Session: 13-11-01: Friction, Fracture, and Damage I

Paper Number: 173396

Adaptive Phase-Field Modeling of Fracture Across Natural and Engineered Systems 

Fracture plays a central role in the failure of both natural and engineered systems. In materials such as concrete, batteries, and architected metamaterials, crack initiation and growth are driven by heterogeneity at the microscale. In natural settings like glaciers, fracture extends over kilometer-scale domains, governed by large-scale stress fields and boundary effects. Modeling such diverse systems within a single framework is challenging, especially when high resolution is required to resolve crack paths accurately.

The phase-field method has emerged as a robust approach to fracture modeling, capable of simulating complex crack topologies without the need for tracking discontinuities or remeshing. However, this robustness comes with a high computational cost. Resolving sharp damage gradients typically requires dense uniform meshes, even in regions where the solution is otherwise smooth. This limitation restricts the method’s scalability to large three-dimensional domains or systems that demand microscale resolution.

This work introduces an adaptive phase-field modeling framework that addresses this computational bottleneck. The key idea is to use mesh refinement strategies that concentrate degrees of freedom only where and when needed, around evolving cracks and damage zones, while keeping the mesh coarse elsewhere. This localized refinement allows the simulation of large-scale problems with millions of degrees of freedom without compromising accuracy.

The framework is first validated using the Sandia–Purdue–LLNL damage mechanics challenge, an experimental benchmark for quasi-brittle fracture. Results show that the adaptive formulation accurately reproduces crack paths, energy dissipation, and load-displacement behavior, while achieving substantial reductions in computational cost compared to uniform mesh approaches.

Subsequent case studies demonstrate the method’s broad applicability. In concrete, the model captures fine-scale fracture features such as aggregate–matrix interactions and crack deflection due to material heterogeneity. In engineered metamaterials and additively manufactured components, the method resolves stress concentrations arising from geometric intricacies, allowing for high-resolution predictions without excessive global mesh refinement.

At the large scale, the framework is applied to model glacier fracture driven by self-weight. The simulations capture stress redistribution and localized crack evolution over kilometer-scale domains, demonstrating the scalability of the approach to geophysical applications. Finally, the framework is extended to battery materials, where it simulates fracture patterns influenced by coupled electrochemical and mechanical fields.

Overall, this work demonstrates that adaptive phase-field modeling offers a practical and scalable solution for high-fidelity fracture simulation across scales. By overcoming the computational limitations of traditional phase-field approaches, this framework enables predictive modeling of failure in a wide range of systems, from microstructural material components to large-scale natural domains, making it a valuable tool for both engineering design and scientific inquiry.

Presenting Author: Abhinav Gupta Vanderbilt University

Presenting Author Biography: Abhinav Gupta is a Research Assistant Professor at Vanderbilt University. He earned his Ph.D. in computational mechanics with a focus on high-performance finite element methods. His research covers adaptive mesh refinement, phase-field fracture modeling, topology optimization, and scalable methods for fracture simulation. His recent work on metamaterials and phase-field methods has been widely cited in the computational mechanics community.

Authors:

Abhinav Gupta Vanderbilt University
Ravindra Duddu Vanderbilt University