Using image recognition to identify vortices for vehicle design
13 Feb 2020



We worked with TotalSim – an aerodynamics and CFD company – to understand vortex formation in Formula One cars and improve aerodynamic performance.





Simulation of blue Formula One car with yellow vortex lines

Credit: Total​Sim


Although well-studied, vortices are still not well-defined and have multiple criteria. A vortex can be described as a spinning spiral flow of fluid or air with closed streamlines, each with distinct properties and behaviour. Many components on a modern Formula One car are designed by the aerodynamicists to shed vortices. These vortices are used to control the air flow around the race car in order to maximise the downforce that can be generated. Understanding the dynamics of vortex structures is important as they have a direct impact on vehicle performance. TotalSim were looking to develop an algorithm that could extract, label and track vortices in vehicle simulation data to help identify the most promising design solutions more efficiently.​


Vortices are well-studied in flow visualisation, however those in 3D vector fields are not well-defined and have multiple criteria. The team approached this problem by first treating simulation results as 2D snapshots, applying image processing techniques to identify significant features. This approach is directly applicable to experimental images, enabling comparison and validation of fluid flow modelling. The team were able to develop an algorithm capable of analysing large image sets and highlighting significant features which ultimately helped improve engineering analyses.​

This work - completed as part of the collaborative Innovation Return on Research (IROR) programme with IBM Research - provided an efficient mechanism of extracting information about wing tip vortices formation in Formula One cars. The algorithm developed is versatile and can be applied to images from both simulations and physical experiments. This work offers detailed insights into the formation of vortex structures, leading to better vehicle designs with enhanced aerodynamic efficiency.​​

"Cross-industry events​ organised at the Hartree Centre have provided us​​ with the opportunity to collaborate with and learn from​ companies with whom we would not otherwise interact​​​. This sharing of ideas is of much benefit to British industry. ​​This has provided opportunities to work with experts outside​ our normal field of experience and allowed us to undertake 'blue sky' research​ projects that would otherwise be too risky or too expensive ​for an SME like us to undertake."

Robert Lewis

At a glance

  • Used image processing techniques to label and extract vortices from Formula One vehicle simulation data
  • Developed an algorithm capable of analysing large image sets, efficiently identifying significant features
  • Offers detailed insights in to vortex formation, helping to identify better vehicle designs
  • Insights from image analysis in both simulation and physical experiments to improve vehicle performance

Download as a PDF