Digital polycrystalline microstructure generation using diffusion probabilistic models

Patxi Fernandez-Zelaia; Jiahao Cheng; Jason Mayeur; Amir Koushyar Ziabari; Michael M. Kirka

doi:10.1016/j.mtla.2023.101976

Digital polycrystalline microstructure generation using diffusion probabilistic models

Patxi Fernandez-Zelaia, Jiahao Cheng, Jason Mayeur, Amir Koushyar Ziabari, Michael M. Kirka

Research output: Contribution to journal › Article › peer-review

4 Scopus citations

Abstract

Accurate micromechanical simulation of polycrystalline materials requires a realistic digital representation of the grain scale microstructure. This work demonstrates the use of a generative diffusion probabilistic model for synthesizing single phase polycrystalline realizations. The model performs well and is capable of producing realistic microstructures consisting of not just simple equiaxed structures but also structures exhibiting more complex spatial arrangements. Masked microstructure generation reveals that the model is context aware of morphological descriptors which may be encoded in the latent space. Training on more diverse data sets, with scaled up architectures, may enable development of future models capable of synthesizing even more complex microstructural features.

Original language	English
Article number	101976
Journal	Materialia
Volume	33
DOIs	https://doi.org/10.1016/j.mtla.2023.101976
State	Published - Mar 2024

Funding

Research was sponsored by the US Department of Energy, Office of Energy Efficiency and Renewable Energy (EERE) , Advanced Manufacturing Office, under contract DE-AC05-00OR22725 with UT-Battelle LLC and performed in partiality at the Oak Ridge National Laboratory’s Manufacturing Demonstration Facility, an Office of Energy Efficiency and Renewable Energy user facility. A.K.Z. was supported by Laboratory Director’s Research and Development (LDRD) grant at ORNL . Notice of Copyright. This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).Research was sponsored by the US Department of Energy, Office of Energy Efficiency and Renewable Energy (EERE), Advanced Manufacturing Office, under contract DE-AC05-00OR22725 with UT-Battelle LLC and performed in partiality at the Oak Ridge National Laboratory's Manufacturing Demonstration Facility, an Office of Energy Efficiency and Renewable Energy user facility. A.K.Z. was supported by Laboratory Director's Research and Development (LDRD) grant at ORNL. Notice of Copyright . This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan ( http://energy.gov/downloads/doe-public-access-plan ).

Funders	Funder number
DOE Public Access Plan
United States Government
U.S. Department of Energy
Advanced Manufacturing Office	DE-AC05-00OR22725
Office of Energy Efficiency and Renewable Energy
Oak Ridge National Laboratory
UT-Battelle

Keywords

Generative modeling
ICME
Machine learning
Microstructure

Access to Document

10.1016/j.mtla.2023.101976

Cite this

@article{2db2a7b095664ce8a71a10f26e9d761c,

title = "Digital polycrystalline microstructure generation using diffusion probabilistic models",

abstract = "Accurate micromechanical simulation of polycrystalline materials requires a realistic digital representation of the grain scale microstructure. This work demonstrates the use of a generative diffusion probabilistic model for synthesizing single phase polycrystalline realizations. The model performs well and is capable of producing realistic microstructures consisting of not just simple equiaxed structures but also structures exhibiting more complex spatial arrangements. Masked microstructure generation reveals that the model is context aware of morphological descriptors which may be encoded in the latent space. Training on more diverse data sets, with scaled up architectures, may enable development of future models capable of synthesizing even more complex microstructural features.",

keywords = "Generative modeling, ICME, Machine learning, Microstructure",

author = "Patxi Fernandez-Zelaia and Jiahao Cheng and Jason Mayeur and Ziabari, {Amir Koushyar} and Kirka, {Michael M.}",

note = "Publisher Copyright: {\textcopyright} 2023",

year = "2024",

month = mar,

doi = "10.1016/j.mtla.2023.101976",

language = "English",

volume = "33",

journal = "Materialia",

issn = "2589-1529",

publisher = "Elsevier",

}

TY - JOUR

T1 - Digital polycrystalline microstructure generation using diffusion probabilistic models

AU - Fernandez-Zelaia, Patxi

AU - Cheng, Jiahao

AU - Mayeur, Jason

AU - Ziabari, Amir Koushyar

AU - Kirka, Michael M.

PY - 2024/3

Y1 - 2024/3

N2 - Accurate micromechanical simulation of polycrystalline materials requires a realistic digital representation of the grain scale microstructure. This work demonstrates the use of a generative diffusion probabilistic model for synthesizing single phase polycrystalline realizations. The model performs well and is capable of producing realistic microstructures consisting of not just simple equiaxed structures but also structures exhibiting more complex spatial arrangements. Masked microstructure generation reveals that the model is context aware of morphological descriptors which may be encoded in the latent space. Training on more diverse data sets, with scaled up architectures, may enable development of future models capable of synthesizing even more complex microstructural features.

AB - Accurate micromechanical simulation of polycrystalline materials requires a realistic digital representation of the grain scale microstructure. This work demonstrates the use of a generative diffusion probabilistic model for synthesizing single phase polycrystalline realizations. The model performs well and is capable of producing realistic microstructures consisting of not just simple equiaxed structures but also structures exhibiting more complex spatial arrangements. Masked microstructure generation reveals that the model is context aware of morphological descriptors which may be encoded in the latent space. Training on more diverse data sets, with scaled up architectures, may enable development of future models capable of synthesizing even more complex microstructural features.

KW - Generative modeling

KW - ICME

KW - Machine learning

KW - Microstructure

UR - http://www.scopus.com/inward/record.url?scp=85179006883&partnerID=8YFLogxK

U2 - 10.1016/j.mtla.2023.101976

DO - 10.1016/j.mtla.2023.101976

M3 - Article

AN - SCOPUS:85179006883

SN - 2589-1529

VL - 33

JO - Materialia

JF - Materialia

M1 - 101976

ER -

Digital polycrystalline microstructure generation using diffusion probabilistic models

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this