External Aerodynamics — CFD for Vehicles and Aircraft

External aerodynamics is the study of airflow around the outside of a body — a car, aircraft wing, building, cyclist, or missile — where the fluid domain extends to "infinity" in all directions away from the body. The key distinction from internal flow is that the boundaries of the fluid domain are artificial (we define a finite computational domain and apply free-stream boundary conditions), and the main quantities of interest are the aerodynamic forces: drag force opposing motion and lift force perpendicular to the free stream.

Drag and lift are reported as dimensionless coefficients. Drag coefficient $C_D = F_D / (0.5 \rho U_\infty^2 A_{ref})$ and lift coefficient $C_L = F_L / (0.5 \rho U_\infty^2 A_{ref})$. The reference area $A_{ref}$ is the frontal projected area for a car, or the wing planform area for an aircraft. These non-dimensional forms let you compare results across different scales and speeds. For a typical modern sedan, $C_D \approx 0.28$–0.32; a racing car with downforce can reach $C_L \approx -3.0$ (negative lift = downforce).

On any surface, there's a thin layer where the flow velocity drops from the free-stream value to zero at the wall — that's the boundary layer. As the boundary layer moves downstream, it grows in thickness. If it encounters an adverse pressure gradient (pressure rising in the flow direction — like the rear of a car body), the momentum in the slow near-wall fluid isn't enough to push through, and the flow detaches from the surface. This creates a large low-pressure recirculation wake behind the body. That wake is the main source of pressure drag for bluff bodies — the front face is at high pressure, the rear wake is at low pressure, and the net pressure difference times frontal area equals your drag force.

Yes, it matters significantly. In a real wind tunnel and on the road, the ground is moving at the same speed as the free stream relative to the car — so there's no relative motion between the air and the ground plane. A slip floor correctly models this (zero wall shear, no boundary layer on the floor). Using a no-slip stationary floor creates a boundary layer that thickens toward the rear, artificially elevates the floor pressure, and can change your underbody drag by 5–15%. For full vehicle simulations, you should also include rotating wheels — wheel rotation dramatically affects the wheelhouse vortex system and can account for 20–25% of total vehicle drag.

Domain sizing is a common mistake for beginners. If your domain is too small, the blockage ratio becomes significant — the model forces the flow to accelerate around it because the domain walls are too close, effectively speeding up the free stream and reducing drag (the same effect that makes wind tunnel results at high blockage ratio unreliable). Standard practice: inlet at least 5 car lengths upstream, outlet 15–20 car lengths downstream (the wake needs space to develop), and sides/top at least 5 car widths/heights away. This keeps blockage below 1% which is the accepted threshold. For mesh independence, always do a 3-mesh study: coarse, medium, fine with roughly 2x refinement each step.

SST k-omega (Shear Stress Transport) is the industry standard for external aerodynamics, and for good reason. Standard k-epsilon performs poorly near walls and in adverse pressure gradients — it typically underpredicts the size of separation zones, so it underestimates the wake and therefore underpredicts drag. SST blends the accurate near-wall behavior of k-omega with the free-stream robustness of k-epsilon, and crucially it includes the Bradshaw stress-intensity correction that improves separated flow prediction. For a typical car, SST gives drag predictions within 3–8% of wind tunnel with a well-resolved mesh. With k-epsilon you might be off by 15–25% on the same mesh.