Detection of the onset of yielding and creep failure from digital image correlation
arXiv:2211.04794 · doi:10.1103/PhysRevMaterials.6.103601
Abstract
There are a multitude of applications in which structural materials would be desired to be nondestructively evaluated, while in a component, for plasticity and failure characteristics. In this way, safety and resilience features can be significantly improved. Nevertheless, while failure can be visible through cracks, plasticity is commonly invisible and highly microstructure-dependent. Here, we show that an equation-free method based on principal component analysis is capable of detecting yielding and tertiary creep onset, directly from strain fields that are obtained by digital image correlation, applicable on components, continuously and nondestructively. We demonstrate the applicability of the method to yielding of Ni-based Haynes 230 metal alloy polycrystalline samples, which are also characterized through electron microscopy and benchmarked using continuum polycrystalline plasticity modeling. Also, we successfully apply this method to yielding during uniaxial tension of Hastelloy X polycrystalline samples, and also to the onset of tertiary creep in quasibrittle fiber composites under uniaxial tension. We conclude that there are key features in the spatiotemporal fluctuations of local strain fields that can be used to infer mechanical properties.
References in corpus (10)
- Discovering Phase Transitions with Unsupervised Learning
- Unsupervised learning of phase transitions: from principal component analysis to variational autoencoders
- Discovering Phases, Phase Transitions and Crossovers through Unsupervised Machine Learning: A critical examination
- Unsupervised Learning of Frustrated Classical Spin Models I: Principle Component Analysis
- Creep failures in heterogeneous materials
- Creep and Fracture of a Protein Gel under Stress
- Visualizing probabilistic models: Intensive Principal Component Analysis
- Probing the transition from dislocation jamming to pinning by machine learning
- Direct detection of plasticity onset through total-strain profile evolution
- Correlations in the shear flow of athermal amorphous solids: A principal component analysis