skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.


Title: Deep learning for quantitative dynamic fragmentation analysis
We have developed an image-based convolutional neural network that is applicable for quantitative time-resolved measurements of the fragmentation behavior of opaque brittle materials using ultra-high speed optical imaging. This model extends previous work on the U-net model. Here we trained binary-, three-, and five-class models using supervised learning on experimentally measured dynamic fracture experiments on various opaque structural ceramic materials that were adhered on transparent polymer (polycarbonate or acrylic) backing materials. Full details of the experimental investigations are outside the scope of this manuscript, but briefly, several different ceramics were loaded using spatially and time-varying mechanical loads to induce inelastic deformation and fracture processes that were recorded at frequencies as high as 5 MHz using high-speed optical imaging. These experiments provided a rich and diverse dataset that includes many of the common fracture modes found in static and dynamic fractures, including cone cracking, median cracking, comminution, and combined complex failure modes that involve effectively simultaneous activation and propagation of multiple fragmentation modes. While the training data presented here were obtained from dynamic fragmentation experiments, this study is applicable to static loading of these materials as the crack speeds are on the order of 1–10 km/s regardless of the loading rate. We believe the methodologies presented here will be useful in quantifying the failure processes in structural materials for protection applications and can be used for direct validation of engineering models used in design.  more » « less
Award ID(s):
2226415 2226414 2226416
PAR ID:
10566732
Author(s) / Creator(s):
;
Publisher / Repository:
American Institute of Physics
Date Published:
Journal Name:
APL Machine Learning
Volume:
3
Issue:
1
ISSN:
2770-9019
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    Abstract. The continuum of behavior that emerges during fracturenetwork development in crystalline rock may be categorized into threeend-member modes: fracture nucleation, isolated fracture propagation, andfracture coalescence. These different modes of fracture growth producefracture networks with distinctive geometric attributes, such as clusteringand connectivity, that exert important controls on permeability and theextent of fluid–rock interactions. To track how these modes of fracturedevelopment vary in dominance throughout loading toward failure and thushow the geometric attributes of fracture networks may vary under theseconditions, we perform in situ X-ray tomography triaxial compressionexperiments on low-porosity crystalline rock (monzonite) under upper-crustalstress conditions. To examine the influence of pore fluid on the varyingdominance of the three modes of growth, we perform two experiments undernominally dry conditions and one under water-saturated conditions with 5 MPa ofpore fluid pressure. We impose a confining pressure of 20–35 MPa and thenincrease the differential stress in steps until the rock failsmacroscopically. After each stress step of 1–5 MPa we acquire athree-dimensional (3D) X-ray adsorption coefficient field from which weextract the 3D fracture network. We develop a novel method of trackingindividual fractures between subsequent tomographic scans that identifieswhether fractures grow from the coalescence and linkage of several fracturesor from the propagation of a single fracture. Throughout loading in all ofthe experiments, the volume of preexisting fractures is larger than that ofnucleating fractures, indicating that the growth of preexisting fracturesdominates the nucleation of new fractures. Throughout loading until close tomacroscopic failure in all of the experiments, the volume of coalescingfractures is smaller than the volume of propagating fractures, indicatingthat fracture propagation dominates coalescence. Immediately precedingfailure, however, the volume of coalescing fractures is at least double thevolume of propagating fractures in the experiments performed at nominallydry conditions. In the water-saturated sample, in contrast, although thevolume of coalescing fractures increases during the stage preceding failure,the volume of propagating fractures remains dominant. The influence ofstress corrosion cracking associated with hydration reactions at fracturetips and/or dilatant hardening may explain the observed difference infracture development under dry and water-saturated conditions. 
    more » « less
  2. Owing to the opaque nature of the laminated structures, traditional high-speed optical camera cannot be used to detect the dynamic process of sub-surface deformation. In this article, we report a study of using high speed X-ray imaging to study the high strain rate deformation in laminated Al structures. We used a Kolsky bar apparatus to apply dynamic compression and a high-speed synchrotron X-ray phase contrast imaging (PCI) setup to conduct the in situ X-ray imaging study. The in situ X-ray imaging captures the shock wave propagation in the laminated structures. After shock compression, we characterized the microstructures by using transmission electron microscopy (TEM), which demonstrates an increase of dislocation density. The micro-pillar compression tests show that the yield strength at 0.2% offset of laminated Al-graphene composite has a significant increase of 67%, from 30 to 50 MPa, compared to laminate Al after shock loading. 
    more » « less
  3. While considerable progress has been made in simulating the overall seismic response of steel structures using nonlinear response history (dynamic) analysis, techniques to simulate fracture propagation under large scale inelastic cyclic loading are not as well developed. This is despite the fact that fracture is often a critical limit state that can precipitate structural failure and collapse. To address this, a new ductile damage-based cohesive zone model is presented. The proposed model is an extension of the established continuum-based local or micromechanical ductile fracture models for evaluating ultra-low cycle fatigue in structural steels. This model is implemented in the finite element program WARP3D, and evaluated against tests of notched bars that fail by ductile crack propagation. The preliminary results indicate that the model is an effective tool for predicting ductile fracture initiation and propagation in structural steels subjected to monotonic and cyclic large scale inelastic loading. Implications of this for characterizing the post-fracture response of structural steel components are discussed, along with limitations of the research. 
    more » « less
  4. While considerable progress has been made in simulating the overall seismic response of steel structures using nonlinear response history (dynamic) analysis, techniques to simulate fracture propagation under large scale inelastic cyclic loading are not as well developed. This is despite the fact that fracture is often a critical limit state that can precipitate structural failure and collapse. To address this, a new ductile damage-based cohesive zone model is presented. The proposed model is an extension of the established continuum-based local or micromechanical ductile fracture models for evaluating ultra-low cycle fatigue in structural steels. This model is implemented in the finite element program WARP3D, and evaluated against tests of notched bars that fail by ductile crack propagation. The preliminary results indicate that the model is an effective tool for predicting ductile fracture initiation and propagation in structural steels subjected to monotonic and cyclic large scale inelastic loading. Implications of this for characterizing the post-fracture response of structural steel components are discussed, along with limitations of the research. 
    more » « less
  5. While considerable progress has been made in simulating the overall seismic response of steel structures using nonlinear response history (dynamic) analysis, techniques to simulate fracture propagation under large scale inelastic cyclic loading are not as well developed. This is despite the fact that fracture is often a critical limit state that can precipitate structural failure and collapse. To address this, a new ductile damage-based cohesive zone model is presented. The proposed model is an extension of the established continuum-based local or micromechanical ductile fracture models for evaluating ultra-low cycle fatigue in structural steels. This model is implemented in the finite element program WARP3D, and evaluated against tests of notched bars that fail by ductile crack propagation. The preliminary results indicate that the model is an effective tool for predicting ductile fracture initiation and propagation in structural steels subjected to monotonic and cyclic large scale inelastic loading. Implications of this for characterizing the post-fracture response of structural steel components are discussed, along with limitations of the research. 
    more » « less