Accelerating Tandem-Airfoil simulations using curriculum-based graph convolutional network for initialization

Wei Xian Lim

doi:10.1016/j.ast.2026.112440

Accelerating Tandem-Airfoil simulations using curriculum-based graph convolutional network for initialization

2026-04-23

0

PRE

AI

DOI：10.1016/j.ast.2026.112440

Original

Original request for help

Save

Abstract

En 中文

• Curriculum-based GCN initializes tandem-airfoil RANS simulations. • Smooth-combined fields and residual learning improve prediction accuracy. • The method reproduces aerodynamic fields with high physical fidelity. • Robust performance is obtained across low- and high-Reynolds-number cases. • Solver convergence time is significantly reduced versus freestream initialization.

Keywords:

Curriculum-based GCN

Tandem-airfoil simulations

RANS equations

Aerodynamic fields

Solver convergence

Accelerating Tandem-Airfoil simulations using curriculum-based graph convolutional network for initialization

Abstract

Journal

Organization

Cited Papers