A Microstructural Model by Space Tessellation for a Sintered Ceramic: Cerine

Coster, Michel; Arnould, Xavier; Chermant, Jean‐Louis; Moataz, Abder El; Chartier, Thierry

doi:10.5566/ias.v24.p105-116

Cited by 18 publications

(6 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Within the materials domain, the earliest computational approaches used to generate microstructural morphologies derive from physics-based models, which have been recognized for both their high level of detail and realistic output. A study by [29] used a Voronoi tessellation model that simulated ceramic grain boundary evolution based on well-established equations developed by [30][31][32]. Additionally, research approaches by [33] combined Monte Carlo simulations and grain growth kinetics to model metal crystallization.…”

Section: Materials-specific Approachesmentioning

confidence: 99%

“…3D ShapeNets [20] 83.54 77.32 VoxNet [69] 92.00 83.00 Geometry image [70] 88.40 83.90 PointNet [71] 77.60 -GIFT [72] 92.35 83.10 FusionNet [73] 93.11 90.80 Method (unsupervised) " " SPH [74] 79.79 68.23 LFD [75] 79.87 75.47 VConv-DAE [26] 80.50 75.50 3D-GAN [27] 91.00 83.30 3D-GAN ( ≈ 100 samples) 90.00 81.30 Ours (M-GAN) (100 samples) 92. 29 85.08…”

Section: Materials Datamentioning

confidence: 99%

See 1 more Smart Citation

3D Grain Shape Generation in Polycrystals Using Generative Adversarial Networks

Jangid

Brodnik

Khan

et al. 2022

Integr Mater Manuf Innov

View full text Add to dashboard Cite

This paper presents a generative adversarial network (GAN) capable of producing realistic microstructure morphology features and demonstrates its capabilities on a dataset of crystalline titanium grain shapes. Alongside this, we present an approach to train deep learning networks to understand material-specific descriptor features, such as grain shapes, based on existing conceptual relationships with established learning spaces, such as functional object shapes. A style-based GAN with Wasserstein loss, called M-GAN, was first trained to recognize distributions of morphology features from function objects in the ShapeNet dataset and was then applied to grain morphologies from a 3D crystallographic dataset of Ti-6Al-4V. Evaluation of feature recognition on objects showed comparable or better performance than state-of-the-art voxel-based network approaches. When applied to experimental data, M-GAN generated realistic grain morphologies comparable to those seen in Ti-6Al-4V. A quantitative comparison of moment invariant distributions showed that the generated grains were similar in shape and structure to the ground truth, but scale invariance learned from object recognition led to difficulty in distinguishing between the physical features of small grains and spatial resolution artifacts. The physical implications of M-GAN's learning capabilities are discussed, as well as the extensibility of this approach to other material characteristics related to grain morphology.

show abstract

Section: Materials-specific Approachesmentioning

confidence: 99%

Section: Materials Datamentioning

confidence: 99%

3D Grain Shape Generation in Polycrystals Using Generative Adversarial Networks

Jangid

Brodnik

Khan

et al. 2022

Integr Mater Manuf Innov

View full text Add to dashboard Cite

show abstract

“…Synthetic Crystalline Microstructures: Physics-based models have been used for microstructure generation due to their high level of detail and realistic morphologies. Work by Coster et al [24] used a Voronoi tessellation model that simulated ceramic grain boundary evolution based on equations developed by Johnson and Mehl [25], Avrami [26], and Kolmogorov [27]. Additionally, the work of Nosonovsky et al [28] combined Monte Carlo simulations and grain growth kinetics to model metal crystallization.…”

Section: Related Workmentioning

confidence: 99%

3DMaterialGAN: Learning 3D Shape Representation from Latent Space for Materials Science Applications

Jangid,

Brodnik,

Khan

et al. 2020

Preprint

View full text Add to dashboard Cite

In the field of computer vision, unsupervised learning for 2D object generation has advanced rapidly in the past few years. However, 3D object generation has not garnered the same attention or success as its predecessor. To facilitate novel progress at the intersection of computer vision and materials science, we propose a 3DMaterialGAN network that is capable of recognizing and synthesizing individual grains whose morphology conforms to a given 3D polycrystalline material microstructure. This Generative Adversarial Network (GAN) architecture yields complex 3D objects from probabilistic latent space vectors with no additional information from 2D rendered images. We show that this method performs comparably or better than state-of-the-art on benchmark annotated 3D datasets, while also being able to distinguish and generate objects that are not easily annotated, such as grain morphologies. The value of our algorithm is demonstrated with analysis on experimental real-world data, namely generating 3D grain structures found in a commercially relevant wrought titanium alloy, which were validated through statistical shape comparison. This framework lays the foundation for the recognition and synthesis of polycrystalline material microstructures, which are used in additive manufacturing, aerospace, and structural design applications.Generative Models: As more comprehensive datasets like ShapeNet from Chang et al. [1] and Wu et al. [2] become available to the community, this will enable us to build better generative models to tackle the difficult problem of 3D object generation. Although there have been previous successful attempts that provide solutions to this task, the quality of the objects generated still require improvement, and many opportunities remain for the application of these models to real-world problems.Preprint. Under review.

show abstract

“…For obtaining realistic geometric shapes for particles with crystalline grains in particle‐level models, numerical tessellation methods (e.g., the Voronoi model) can be applied. In such tessellation methods, the grain boundaries are determined by the positions of individual grains . In addition to a particle geometry, a fixed initial particle size distribution, as well as, fixed values for the specific energy values and diffusion coefficients are assigned to the simulated particles.…”

Section: Conversion Of Highly Filled Papers Into Paper‐derived Sintermentioning

confidence: 99%

Highly Filled Papers, on their Manufacturing, Processing, and Applications

et al. 2019

View full text Add to dashboard Cite

Since more than a decade ago, the research on highly filled papers, as well as paper-derived inorganic materials, has greatly intensified. As presented in this review, highly filled papers as preforms allow for the design of porous or dense, multilayered, and geometrically complex structures. These paperderived ceramic-or metal-based materials are generated by the heattreatment of highly filled papers. Paper-derived materials are potential materials of choice for applications in transportation, energy-generation, environmental conservation, support structures, medical uses, and electronic components. Due to the adjustability of the filler content and the good machinability of highly filled papers, paper-derived sheets or multilayers may include intricate structures and tailored gradients in phase structure or porosity. Paper-derived multilayers also may contain cast ceramic tapes or other functionalized layers, as presented in some examples. Computer-aided manufacturing processes for paper-derived materials can be supplemented by prediction models for the sintering shrinkage in order to identify optimal post-processing steps, stacking orders and orientations for highly filled paper layers within multilayer green bodies. The accuracy of established component-level sintering models can be significantly increased by microstructure models of the highly filled paper. ParameterName Unit α Mismatch parameter for fiber cross-sections, Equation (5) Dimensionless α p Layer position in viscoelastic paper structure model, Equation (2) Dimensionless A 0 Initial amplitude of perturbation, Equations (15) 10 À6 m A(f) Parameter for evolution of instantaneous carbon conversion rate, Equation (7) Dimensionless A diff Source controlled diffusion parameter, Equations (9) 10 À7 N A match Modeling parameter, Equations (8) Dimensionless a Inter-particle contact-area radius, Equations (19) 10 À8 m a Local average pore area, Equation (4) 10 À6 m 2 b Pore geometry constant, Equation (20) Dimensionless ß evo , m evo Empirical constants for grain evolution in SOVS model, Equation (8) Dimensionless ß r(c)p Proportionality constant for bimodal packing fraction determination, Equation (18) Dimensionless C Modeling constant for debindering pressure calculation, Equation (7) 10 À6 (m 2 Á s)/K C 1 , C 2 , C 3 , C 4 , C 5 Parameters in the Riedel model that are determined by the dihedral angle, Equations (9) Dimensionless c glue Heat capacity of the adhesive layers, Equation (3) 10 À6 s c L , c s Volume fraction of larger-sized/smaller-sized particles in a multimodal mixture of particles, Equation (17)and (18) 10 0 vol% c vol Volume concentration of pulp fibers, Equation (4) 10 0 vol% γ, γ 0 Strain relaxation rate of calendered paper with initial value, Equation (2) Dimensionless γ B Grain boundary energy density, Equation (13) 10 0 J m À2 γ b Specific energy for grain boundary diffusion, Equation (9) 10 3 J mol À1 γ S Material surface energy, Equation (8, 11, and 13) 10 0 J m À2 Δ Half of the inter-particle boundary thickness, Equation (19) 10 À8...

show abstract

A Microstructural Model by Space Tessellation for a Sintered Ceramic: Cerine

Cited by 18 publications

References 31 publications

3D Grain Shape Generation in Polycrystals Using Generative Adversarial Networks

3D Grain Shape Generation in Polycrystals Using Generative Adversarial Networks

3DMaterialGAN: Learning 3D Shape Representation from Latent Space for Materials Science Applications

Highly Filled Papers, on their Manufacturing, Processing, and Applications

Contact Info

Product

Resources

About