Aerodynamic study of a formula student car by means of high fidelity simulations
In modern single-seater racing, and particularly in Formula Student, aerodynamics is recognized as one of the most decisive performance factors. To optimize aerodynamic efficiency and gain competitive advantage, motorsport traditionally relies on a combination of Computational Fluid Dynamics (CFD) a...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/457427 |
| Acceso en línea: | https://hdl.handle.net/2117/457427 |
| Access Level: | acceso embargado |
| Palabra clave: | Automobiles, Racing -- Aerodynamics Aerodinàmica CFD LES WMLES SOD2D Automòbils de competició -- Aerodinàmica Àrees temàtiques de la UPC::Aeronàutica i espai::Aerodinàmica |
| Sumario: | In modern single-seater racing, and particularly in Formula Student, aerodynamics is recognized as one of the most decisive performance factors. To optimize aerodynamic efficiency and gain competitive advantage, motorsport traditionally relies on a combination of Computational Fluid Dynamics (CFD) and wind tunnel testing. Today, the emergence of supercomputing computing architectures and high-order numerical methods has enabled the adoption of scale-resolving approaches, most notably Large Eddy Simulation (LES), which offer superior accuracy over Reynolds-Averaged Navier–Stokes (RANS) models in regions of highly unsteady flow. This work investigates the aerodynamics of a Formula Student race car through the application of a wallmodelled Large Eddy Simulation (WMLES) and its comparison to a Reynolds-Averaged Navier–Stokes (RANS) simulation. A second-order hexahedral half-car mesh was generated for the high-order solver SOD2D by splitting a linear tetrahedral mesh into hexahedra in ANSA, before proceeding with the a-posteriori conversion to second order. The meshing process presented several challenges, including the occurrence of tangled or invalid elements and the lack of robust quality criteria to guarantee proper solution convergence. The WMLES was run for 6 TU, with the final 2 TU used for time-averaging, and the results were contrasted with a RANS simulation. The WMLES consistently predicted higher aerodynamic loads, overestimating SCL by 10–12% and SCD by 5–12%, likely due to limitations of the wall model in the accurate prediction of boundary layer detachment. Nevertheless, the LES, particularly when combined with the low dissipation numerical schemes of SOD2D, demonstrated a significant advantage over RANS in capturing the complex flow phenomena around the geometry, such as vortices. The improved fidelity of LES opens new opportunities in the accurate prediction of aerodynamic performance for race-car applications, highlighting the potential of scale-resolving simulations |
|---|