Results are presented from a study on the use of Computational Fluid Dynamics (CFD) for automotive underbody design. A diffuser-equipped bluff body with endplates was examined in ground effect at varying ride heights in configurations with and without wheels. The study was performed using commercial CFD, Fluent© 6.3.26. CFD data is compared to experimental work done with similar bodies by Cooper et al. [1, 2], George et al. [3, 4], Zhang et al. [5, 6], and others [7, 8, 9]. Emphasis is made on the study of vortex structures in bluff body flow. Various mesh geometries and solvers were explored with computational models designed to operate on single-processor workstations or small networks. Steady-state solutions were modeled for all cases; boundary layers were approximated with wall functions.
CFD results for lift coefficient measured within 15-25% of experimental cases, dependent on solver. Qualitative results matched well with experimentally measured flow structures. Downforce reduction due to stall was found at ride heights similar to those established experimentally, attributed to trailing-edge separation of both the underbody and ground boundary layers at the diffuser, as well as to vortex deterioration or breakdown. The effects of underbody vortices on downforce generation and stall prevention at low ride heights are discussed in detail from insights derived using the CFD models.
After validating computational models against experimental baselines, CFD was explored for multi-body flow applications. Effects of wheel-shaped objects along side of, ahead of, and behind the bluff body in ground effect were examined. A decrease in force generation from the bluff body was found in certain configurations. This was found to relate to the disruption of vortex formation along the diffuser and underbody region, as well as from blockage. Similar experiments conducted by Breslouer & George [10] extend upon these findings.
From the study, opportunities for practical automotive development using commercial CFD are discussed while indicating likely obstacles and limitations. Recommendations are made relating to solver choice and mesh generation practice with the aim to optimize model creation and run time while still realizing useful qualitative and quantitative results.