In this study, experimental tests and Computational Fluid Dynamics are used to investigate the aerodynamic performance of two types of track-based racing cars. After the literature study, where automotive aerodynamics is discussed in very general terms, the air flow beneath a Formula One Grand Prix Racing Car is investigated. This is achieved by fitting the under-tray of a 30% scale model of the Parmalat Forti Ford FGO 1-95 with surface-static pressure ports and testing the model in a rolling-road wind tunnel. By varying a number of model parameters, it is found that the wheels significantly alter the pressure distribution under the floor of the racing car at positions away from the centre-line. It is shown that the front or rear wheel sets are independently sufficient to induce the flow changes. The addition of the other set then only produces milder and more local changes. The numerical part of the floor investigation is aimed at reproducing the centre-line flow pattern by solving the full Reynolds-Average Navier-Stokes equations over a two-dimensional curvilinear grid of the isolated floor. Two algorithms, Roe's flux-difference splitting method and the commercial package, STAR-CD which employs the SIMPLE algorithm and a two-equation turbulence model, are used to solve the governing equations. It is found that although the correct trends are observed when two different ride heights are simulated, absolute correlation is inadequate despite the use of experimentally-controlled boundary conditions. The simulations are however used to demonstrate the saturation in downforce with increasing vehicle speed. In order to improve numerical accuracy, a second study was launched where the effect of including the centre-line profile of the complete vehicle is investigated. To reduce the amount of detail a 1/12th scale model of a generic BMW Touring Car is used. Experimental data in the form of centre-line surface-static pressure coefficients are used for numerical correlation. The data is obtained by testing the three-dimensional model in a wind tunnel fitted with a stationary-road raised-platform floor. To establish continuity, the experimental data is used to show the similarities between the pressure distribution on the centre line of the open-wheel and the closed-wheel racing car. The effect of a rear-mounted aerodynamic device on the downforce is also discussed. The numerical investigation using the SIMPLE algorithm of STAR-CD and three high Reynolds-Number turbulence models, is based on the centre-line profile of the experimental model. It is seen that although qualitative correlation exists in areas around the car, quantitative agreement is less positive. Discrepancies are found to be most significant under the floor. It is shown that the influence of the three dimensional flow field on the experimental results are unlikely to cause satisfactory correlation. It is suggested that, in order to improve correlation, a new investigation is launched aimed at refining the numerical model. An outline for the new study is presented and includes simulations indicating the dependence of the computational solution on the density of the grid and on the user-definable turbulence parameters. / Dissertation (M Eng (Mechanical Engineering))--University of Pretoria, 1999. / Mechanical and Aeronautical Engineering / unrestricted