Toward quantitative fractography using convolutional neural networks

Tsopanidis, Stylianos; Moreno, Raúl; Osovski, S.

doi:10.1016/j.engfracmech.2020.106992

Cited by 40 publications

(20 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…By combining tools such as the one presented here, with existing experimental methodologies and other advanced analysis techniques, including quantitative fractography Tsopanidis et al [54] and surrogate models Alabort et al [19], the authors believe that the fracture mechanics community, and specifically researchers in the community aiming to correlate materials microstructure with fracture experiments, are facing an exciting and promising future.…”

Section: Discussionmentioning

confidence: 99%

A graph based workflow for extracting grain-scale toughness from meso-scale experiments

Tsopanidis¹,

Osovski²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

We introduce a novel machine learning computational framework that aims to compute the material toughness, after subjected to a short training process on a limited meso-scale experimental dataset. The three part computational framework relies on the ability of a graph neural network to perform high accuracy predictions of the micro-scale material toughness, utilizing a limited size dataset that can be obtained from meso-scale fracture experiments. We analyze the functionality of the different components of the framework, but the focus is on the capabilities of the neural network. The minimum size of the dataset required for the network training is investigated. The results demonstrate the high efficiency of the algorithm in predicting the crack growth resistance in micro-scale level, using a crack path trajectory limited to 200-300 grains for the network training. The merit of the proposed framework arises from the capacity to enhance its performance in different material systems with a limited additional training on data obtained from experiments that do not require complex or cumbersome measurements. The main objective is the development of an efficient computational tool that enables the study of a wide range of material microstructure properties and the investigation of their influence on the material toughness.

show abstract

Section: Discussionmentioning

confidence: 99%

A graph based workflow for extracting grain-scale toughness from meso-scale experiments

Tsopanidis¹,

Osovski²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…To compensate for the limited data, we used the TL method, which allowed us to construct high-performance prediction models even with a small dataset by transferring machine learning models trained using other large datasets. TL can be used to identify fracture surfaces using DNNs [8][9][10][11][12]. Thus, it is expected that by using a DNN and TL, a framework for directly estimating K Ic can be established.…”

Section: Introductionmentioning

confidence: 99%

Quantitative Prediction of Fracture Toughness $(K_{{\rm I}c})$ of Polymer by Fractography Using Deep Neural Networks

Mototake¹,

Ito²,

Demura³

2022

Preprint

View full text Add to dashboard Cite

Fracture surfaces provide various types of information about fracture. The fracture toughness KIc, which represents the resistance to fracture, can be estimated using the three-dimensional (3D) information of a fracture surface, i.e., its roughness. However, this is time-consuming and expensive to obtain the 3D information of a fracture surface; thus, it is desirable to estimate KIc from a two-dimensional (2D) image, which can be easily obtained. In recent years, methods of estimating a 3D structure from its 2D image using deep learning have been rapidly developed. In this study, we propose a framework for fractography that directly estimates KIc from a 2D fracture surface image using deep neural networks (DNNs). Typically, image recognition using a DNN requires a tremendous amount of image data, which is difficult to acquire for fractography owing to the high experimental cost. To compensate for the limited data, in this study, we used the transfer learning (TL) method, and constructed high-performance prediction models even with a small dataset by transferring machine learning models trained using other large datasets. We found that the regression model obtained using our proposed framework can predict KIc in the range of approximately 1-5 [MPa √ m] with a standard deviation of the estimation error of approximately ±0.37 [MPa √ m]. The present results demonstrate that the DNN trained with TL opens a new route for quantitative fractography by which parameters of fracture process can be estimated from a fracture surface even with a small dataset. The proposed framework also enables the building of regression models in a few hours. Therefore, our framework enables us to screen a large number of image datasets available in the field of materials science and find candidates that are worth expensive machine learning analysis.

show abstract

“…Fourier transform is widely used for studies of fatigue crack growth (FCG) in materials with interactive analysis of fatigue striation (FS) images [11]. For a quantitative assessment of the characteristic elements of fracture surfaces, fractographic images are analyzed using convolutional neural networks [12] and unsupervised machine learning [13]. At the same time, pattern recognition methods based on a modified support vector machine are being applied to fracture surface analysis [14].…”

Section: Introductionmentioning

confidence: 99%

Estimation of Fatigue Crack Growth Rate in Heat-Resistant Steel by Processing of Digital Images of Fracture Surfaces

Maruschak

Vorobel

Student

et al. 2021

Metals

View full text Add to dashboard Cite

The micro- and macroscopic fatigue crack growth (FCG) rates of a wide class of structural materials were analyzed and it was concluded that both rates coincide either during high-temperature tests or at high stress intensity factor (SIF) values. Their coincidence requires a high level of cyclic deformation of the metal along the entire crack front as a necessary condition for the formation of fatigue striations (FS). Based on the analysis of digital fractographic images of the fatigue fracture surfaces, a method for the quantitative assessment of the spacing of FS has been developed. The method includes the detection of FS by binarization of the image based on the principle of local minima, rotation of the highlighted fragments of the image using the Hough transform, and the calculation of the distances between continuous lines. The method was tested on 34KhN3M steel in the initial state and after long-term operation (~3 × 105 h) in the rotor disk of a steam turbine at a thermal power plant (TPP). Good agreement was confirmed between FCG rates (both macro and microscopic, determined manually or using digital imaging techniques) at high SIF ranges and their noticeable discrepancy at low SIF ranges. Possible reasons for the discrepancy between the micro- and macroscopic FCG rates at low values of the SIF are analyzed. It has also been noted that FS is easier to detect on the fracture surface of degraded steel. Hydrogen embrittlement of steel during operation promotes secondary cracking along the FS, making them easier to detect and quantify. It is shown that the invariable value of the microscopic FCG rate at a low SIF range in the operated steel is lower than observable for the steel in the initial state. Secondary cracking of the operated steel may have contributed to the formation of a typical FS pattern along the entire crack front at a lower FCG rate than in unoperated steel.

show abstract

Toward quantitative fractography using convolutional neural networks

Cited by 40 publications

References 37 publications

A graph based workflow for extracting grain-scale toughness from meso-scale experiments

A graph based workflow for extracting grain-scale toughness from meso-scale experiments

Quantitative Prediction of Fracture Toughness $(K_{{\rm I}c})$ of Polymer by Fractography Using Deep Neural Networks

Estimation of Fatigue Crack Growth Rate in Heat-Resistant Steel by Processing of Digital Images of Fracture Surfaces

Contact Info

Product

Resources

About