CoPiT transformer improves aerodynamic predictions on unstructured meshes
CoPiT: A conditional Position-induced Transformer for aerodynamic field prediction on unstructured meshes
Machine Learning
Summary
High-quality aerodynamic simulations take a long time because computers must solve complex physics on irregular shapes. The authors created CoPiT, a new type of AI model that better understands both the shapes and global conditions like wind speed by adjusting its internal parts explicitly. This makes predictions of airflow around objects faster and more accurate, especially on complicated surfaces and when data are limited. Their tests show CoPiT outperforms prior methods, helping engineers explore designs more efficiently.
What this means in practice
- •For aerospace design teams: Reduce time needed to predict airflow in different design and operating conditions by using CoPiT as a surrogate for expensive simulations.
- •For automotive engineers: Speed up aerodynamic shape optimization by efficiently predicting flow fields on complex vehicle surfaces without running full simulations.
Authors
Junfeng Chen, Yongchen Fan, Zhonghua Qiao, Kailiang Wu
Abstract
High-fidelity Reynolds-averaged Navier--Stokes (RANS) simulations remain a standard tool for aerodynamic analysis, yet each solve on an unstructured mesh is computationally expensive, making many-query tasks such as design exploration, optimization, and multi-condition evaluation costly. Neural operators provide mesh-flexible surrogates for mapping geometry and operating conditions to flow fields; however, in many existing architectures, global parameters such as the freestream state and design variables are concatenated with pointwise inputs, leaving their influence on intermediate features implicit. This paper proposes CoPiT, a \textbf{Co}nditional \textbf{P}osition-induced \textbf{T}ransformer that builds on the Position-induced Transformer (PiT) as its geometric backbone and injects a global condition vector through multi-stage feature-wise linear modulation (FiLM), thereby making parametric modulation explicit while preserving geometry-based attention weights. The model is evaluated on two-dimensional AirfRANS airfoil fields, including scarce-data and extrapolation regimes, and on three-dimensional BlendedNet++ blended-wing-body surfaces. On the AirfRANS full-data test set, CoPiT reduces surface and volume losses by approximately 54\% and 24\% relative to Transolver, respectively, and achieves the lowest overall normalized test MSE among the compared baselines on BlendedNet++. Matched ablations show that FiLM conditioning generally improves upon input concatenation, while the preferred placement is task-dependent. CoPiT thus provides a practical operator-learning surrogate for parametric aerodynamics on unstructured meshes.