- Physical Review Journals
All JournalsPhysics Magazine
Access Provided by Xinjiang University

Physical Review Fluids

Export citation

Export citation

Choose format for download:

Download Citation
  • Access by Xinjiang University

Data-driven turbulent heat flux modeling with inputs of multiple fidelity

Matilde Fiore1,*, Enrico Saccaggi1,2, Lilla Koloszar1, Yann Bartosiewicz3, and Miguel A. Mendez1

  • 1Environmental and Applied Fluid Dynamics Department, von Karman Institute for Fluid Dynamics, Belgium, Waterloosesteenweg 72, 1640 Sint-Genesius-Rode
  • 2Department of Mechanical and Aerospace Engineering, Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, Italy
  • 3Institute of Mechanics, Materials and Civil Engineering (IMMC), Universite catholique de Louvain (UCLouvain), Place du Levant 2, 1348 Louvain-la-Neuve, Belgium

  • *Contact author: matilde.fiore@vki.ac.be
PDF

Phys. Rev. Fluids 10, 034606 – Published 17 March, 2025

DOI: https://doi.org/10.1103/PhysRevFluids.10.034606

Abstract

Data-driven RANS modeling is emerging as a promising methodology to exploit the information provided by high-fidelity data. However, its widespread application is limited by challenges in generalization and robustness to inconsistencies between input data of varying fidelity levels. This is especially true for thermal turbulent closures, which inherently depend on momentum statistics provided by low- or high-fidelity turbulence momentum models. This work investigates the impact of momentum modeling inconsistencies on a data-driven thermal closure trained with a dataset with multiple fidelity (DNS and RANS). The analysis of the model inputs shows that the two fidelity levels correspond to separate regions in the input space. It is here demonstrated that such separation can be exploited by training with heterogeneous data, allowing the model to detect the level of fidelity in its inputs and adjust its prediction accordingly. In particular, sensitivity analysis and verification show that such a model can leverage data inconsistencies to increase its robustness. Finally, the verification with a CFD simulation shows the potential of this multifidelity training approach for flows in which momentum statistics provided by traditional models are affected by model uncertainties.

View figure in article

Physics Subject Headings (PhySH)

Article Text

References (40)

  1. J. Ling, A. Kurzawski, and J. Templeton, Reynolds averaged turbulence modelling using deep neural networks with embedded invariance, J. Fluid Mech. 807, 155 (2016).
  2. C. Jiang, R. Vinuesa, R. Chen, J. Mi, S. Laima, and H. Li, An interpretable framework of data-driven turbulence modeling using deep neural networks, Phys. Fluids 33, 055133 (2021).
  3. Y. Yin, Z. Shen, Y. Zhang, H. Chen, and S. Fu, An iterative data-driven turbulence modeling framework based on Reynolds stress representation, Theor. Appl. Mech. Lett. 12, 100381 (2022).
  4. R. McConkey, E. Yee, and F.-S. Lien, On the generalizability of data-driven turbulence closure models, Int. J. Comput. Fluid Dynam. 36, 555 (2022).
  5. N. Geneva and N. Zabaras, Quantifying model form uncertainty in Reynolds-averaged turbulence models with Bayesian deep neural networks, J. Comput. Phys. 383, 125 (2019).
  6. S. Berrone and D. Oberto, An invariances-preserving vector basis neural network for the closure of Reynolds-averaged Navier–Stokes equations by the divergence of the Reynolds stress tensor, Phys. Fluids 34, 095136 (2022).
  7. J. Cai, P.-E. Angeli, J.-M. Martinez, G. Damblin, and D. Lucor, Revisiting Tensor Basis Neural Network for Reynolds stress modeling: Application to plane channel and square duct flows, Comput. Fluids 275, 106246 (2024).
  8. C. Sotgiu, B. Weigand, K. Semmler, and P. Wellinger, Towards a general data-driven explicit algebraic Reynolds stress prediction framework, Int. J. Heat Fluid Flow 79, 108454 (2019).
  9. Y. Zhang, D. Zhang, and H. Jiang, Review of challenges and opportunities in turbulence modeling: A comparative analysis of data-driven machine learning approaches, J. Mar. Sci. Eng. 11, 1440 (2023).
  10. V. Srivastava and K. Duraisamy, Generalizable physics-constrained modeling using learning and inference assisted by feature-space engineering, Phys. Rev. Fluids 6, 124602 (2021).
  11. P. M. Milani et al., A machine learning approach for determining the turbulent diffusivity in film cooling flows, J. Turbomach. 140, 021006 (2018).
  12. Y. Zhao, H. D. Akolekar, J. Weatheritt, V. Michelassi, and R. D. Sandberg, RANS turbulence model development using CFD-driven machine learning, J. Comput. Phys. 411, 109413 (2020).
  13. S. S. Girimaji, Turbulence closure modeling with machine learning: A foundational physics perspective, New J. Phys. 26, 071201 (2024).
  14. H. Mandler and B. Weigand, On frozen-rans approaches in data-driven turbulence modeling: Practical relevance of turbulent scale consistency during closure inference and application, Int. J. Heat Fluid Flow 97, 109017 (2022).
  15. M. Schmelzer, R. P. Dwight, and P. Cinnella, Discovery of algebraic Reynolds-stress models using sparse symbolic regression, Flow, Turbul. Combust. 104, 579 (2020).
  16. J. Weatheritt and R. D. Sandberg, The development of algebraic stress models using a novel evolutionary algorithm, Int. J. Heat Fluid Flow 68, 298 (2017).
  17. M. Fiore, L. Koloszar, C. Fare, M. A. Mendez, M. Duponcheel, and Y. Bartosiewicz, Physics-constrained machine learning for thermal turbulence modelling at low Prandtl numbers, Int. J. Heat Mass Transf. 194, 122998 (2022).
  18. B. J. Daly and F. H. Harlow, Transport equations in turbulence, Phys. Fluids 13, 2634 (1970).
  19. K. Suga and K. Abe, Nonlinear eddy viscosity modelling for turbulence and heat transfer near wall and shear-free boundaries, Int. J. Heat Fluid Flow 21, 37 (2000).
  20. P. Wikström, S. Wallin, and A. V. Johansson, Derivation and investigation of a new explicit algebraic model for the passive scalar flux, Phys. Fluids 12, 688 (2000).
  21. S. Manservisi and F. Menghini, A CFD four parameter heat transfer turbulence model for engineering applications in heavy liquid metals, Int. J. Heat Mass Transf. 69, 312 (2014).
  22. S. Manservisi and F. Menghini, CFD simulations in heavy liquid metal flows for square lattice bare rod bundle geometries with a four parameter heat transfer turbulence model, Nucl. Eng. Des. 295, 251 (2015).
  23. A. Shams and A. De Santis, Towards the accurate prediction of the turbulent flow and heat transfer in low-Prandtl fluids, Int. J. Heat Mass Transf. 130, 290 (2019).
  24. M. Fiore, L. Koloszar, M. A. Mendez, M. Duponcheel, and Y. Bartosiewicz, Turbulent heat flux modelling in forced convection flows using artificial neural networks, Nucl. Eng. Des. 399, 112005 (2022).
  25. W. M. Kays, Turbulent Prandtl number. Where are we? ASME Trans. J. Heat Transf. 116, 284 (1994).
  26. R. Manceau and K. Hanjalić, Elliptic blending model: A new near-wall Reynolds-stress turbulence closure, Phys. Fluids 14, 744 (2002).
  27. H. Kawamura, K. Ohsaka, H. Abe, and K Yamamoto, DNS of turbulent heat transfer in channel flow with low to medium-high Prandtl number fluid, Int. J. Heat Fluid Flow 19, 482 (1998).
  28. I. Tiselj, R. Bergant, B. Mavko, I. Bajsić, and G. Hetsroni, DNS of turbulent heat transfer in channel flow with heat conduction in the solid wall, J. Heat Transfer 123, 849 (2001).
  29. J. Oder, A. Shams, L. Cizelj, and I. Tiselj, Direct numerical simulation of low-Prandtl fluid flow over a confined backward facing step, Int. J. Heat Mass Transf. 142, 118436 (2019).
  30. M. Duponcheel and Y. Bartosiewicz, Direct numerical simulation of turbulent heat transfer at low Prandtl numbers in planar impinging jets, Int. J. Heat Mass Transf. 173, 121179 (2021).
  31. OpenFOAM v6, https://www.openfoam.com/releases/openfoam-v6.
  32. M. Duponcheel, L. Bricteux, M. Manconi, G. Winckelmans, and Y. Bartosiewicz, Assessment of RANS and improved near-wall modeling for forced convection at low Prandtl numbers based on LES up to Reτ=2000, Int. J. Heat Mass Transf. 75, 470 (2014).
  33. S. Banerjee, R. Krahl, F. Durst, and C. Zenger, Presentation of anisotropy properties of turbulence, invariants versus eigenvalue approaches, J. Turbul. 8, N32 (2007).
  34. C. M. Bishop and M. N. Nasser, Pattern Recognition and Machine Learning, Vol. 4 (Springer, New York, 2006).
  35. S. M. Lundberg and S.-I. Lee, A unified approach to interpreting model predictions, Adv. Neural Info. Proc. Syst. 30, 4768 (2017).
  36. M. Sundararajan, A. Taly, and Q. Yan, Axiomatic attribution for deep networks, International Conference on Machine Learning (PMLR, 2017).
  37. R. Maulik, D. K. Fytanidis, B. Lusch, V. Vishwanath, and S. Patel, PythonFOAM: In-situ data analyses with OpenFOAM and Python, J. Comput. Sci. 62, 101750 (2022).
  38. M. Fiore, Thermal turbulence modeling of heavy liquid metals via data analysis and machine learning, Ph.D. thesis, Université catholique de Louvain, Louvain-la-Neuve, Belgium, 2023.
  39. G. Grötzbach, Challenges in low-Prandtl number heat transfer simulation and modelling, Nucl. Eng. Des. 264, 41 (2013).
  40. Note that the sign of the coefficients ai does not alter the final output, since A=∑aiTi is the Cholesky factorization of the symmetric part of D, as indicated by Eq. (6).

Outline

Information

Phys. Rev. Fluids 10, 034606– Published 17 March, 2025
  • Received 16 August 2024
  • Accepted 28 January 2025
Crossmark - Check for updates button

©2025 American Physical Society

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation