Real-time simulation of parameter-dependent fluid flows through deep learning-based reduced order models

  title={Real-time simulation of parameter-dependent fluid flows through deep learning-based reduced order models},
  author={Stefania Fresca and Andrea Manzoni},
Simulating fluid flows in different virtual scenarios is of key importance in engineering applications. However, high-fidelity, full-order models relying, e.g., on the finite element method, are unaffordable whenever fluid flows must be simulated in almost real-time. Reduced order models (ROMs) relying, e.g., on proper orthogonal decomposition (POD) provide reliable approximations to parameter-dependent fluid dynamics problems in rapid times. However, they might require expensive hyper… 

POD-Enhanced Deep Learning-Based Reduced Order Models for the Real-Time Simulation of Cardiac Electrophysiology in the Left Atrium

It is shown that performing a prior dimensionality reduction on FOM snapshots through randomized proper orthogonal decomposition (POD) enables to speed up training times and to decrease networks complexity.

Three-dimensional deep learning-based reduced order model for unsteady flow dynamics with variable Reynolds number

The proposed DL-ROM framework aligns with the development of a digital twin for 3D unsteady flow field and instantaneous force predictions with variable Re-based effects and an assessment of the computing requirements in terms of the memory usage, training, and testing cost of the 3D CRAN framework.

Long-time prediction of nonlinear parametrized dynamical systems by deep learning-based reduced order models

This work aims at taking a further step towards the use of DL algorithms for the efficient numerical approximation of parametrized PDEs by introducing the μt-POD-LSTMROM framework, a novel technique extending the POD-DL-ROM framework by adding a two-fold architecture taking advantage of long short-term memory cells.

Multi-fidelity surrogate modeling using long short-term memory networks

A novel data-driven framework of multi-fidelity surrogate modeling for parametrized, time-dependent problems using long short-term memory (LSTM) networks, to enhance output predictions both for unseen parameter values and forward in time simultaneously – a task known to be particularly challenging for data- driven models.

Predicting parametric spatiotemporal dynamics by multi-resolution PDE structure-preserved deep learning

A novel PDE-preserved neural network (PPNN) for rapidly predicting parametric spatiotemporal dynamics, given the governing PDEs are (partially) known, and this physics-inspired learning architecture design endows PPNN with excellent generalizability and long-term prediction accuracy compared to the state-of-the-art blackbox ConvResNet baseline.

A Direct-Forcing Immersed Boundary Method for Incompressible Flows Based on Physics-Informed Neural Network

The application of physics-informed neural networks (PINNs) to computational fluid dynamics simulations has recently attracted tremendous attention. In the simulations of PINNs, the collocation

Proper Orthogonal Decomposition for Fluid Mechanics Problems

  • 2021
The goal of this thesis is the reduced order modelling (ROM) of fluid mechanics problems by using the proper orthogonal decomposition (POD). The main motivation of this work is the need for real-time

Hybridized Discontinuous Galerkin Methods for a Multiple Network Poroelasticity Model with Medical Applications

A new e-cient parameter-robust hybridized discontinuous Galerkin method, which also features first introduced in the context of geomechanics, is proposed for the MPET model and cost-efficient fast parameter-Robust preconditioners are derived.

A Galerkin/POD Reduced-Order Model from Eigenfunctions of Non-Converged Time Evolution Solutions in a Convection Problem

A Galerkin/POD reduced-order model from eigenfunctions of non-converged time evolution transitory states in a problem of Rayleigh–Bénard is presented. The problem is modeled in a rectangular box with

Towards Incompressible Laminar Flow Estimation Based on Interpolated Feature Generation and Deep Learning

This work suggests a rapid laminar flow prediction model for inference of Naiver–Stokes solutions based on interpolated feature data generation and a deep U-Net learning model, which can improve the accuracy of estimating velocities and pressure fields with minimal relative errors.



A comprehensive deep learning-based approach to reduced order modeling of nonlinear time-dependent parametrized PDEs

Numerical results indicate that DL-ROMs whose dimension is equal to the intrinsic dimensionality of the PDE solutions manifold are able to efficiently approximate the solution of parametrized PDEs, especially in cases for which a huge number of POD modes would have been necessary to achieve the same degree of accuracy.

Assessment of unsteady flow predictions using hybrid deep learning based reduced-order models

Two deep learning-based hybrid data-driven reduced order models for the prediction of unsteady fluid flows are presented, and an observer-corrector method for the calculation of integrated pressure force coefficients on the fluid-solid boundary on a reference grid is introduced.

POD-DL-ROM: enhancing deep learning-based reduced order models for nonlinear parametrized PDEs by proper orthogonal decomposition

Construction of reduced-order models for fluid flows using deep feedforward neural networks

The DNN approach is able to learn transient features of the flow and presents more accurate and stable long-term predictions compared to sparse regression and the robustness of the current ROMs constructed via DNNs is demonstrated through a comparison with sparse regression.

An artificial neural network framework for reduced order modeling of transient flows

Deep learning observables in computational fluid dynamics

Efficient Reduction Techniques for the Simulation and Optimization of Parametrized Systems

This thesis exploits the MDEIM to develop fast and efficient ROMs for nonaffinely parametrized elliptic and parabolic PDEs, as well as for the time-dependent Navier-Stokes equations.

Model identification of reduced order fluid dynamics systems using deep learning

A novel model reduction method is presented: deep learning reduced order model, which is based on proper orthogonal decomposition and deep learning methods, and is independent of the source code of the full physical system.

Deep learning in fluid dynamics

  • J. Kutz
  • Computer Science
    Journal of Fluid Mechanics
  • 2017
Although neural networks have been applied previously to complex fluid flows, the article featured here is the first to apply a true DNN architecture, specifically to Reynolds averaged Navier Stokes turbulence models, suggesting that DNNs may play a critically enabling role in the future of modelling complex flows.