Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
169 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

A probabilistic, data-driven closure model for RANS simulations with aleatoric, model uncertainty (2307.02432v2)

Published 5 Jul 2023 in physics.flu-dyn, cs.LG, physics.comp-ph, and stat.ML

Abstract: We propose a data-driven, closure model for Reynolds-averaged Navier-Stokes (RANS) simulations that incorporates aleatoric, model uncertainty. The proposed closure consists of two parts. A parametric one, which utilizes previously proposed, neural-network-based tensor basis functions dependent on the rate of strain and rotation tensor invariants. This is complemented by latent, random variables which account for aleatoric model errors. A fully Bayesian formulation is proposed, combined with a sparsity-inducing prior in order to identify regions in the problem domain where the parametric closure is insufficient and where stochastic corrections to the Reynolds stress tensor are needed. Training is performed using sparse, indirect data, such as mean velocities and pressures, in contrast to the majority of alternatives that require direct Reynolds stress data. For inference and learning, a Stochastic Variational Inference scheme is employed, which is based on Monte Carlo estimates of the pertinent objective in conjunction with the reparametrization trick. This necessitates derivatives of the output of the RANS solver, for which we developed an adjoint-based formulation. In this manner, the parametric sensitivities from the differentiable solver can be combined with the built-in, automatic differentiation capability of the neural network library in order to enable an end-to-end differentiable framework. We demonstrate the capability of the proposed model to produce accurate, probabilistic, predictive estimates for all flow quantities, even in regions where model errors are present, on a separated flow in the backward-facing step benchmark problem.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (87)
  1. On closures for reduced order models—A spectrum of first-principle to machine-learned avenues. Physics of Fluids 33. URL: https://doi.org/10.1063/5.0061577, doi:10.1063/5.0061577. 091301.
  2. Reynolds-Averaged Navier-Stokes Equations for Turbulence Modeling. Applied Mechanics Reviews - APPL MECH REV 62. doi:10.1115/1.3124648.
  3. The fenics project version 1.5. Archive of Numerical Software 3.
  4. Experimental and theoretical investigation of backward-facing step flow. Journal of fluid Mechanics 127, 473–496.
  5. Gaussian markov random field priors for inverse problems. Inverse Problems and Imaging 7, 397–416. doi:10.3934/ipi.2013.7.397.
  6. A fully-differentiable compressible high-order computational fluid dynamics solver. arXiv preprint arXiv:2112.04979 .
  7. An Open-source Adjoint-based Field Inversion Tool for Data-driven RANS Modelling. doi:10.2514/6.2022-4125.
  8. Pyro: Deep universal probabilistic programming. The Journal of Machine Learning Research 20, 973–978.
  9. Pattern recognition and machine learning. volume 4. Springer.
  10. Backward-facing step flows for various expansion ratios at low and moderate reynolds numbers. J. Fluids Eng. 126, 362–374.
  11. Variational inference: A review for statisticians. Journal of the American statistical Association 112, 859–877.
  12. Efficient assimilation of sparse data into rans-based turbulent flow simulations using a discrete adjoint method. Journal of Computational Physics 471, 111667.
  13. Machine learning for fluid mechanics. Annual Review of Fluid Mechanics 52, 477–508.
  14. Discovering governing equations from data by sparse identification of nonlinear dynamical systems. Proceedings of the national academy of sciences 113, 3932–3937. Publisher: National Acad Sciences.
  15. The frontier of simulation-based inference. Proceedings of the National Academy of Sciences 117, 30055–30062.
  16. Newton methods for nonlinear problems: affine invariance and adaptive algorithms. volume 35. Springer Science & Business Media.
  17. Finite element methods for flow problems. John Wiley & Sons.
  18. Perspectives on Machine Learning-augmented Reynolds-averaged and Large Eddy Simulation Models of Turbulence. Physical Review Fluids 6, 050504. URL: http://arxiv.org/abs/2009.10675, doi:10.1103/PhysRevFluids.6.050504. arXiv:2009.10675 [physics].
  19. Turbulence modeling in the age of data. Annual Review of Fluid Mechanics 51, 357–377.
  20. Bayesian estimates of parameter variability in the k–ε𝜀\varepsilonitalic_ε turbulence model. Journal of Computational Physics 258, 73–94.
  21. Data-free and data-driven rans predictions with quantified uncertainty. Flow, Turbulence and Combustion 100, 593–616.
  22. Modeling of structural uncertainties in reynolds-averaged navier-stokes closures. Physics of Fluids 25, 110822.
  23. Physics-constrained, data-driven discovery of coarse-grained dynamics. Communications in Computational Physics 25, 1259–1301. doi:10.4208/cicp.OA-2018-0174.
  24. The galerkin gradient least-squares method. Computer Methods in Applied Mechanics and Engineering 74, 41–54.
  25. Quantifying model form uncertainty in reynolds-averaged turbulence models with bayesian deep neural networks. Journal of Computational Physics 383, 125–147.
  26. Algorithm developments for discrete adjoint methods. AIAA journal 41, 198–205.
  27. An introduction to the adjoint approach to design. Flow, turbulence and combustion 65, 393–415.
  28. A framework for epistemic uncertainty quantification of turbulent scalar flux models for reynolds-averaged navier-stokes simulations. Physics of Fluids 25, 055105.
  29. Is the steady viscous incompressible two-dimensional flow over a backward-facing step at re= 800 stable? International Journal for Numerical Methods in Fluids 17, 501–541.
  30. Vcnn-e: A vector-cloud neural network with equivariance for emulating reynolds stress transport equations. arXiv preprint arXiv:2201.01287 .
  31. Stochastic variational inference. Journal of Machine Learning Research .
  32. Towards Integrated Field Inversion and Machine Learning With Embedded Neural Networks for RANS Modeling, in: AIAA Scitech 2019 Forum. American Institute of Aeronautics and Astronautics. URL: https://arc.aiaa.org/doi/abs/10.2514/6.2019-1884, doi:10.2514/6.2019-1884. _eprint: https://arc.aiaa.org/doi/pdf/10.2514/6.2019-1884.
  33. Eigenspace perturbations for uncertainty estimation of single-point turbulence closures. Physical Review Fluids 2. doi:10.1103/PhysRevFluids.2.024605.
  34. Aerodynamic design via control theory. Journal of scientific computing 3, 233–260.
  35. Openfoam: A c++ library for complex physics simulations, in: International workshop on coupled methods in numerical dynamics, pp. 1–20.
  36. Data-driven modelling of the reynolds stress tensor using random forests with invariance. Computers & Fluids 202, 104497.
  37. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 .
  38. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114 .
  39. Machine learning–accelerated computationalfluid dynamics .
  40. Big data and predictive computational modeling. Journal of Computational Physics 321, 1252–1254.
  41. Progress in the development of a reynolds-stress turbulence closure. Journal of fluid mechanics 68, 537–566.
  42. Application of the energy-dissipation model of turbulence to the calculation of flow near a spinning disc. Letters in Heat and Mass Transfer 1, 131–137. doi:10.1016/0094-4548(74)90150-7.
  43. Machine learning strategies for systems with invariance properties. Journal of Computational Physics 318, 22–35. URL: http://dx.doi.org/10.1016/j.jcp.2016.05.003, doi:10.1016/j.jcp.2016.05.003. publisher: Elsevier Inc.
  44. Reynolds averaged turbulence modelling using deep neural networks with embedded invariance. Journal of Fluid Mechanics 807, 155–166.
  45. Learned turbulence modelling with differentiable fluid solvers. arXiv preprint arXiv:2202.06988 .
  46. Simple computational strategies for more effective physics-informed neural networks modeling of turbulent natural convection. Journal of Computational Physics 456, 111022. URL: https://doi.org/10.1016/j.jcp.2022.111022, doi:10.1016/j.jcp.2022.111022. publisher: Elsevier Inc.
  47. Leonardo da vinci and fluid mechanics. Annual Review of Fluid Mechanics 53, 1–25. doi:10.1146/annurev-fluid-022620-122816.
  48. Uncertainty estimation for reynolds-averaged navier-stokes predictions of high-speed aircraft nozzle jets. AIAA Journal 55, 3999–4004. doi:10.2514/1.J056059.
  49. dolfin-adjoint 2018.1: automated adjoints for fenics and firedrake. Journal of Open Source Software 4, 1292.
  50. High reynolds number flow over a backward-facing step: structure of the mean separation bubble. Experiments in fluids 55, 1–22.
  51. Bayesian learning for neural networks. volume 118. Springer Science & Business Media.
  52. Subgrid-scale stress modelling based on the square of the velocity gradient tensor. Flow, turbulence and Combustion 62, 183–200.
  53. Uncertainty quantification for rans turbulence model predictions, in: APS division of fluid dynamics meeting abstracts, pp. LC–004.
  54. Bayesian uncertainty quantification applied to RANS turbulence models, in: Journal of Physics: Conference Series, Institute of Physics Publishing. p. 042032. URL: https://iopscience.iop.org/article/10.1088/1742-6596/318/4/042032, doi:10.1088/1742-6596/318/4/042032.
  55. Quantification of turbulence modeling uncertainties using full field inversion, in: 22nd AIAA Computational Fluid Dynamics Conference, p. 2459.
  56. A paradigm for data-driven predictive modeling using field inversion and machine learning. Journal of Computational Physics 305, 758–774.
  57. Non-markovian closure models for large eddy simulations using the mori-zwanzig formalism. Physical Review Fluids 2, 014604.
  58. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems 32.
  59. Turbulence Modeling for Physics-Informed Neural Networks: Comparison of Different RANS Models for the Backward-Facing Step Flow. Fluids 8, 43. URL: https://www.mdpi.com/2311-5521/8/2/43, doi:10.3390/fluids8020043. number: 2 Publisher: Multidisciplinary Digital Publishing Institute.
  60. A more general effective-viscosity hypothesis. Journal of Fluid Mechanics 72, 331–340.
  61. Turbulent flows. Cambridge university press.
  62. Neural network closures for nonlinear model order reduction URL: https://arxiv.org/abs/1705.08532v1, doi:10.48550/arXiv.1705.08532.
  63. Discovery of algebraic reynolds-stress models using sparse symbolic regression. Flow, Turbulence and Combustion 104, 579–603.
  64. Machine-learning-augmented predictive modeling of turbulent separated flows over airfoils. AIAA journal 55, 2215–2227.
  65. CFD Vision 2030 Study: A Path to Revolutionary Computational Aerosciences. NNASA/CR-2014-218178 .
  66. Reduced Order Model Closures: A Brief Tutorial. arXiv:2202.14017 [physics] URL: http://arxiv.org/abs/2202.14017. arXiv: 2202.14017.
  67. A comprehensive overview of a non-parametric probabilistic approach of model uncertainties for predictive models in structural dynamics. Journal of sound and vibration 288, 623–652.
  68. On nonlinear K-l and K-ϵitalic-ϵ\epsilonitalic_ϵ models of turbulence. Journal of Fluid Mechanics 178, 459–475. doi:10.1017/S0022112087001319.
  69. End-to-end differentiable learning of turbulence models from indirect observations. arXiv preprint arXiv:2104.04821 .
  70. Turbulence closure modeling with data-driven techniques: physical compatibility and consistency considerations. New Journal of Physics 22, 093023.
  71. Eigenvector perturbation methodology for uncertainty quantification of turbulence models. Physical Review Fluids 4, 044603.
  72. A methodology to evaluate statistical errors in dns data of plane channel flows. Computers & Fluids 130, 1–7.
  73. The Relevance Vector Machine, in: Solla, S.A., Leen, T.K., Müller, K.R. (Eds.), Advances in Neural Information Processing Systems 12, MIT Press. pp. 652–658.
  74. A machine learning strategy to assist turbulence model development, in: 53rd AIAA aerospace sciences meeting, p. 1287.
  75. Solver-in-the-Loop: Learning from Differentiable Physics to Interact with Iterative PDE-Solvers. arXiv:2007.00016 [physics] URL: http://arxiv.org/abs/2007.00016. arXiv: 2007.00016.
  76. Emerging trends in machine learning for computational fluid dynamics. URL: http://arxiv.org/abs/2211.15145. arXiv:2211.15145 [physics].
  77. A comprehensive physics-informed machine learning framework for predictive turbulence modeling. arXiv preprint arXiv:1701.07102 .
  78. Physics-informed machine learning approach for reconstructing reynolds stress modeling discrepancies based on dns data. Physical Review Fluids 2, 034603.
  79. Formulation of the k-ω𝜔\omegaitalic_ω turbulence model revisited, in: AIAA Journal, pp. 2823–2838. URL: https://arc.aiaa.org/doi/abs/10.2514/1.36541, doi:10.2514/1.36541.
  80. Turbulence modeling for CFD. volume 2. DCW industries La Canada, CA.
  81. Sparse bayesian structure learning with dependent relevance determination prior, in: Proceedings of the 27th International Conference on Neural Information Processing Systems - Volume 1, MIT Press, Cambridge, MA, USA. p. 1628–1636.
  82. Reynolds-averaged navier–stokes equations with explicit data-driven reynolds stress closure can be ill-conditioned. Journal of Fluid Mechanics 869, 553–586.
  83. A priori assessment of prediction confidence for data-driven turbulence modeling. Flow, Turbulence and Combustion 99, 25–46.
  84. Quantification of model uncertainty in rans simulations: A review. Progress in Aerospace Sciences 108, 1–31.
  85. Quantifying and reducing model-form uncertainties in reynolds-averaged navier–stokes simulations: A data-driven, physics-informed bayesian approach. Journal of Computational Physics 324, 115–136.
  86. Ensemble Kalman method for learning turbulence models from indirect observation data. Journal of Fluid Mechanics 949, A26. doi:10.1017/jfm.2022.744. publisher: Cambridge University Press.
  87. Machine learning methods for data-driven turbulence modeling, in: 22nd AIAA computational fluid dynamics conference, p. 2460.
Citations (11)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now