Machine learning for turbulence modeling and predictions

S. Bhushan; Greg W. Burgreen; D. Martinez; Wes Brewer

doi:10.1115/FEDSM2020-20038

Machine learning for turbulence modeling and predictions

S. Bhushan, Greg W. Burgreen, D. Martinez, Wes Brewer

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

4 Scopus citations

Abstract

A stand-alone machine learned turbulence model is applied for the solution of integral boundary layer equations, and issues and constraints associated with the model are discussed. The results demonstrate that grouping flow variables into a problem relevant parameter for input during machine learning is desirable to improve accuracy of the model. Further, the accuracy of the model can be improved significantly by incorporation of physics-based constraints during training. Data driven machine learning training requires trial-and-error approach, shows oscillations in a posteriori predictions, and shows unphysical results when used with arbitrary initial condition, as the query is essentially extrapolations. Physics informed machine learning addresses the above limitations, and is identified to be a viable approach for development of machine learned turbulence model.

Original language	English
Title of host publication	Computational Fluid Dynamics; Micro and Nano Fluid Dynamics
Publisher	American Society of Mechanical Engineers (ASME)
ISBN (Electronic)	9780791883730
DOIs	https://doi.org/10.1115/FEDSM2020-20038
State	Published - 2020
Externally published	Yes
Event	ASME 2020 Fluids Engineering Division Summer Meeting, FEDSM 2020, collocated with the ASME 2020 Heat Transfer Summer Conference and the ASME 2020 18th International Conference on Nanochannels, Microchannels, and Minichannels - Virtual, Online Duration: Jul 13 2020 → Jul 15 2020

Publication series

Name	American Society of Mechanical Engineers, Fluids Engineering Division (Publication) FEDSM
Volume	3
ISSN (Print)	0888-8116

Conference

Conference	ASME 2020 Fluids Engineering Division Summer Meeting, FEDSM 2020, collocated with the ASME 2020 Heat Transfer Summer Conference and the ASME 2020 18th International Conference on Nanochannels, Microchannels, and Minichannels
City	Virtual, Online
Period	07/13/20 → 07/15/20

Funding

This material is also based upon work supported by, or in part by, the Department of Defense (DoD) High Performance Computing Modernization Program (HPCMP) under User Productivity Enhancement, Technology Transfer, and Training (PET) contract #47QFSA18K0111, TO# ID04180146. Effort at Mississippi State University was sponsored by the Engineering Research & Development Center under Cooperative Agreement number W912HZ-17-2-0014. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the Engineering Research & Development Center or the U.S. Government.

Funders	Funder number
Engineering Research & Development Center	W912HZ-17-2-0014
U.S. Department of Defense	47QFSA18K0111, ID04180146

Keywords

Machine Learning
Turbulence Modeling

Access to Document

10.1115/FEDSM2020-20038

Cite this

Bhushan, S., Burgreen, G. W., Martinez, D., & Brewer, W. (2020). Machine learning for turbulence modeling and predictions. In Computational Fluid Dynamics; Micro and Nano Fluid Dynamics Article V003T05A008 (American Society of Mechanical Engineers, Fluids Engineering Division (Publication) FEDSM; Vol. 3). American Society of Mechanical Engineers (ASME). https://doi.org/10.1115/FEDSM2020-20038

@inproceedings{d34213f43c304b349a2f70b9161edac6,

title = "Machine learning for turbulence modeling and predictions",

abstract = "A stand-alone machine learned turbulence model is applied for the solution of integral boundary layer equations, and issues and constraints associated with the model are discussed. The results demonstrate that grouping flow variables into a problem relevant parameter for input during machine learning is desirable to improve accuracy of the model. Further, the accuracy of the model can be improved significantly by incorporation of physics-based constraints during training. Data driven machine learning training requires trial-and-error approach, shows oscillations in a posteriori predictions, and shows unphysical results when used with arbitrary initial condition, as the query is essentially extrapolations. Physics informed machine learning addresses the above limitations, and is identified to be a viable approach for development of machine learned turbulence model.",

keywords = "Machine Learning, Turbulence Modeling",

author = "S. Bhushan and Burgreen, {Greg W.} and D. Martinez and Wes Brewer",

note = "Publisher Copyright: {\textcopyright} 2020 American Society of Mechanical Engineers (ASME). All rights reserved.; ASME 2020 Fluids Engineering Division Summer Meeting, FEDSM 2020, collocated with the ASME 2020 Heat Transfer Summer Conference and the ASME 2020 18th International Conference on Nanochannels, Microchannels, and Minichannels ; Conference date: 13-07-2020 Through 15-07-2020",

year = "2020",

doi = "10.1115/FEDSM2020-20038",

language = "English",

series = "American Society of Mechanical Engineers, Fluids Engineering Division (Publication) FEDSM",

publisher = "American Society of Mechanical Engineers (ASME)",

booktitle = "Computational Fluid Dynamics; Micro and Nano Fluid Dynamics",

}

Bhushan, S, Burgreen, GW, Martinez, D & Brewer, W 2020, Machine learning for turbulence modeling and predictions. in Computational Fluid Dynamics; Micro and Nano Fluid Dynamics., V003T05A008, American Society of Mechanical Engineers, Fluids Engineering Division (Publication) FEDSM, vol. 3, American Society of Mechanical Engineers (ASME), ASME 2020 Fluids Engineering Division Summer Meeting, FEDSM 2020, collocated with the ASME 2020 Heat Transfer Summer Conference and the ASME 2020 18th International Conference on Nanochannels, Microchannels, and Minichannels, Virtual, Online, 07/13/20. https://doi.org/10.1115/FEDSM2020-20038

Machine learning for turbulence modeling and predictions. / Bhushan, S.; Burgreen, Greg W.; Martinez, D. et al.
Computational Fluid Dynamics; Micro and Nano Fluid Dynamics. American Society of Mechanical Engineers (ASME), 2020. V003T05A008 (American Society of Mechanical Engineers, Fluids Engineering Division (Publication) FEDSM; Vol. 3).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Machine learning for turbulence modeling and predictions

AU - Bhushan, S.

AU - Burgreen, Greg W.

AU - Martinez, D.

AU - Brewer, Wes

PY - 2020

Y1 - 2020

N2 - A stand-alone machine learned turbulence model is applied for the solution of integral boundary layer equations, and issues and constraints associated with the model are discussed. The results demonstrate that grouping flow variables into a problem relevant parameter for input during machine learning is desirable to improve accuracy of the model. Further, the accuracy of the model can be improved significantly by incorporation of physics-based constraints during training. Data driven machine learning training requires trial-and-error approach, shows oscillations in a posteriori predictions, and shows unphysical results when used with arbitrary initial condition, as the query is essentially extrapolations. Physics informed machine learning addresses the above limitations, and is identified to be a viable approach for development of machine learned turbulence model.

AB - A stand-alone machine learned turbulence model is applied for the solution of integral boundary layer equations, and issues and constraints associated with the model are discussed. The results demonstrate that grouping flow variables into a problem relevant parameter for input during machine learning is desirable to improve accuracy of the model. Further, the accuracy of the model can be improved significantly by incorporation of physics-based constraints during training. Data driven machine learning training requires trial-and-error approach, shows oscillations in a posteriori predictions, and shows unphysical results when used with arbitrary initial condition, as the query is essentially extrapolations. Physics informed machine learning addresses the above limitations, and is identified to be a viable approach for development of machine learned turbulence model.

KW - Machine Learning

KW - Turbulence Modeling

UR - http://www.scopus.com/inward/record.url?scp=85094929527&partnerID=8YFLogxK

U2 - 10.1115/FEDSM2020-20038

DO - 10.1115/FEDSM2020-20038

M3 - Conference contribution

AN - SCOPUS:85094929527

T3 - American Society of Mechanical Engineers, Fluids Engineering Division (Publication) FEDSM

BT - Computational Fluid Dynamics; Micro and Nano Fluid Dynamics

PB - American Society of Mechanical Engineers (ASME)

T2 - ASME 2020 Fluids Engineering Division Summer Meeting, FEDSM 2020, collocated with the ASME 2020 Heat Transfer Summer Conference and the ASME 2020 18th International Conference on Nanochannels, Microchannels, and Minichannels

Y2 - 13 July 2020 through 15 July 2020

ER -

Machine learning for turbulence modeling and predictions

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this