Fracture-Bench offers benchmark for machine learning on fracture simulations
Fracture-Bench: A Controlled Benchmark Suite for Machine Learning Surrogates in Phase-Field Fracture Mechanics
Computational Engineering, Finance, and Science
Summary
Simulating how cracks grow in materials is complex and slow, so some try using machine learning to speed it up. The authors created Fracture-Bench, a set of two detailed datasets to test different machine learning methods on fracture problems. They found that the best-performing methods vary by problem type, and most errors happen when cracks start to form sharply. Their benchmark helps show when and how well different AI tools can predict crack growth.
What this means in practice
- •For materials engineers: Evaluate AI models for predicting crack propagation in materials using standardized fracture simulation datasets to improve design accuracy.
- •For software developers in simulation: Test and compare machine learning architectures for fracture prediction to choose the best approach for fast surrogate modeling in engineering software.
Authors
Meenu Krishnan U., Maryam Hakimzadeh, Indrashish Saha, Meer Mehran Rashid, Lori Graham-Brady, Somdatta Goswami
Abstract
Machine-learning surrogates are increasingly proposed to replace phase-field fracture simulations in many-query tasks such as uncertainty quantification, inverse analysis and design. Little controlled evidence shows which architectures are reliable, for which fracture problems and how far beyond their training data. We introduce Fracture-Bench, an open benchmark of two phase-field datasets with fixed splits, common test metrics and a description of each dataset's physical regime. Dataset~1 contains 1,000 AT2 simulations of multiple cracks growing from a notched hole in a hyperelastic plate under random non-uniform biaxial boundary displacements (44 load steps, $128\times128$ grid). Dataset~2 contains 4,000 PF-CZM simulations of a pre-cracked plate with two inclusions, each ten times stiffer or more compliant than the matrix ($128\times256$ grid). We train a Fourier neural operator (FNO), DeepONet, Latent DeepONet, a UNet and a conditional diffusion model on identical splits and compare one test metric per dataset, training time, inference time and model size. On Dataset~1, FNO and Latent DeepONet attain mean test errors of order $10^{-6}$, about ten times lower than UNet and DeepONet. On Dataset~2, the diffusion model is most accurate, with about one sixth of the UNet error, but requires \SI{706}{s} per prediction versus \SI{6}{ms} for UNet. For all architectures, most of the error occurs in the final load steps, after crack localization. Dataset~2 contains initial cracks less than one grid cell apart that produce qualitatively different paths. Architecture rankings therefore depend on the task, and errors averaged over densely sampled inputs reveal little about near-bifurcation cases. Data, codes, splits, regime descriptors and generalization splits are public, with a protocol for adding under-represented regimes such as mixed-mode, three-dimensional and experimentally paired fracture.