IBM and Dallara Cut Aerodynamic Simulation Times From Hours to 10 Seconds
IBM and Dallara are co-developing physics-based AI foundation models trained on validated aerodynamic data which reduce simulation times from hours to under ten seconds during early design stages. The partners plan to integrate quantum and hybrid computing approaches to further boost aerodynamic simulation fidelity and accelerate high-performance vehicle design workflows.
1. Collaboration Focus
IBM and Dallara are collaborating to develop physics-based AI foundation models using Dallara’s high-fidelity aerodynamic datasets. The initiative aims to accelerate aerodynamic design workflows by combining IBM’s AI expertise with Dallara’s racing vehicle engineering prowess.
2. AI-driven Simulation Speedup
Early tests on a conceptual LMP2-style rear diffuser demonstrated that the AI model matched traditional CFD error margins while completing multiple geometry evaluations in about 10 seconds versus hours required by CFD. This speedup could reduce full vehicle development simulation from weeks to minutes for hundreds of configurations.
3. Quantum Computing and Future Applications
Both teams plan to explore integration of quantum and hybrid quantum-classical computing to further enhance simulation fidelity for complex aerodynamic problems. The collaboration may expand AI models across varied maneuvers and vehicle types, potentially benefiting automotive, aerospace, and passenger transport sectors.