# Platform control & ground effect: diffuser, stall, porpoising, plank

> Underbody downforce rises as the car gets closer to the ground — until the flow stalls. Controlling the dynamic platform (heave, pitch, roll) to stay near the peak without crossing the cliff is the core of modern aero setup, and the source of porpoising.

Area: Aerodynamics · Starting level: L2 Practitioner · Updated 2026-11-16
Source page: https://race-engineer-academy.vercel.app/lesson/aero-platform-ground-effect

## L1 · Foundation

**How the underbody makes downforce.** Air entering the gap under the car is accelerated because the gap is narrow (Venturi effect); faster air has lower static pressure (Bernoulli). The **diffuser** at the rear lets that air slow down and recover to ambient pressure gradually, which allows the floor to run a *lower* pressure than it could otherwise. The result is suction pulling the car onto the road, with very little drag cost compared with wings — the most efficient downforce on the car.

**Closer is better — to a point.** As the ride height falls, the air under the floor speeds up and downforce rises. Below some height, the floor's own boundary layer can no longer cope with the pressure recovery the diffuser demands: the flow separates in the diffuser and the edge vortices that seal the floor break down, and the downforce **stalls**, often dropping sharply. (In the car's frame the road moves at freestream speed, so unlike a fixed tunnel floor it brings no boundary layer of its own into the gap; that is why tunnels need a rolling road.) On ground-effect formula cars this cliff can be a 10–30 % loss over a few millimetres.

**Platform** is the engineer's word for the car's attitude relative to the road: ride heights, pitch and roll. **Platform control** means using springs, heave/third elements, packers and bump rubbers, anti-dive/anti-squat and dampers to keep the floor in the right window across all speeds, under braking and over bumps.

**Porpoising** is a self-sustained vertical oscillation at speed, typically a few hertz, on cars with strong ground effect: the car sinks, the floor stalls, downforce drops, the car rises, the flow re-attaches, downforce returns, the car sinks again. Drivers feel it as violent bouncing on the straights, getting worse with speed. It is different from ordinary *bouncing* over bumps, which comes from the track input, not the aero.

**The plank (skid block).** Many single-seaters and prototypes carry a wooden or composite plank under the floor whose thickness is measured after the session (e.g. a maximum wear allowance of 1–2 mm in some series). It prevents teams running the car on the ground. Plank wear is therefore a legal constraint on how low you can run — plus a useful indicator of where the car touches.

Typical dynamic ride heights at the floor reference: formula cars a few tens of millimetres at speed; GT3 floors higher (40–70 mm), and correspondingly less sensitive.

## L2 · Practitioner

**Trackside tools for platform control:**

| Tool | Effect on platform |
|---|---|
| Heave/third spring (front, rear) | Resists aero squat without stiffening roll; sets RH drop with speed |
| Packers / bump stops | Hard limit on travel; fixes minimum RH in braking/at speed |
| Static ride height and rake | Places the platform in the map |
| Front/rear anti-dive & anti-squat (if adjustable) | Reduces pitch under braking/traction |
| Low-speed damping (heave) | Controls RH transients, e.g. after kerbs and on pitch change |
| Tyre pressure | Tyre vertical rate ~ a large share of total stiffness on stiff cars |

**Reading the data.**
- Plot front and rear dynamic RH vs speed for straight-line sections. A well-controlled platform shows a smooth $V^2$-shaped fall; an unstable one shows growing oscillations above some speed.
- Spectral analysis (FFT) of ride height or vertical acceleration on straights: porpoising appears as a narrow peak (typically 3–8 Hz) whose amplitude grows with speed; road-induced bouncing is broadband and tied to track location.
- Plank wear vs location: check which part of the plank wears (front, rear, one side — roll in a particular corner).
- Ride-height minima histogram per lap: how often do you reach the packers / the ground?

**Diagnosing a stall.** Symptoms: sudden loss of rear (if diffuser) or front grip at the highest speeds or over a crest/bump, often with a speed-trap *increase* (stalled floor = less drag). Correlate with RH at the event: if the car loses grip only when $h_R$ falls below, say, 45 mm, you are at the diffuser cliff. Fix: raise static height, stiffer heave, or packers to stop the car reaching that height.

**Porpoising fixes in order of cost:** raise ride height (costs downforce, always works), stiffen heave/raise tyre pressure (costs mechanical grip and tyre contact over bumps), add heave damping (helps a marginal case), change floor/diffuser parts if homologated options exist.

**The trade-off.** Stiffer platform = stable aero but worse ride over bumps, kerbs and less mechanical grip. On bumpy tracks (Sebring, street circuits) the optimum moves to softer and higher; on smooth fast tracks, lower and stiffer.

**Misconception:** "porpoising is a damper problem." Dampers can reduce a marginal oscillation but cannot remove an aero-driven instability whose energy source is the floor; the root cause is the aero load characteristic around the downforce peak (the gradient changing sign, plus the lag of the floor flow), not the dampers.

## L3 · Advanced

**Aero stiffness.** Linearise the downforce on an axle about the working height: $\partial L/\partial h$ (N/mm). In the normal operating region $\partial L/\partial h < 0$ (lower → more downforce). For a single-DOF heave model with mass $m$, suspension+tyre stiffness $k$ and damping $c$:

$$m\ddot z + c\dot z + \left(k + \frac{\partial L}{\partial h}\right) z = F_{road}$$

The **effective stiffness** $k_{eff} = k + \partial L/\partial h$ falls as speed rises, because $\partial L/\partial h = q\,\partial C_LA/\partial h$. Example: a formula-car front axle with $k = 300$ N/mm and a map gradient of $-0.02$ m²/mm at $q = 4000$ Pa gives $\partial L/\partial h = -80$ N/mm, so $k_{eff} = 220$ N/mm: ride frequency drops by ~14 % and the ride-height drop at speed is larger than a pure-spring calculation predicts. If $k_{eff} \to 0$ the platform is statically unstable — the car is pulled onto its bump stops.

**Why porpoising is a dynamic instability.** Static analysis alone does not explain a limit cycle. Two ingredients make it self-excited:
1. **Lag/hysteresis.** The floor load responds to height changes with a delay $\tau$ (separation and re-attachment take time). With a first-order lag, $L' = \partial L/\partial h$ contributes an effective damping $c_{eff} = c - L'\tau/(1+\omega^2\tau^2)$. In the normal branch ($L' < 0$, lower = more load) the lag actually *adds* damping. Past the peak ($L' > 0$, lower = less load: the stall branch) the aero force is restoring, and a delayed restoring force feeds energy in: $c_{eff} < c$, possibly negative.
2. **Stall nonlinearity.** Below the stall height the gradient changes sign. The car drops into the stall, loses load, springs up, re-attaches and falls again. The limit cycle forms because the car keeps crossing between the two branches; above a threshold speed the aero energy input per cycle exceeds damper dissipation.

The frequency is close to the heave (or pitch) natural frequency of the car on its tyres and springs, typically 3–8 Hz for stiff ground-effect cars. Remedies follow directly: raise the stall height margin, raise $k$ (stiffness), increase heave damping, or reduce the aero gradient (floor design).

**Diffuser physics.** The diffuser's pressure recovery sets the floor's base pressure. Its performance depends on expansion angle, boundary-layer health (vortices from floor edges and strakes re-energise it) and the upwash from the rear wing/beam wing, which "pulls" the diffuser flow. That is why the **rear wing interacts with the diffuser**: lowering the rear wing angle can cost underfloor downforce too, not just wing load.

**Roll and sealing.** In roll, one floor edge approaches the ground. Edge vortices that "seal" the floor are sensitive to that gap; asymmetric RH in high-speed corners can unseal one side and drop downforce. Roll stiffness distribution therefore has an aero dimension on ground-effect cars, not just a mechanical-balance one.

## L4 · Expert

**Coupled heave–pitch with aero.** Writing the platform as a 2-DOF system (heave $z$, pitch $\theta$) with front and rear aero gradients $\partial L_F/\partial h_F$, $\partial L_F/\partial h_R$, etc., the aero contributes a *non-symmetric* stiffness matrix (front load depends on rear height and vice versa). Non-symmetric stiffness can create flutter-like coalescence of the heave and pitch modes — instability even when each mode alone is stable. This is why porpoising can be cured on one car by a rear change and on another by a front change: what matters is the eigenstructure, not one axle.

**Modelling.** A practical expert workflow:
1. Fit $C_LA$ and AB maps (or $C_{LF}A$, $C_{LR}A$) from tunnel/CFD with local gradients.
2. Build a 2-DOF (or 7-post-like 4-DOF) model including tyre vertical stiffness/damping, heave elements, packers (nonlinear), dampers (piecewise-linear).
3. Add a first-order lag to the aero force ($\tau$ of order a few to tens of ms from CFD/track identification).
4. Sweep speed: compute eigenvalues of the linearised model; find the speed where a complex pair crosses into the right half-plane. Validate against the onset speed seen in track FFTs.

**Bump-induced stall vs self-excited porpoising.** A crest or bump can push the car through the stall height once (single event, transient), then the car recovers. This shows as a single downforce loss at a fixed track location — a ride/packer issue. True porpoising persists on a smooth straight and grows with speed. Distinguish them before changing anything.

**Limits.** Quasi-steady maps assume the flow adapts instantly. At the frequencies of interest (5 Hz → 0.2 s period; flow-through time under a 3 m floor at 80 m/s is ~0.04 s) that is only marginally true; the flow history matters. Also, tyre sidewall deformation under load changes the floor-edge seal and is poorly represented in tunnels. Expect your model to predict trends, not exact onset speeds — then calibrate with on-track runs at controlled speeds.

## Key formulas

**Effective platform stiffness**

$$k_{eff} = k + \frac{\partial L}{\partial h} = k + q\,\frac{\partial C_LA}{\partial h}$$

- k: suspension + tyre stiffness of the axle (N/mm)
- \partial L/\partial h: aero load gradient, negative in normal range (N/mm)

**Heave equation with aero**

$$m\ddot z + c\dot z + \left(k + \frac{\partial L}{\partial h}\right) z = F_{road}$$


## Trackside takeaways

- The floor is the most efficient downforce on the car, and the most ride-height sensitive.
- Below the stall height downforce collapses: look for grip loss tied to a ride-height threshold, often with a trap-speed increase.
- Aero stiffness is negative: effective platform stiffness falls with speed squared.
- Porpoising = self-excited aero oscillation (3–8 Hz, grows with speed); bouncing = road-driven. Use FFT on straights to tell them apart.
- Platform tools: heave springs, packers, static RH/rake, tyre pressure, heave damping. Raising ride height always works but costs downforce.
- Plank wear is both a legal limit and a free sensor of where the car touches.

## By car category

### FIA F2

The current F2 car uses an underfloor with tunnels and a large diffuser, making it ride-height sensitive; platform control with heave elements and packers is a big part of setup, and skid wear is checked.

### FIA F3

F3 has significant underfloor downforce; front ride height under braking (front wing in ground effect plus floor) sets entry balance. Watch skid wear after kerb-heavy tracks.

### Super Formula

SF23 was designed to reduce wake sensitivity with more underbody downforce; platform control at the front and plank wear are recurring setup topics.

### LMP2

The Oreca 07 floor works in a ride-height window; Le Mans bumps (Mulsanne, Porsche Curves) force a compromise between plank wear/bottoming and downforce.

### Hypercar (LMH/LMDh)

Hypercars are heavy with large floors: heave elements, packers and the platform under braking (hybrid regen adds pitch on some cars) drive aero consistency. Plank and ride-height checks apply.

### GT3

GT3 floors are flatter and run higher, so classical porpoising is rare, but splitter stall at low front ride height (big kerbs, heavy braking) and rear diffuser sensitivity in roll are real.

### IndyCar

The IR18 underwing tunnels make it RH-sensitive; on ovals, high sustained loads require stiff platforms, and short-oval bumps force softer, higher compromises.

## References

- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 15 Applied Aerodynamics (ground effect, underbody)
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 16 Ride and Roll Rates
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §13.4 Ride Height Measurement
- Other: J. Katz, Race Car Aerodynamics (Bentley, 1995) – underbody and diffuser flow
