Convolutional Neural Networks for Steady Flow Approximation

TLDR

CFD solvers are computationally expensive, memory demanding, and time consuming, limiting design space exploration and preventing interactive design. The authors propose a general, flexible CNN‑based model for real‑time prediction of non‑uniform steady laminar flow in 2D or 3D domains. They explored various geometry representations and CNN architectures to construct the approximation model. The CNN estimates the velocity field up to two orders of magnitude faster than a GPU‑accelerated CFD solver and four orders faster than a CPU‑based solver, with low error, enabling immediate feedback for real‑time design iterations and efficient whole‑field estimation without requiring additional surrogate models.

Abstract

In aerodynamics related design, analysis and optimization problems, flow fields are simulated using computational fluid dynamics (CFD) solvers. However, CFD simulation is usually a computationally expensive, memory demanding and time consuming iterative process. These drawbacks of CFD limit opportunities for design space exploration and forbid interactive design. We propose 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). We explored alternatives for the geometry representation and the network architecture of CNNs. We show that convolutional neural networks can estimate the velocity field two orders of magnitude faster than a GPU-accelerated CFD solver and four orders of magnitude faster than a CPU-based CFD solver at a cost of a low error rate. This approach can provide immediate feedback for real-time design iterations at the early stage of design. Compared with existing approximation models in the aerodynamics domain, CNNs enable an efficient estimation for the entire velocity field. Furthermore, designers and engineers can directly apply the CNN approximation model in their design space exploration algorithms without training extra lower-dimensional surrogate models.