The characterization of a racecar’s aerodynamic behavior at various yaw and pitch configurations has always been an integral part of its on-track performance evaluation in terms of lap time predictions. Although computational fluid dynamics has emerged as the ubiquitous tool in motorsports industry, a clarity is still lacking about the prediction veracity dependence on the choice of turbulence models, which is central to the prediction variability and unreliability for the Reynolds Averaged Navier–Stokes simulations, which is by far the most widely used computational fluid dynamics methodology in this industry. Subsequently, this paper presents a comprehensive assessment of three commonly used eddy viscosity turbulence models, namely, the realizable [Formula: see text] (RKE), Abe–Kondoh–Nagano [Formula: see text], and shear stress transport [Formula: see text], in predicting the aerodynamic characteristics of a full-scale NASCAR Monster Energy Cup racecar under various yaw and pitch configurations, which was never been explored before. The simulations are conducted using the steady Reynolds Averaged Navier–Stokes approach with unstructured trimmer cells. The tested yaw and pitch configurations were chosen in consultation with the race teams such that they reflect true representations of the racecar orientations during cornering, braking, and accelerating scenarios. The study reiterated that the prediction discrepancies between the turbulence models are mainly due to the differences in the predictions of flow recirculation and separation, caused by the individual model’s effectiveness in capturing the evolution of adverse pressure gradient flows, and predicting the onset of separation and subsequent reattachment (if there be any). This paper showed that the prediction discrepancies are linked to the computation of the turbulent eddy viscosity in the separated flow region, and using flow-visualizations identified the areas on the car body which are critical to this analysis. In terms of racecar aerodynamic performance parameter predictions, it can be reasonably argued that, excluding the prediction of the %Front prediction, shear stress transport is the best choice between the three tested models for stock-car type racecar Reynolds Averaged Navier–Stokes computational fluid dynamics simulations as it is the only model that predicted directionally correct changes of all aerodynamic parameters as the racecar is either yawed from the 0° to 3° or pitched from a high splitter-ground clearance to a low one. Furthermore, the magnitude of the shear stress transport predicted delta force coefficients also agreed reasonably well with test results.

<div class="section abstract"><div class="htmlview paragraph">Faster turn-around times and cost-effectiveness make the Reynolds Averaged Navier-Stokes (RANS) simulation approach still a widely utilized tool in racecar aerodynamic development, an industry where a large volume of simulations and short development cycles are constantly demanded. However, a well-known flaw of the RANS methodology is its inability to properly characterize the separated and wake flow associated with complex automotive geometries using the existing turbulence models. Experience suggests that this limitation cannot be overcome by simply refining the meshing schemes alone. Some earlier researches have shown that the closure coefficients involved in the RANS turbulence modeling transport equations most times influence the simulation prediction results. The current study explores the possibility of improving the performance of the SST <i>k</i> − <i>ω</i> turbulence model, one of the most popular turbulence models in motorsports aerodynamic applications, by re-evaluating the values of certain model closure constants. A detailed full-scale current generation NASCAR Cup racecar was used for the investigation. The simulations were run using a commercial CFD package STAR-CCM+ (version 13.04.010). Five different closure coefficients in the SST <i>k</i> − <i>ω</i> model, <i>σ</i>_<i>k</i>1, <i>σ</i>_<i>k</i>2, <i>σ</i>_<i>ω</i>1, <i>σ</i>_<i>ω</i>2 and <i>β</i>^∗, were examined. The investigation suggests the influence of each closure coefficient on the simulation prediction results are significantly different. <i>β</i>^∗ appeared to be the most sensitive closure coefficient whereas both <i>σ</i>_<i>k</i>1 and <i>σ</i>_<i>k</i>2 had almost no effect on the NASCAR Cup racecar aerodynamic predictions. This study proposes a new set of SST <i>k</i> − <i>ω</i> turbulence model closure coefficients which has the potential of providing better-correlated aerodynamic predictions of a NASCAR Cup racecar under a range of different operating conditions.</div></div>

No abstract

Transient Scale Resolved Simulations, like the Detached Eddy Simulation, are currently seen to be the preferred modeling approach over the steady-state Reynolds Averaged Navier-Stokes (RANS) simulations for numerical investigations of external flow due to the former’s perceived capability of providing a more realistic flow field prediction. However, the latter approach is still a widely used methodology in road vehicle aerodynamic developments because of its faster turn-around time and cost-effectiveness. However, RANS models, like the SST k–ω, generally fail to produce well-correlated predictions. Studies reveal that good correlations with experiment cannot be achieved by simply refining the mesh when using the SST k–ω model. As such, this study explores the possibility of improving the prediction veracity by investigating the influence of a few selected model closure coefficients on the CFD prediction. This involves first identifying the effect of each individual model parameter on the prediction, and then formulating the best combination of the model closure coefficient values that yield the best correlation with the experiment. This procedure is applied to three different test objects: NACA 4412 airfoil at 12 degree angle of attack, the 25 degree slant angle Ahmed body, and a full-scale passenger road vehicle. Although some closure coefficients do not influence the CFD results much, the predictions are very sensitive to the choice of certain model constants, irrespective of the test object geometry. The study also shows that it is possible to formulate a combination of closure model coefficients that can produce very well correlated CFD predictions.