Comparison of reduced order models based on dynamic mode decomposition and deep learning for predicting chaotic flow in a random arrangement of cylinders

Three reduced order models are evaluated in their capacity to predict the future state of an unsteady chaotic flow field. A spatially fully developed flow generated in a random packing of cylinders at a solid fraction of 0.1 and a nominal Reynolds number of 50 is investigated. For deep learning (DL), convolutional autoencoders are used to reduce the high-dimensional data to lower dimensional latent space representations of size 16, which were then used for training the temporal architectures. To predict the future states, two DL based methods, long short-term memory and temporal convolutional neural networks, are used and compared to the linear dynamic mode decomposition (DMD). The predictions are tested in their capability to predict the spatiotemporal variations of velocity and pressure, flow statistics such as root mean squared values, and the capability to predict fluid forces on the cylinders. Relative errors between 15% and 20% are evident in predicting instantaneous velocities, chiefly resulting from phase differences between predictions and ground truth. The spatial distribution of statistical second moments is predicted to be within a maximum of 5%–10% of the ground truth with mean error in the range of 1%–2%. Using the predicted fields, instantaneous fluid drag force predictions on individual particles exhibit a mean relative error within 20%, time-averaged drag force predictions to within 5%, and total drag force over all particles to within 1% of the ground truth values. It is found that overall, the non-linear DL models are more accurate than the linear DMD algorithm for the prediction of future states.

[1]  Peter J. Baddoo,et al.  Physics-informed dynamic mode decomposition , 2023, Proceedings of the Royal Society A.

[2]  Mengjia Wang,et al.  A hierarchical autoencoder and temporal convolutional neural network reduced-order model for the turbulent wake of a three-dimensional bluff body , 2023, Physics of Fluids.

[3]  A. Mohan,et al.  Embedding hard physical constraints in neural network coarse-graining of three-dimensional turbulence , 2023, Physical Review Fluids.

[4]  Hui Li,et al.  DeepTRNet: Time-resolved reconstruction of flow around a circular cylinder via spatiotemporal deep neural networks , 2022, Physics of Fluids.

[5]  Qiuwan Wang,et al.  Predictions of flow and temperature field in a T-junction based on dynamic mode decomposition and deep learning , 2022, Energy.

[6]  Soledad Le Clainche,et al.  A predictive hybrid reduced order model based on proper orthogonal decomposition combined with deep learning architectures , 2022, Expert Syst. Appl..

[7]  R. Jaiman,et al.  Three-dimensional deep learning-based reduced order model for unsteady flow dynamics with variable Reynolds number , 2021, Physics of Fluids.

[8]  Yoshihide Tominaga,et al.  Multi-fidelity shape optimization methodology for pedestrian-level wind environment , 2021 .

[9]  Peter J. Schmid,et al.  Dynamic Mode Decomposition and Its Variants , 2021, Annual Review of Fluid Mechanics.

[10]  Wendong Liang,et al.  Decomposition of unsteady sheet/cloud cavitation dynamics in fluid-structure interaction via POD and DMD methods , 2021 .

[11]  M. Krajecki,et al.  Convolutional neural networks and temporal CNNs for COVID-19 forecasting in France , 2021, Applied Intelligence.

[12]  Mingsheng Long,et al.  MotionRNN: A Flexible Model for Video Prediction with Spacetime-Varying Motions , 2021, 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[13]  Kai Fukami,et al.  Convolutional neural networks for fluid flow analysis: toward effective metamodeling and low dimensionalization , 2021, Theoretical and Computational Fluid Dynamics.

[14]  Koji Fukagata,et al.  CNN-LSTM based reduced order modeling of two-dimensional unsteady flows around a circular cylinder at different Reynolds numbers , 2020, Fluid Dynamics Research.

[15]  Muhammad Arsalan,et al.  Novel Methods for Production Data Forecast Utilizing Machine Learning and Dynamic Mode Decomposition , 2020 .

[16]  Koji Fukagata,et al.  Convolutional neural network and long short-term memory based reduced order surrogate for minimal turbulent channel flow , 2020, Physics of Fluids.

[17]  Koji Fukagata,et al.  Sparse identification of nonlinear dynamics with low-dimensionalized flow representations , 2020, Journal of Fluid Mechanics.

[18]  Allan Ross Magee,et al.  Assessment of unsteady flow predictions using hybrid deep learning based reduced-order models , 2020, 2009.04396.

[19]  J. Ringwood,et al.  Boundary element and integral methods in potential flow theory: a review with a focus on wave energy applications , 2020, Journal of Ocean Engineering and Marine Energy.

[20]  R. Maulik,et al.  Latent-space time evolution of non-intrusive reduced-order models using Gaussian process emulation , 2020, Physica D: Nonlinear Phenomena.

[21]  Hadi Veisi,et al.  Deep Neural Networks for Nonlinear Model Order Reduction of Unsteady Flows , 2020, Physics of Fluids.

[22]  Kai Fukami,et al.  Convolutional neural network based hierarchical autoencoder for nonlinear mode decomposition of fluid field data , 2020, Physics of Fluids.

[23]  Albert Y. Zomaya,et al.  Temporal Convolutional Networks for the Advance Prediction of ENSO , 2020, Scientific Reports.

[24]  Kai Fukami,et al.  Probabilistic neural networks for fluid flow surrogate modeling and data recovery , 2020 .

[25]  Kurt L. Mudie,et al.  Lower Limb Kinematics Trajectory Prediction Using Long Short-Term Memory Neural Networks , 2020, Frontiers in Bioengineering and Biotechnology.

[26]  Takaaki Murata,et al.  Machine-learning-based reduced-order modeling for unsteady flows around bluff bodies of various shapes , 2020, Theoretical and Computational Fluid Dynamics.

[27]  José C. Riquelme,et al.  Temporal Convolutional Networks Applied to Energy-Related Time Series Forecasting , 2020, Applied Sciences.

[28]  Wenjie Zhang,et al.  Data-driven reduced order model with temporal convolutional neural network , 2020 .

[29]  Prasanna Balaprakash,et al.  Reduced-order modeling of advection-dominated systems with recurrent neural networks and convolutional autoencoders , 2020, Physics of Fluids.

[30]  K. Taira,et al.  Assessment of supervised machine learning methods for fluid flows , 2020, Theoretical and Computational Fluid Dynamics.

[31]  Jian-Xun Wang,et al.  Physics-constrained bayesian neural network for fluid flow reconstruction with sparse and noisy data , 2020, Theoretical and Applied Mechanics Letters.

[32]  M. P. Brenner,et al.  Perspective on machine learning for advancing fluid mechanics , 2019, Physical Review Fluids.

[33]  Jun Wang,et al.  Multivariate Temporal Convolutional Network: A Deep Neural Networks Approach for Multivariate Time Series Forecasting , 2019, Electronics.

[34]  Gang Chen,et al.  A novel spatial-temporal prediction method for unsteady wake flows based on hybrid deep neural network , 2019, Physics of Fluids.

[35]  Ming Liu,et al.  Dynamic mode decomposition of gas-liquid flow in a rotodynamic multiphase pump , 2019, Renewable Energy.

[36]  Kai Fukami,et al.  Nonlinear mode decomposition with convolutional neural networks for fluid dynamics , 2019, Journal of Fluid Mechanics.

[37]  Petros Koumoutsakos,et al.  Machine Learning for Fluid Mechanics , 2019, Annual Review of Fluid Mechanics.

[38]  Hossein Azizpour,et al.  Predictions of turbulent shear flows using deep neural networks , 2019, Physical Review Fluids.

[39]  Xuelong Li,et al.  A CNN-RNN architecture for multi-label weather recognition , 2018, Neurocomputing.

[40]  D. Tafti,et al.  Investigation of drag, lift and torque for fluid flow past a low aspect ratio (1:4) cylinder , 2018, Computers & Fluids.

[41]  Koji Fukagata,et al.  Synthetic turbulent inflow generator using machine learning , 2018, Physical Review Fluids.

[42]  An-Shik Yang,et al.  Optimization procedures for enhancement of city breathability using arcade design in a realistic high-rise urban area , 2017 .

[43]  Dit-Yan Yeung,et al.  Deep Learning for Precipitation Nowcasting: A Benchmark and A New Model , 2017, NIPS.

[44]  Danesh K. Tafti,et al.  Evaluation of drag correlations using particle resolved simulations of spheres and ellipsoids in assembly , 2017 .

[45]  Zhiyong Luo,et al.  Combination of Convolutional and Recurrent Neural Network for Sentiment Analysis of Short Texts , 2016, COLING.

[46]  Gregory D. Hager,et al.  Temporal Convolutional Networks for Action Segmentation and Detection , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[47]  Gregory D. Hager,et al.  Temporal Convolutional Networks: A Unified Approach to Action Segmentation , 2016, ECCV Workshops.

[48]  Vladlen Koltun,et al.  Multi-Scale Context Aggregation by Dilated Convolutions , 2015, ICLR.

[49]  Kai Chen,et al.  A LSTM-based method for stock returns prediction: A case study of China stock market , 2015, 2015 IEEE International Conference on Big Data (Big Data).

[50]  S. Brunton,et al.  Discovering governing equations from data by sparse identification of nonlinear dynamical systems , 2015, Proceedings of the National Academy of Sciences.

[51]  Steven L. Brunton,et al.  Multiresolution Dynamic Mode Decomposition , 2015, SIAM J. Appl. Dyn. Syst..

[52]  Marie-Francine Moens,et al.  A survey on the application of recurrent neural networks to statistical language modeling , 2015, Comput. Speech Lang..

[53]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[54]  Danesh K. Tafti,et al.  A new approach for conjugate heat transfer problems using immersed boundary method for curvilinear grid based solvers , 2014, J. Comput. Phys..

[55]  Jonathan H. Tu,et al.  On dynamic mode decomposition: Theory and applications , 2013, 1312.0041.

[56]  S. M. Ghiaasiaan,et al.  The effect of flow pulsation on drag and heat transfer in an array of heated square cylinders , 2013 .

[57]  P. Schmid,et al.  Applications of the dynamic mode decomposition , 2011 .

[58]  Sin Chien Siw,et al.  Effects of Pin Detached Space on Heat Transfer and From Pin Fin Arrays , 2010 .

[59]  Geoffrey E. Hinton,et al.  Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.

[60]  H. Mu,et al.  Numerical simulation of pollutant flow and dispersion in different street layouts , 2010 .

[61]  P. Schmid,et al.  Dynamic mode decomposition of numerical and experimental data , 2008, Journal of Fluid Mechanics.

[62]  Fue-Sang Lien,et al.  Partially resolved numerical simulation and RANS modeling of flow and passive scalar transport in an urban environment , 2008 .

[63]  M. D. Lemos,et al.  A Correlation for Interfacial Heat Transfer Coefficient for Turbulent Flow Over an Array of Square Rods , 2006 .

[64]  Michele Milano,et al.  Neural network modeling for near wall turbulent flow , 2002 .

[65]  D. Tafti.  GenIDLEST: A Scalable Parallel Computational Tool for Simulating Complex Turbulent Flows , 2001, Fluids Engineering.

[66]  Kwan-Soo Lee,et al.  Optimal shape and arrangement of staggered pins in the channel of a plate heat exchanger , 2001 .

[67]  Agnieszka J. Klemm,et al.  Multicriteria optimisation of the building arrangement with application of numerical simulation , 2000 .

[68]  S. Hochreiter,et al.  Long Short-Term Memory , 1997, Neural Computation.

[69]  M. Chyu,et al.  Heat transfer on the base surface of threedimensional protruding elements , 1996 .

[70]  Ching-Jen Chen,et al.  Finite Analytic Solution of Convective Heat Transfer for Tube Arrays in Crossflow: Part I—Flow Field Analysis , 1989 .

[71]  Geoffrey E. Hinton,et al.  Learning internal representations by error propagation , 1986 .

[72]  Andreas Acrivos,et al.  Slow flow past periodic arrays of cylinders with application to heat transfer , 1982 .

[73]  Guangwei Bai,et al.  Deep Temporal Convolutional Networks for Short-Term Traffic Flow Forecasting , 2019, IEEE Access.

[74]  Bert Blocken,et al.  CFD simulation of near-field pollutant dispersion on a high-resolution grid : a case study by LES and RANS for a building group in downtown Montreal , 2011 .