• Corpus ID: 232046152

Machine Learning-Based Optimal Mesh Generation in Computational Fluid Dynamics

  title={Machine Learning-Based Optimal Mesh Generation in Computational Fluid Dynamics},
  author={Keefe Huang and Moritz Kr{\"u}gener and Alistair Brown and Friedrich Menhorn and Hans-Joachim Bungartz and Dirk Hartmann},
Computational Fluid Dynamics (CFD) is a major sub-field of engineering. Corresponding flow simulations are typically characterized by heavy computational resource requirements. Often, very fine and complex meshes are required to resolve physical effects in an appropriate manner. Since all CFD algorithms scale at least linearly with the size of the underlying mesh discretization, finding an optimal mesh is key for computational efficiency. One methodology used to find optimal meshes is goal… 
E2N: Error Estimation Networks for Goal-Oriented Mesh Adaptation
This paper develops a “data-driven” goal-oriented mesh adaptation approach through the selective replacement of the expensive error estimation step with an appropriately configured and trained neural network, and demonstrates that the element-by-element approach implies reasonably low training costs.
Deep Learning for Computational Hemodynamics: A Brief Review of Recent Advances
It is suggested that incorporating physiologic understandings and underlying fluid mechanics laws in deep learning models will soon lead to a paradigm shift in the development novel non-invasive computational medical decisions.
Learning robust marking policies for adaptive mesh refinement
This work recast AMR as a Markov decision process in which refinement parameters can be selected on-the-fly at run time, without the need for pre-tuning by expert users, and uses the Poisson equation to demonstrate the techniques.
M2N: Mesh Movement Networks for PDE Solvers
This approach can greatly accelerate the mesh adaptation process, whilst achieving comparable numerical error reduction, and validate the methods on stationary and time-dependent, linear and non-linear equations, as well as regularly and irregularly shaped domains.


Convolutional Neural Networks for Steady Flow Approximation
This work proposes a general and flexible approximation model for real-time prediction of non-uniform steady laminar flow in a 2D or 3D domain based on convolutional neural networks (CNNs), and shows that convolutionAL neural networks can estimate the velocity field two orders of magnitude faster than a GPU-accelerated CFD solver and four orders of order than a CPU-based CFDsolver at a cost of a low error rate.
Accelerating Eulerian Fluid Simulation With Convolutional Networks
This work proposes a data-driven approach that leverages the approximation power of deep-learning with the precision of standard solvers to obtain fast and highly realistic simulations of the Navier-Stokes equations.
Well, how accurate is it? A Study of Deep Learning Methods for Reynolds-Averaged Navier-Stokes Simulations
This work focuses on a modernized Unet architecture, and evaluates a large number of trained neural networks with respect to their accuracy for the calculation of pressure and velocity distributions, illustrating how training data size and the number of weights influence the accuracy of the solutions.
An optimal control approach to adaptivity in computational fluid mechanics
We consider a typical design cycle in computational fluid mechanics: First, an output value such as the drag coefficient is computed. Then, this value is optimized by varying certain control
Adaptive Computation of Reactive Flows with Local Mesh Refinement and Model Adaptation
An adaptive method for reactive flows involving locally refined meshes and different types of diffusion models is proposed. Starting with a less exact diffusion model, the model is changed locally
Machine Learning for Fluid Mechanics
Fundamental ML methodologies are outlined and their uses for understanding, modeling, optimizing, and controlling fluid flows are discussed and the strengths and limitations of these methods are addressed.
Error Estimates for Adaptive Finite Element Computations
The main theorem gives an error estimate in terms of localized quantities which can be computed approximately, and the estimate is optimal in the sense that, up to multiplicative constants which are independent of the mesh and solution, the upper and lower error bounds are the same.
Weighted a Posteriori Error Control in Fe Methods
A reened approach to residual{based error estimation is proposed where certain quantities involving the dual solution are used as weight-factors in order to capture information about the local error propagation.
An optimal control approach to a posteriori error estimation in finite element methods
The ‘dual-weighted-residual method’ is introduced initially within an abstract functional analytic setting, and is then developed in detail for several model situations featuring the characteristic properties of elliptic, parabolic and hyperbolic problems.