CCP-NTH Webinar – OpenFOAMGPT: a LLM based agent for computational fluid dynamics

Topic: CCP-NTH Webinar – OpenFOAMGPT: a LLM based agent for computational fluid dynamics

Time: Apr 25, 2025 01:30 PM London

Join Zoom Meeting

https://ukri.zoom.us/j/98041354077?pwd=t59cOQKDDvMxCrPsf7ipYPWQWFs46n.1

Password: 246884

Speaker: Dr. Xu Chu, University of Exeter

Topic: OpenFOAMGPT: a LLM based agent for computational fluid dynamics

Abstract: 

We developed a Large Language Model (LLM)-based agent to advance Computational Fluid Dynamics (CFD) workflows. CFD simulations often involve intricate configurations, including solver selection, initial- and boundary condition setup, and iterative adjustments, which require substantial expertise and manual effort. The proposed LLM-CFD framework integrates natural language processing capabilities with CFD tools, such as OpenFOAM, to assist users in navigating these complexities more efficiently. The LLM agent employs retrieval-augmented generation (RAG) on domain-specific datasets to provide accurate, context-aware guidance. Key functionalities include interactive solver configuration, and error diagnostics, allowing users to focus on analysis and decision-making. Reliability is a central focus of this work, addressing challenges such as potential hallucinations and uncertainty in LLM-generated outputs. The framework incorporates validations and benchmarks to ensure dependable performance. \\ The integration of LLM technology into CFD workflows has the potential to enhance productivity by automating routine tasks, reducing human error, and improving accessibility for non-specialists. It also provides opportunities for interdisciplinary collaboration by bridging the gap between domain experts and computational tools. The presentation will highlight use cases demonstrating the agent’s capabilities, discuss methods to mitigate reliability risks, and outline future directions for applying LLMs in CFD research and industry. This work aims to contribute to the ongoing discourse on LLM applications in CFD, fostering innovation in simulation-based engineering and expanding the scope of fluid dynamics research.

A short bio of the speaker:

Priv.-Doz. Dr.-Ing. Xu Chu has served as Senior Lecturer in Data-Centric Engineering (combined with thermal fluids) at the University of Exeter. Dr. Xu Chu received his B.S. from Tongji University in China and both his Dipl.-Ing. and Dr.-Ing. from the University of Stuttgart, Germany. Following his Ph.D. at the Institute of Nuclear Technology and Energy Systems, he led a research group at the Institute of Aerospace Thermodynamics, University of Stuttgart, and completed his Habilitation (comparable to tenure track). His research focuses on the integration of numerical, experimental, and data-driven approaches in multi-X thermofluids, with applications in energy, aerospace, and biology. Dr. Chu has supervised and co-supervised over 10 Ph.D. students and postdoctoral researchers, bringing a strong mentoring and academic leadership record to his role.