Co-optimized machine-learned manifold models for large eddy simulation of turbulent combustion
Abstract
Many modeling approaches in large eddy simulation (LES) of turbulent combustion employ a projection of the thermochemical state onto a low-dimensional manifold within state space to reduce the number of transported variables and hence computational cost. Flamelet-generated manifolds (FGM) is an example of a well-established, physics-based approach, but increasingly, principal component analysis (PCA) is being used as a data-driven method for generating manifold models. For both approaches, the nonlinear relationship between the location on the predefined manifold and the outputs of interest, such as reaction rates, can be tabulated or encoded in a neural network. This work proposes a new approach for manifold modeling that extends these existing approaches. A modified neural network structure simultaneously encodes the definition of the manifold variables, the nonlinear mapping, and the subfilter closure for LES. This allows all three of these aspects of the model to be co-optimized, generating a model from any source of combustion thermochemical state data. The manifold parameterizing variables are constrained to be linear combinations of species, as in FGM and PCA-based models, to aid in interpretability and implementation. For LES, subfilter variances of the manifold variables are also included as inputs. Two types of a priori analysis are performedmore »
- Authors:
-
- National Renewable Energy Lab. (NREL), Golden, CO (United States)
- Publication Date:
- Research Org.:
- National Renewable Energy Lab. (NREL), Golden, CO (United States)
- Sponsoring Org.:
- USDOE Office of Energy Efficiency and Renewable Energy (EERE), Transportation Office. Vehicle Technologies Office
- OSTI Identifier:
- 1877598
- Report Number(s):
- NREL/JA-2C00-81425
Journal ID: ISSN 0010-2180; MainId:82198;UUID:ef159f2c-3929-41b2-8f0c-89af7a4e8333;MainAdminID:64851
- Grant/Contract Number:
- AC36-08GO28308
- Resource Type:
- Accepted Manuscript
- Journal Name:
- Combustion and Flame
- Additional Journal Information:
- Journal Volume: 244; Journal ID: ISSN 0010-2180
- Publisher:
- Elsevier
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; artificial neural network; machine learning; principal component analysis; reduced-order manifold models
Citation Formats
Perry, Bruce A., Henry de Frahan, Marc T., and Yellapantula, Shashank. Co-optimized machine-learned manifold models for large eddy simulation of turbulent combustion. United States: N. p., 2022.
Web. doi:10.1016/j.combustflame.2022.112286.
Perry, Bruce A., Henry de Frahan, Marc T., & Yellapantula, Shashank. Co-optimized machine-learned manifold models for large eddy simulation of turbulent combustion. United States. https://doi.org/10.1016/j.combustflame.2022.112286
Perry, Bruce A., Henry de Frahan, Marc T., and Yellapantula, Shashank. Sat .
"Co-optimized machine-learned manifold models for large eddy simulation of turbulent combustion". United States. https://doi.org/10.1016/j.combustflame.2022.112286. https://www.osti.gov/servlets/purl/1877598.
@article{osti_1877598,
title = {Co-optimized machine-learned manifold models for large eddy simulation of turbulent combustion},
author = {Perry, Bruce A. and Henry de Frahan, Marc T. and Yellapantula, Shashank},
abstractNote = {Many modeling approaches in large eddy simulation (LES) of turbulent combustion employ a projection of the thermochemical state onto a low-dimensional manifold within state space to reduce the number of transported variables and hence computational cost. Flamelet-generated manifolds (FGM) is an example of a well-established, physics-based approach, but increasingly, principal component analysis (PCA) is being used as a data-driven method for generating manifold models. For both approaches, the nonlinear relationship between the location on the predefined manifold and the outputs of interest, such as reaction rates, can be tabulated or encoded in a neural network. This work proposes a new approach for manifold modeling that extends these existing approaches. A modified neural network structure simultaneously encodes the definition of the manifold variables, the nonlinear mapping, and the subfilter closure for LES. This allows all three of these aspects of the model to be co-optimized, generating a model from any source of combustion thermochemical state data. The manifold parameterizing variables are constrained to be linear combinations of species, as in FGM and PCA-based models, to aid in interpretability and implementation. For LES, subfilter variances of the manifold variables are also included as inputs. Two types of a priori analysis are performed to evaluate the new approach. In the first, the model is trained on data from one-dimensional premixed flames. In this case, the approach recovers the behavior of flamelet-based manifold approaches, and in fact slightly improves performance by identifying an optimized progress variable. The approach is also applied to data from direct numerical simulations of spherical ignition kernels in isotropic turbulence. For any specified manifold dimensionality, the new approach provides substantially lower prediction errors than a PCA-based model developed from the same data set. Additionally, the LES formulation of the new approach can provide accurate predictions for filtered reaction rates across a variety of filter widths.},
doi = {10.1016/j.combustflame.2022.112286},
journal = {Combustion and Flame},
number = ,
volume = 244,
place = {United States},
year = {Sat Jul 16 00:00:00 EDT 2022},
month = {Sat Jul 16 00:00:00 EDT 2022}
}
Works referenced in this record:
Comparative analysis of methods for heat losses in turbulent premixed flames using physically-derived reduced-order manifolds
journal, June 2018
- Nunno, A. Cody; Grenga, Temistocle; Mueller, Michael E.
- Combustion Theory and Modelling, Vol. 23, Issue 1
Training convolutional neural networks to estimate turbulent sub-grid scale reaction rates
journal, May 2019
- Lapeyre, Corentin J.; Misdariis, Antony; Cazard, Nicolas
- Combustion and Flame, Vol. 203
A novel principal component analysis-based acceleration scheme for LES–ODT: An a priori study
journal, May 2013
- Mirgolbabaei, Hessam; Echekki, Tarek
- Combustion and Flame, Vol. 160, Issue 5
The role of differential diffusion during early flame kernel development under engine conditions - part I: Analysis of the heat-release-rate response
journal, November 2020
- Falkenstein, Tobias; Rezchikova, Aleksandra; Langer, Raymond
- Combustion and Flame, Vol. 221
A framework for data-based turbulent combustion closure: A priori validation
journal, August 2019
- Ranade, Rishikesh; Echekki, Tarek
- Combustion and Flame, Vol. 206
Efficient bifurcation and tabulation of multi-dimensional combustion manifolds using deep mixture of experts: An a priori study
journal, January 2021
- Owoyele, Opeoluwa; Kundu, Prithwish; Pal, Pinaki
- Proceedings of the Combustion Institute, Vol. 38, Issue 4
An optimization-based approach to detailed chemistry tabulation: Automated progress variable definition
journal, April 2013
- Niu, Yi-Shuai; Vervisch, Luc; Tao, Pham Dinh
- Combustion and Flame, Vol. 160, Issue 4
Progress-variable approach for large-eddy simulation of non-premixed turbulent combustion
journal, January 1999
- Pierce, Charles D.; Moin, Parviz
- Journal of Fluid Mechanics, Vol. 504
Liminar premixed hydrogen/air counterflow flame simulations using flame prolongation of ILDM with differential diffusion
journal, January 2000
- Gicquel, Olivier; Darabiha, Nasser; Thévenin, Dominique
- Proceedings of the Combustion Institute, Vol. 28, Issue 2
A two mixture fraction flamelet model for large eddy simulation of turbulent flames with inhomogeneous inlets
journal, January 2017
- Perry, Bruce A.; Mueller, Michael E.; Masri, Assaad R.
- Proceedings of the Combustion Institute, Vol. 36, Issue 2
Low-dimensional manifolds in direct numerical simulations of premixed turbulent flames
journal, January 2007
- van Oijen, J. A.; Bastiaans, R. J. M.; de Goey, L. P. H.
- Proceedings of the Combustion Institute, Vol. 31, Issue 1
A comparative study of presumed PDFs for premixed turbulent combustion modeling based on progress variable and its variance
journal, November 2015
- Mukhopadhyay, S.; van Oijen, J. A.; de Goey, L. P. H.
- Fuel, Vol. 159
A nonlinear principal component analysis approach for turbulent combustion composition space
journal, March 2014
- Mirgolbabaei, Hessam; Echekki, Tarek; Smaoui, Nejib
- International Journal of Hydrogen Energy, Vol. 39, Issue 9
Regularization of reaction progress variable for application to flamelet-based combustion models
journal, October 2012
- Ihme, Matthias; Shunn, Lee; Zhang, Jian
- Journal of Computational Physics, Vol. 231, Issue 23
Machine learning for integrating combustion chemistry in numerical simulations
journal, September 2021
- Nguyen, Huu-Tri; Domingo, Pascale; Vervisch, Luc
- Energy and AI, Vol. 5
A filter-independent model identification technique for turbulent combustion modeling
journal, May 2012
- Biglari, Amir; Sutherland, James C.
- Combustion and Flame, Vol. 159, Issue 5
Assumed β-pdf Model for Turbulent Mixing: Validation and Extension to Multiple Scalar Mixing
journal, August 1991
- Girimaji, S. S.
- Combustion Science and Technology, Vol. 78, Issue 4-6
Optimizing progress variable definition in flamelet-based dimension reduction in combustion
journal, November 2015
- Chen, Jing; Liu, Minghou; Chen, Yiliang
- Applied Mathematics and Mechanics, Vol. 36, Issue 11
LES of a premixed jet flame DNS using a strained flamelet model
journal, December 2013
- Knudsen, Edward; Kolla, Hemanth; Hawkes, Evatt R.
- Combustion and Flame, Vol. 160, Issue 12
An a-posteriori evaluation of principal component analysis-based models for turbulent combustion simulations
journal, October 2015
- Biglari, Amir; Sutherland, James C.
- Combustion and Flame, Vol. 162, Issue 10
An a priori evaluation of a principal component and artificial neural network based combustion model in diesel engine conditions
journal, January 2021
- Dalakoti, Deepak K.; Wehrfritz, Armin; Savard, Bruno
- Proceedings of the Combustion Institute, Vol. 38, Issue 2
Artificial Neural Networks for Chemistry Representation in Numerical Simulation of the Flamelet-Based Models for Turbulent Combustion
journal, January 2020
- Zhang, Jiarui; Huang, Honglan; Xia, Zhixun
- IEEE Access, Vol. 8
The reconstruction of thermo-chemical scalars in combustion from a reduced set of their principal components
journal, May 2015
- Mirgolbabaei, Hessam; Echekki, Tarek
- Combustion and Flame, Vol. 162, Issue 5
Modelling of Premixed Laminar Flames using Flamelet-Generated Manifolds
journal, December 2000
- Oijen, J. A. Van; Goey, L. P. H. De
- Combustion Science and Technology, Vol. 161, Issue 1
LES flamelet modeling of a three-stream MILD combustor: Analysis of flame sensitivity to scalar inflow conditions
journal, January 2011
- Ihme, Matthias; See, Yee Chee
- Proceedings of the Combustion Institute, Vol. 33, Issue 1
A filtered tabulated chemistry model for LES of premixed combustion
journal, March 2010
- Fiorina, B.; Vicquelin, R.; Auzillon, P.
- Combustion and Flame, Vol. 157, Issue 3
Adaptive mesh based combustion simulations of direct fuel injection effects in a supersonic cavity flame-holder
journal, October 2021
- Sitaraman, Hariswaran; Yellapantula, Shashank; Henry de Frahan, Marc T.
- Combustion and Flame, Vol. 232
Principal component transport in turbulent combustion: A posteriori analysis
journal, May 2015
- Echekki, Tarek; Mirgolbabaei, Hessam
- Combustion and Flame, Vol. 162, Issue 5
A framework for data-based turbulent combustion closure: A posteriori validation
journal, December 2019
- Ranade, Rishikesh; Echekki, Tarek
- Combustion and Flame, Vol. 210
Effect of multiscalar subfilter PDF models in LES of turbulent flames with inhomogeneous inlets
journal, January 2019
- Perry, Bruce A.; Mueller, Michael E.
- Proceedings of the Combustion Institute, Vol. 37, Issue 2
Problem adapted reduced models based on Reaction–Diffusion Manifolds (REDIMs)
journal, January 2009
- Bykov, V.; Maas, U.
- Proceedings of the Combustion Institute, Vol. 32, Issue 1
Modelling of premixed counterflow flames using the flamelet-generated manifold method
journal, September 2002
- van Oijen, J. A.; de Goey, L. P. H.
- Combustion Theory and Modelling, Vol. 6, Issue 3
Optimal artificial neural networks and tabulation methods for chemistry representation in LES of a bluff-body swirl-stabilized flame
journal, January 2009
- Ihme, Matthias; Schmitt, Christoph; Pitsch, Heinz
- Proceedings of the Combustion Institute, Vol. 32, Issue 1
Combustion modeling using principal component analysis
journal, January 2009
- Sutherland, James C.; Parente, Alessandro
- Proceedings of the Combustion Institute, Vol. 32, Issue 1
Investigation of lengthscales, scalar dissipation, and flame orientation in a piloted diffusion flame by LES
journal, January 2005
- Kempf, A.; Flemming, F.; Janicka, J.
- Proceedings of the Combustion Institute, Vol. 30, Issue 1
Small scales, many species and the manifold challenges of turbulent combustion
journal, January 2013
- Pope, Stephen B.
- Proceedings of the Combustion Institute, Vol. 34, Issue 1
Tabulation of combustion chemistry via Artificial Neural Networks (ANNs): Methodology and application to LES-PDF simulation of Sydney flame L
journal, November 2017
- Franke, Lucas L. C.; Chatzopoulos, Athanasios K.; Rigopoulos, Stelios
- Combustion and Flame, Vol. 185
Deep learning for presumed probability density function models
journal, October 2019
- T. Henry de Frahan, Marc; Yellapantula, Shashank; King, Ryan
- Combustion and Flame, Vol. 208
Combustion modeling using Principal Component Analysis: A posteriori validation on Sandia flames D, E and F
journal, January 2021
- Malik, Mohammad Rafi; Obando Vega, Pedro; Coussement, Axel
- Proceedings of the Combustion Institute, Vol. 38, Issue 2
Numerically accurate computational techniques for optimal estimator analyses of multi-parameter models
journal, February 2018
- Berger, Lukas; Kleinheinz, Konstantin; Attili, Antonio
- Combustion Theory and Modelling, Vol. 22, Issue 3
Principal component analysis of turbulent combustion data: Data pre-processing and manifold sensitivity
journal, February 2013
- Parente, Alessandro; Sutherland, James C.
- Combustion and Flame, Vol. 160, Issue 2
Approximation capabilities of multilayer feedforward networks
journal, January 1991
- Hornik, Kurt
- Neural Networks, Vol. 4, Issue 2
A Constrained Control Approach for the Automated Choice of an Optimal Progress Variable for Chemistry Tabulation
journal, February 2015
- Prüfert, Uwe; Hartl, Sandra; Hunger, Franziska
- Flow, Turbulence and Combustion, Vol. 94, Issue 3
Integrated analysis of CFD simulation data with K-means clustering algorithm for soot formation under varied combustion conditions
journal, May 2019
- Yu, Wenbin; Zhao, Feiyang; Yang, Wenming
- Applied Thermal Engineering, Vol. 153
Empirical low-dimensional manifolds in composition space
journal, October 2013
- Yang, Yue; Pope, Stephen B.; Chen, Jacqueline H.
- Combustion and Flame, Vol. 160, Issue 10
Identification of low-dimensional manifolds in turbulent flames
journal, January 2009
- Parente, A.; Sutherland, J. C.; Tognotti, L.
- Proceedings of the Combustion Institute, Vol. 32, Issue 1
Advanced regression methods for combustion modelling using principal components
journal, June 2015
- Isaac, Benjamin J.; Thornock, Jeremy N.; Sutherland, James
- Combustion and Flame, Vol. 162, Issue 6
From Large-Eddy Simulation to Direct Numerical Simulation of a lean premixed swirl flame: Filtered laminar flame-PDF modeling
journal, July 2011
- Moureau, V.; Domingo, P.; Vervisch, L.
- Combustion and Flame, Vol. 158, Issue 7
Deep learning-based model for progress variable dissipation rate in turbulent premixed flames
journal, January 2021
- Yellapantula, Shashank; Perry, Bruce A.; Grout, Ray W.
- Proceedings of the Combustion Institute, Vol. 38, Issue 2
Multidimensional flamelet-generated manifolds for partially premixed combustion
journal, January 2010
- Nguyen, Phuc-Danh; Vervisch, Luc; Subramanian, Vallinayagam
- Combustion and Flame, Vol. 157, Issue 1
Two-dimensional manifold equations for multi-modal turbulent combustion: Nonpremixed combustion limit and scalar dissipation rates
journal, September 2021
- Novoselov, Alex G.; Perry, Bruce A.; Mueller, Michael E.
- Combustion and Flame, Vol. 231
Assessment of different chemistry reduction methods based on principal component analysis: Comparison of the MG-PCA and score-PCA approaches
journal, June 2016
- Coussement, Axel; Isaac, Benjamin J.; Gicquel, Olivier
- Combustion and Flame, Vol. 168
Application of deep artificial neural networks to multi-dimensional flamelet libraries and spray flames
journal, March 2019
- Owoyele, Opeoluwa; Kundu, Prithwish; Ameen, Muhsin M.
- International Journal of Engine Research, Vol. 21, Issue 1
A physics-based approach to modeling real-fuel combustion chemistry - I. Evidence from experiments, and thermodynamic, chemical kinetic and statistical considerations
journal, July 2018
- Wang, Hai; Xu, Rui; Wang, Kun
- Combustion and Flame, Vol. 193
Modeling of radiation and nitric oxide formation in turbulent nonpremixed flames using a flamelet/progress variable formulation
journal, May 2008
- Ihme, Matthias; Pitsch, Heinz
- Physics of Fluids, Vol. 20, Issue 5
Large-Eddy Simulation of Turbulent Combustion
journal, January 2006
- Pitsch, Heinz
- Annual Review of Fluid Mechanics, Vol. 38, Issue 1
Laminar diffusion flamelet models in non-premixed turbulent combustion
journal, January 1984
- Peters, N.
- Progress in Energy and Combustion Science, Vol. 10, Issue 3
Principal component analysis coupled with nonlinear regression for chemistry reduction
journal, January 2018
- Malik, Mohammad Rafi; Isaac, Benjamin J.; Coussement, Axel
- Combustion and Flame, Vol. 187