Honda faces the challenge of balancing vehicle design and performance amid increasingly compressed development timelines. Key concerns include wind noise from front grille and A-pillar designs, and mirror visibility in bad weather. Airflow separation around the A-pillar, located on either side of the windshield, can strengthen vortices and increase wind noise. While full-scale and wind tunnel testing remain valuable, they can be cumbersome and costly.
To address this, Honda decided to replace portions of traditional wind tunnel testing with CFD using Ansys Fluent and Ansys Discovery. However, preparing a separate computational model for each design became a bottleneck. Honda is evaluating Ansys GeomAI, which learns from existing designs to generate new geometry variations. Using four A-pillar geometries with different widths and gaps, it generated consistent variations, including useful shapes beyond its training data. Although initial training is required, further variants take approximately a minute to generate, with virtually no additional effort. The approach aims to shift engineers’ time from repetitive modelling to detailed flow analysis and design optimisation.
Image generated by: ChatGPT
