Towards scientific machine learning for granular material simulations -- challenges and opportunities

TL;DR

This position paper identifies seven core challenges in modeling granular materials across micro, meso, and macro scales and across solid-like, fluid-like, and transitional regimes. It surveys state-of-the-art ML approaches—including LSTM-based surrogates, graph neural networks, neural operators, and reduced-order models—and proposes a GranML workflow to unify data structures, modelling pipelines, and deployment for granular simulations. The authors illustrate the workflow with two demonstrations: quasi-static solid-like compression/shear using LSTM-DEM and fluid-like granular column collapse using GNS on MPM data, while also outlining open infrastructure and best practices for data management, software, and training. The work emphasizes hybrid physics-driven and data-driven strategies, uncertainty quantification, interpretability, and open-science infrastructure to enable robust digital twins for industrial granular systems. Overall, the paper outlines a practical, collaborative roadmap for leveraging ML to accelerate granular material simulations and to develop trustworthy, scalable digital twins across diverse regimes and applications.

Abstract

Micro-scale mechanisms, such as inter-particle and particle-fluid interactions, govern the behaviour of granular systems. While particle-scale simulations provide detailed insights into these interactions, their computational cost is often prohibitive. Attended by researchers from both the granular materials (GM) and machine learning (ML) communities, a recent Lorentz Center Workshop on "Machine Learning for Discrete Granular Media" brought the ML community up to date with GM challenges. This position paper emerged from the workshop discussions. We define granular materials and identify seven key challenges that characterise their distinctive behaviour across various scales and regimes, ranging from gas-like to fluid-like and solid-like. Addressing these challenges is essential for developing robust and efficient digital twins for granular systems in various industrial applications. To showcase the potential of ML to the GM community, we present classical and emerging machine/deep learning techniques that have been, or could be, applied to granular materials. We reviewed sequence-based learning models for path-dependent constitutive behaviour, followed by encoder-decoder type models for representing high-dimensional data. We then explore graph neural networks and recent advances in neural operator learning. Lastly, we discuss model-order reduction and probabilistic learning techniques for high-dimensional parameterised systems, which are crucial for quantifying uncertainties arising from physics-based and data-driven models. We present a workflow aimed at unifying data structures and modelling pipelines and guiding readers through the selection, training, and deployment of ML surrogates for granular material simulations. Finally, we illustrate the workflow's practical use with two representative examples, focusing on granular materials in solid-like and fluid-like regimes.

Figures (18)

• Figure 1: Examples of laboratory devices for granular materials in (a) triaxial compression and (b) continuous ring-shear conditions. Copyright: https://www.dietmar-schulze.de/ringschergeraete_e.html.
• Figure 2: Examples of granular materials of various sizes and shapes.
• Figure 3: Schematic of (a) coarse-grained particles and (b) original particles. Reprinted with permission from Nakamura2020.
• Figure 4: Micro-macro relationship between a representative volume element (DEM) and a phenomenological representation (constitutive law) of granular materials. Figure adapted from Zhao2013.
• Figure 5: Multi-scale modelling for granular materials: (a) concurrent and (b) hierarchical methods. Reprinted with permission from Cheng2023 and Guo2014.