explicit thermal fluctuations. ### Objectives - **Improve Computational Efficiency**: Make phase field simulations more efficient by reducing computation time and resource consumption. - **Maintain Simulation Accuracy**: Ensure the accuracy of neural network predictions so that simulation results are consistent with traditional phase field simulation results. ### Experimental Validation - **Acceleration Effect**: Experimental results show that using the LeapFrog algorithm can significantly reduce computation time, especially in long-term simulations. - **Results Comparison**: Compared to traditional AMR phase field simulations, the LeapFrog algorithm significantly reduces computation time while maintaining similar accuracy. ### Application Prospects - **Multi-Scale Material Design**: This algorithm can be applied to the design and optimization of various materials, particularly in complex multi-scale simulations in fields such as alloy solidification and additive manufacturing. - **Further Expansion**: Future work can further optimize the algorithm to improve its applicability to larger systems and longer time scales. By combining AMR and machine learning, the LeapFrog algorithm provides a new, efficient solution for multi-scale material simulations.

LeapFrog: Getting the Jump on Multi-Scale Materials Simulations Using Machine Learning

Accelerating phase-field-based microstructure evolution predictions via surrogate models trained by machine learning methods

Accelerating phase-field simulation of three-dimensional microstructure evolution in laser powder bed fusion with composable machine learning predictions

Numerical simulation of dendritic growth during solidification process using multiphase-field model aided with machine learning method

Rethinking materials simulations: Blending direct numerical simulations with neural operators

Time series forecasting of multiphase microstructure evolution using deep learning

Accelerating microstructure modelling via machine learning: a new method combining Autoencoder and ConvLSTM

Quasi-Phase Equilibrium Prediction of Multi-Element Alloys Based on Machine Learning and Deep Learning

Accelerated Solutions of Coupled Phase-Field Problems using Generative Adversarial Networks

Spatiotemporal Evolution of Grain Microstructure: A CNN Perspective

Accelerate Microstructure Evolution Simulation Using Graph Neural Networks with Adaptive Spatiotemporal Resolution

Accelerating three-dimensional phase-field simulations via deep learning approaches

Towards High-Throughput Microstructure Simulation in Compositionally Complex Alloys Via Machine Learning

Accelerating Phase Field Simulations Through a Hybrid Adaptive Fourier Neural Operator with U-Net Backbone

A Machine Learning Method of Accelerating Multiscale Analysis for Spatially Varying Microstructures

Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing

A computational building block approach towards multiscale architected materials analysis and design with application to hierarchical metal metamaterials

Rapid and Automated Alloy Design with Graph Neural Network-Powered LLM-Driven Multi-Agent Systems

Learning two-phase microstructure evolution using neural operators and autoencoder architectures

Neural Network Accelerated Process Design of Polycrystalline Microstructures

A Microstructure-based Graph Neural Network for Accelerating Multiscale Simulations