BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA

Carl Scarth; Andrew T. Rhead; Richard Butler

doi:10.5075/epfl-298799_978-2-9701614-0-0

BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA

Carl Scarth, Andrew T. Rhead, Richard Butler

Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding

Abstract

Bayesian calibration is a statistical framework for combining experimental and numerical data to make predictions using computer models, while formally accounting for uncertainty associated with: i) experimental observation error, ii) inaccuracies in the model and assumed physics, and iii) variability in the physical parameters of the test specimen. The ability to accurately quantify such uncertainty will enable certification processes to be better informed by mathematical models and component-level structural tests. In this paper, Bayesian calibration is applied to a finite element stiffened panel model implemented in Abaqus, using strain gauge data from experimental compression tests. In this preliminary investigation, the elastic properties of the AS4/8552 prepreg lamina from which the panel was manufactured are inferred. Such parameters cannot be precisely controlled in the experiment during to inherent variability in manufacturing processes. A Gaussian Process Emulator is used in the calibration for the sake of computational efficiency.

Original language	English
Title of host publication	Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland
Editors	Anastasios P. Vassilopoulos, Veronique Michaud
Publisher	Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL)
Pages	875-882
Number of pages	8
Volume	4
ISBN (Electronic)	9782970161400
DOIs	https://doi.org/10.5075/epfl-298799_978-2-9701614-0-0
Publication status	Published - 30 Jun 2022
Event	20th European Conference on Composite Materials: Composites Meet Sustainability, ECCM 2022 - Lausanne, Switzerland Duration: 26 Jun 2022 → 30 Jun 2022

Publication series

Name	Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability
Volume	1-6

Conference

Conference	20th European Conference on Composite Materials: Composites Meet Sustainability, ECCM 2022
Country/Territory	Switzerland
City	Lausanne
Period	26/06/22 → 30/06/22

Bibliographical note

Funding Information:
The presented research was supported by?the? EPSRC? Programme?Grant?‘Certification?for? Design? – Reshaping?the?Testing?Pyramid’?(CerTest,?EP/S017038/1).

Funding

The presented research was supported by?the? EPSRC? Programme?Grant?‘Certification?for? Design? – Reshaping?the?Testing?Pyramid’?(CerTest,?EP/S017038/1).

Keywords

Bayesian inference
Finite Element Analysis
Gaussian Process Emulators
Model calibration
Uncertainty Quantification

ASJC Scopus subject areas

Ceramics and Composites

Access to Document

10.5075/epfl-298799_978-2-9701614-0-0Licence: CC BY-NC

Cite this

Scarth, C., Rhead, A. T., & Butler, R. (2022). BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA. In A. P. Vassilopoulos, & V. Michaud (Eds.), Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland (Vol. 4, pp. 875-882). (Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability; Vol. 1-6). Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL). https://doi.org/10.5075/epfl-298799_978-2-9701614-0-0

BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA. / Scarth, Carl ; Rhead, Andrew T.; Butler, Richard.
Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland . ed. / Anastasios P. Vassilopoulos; Veronique Michaud. Vol. 4 Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL), 2022. p. 875-882 (Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability; Vol. 1-6).

Research output: Chapter or section in a book/report/conference proceeding › Chapter in a published conference proceeding

Scarth, C , Rhead, AT & Butler, R 2022, BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA. in AP Vassilopoulos & V Michaud (eds), Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland . vol. 4, Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability, vol. 1-6, Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL), pp. 875-882, 20th European Conference on Composite Materials: Composites Meet Sustainability, ECCM 2022, Lausanne, Switzerland, 26/06/22. https://doi.org/10.5075/epfl-298799_978-2-9701614-0-0

Scarth C , Rhead AT , Butler R. BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA. In Vassilopoulos AP, Michaud V, editors, Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland . Vol. 4. Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL). 2022. p. 875-882. (Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability). doi: 10.5075/epfl-298799_978-2-9701614-0-0

Scarth, Carl ; Rhead, Andrew T. ; Butler, Richard. / BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA. Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland . editor / Anastasios P. Vassilopoulos ; Veronique Michaud. Vol. 4 Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL), 2022. pp. 875-882 (Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability).

@inproceedings{3e2caa19b0b04998a6c559f52ada0350,

title = "BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA",

abstract = "Bayesian calibration is a statistical framework for combining experimental and numerical data to make predictions using computer models, while formally accounting for uncertainty associated with: i) experimental observation error, ii) inaccuracies in the model and assumed physics, and iii) variability in the physical parameters of the test specimen. The ability to accurately quantify such uncertainty will enable certification processes to be better informed by mathematical models and component-level structural tests. In this paper, Bayesian calibration is applied to a finite element stiffened panel model implemented in Abaqus, using strain gauge data from experimental compression tests. In this preliminary investigation, the elastic properties of the AS4/8552 prepreg lamina from which the panel was manufactured are inferred. Such parameters cannot be precisely controlled in the experiment during to inherent variability in manufacturing processes. A Gaussian Process Emulator is used in the calibration for the sake of computational efficiency.",

keywords = "Bayesian inference, Finite Element Analysis, Gaussian Process Emulators, Model calibration, Uncertainty Quantification",

author = "Carl Scarth and Rhead, {Andrew T.} and Richard Butler",

note = "Funding Information: The presented research was supported by?the? EPSRC? Programme?Grant?{\textquoteleft}Certification?for? Design? – Reshaping?the?Testing?Pyramid{\textquoteright}?(CerTest,?EP/S017038/1). ; 20th European Conference on Composite Materials: Composites Meet Sustainability, ECCM 2022 ; Conference date: 26-06-2022 Through 30-06-2022",

year = "2022",

month = jun,

day = "30",

doi = "10.5075/epfl-298799_978-2-9701614-0-0",

language = "English",

volume = "4",

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

publisher = "Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL)",

pages = "875--882",

editor = "Vassilopoulos, {Anastasios P.} and Veronique Michaud",

booktitle = "Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland",

}

TY - GEN

T1 - BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA

AU - Scarth, Carl

AU - Rhead, Andrew T.

AU - Butler, Richard

N1 - Funding Information: The presented research was supported by?the? EPSRC? Programme?Grant?‘Certification?for? Design? – Reshaping?the?Testing?Pyramid’?(CerTest,?EP/S017038/1).

PY - 2022/6/30

Y1 - 2022/6/30

N2 - Bayesian calibration is a statistical framework for combining experimental and numerical data to make predictions using computer models, while formally accounting for uncertainty associated with: i) experimental observation error, ii) inaccuracies in the model and assumed physics, and iii) variability in the physical parameters of the test specimen. The ability to accurately quantify such uncertainty will enable certification processes to be better informed by mathematical models and component-level structural tests. In this paper, Bayesian calibration is applied to a finite element stiffened panel model implemented in Abaqus, using strain gauge data from experimental compression tests. In this preliminary investigation, the elastic properties of the AS4/8552 prepreg lamina from which the panel was manufactured are inferred. Such parameters cannot be precisely controlled in the experiment during to inherent variability in manufacturing processes. A Gaussian Process Emulator is used in the calibration for the sake of computational efficiency.

AB - Bayesian calibration is a statistical framework for combining experimental and numerical data to make predictions using computer models, while formally accounting for uncertainty associated with: i) experimental observation error, ii) inaccuracies in the model and assumed physics, and iii) variability in the physical parameters of the test specimen. The ability to accurately quantify such uncertainty will enable certification processes to be better informed by mathematical models and component-level structural tests. In this paper, Bayesian calibration is applied to a finite element stiffened panel model implemented in Abaqus, using strain gauge data from experimental compression tests. In this preliminary investigation, the elastic properties of the AS4/8552 prepreg lamina from which the panel was manufactured are inferred. Such parameters cannot be precisely controlled in the experiment during to inherent variability in manufacturing processes. A Gaussian Process Emulator is used in the calibration for the sake of computational efficiency.

KW - Bayesian inference

KW - Finite Element Analysis

KW - Gaussian Process Emulators

KW - Model calibration

KW - Uncertainty Quantification

UR - http://www.scopus.com/inward/record.url?scp=85149375333&partnerID=8YFLogxK

U2 - 10.5075/epfl-298799_978-2-9701614-0-0

DO - 10.5075/epfl-298799_978-2-9701614-0-0

M3 - Chapter in a published conference proceeding

AN - SCOPUS:85149375333

VL - 4

T3 - Proceedings of the 20th European Conference on Composite Materials: Composites Meet Sustainability

SP - 875

EP - 882

BT - Composites Meet Sustainability – Proceedings of the 20 th European Conference on Composite Materials, ECCM20. 26-30 June, 2022, Lausanne, Switzerland

A2 - Vassilopoulos, Anastasios P.

A2 - Michaud, Veronique

PB - Composite Construction Laboratory (CCLab), Ecole Polytechnique Federale de Lausanne (EPFL)

T2 - 20th European Conference on Composite Materials: Composites Meet Sustainability, ECCM 2022

Y2 - 26 June 2022 through 30 June 2022

ER -

BAYESIAN CALIBRATION OF A FINITE ELEMENT STIFFENED PANEL MODEL USING EXPERIMENTAL COMPRESSION TEST DATA

Abstract

Publication series

Conference

Bibliographical note

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this