Direct Prediction of Steady-State Flow Fields in Meshed Domain with Graph Networks

05/06/2021
by   Lukas Harsch, et al.
0

We propose a model to directly predict the steady-state flow field for a given geometry setup. The setup is an Eulerian representation of the fluid flow as a meshed domain. We introduce a graph network architecture to process the mesh-space simulation as a graph. The benefit of our model is a strong understanding of the global physical system, while being able to explore the local structure. This is essential to perform direct prediction and is thus superior to other existing methods.

READ FULL TEXT

page 3

page 9

research
04/22/2020

ML-LBM: Machine Learning Aided Flow Simulation in Porous Media

Simulation of fluid flow in porous media has many applications, from the...
research
08/04/2021

Partial Reuse AMG Setup Cost Amortization Strategy for the Solution of Non-Steady State Problems

The partial reuse algebraic multigrid (AMG) setup cost amortization stra...
research
07/25/2023

Multi-GPU Approach for Training of Graph ML Models on large CFD Meshes

Mesh-based numerical solvers are an important part in many design tool c...
research
11/25/2019

Towards Sustainable Architecture: 3D Convolutional Neural Networks for Computational Fluid Dynamics Simulation and Reverse DesignWorkflow

We present a general and flexible approximation model for near real-time...
research
10/09/2022

Quasi-Monolithic Graph Neural Network for Fluid-Structure Interaction

Using convolutional neural networks, deep learning-based reduced-order m...
research
02/06/2023

Impact of artificial topological changes on flow and transport through fractured media due to mesh resolution

We performed a set of numerical simulations to characterize the interpla...
research
02/10/2023

Graph-Theoretic Analyses and Model Reduction for an Open Jackson Queueing Network

A graph-theoretic analysis of the steady-state behavior of an open Jacks...

Please sign up or login with your details

Forgot password? Click here to reset