PhD in Computer Science

The Aerodynamic Translator is a technology that you'll develop to efficiently predict the aerodynamic performance of a ground vehicle, whether the shape of that object. This doctoral research focuses on advancing computational fluid dynamics (CFD) surrogates to optimize the aerodynamic efficiency of ground vehicles. The core objective is the development of a Physics-Informed Artificial Intelligence (AI) framework designed for highly accurate, near-real-time prediction of aerodynamic flow fields around generic car geometries. The project leverages a unique database of validated aerodynamic data to train this system. By embedding the governing physical laws of fluid dynamics directly into the machine learning architecture, this Physics AI approach ensures physical plausibility and overcomes the resource intensity of traditional CFD solvers. This methodology delivers higher predictive accuracy with equivalent data resources, or maintains accuracy using a dramatically smaller computational footprint, heavily exploiting modern GPU computing. The key scientific contribution is the creation of an Aerodynamic Translator. This sophisticated model is designed for high generalization, allowing engineers to re-train the model for a completely new vehicle shape with minimal effort and at a fraction of the cost. The successful delivery of this research will provide with a transformative tool to rapidly accelerate the design process for more energy-efficient cars and HGVs, directly supporting global efforts to reduce carbon emissions and operational energy consumption. You'll be given specialist training on Machine Learning that the supervisor has developed for engineers.