Aerodynamics of Racing Cars

The fundamental objectives are different. For production cars, the main goal is reducing aerodynamic drag, but for racing cars, the goal is to maximize downforce (negative lift) while keeping drag within acceptable limits.

An F1 car generates about 3.5G of downforce at 300 km/h. This pressing force, significantly exceeding the car's weight, theoretically allows it to drive upside down on a ceiling.

This downforce determines cornering speed. Since tire grip is proportional to vertical load, increasing downforce raises the cornering limit.

Downforce and drag are expressed using dimensionless coefficients.

Here $A$ is the frontal projected area. For F1 cars, $C_L \approx -3.0$--$-5.0$, $C_D \approx 0.7$--$1.2$, and $L/D \approx 3$--$5$.

Correct. A higher L/D means obtaining large downforce with low drag. Floor design utilizing ground effect is key to improving L/D.

Let's organize the main sources of downforce.

Ground effect increases flow velocity in the narrow gap with the ground, creating a large low-pressure area via Bernoulli's principle. It has very high L/D because drag increase is small. Since the 2022 F1 regulations, designs actively utilize ground effect.

The typical Reynolds number around a racing car, based on car length, is $Re \approx 10^7$.

Flow characteristics:

• Front: Bluff body flow: Stagnation point, large-scale separation
• Wing: Flow around airfoils (high angle of attack)
• Wheel: Unsteady vortices from rotating bluff body
• Diffuser: Flow in an expanding duct under adverse pressure gradient
• Wake: Interference of multiple vortex structures

RANS, DES/DDES, and LES are all used appropriately.

FIA regulations limit CFD usage (ATR: Aerodynamic Testing Restrictions). As of 2024, the champion team has an annual limit of 25 Teraflop-hours. This roughly corresponds to a 2000-3000 core HPC cluster running at full capacity for one year.

The following are important for full-car mesh generation.

• Wall Prism Layers: $y^+ \approx 1$, 20--30 layers. Especially critical for wings and floor
• MRF (Moving Reference Frame): Models wheel rotation
• Ground Boundary: Moving wall condition. Moves at the same speed as the car
• Refinement Zones: Wing tips, diffuser exit, wake region
• Total Cell Count: RANS 50-100 million, DDES 200-500 million

In actual driving, the car moves forward, but in CFD, the car is fixed and the ground is moved at the flow velocity. If you forget this and set the ground as a fixed wall, the ground boundary layer develops and ground effect is not correctly reproduced.

SST k-omega is the industry standard. The SA model is also used, but SST is superior for predicting separation under adverse pressure gradients.

Detailed wing analysis:

• SST k-omega: Steady RANS. Good for predicting wing surface pressure distribution and separation location
• SST k-omega + gamma-Re_theta: When transition prediction is needed (low Re airfoils)
• DDES (SST based): Analysis of unsteady vortex structures in wing wake

There are three methods.

MRF is common in the design exploration phase with steady RANS. To accurately capture unsteady vortex structures in the tire wake, use Sliding Mesh or Overset Mesh. STAR-CCM+'s Rigid Body Motion is user-friendly.

Tire deformation is typically not modeled in aerodynamic CFD. A rigid tire geometry is used from previous test data or structural simulations. Some advanced analyses use a moving contact patch based on tire load and properties.