This research demonstrates how fine-tuned Large Language Models can automate complex computational fluid dynamics (CFD) workflows, reducing the need for specialized expertise.

• Created NL2FOAM: a custom dataset of 28,716 natural language-to-OpenFOAM configuration pairs
• Fine-tuned Qwen2.5-7B-Instruct model with chain-of-thought annotations
• Enables engineers to set up CFD simulations using natural language commands
• Democratizes access to powerful simulation tools for engineering applications

This innovation has significant implications for engineering teams by lowering barriers to advanced fluid dynamics modeling, accelerating prototyping processes, and reducing the expertise needed to configure complex simulations.

Fine-tuning an Large Language Model for Automating Computational Fluid Dynamics Simulations