Deep learning based prediction of fibrous microstructure permeability

Baris Caglar; G.C. Broggi; Muhammad A. Ali; Laurent Orgéas; Véronique Michaud

Deep learning based prediction of fibrous microstructure permeability

Baris Caglar, G.C. Broggi, Muhammad A. Ali, Laurent Orgéas, Véronique Michaud

Aerospace Manufacturing Technologies

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

6 Downloads (Pure)

Abstract

Knowledge of permeability of fibrous microstructures is crucial for predicting the mold fill times and resin flow path in composite manufacturing. Herein we report a method to rapidly predict the permeability of 3D fibrous microstructures. Our method relies on predicting the permeability of 2D cross-sections via deep neural networks and extending this capability to 3D microstructures via circuit analogy as a means of reduced order modeling. Approximately 50% of the permeability predictions of 2D cross-sections have 10% or less deviation from the permeability results obtained via flow simulations in Geodict. Computational time required for predicting the permeability of 3D microstructures is reduced from hours to less than 10 seconds. This framework enables fast and accurate prediction of micro-permeability and serves as the first building block towards prediction of fabric mesostructures’ permeability via deep learning based methods.

Original language	English
Title of host publication	Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability
Subtitle of host publication	Vol 4 – Modeling and Prediction
Editors	Anastasios P. Vassilopoulos , Véronique Michaud
Place of Publication	Lausanne
Publisher	EPFL Lausanne, Composite Construction Laboratory
Pages	807-814
Number of pages	8
ISBN (Electronic)	978-2-9701614-0-0
Publication status	Published - 2022
Event	20th European Conference on Composite Materials: Composites Meet Sustainability - Lausanne, Switzerland Duration: 26 Jun 2022 → 30 Jun 2022 Conference number: 20

Conference

Conference	20th European Conference on Composite Materials
Abbreviated title	ECCM20
Country/Territory	Switzerland
City	Lausanne
Period	26/06/22 → 30/06/22

Keywords

Deep Learning
Permeability
Microstructures
Numerical analysis

Access to Document

ECCM2022Final published version, 11.6 MBLicence: CC BY

Cite this

Caglar, B., Broggi, G. C., Ali, M. A., Orgéas, L., & Michaud, V. (2022). Deep learning based prediction of fibrous microstructure permeability. In A. P. Vassilopoulos , & V. Michaud (Eds.), Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability: Vol 4 – Modeling and Prediction (pp. 807-814). EPFL Lausanne, Composite Construction Laboratory.

Caglar, Baris ; Broggi, G.C. ; Ali, Muhammad A. ; Orgéas, Laurent ; Michaud, Véronique. / Deep learning based prediction of fibrous microstructure permeability. Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability: Vol 4 – Modeling and Prediction. editor / Anastasios P. Vassilopoulos ; Véronique Michaud . Lausanne : EPFL Lausanne, Composite Construction Laboratory, 2022. pp. 807-814

@inproceedings{4e2cdfab8ec04623a82559bedf3417de,

title = "Deep learning based prediction of fibrous microstructure permeability",

abstract = "Knowledge of permeability of fibrous microstructures is crucial for predicting the mold fill times and resin flow path in composite manufacturing. Herein we report a method to rapidly predict the permeability of 3D fibrous microstructures. Our method relies on predicting the permeability of 2D cross-sections via deep neural networks and extending this capability to 3D microstructures via circuit analogy as a means of reduced order modeling. Approximately 50% of the permeability predictions of 2D cross-sections have 10% or less deviation from the permeability results obtained via flow simulations in Geodict. Computational time required for predicting the permeability of 3D microstructures is reduced from hours to less than 10 seconds. This framework enables fast and accurate prediction of micro-permeability and serves as the first building block towards prediction of fabric mesostructures{\textquoteright} permeability via deep learning based methods. ",

keywords = "Deep Learning, Permeability, Microstructures, Numerical analysis",

author = "Baris Caglar and G.C. Broggi and Ali, {Muhammad A.} and Laurent Org{\'e}as and V{\'e}ronique Michaud",

year = "2022",

language = "English",

pages = "807--814",

editor = "{Vassilopoulos }, {Anastasios P. } and {Michaud }, {V{\'e}ronique }",

booktitle = "Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability",

publisher = "EPFL Lausanne, Composite Construction Laboratory",

note = "20th European Conference on Composite Materials : Composites Meet Sustainability, ECCM20 ; Conference date: 26-06-2022 Through 30-06-2022",

}

Caglar, B , Broggi, GC, Ali, MA, Orgéas, L & Michaud, V 2022, Deep learning based prediction of fibrous microstructure permeability. in AP Vassilopoulos & V Michaud (eds), Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability: Vol 4 – Modeling and Prediction. EPFL Lausanne, Composite Construction Laboratory, Lausanne, pp. 807-814, 20th European Conference on Composite Materials, Lausanne, Switzerland, 26/06/22.

Deep learning based prediction of fibrous microstructure permeability. / Caglar, Baris ; Broggi, G.C.; Ali, Muhammad A.; Orgéas, Laurent; Michaud, Véronique.

Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability: Vol 4 – Modeling and Prediction. ed. / Anastasios P. Vassilopoulos ; Véronique Michaud . Lausanne : EPFL Lausanne, Composite Construction Laboratory, 2022. p. 807-814.

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Deep learning based prediction of fibrous microstructure permeability

AU - Caglar, Baris

AU - Broggi, G.C.

AU - Ali, Muhammad A.

AU - Orgéas, Laurent

AU - Michaud, Véronique

N1 - Conference code: 20

PY - 2022

Y1 - 2022

N2 - Knowledge of permeability of fibrous microstructures is crucial for predicting the mold fill times and resin flow path in composite manufacturing. Herein we report a method to rapidly predict the permeability of 3D fibrous microstructures. Our method relies on predicting the permeability of 2D cross-sections via deep neural networks and extending this capability to 3D microstructures via circuit analogy as a means of reduced order modeling. Approximately 50% of the permeability predictions of 2D cross-sections have 10% or less deviation from the permeability results obtained via flow simulations in Geodict. Computational time required for predicting the permeability of 3D microstructures is reduced from hours to less than 10 seconds. This framework enables fast and accurate prediction of micro-permeability and serves as the first building block towards prediction of fabric mesostructures’ permeability via deep learning based methods.

AB - Knowledge of permeability of fibrous microstructures is crucial for predicting the mold fill times and resin flow path in composite manufacturing. Herein we report a method to rapidly predict the permeability of 3D fibrous microstructures. Our method relies on predicting the permeability of 2D cross-sections via deep neural networks and extending this capability to 3D microstructures via circuit analogy as a means of reduced order modeling. Approximately 50% of the permeability predictions of 2D cross-sections have 10% or less deviation from the permeability results obtained via flow simulations in Geodict. Computational time required for predicting the permeability of 3D microstructures is reduced from hours to less than 10 seconds. This framework enables fast and accurate prediction of micro-permeability and serves as the first building block towards prediction of fabric mesostructures’ permeability via deep learning based methods.

KW - Deep Learning

KW - Permeability

KW - Microstructures

KW - Numerical analysis

M3 - Conference contribution

SP - 807

EP - 814

BT - Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability

A2 - Vassilopoulos , Anastasios P.

A2 - Michaud , Véronique

PB - EPFL Lausanne, Composite Construction Laboratory

CY - Lausanne

T2 - 20th European Conference on Composite Materials

Y2 - 26 June 2022 through 30 June 2022

ER -

Caglar B , Broggi GC, Ali MA, Orgéas L, Michaud V. Deep learning based prediction of fibrous microstructure permeability. In Vassilopoulos AP, Michaud V, editors, Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability: Vol 4 – Modeling and Prediction. Lausanne: EPFL Lausanne, Composite Construction Laboratory. 2022. p. 807-814

Deep learning based prediction of fibrous microstructure permeability

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this