Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework

Majdi I. Radaideh, Katarzyna Borowiec, Tomasz Kozlowski

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Bayesian-based model selection and UQ methodology is developed and applied to nuclear thermal-hydraulics codes and void fraction data in this paper. Uncertainties inherent in the experimental data along with the predictive and model-form uncertainty are quantified to construct a composite/hybrid model based on the competent models to predict the response with confidence. The predictive uncertainty or model discrepancy dominates the model-form uncertainty for the void fraction at low axial locations. Improvements in composite predictions are observed at higher axial locations at which model-form uncertainty plays a major role. It is found also that including the measured data uncertainty during the UQ process improves the model prediction performance, instead of assuming perfect data and penalizing the models. The proposed methodology is flexible and extendable to other types of physics, models, and data. Developing the underlying methodology of estimating the model weights will be the main focus of the subsequent studies.

Original languageEnglish (US)
Title of host publicationInternational Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019
PublisherAmerican Nuclear Society
Pages2775-2784
Number of pages10
ISBN (Electronic)9780894487699
StatePublished - Jan 1 2019
Event2019 International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019 - Portland, United States
Duration: Aug 25 2019Aug 29 2019

Publication series

NameInternational Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019

Conference

Conference2019 International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019
CountryUnited States
CityPortland
Period8/25/198/29/19

Fingerprint

Uncertainty Quantification
Uncertainty
Void Fraction
Model
Methodology
Void fraction
Composite
Framework
Form
Performance Prediction
Hybrid Model
Model Selection
Hydraulics
Prediction Model
Confidence
Discrepancy
Composite materials
Physics
Experimental Data
Model-based

Keywords

  • Bayesian model averaging
  • Model-form
  • Nuclear codes
  • Uncertainty quantification

ASJC Scopus subject areas

  • Applied Mathematics
  • Nuclear Energy and Engineering

Cite this

Radaideh, M. I., Borowiec, K., & Kozlowski, T. (2019). Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework. In International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019 (pp. 2775-2784). (International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019). American Nuclear Society.

Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework. / Radaideh, Majdi I.; Borowiec, Katarzyna; Kozlowski, Tomasz.

International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019. American Nuclear Society, 2019. p. 2775-2784 (International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Radaideh, MI, Borowiec, K & Kozlowski, T 2019, Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework. in International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019. International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019, American Nuclear Society, pp. 2775-2784, 2019 International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019, Portland, United States, 8/25/19.
Radaideh MI, Borowiec K, Kozlowski T. Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework. In International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019. American Nuclear Society. 2019. p. 2775-2784. (International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019).
Radaideh, Majdi I. ; Borowiec, Katarzyna ; Kozlowski, Tomasz. / Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework. International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019. American Nuclear Society, 2019. pp. 2775-2784 (International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019).
@inproceedings{68ab6998658f43ff825619ed7cff6617,
title = "Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework",
abstract = "Bayesian-based model selection and UQ methodology is developed and applied to nuclear thermal-hydraulics codes and void fraction data in this paper. Uncertainties inherent in the experimental data along with the predictive and model-form uncertainty are quantified to construct a composite/hybrid model based on the competent models to predict the response with confidence. The predictive uncertainty or model discrepancy dominates the model-form uncertainty for the void fraction at low axial locations. Improvements in composite predictions are observed at higher axial locations at which model-form uncertainty plays a major role. It is found also that including the measured data uncertainty during the UQ process improves the model prediction performance, instead of assuming perfect data and penalizing the models. The proposed methodology is flexible and extendable to other types of physics, models, and data. Developing the underlying methodology of estimating the model weights will be the main focus of the subsequent studies.",
keywords = "Bayesian model averaging, Model-form, Nuclear codes, Uncertainty quantification",
author = "Radaideh, {Majdi I.} and Katarzyna Borowiec and Tomasz Kozlowski",
year = "2019",
month = "1",
day = "1",
language = "English (US)",
series = "International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019",
publisher = "American Nuclear Society",
pages = "2775--2784",
booktitle = "International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019",

}

TY - GEN

T1 - Uncertainty quantification of model-form and predictive uncertainties in nuclear codes using Bayesian framework

AU - Radaideh, Majdi I.

AU - Borowiec, Katarzyna

AU - Kozlowski, Tomasz

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Bayesian-based model selection and UQ methodology is developed and applied to nuclear thermal-hydraulics codes and void fraction data in this paper. Uncertainties inherent in the experimental data along with the predictive and model-form uncertainty are quantified to construct a composite/hybrid model based on the competent models to predict the response with confidence. The predictive uncertainty or model discrepancy dominates the model-form uncertainty for the void fraction at low axial locations. Improvements in composite predictions are observed at higher axial locations at which model-form uncertainty plays a major role. It is found also that including the measured data uncertainty during the UQ process improves the model prediction performance, instead of assuming perfect data and penalizing the models. The proposed methodology is flexible and extendable to other types of physics, models, and data. Developing the underlying methodology of estimating the model weights will be the main focus of the subsequent studies.

AB - Bayesian-based model selection and UQ methodology is developed and applied to nuclear thermal-hydraulics codes and void fraction data in this paper. Uncertainties inherent in the experimental data along with the predictive and model-form uncertainty are quantified to construct a composite/hybrid model based on the competent models to predict the response with confidence. The predictive uncertainty or model discrepancy dominates the model-form uncertainty for the void fraction at low axial locations. Improvements in composite predictions are observed at higher axial locations at which model-form uncertainty plays a major role. It is found also that including the measured data uncertainty during the UQ process improves the model prediction performance, instead of assuming perfect data and penalizing the models. The proposed methodology is flexible and extendable to other types of physics, models, and data. Developing the underlying methodology of estimating the model weights will be the main focus of the subsequent studies.

KW - Bayesian model averaging

KW - Model-form

KW - Nuclear codes

KW - Uncertainty quantification

UR - http://www.scopus.com/inward/record.url?scp=85075372032&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85075372032&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85075372032

T3 - International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019

SP - 2775

EP - 2784

BT - International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019

PB - American Nuclear Society

ER -