Neural Network-Based Parametric Model Reduction for Predicting Turbulent Flow for Different Vehicle Geometries

Kazuto Ando

doi:10.1016/j.compfluid.2026.107196

Neural Network-Based Parametric Model Reduction for Predicting Turbulent Flow for Different Vehicle Geometries

2026-06-25

0

PRE

AI

K

Kazuto Ando

R

Rahul Bale

A

Akiyoshi Kuroda

M

Makoto Tsubokura

DOI：10.1016/j.compfluid.2026.107196

Original

Original request for help

Save

Abstract

En 中文

• NN-based parametric ROM for 3D turbulent flows. • VAE improved reconstruction accuracy over MD-CNN-AE. • Captured complex nonlinear wake structures with 64 latent variables. • Validated on 11 realistic vehicle models at Re ≈ 8.4 × 10^6. • Predicted unseen high-Re turbulent flows around complex geometries.

Neural Network-Based Parametric Model Reduction for Predicting Turbulent Flow for Different Vehicle Geometries

Abstract

Journal

Organization