Rolling Road Wind Tunnels and Ground Simulation

On the road, the car moves and the ground stands still. In a wind tunnel, the air moves and both the car and the ground stand still — and that single inversion is responsible for most of the difference between tunnel numbers and reality.

A rolling road wind tunnel exists to remove it. Everything else in automotive aerodynamics — underbody flow, wheel wakes, cooling drag, rear-end pressure recovery — is downstream of getting the ground right.

1. What a stationary floor does to your data

With a fixed floor, a boundary layer grows along the tunnel floor and arrives at the car. Instead of clean, full-speed air under the vehicle, the underbody sees slowed, thickened flow. Three consequences follow immediately:

Underbody flow is wrong. Ride-height sensitivity, diffuser performance and floor-edge vortices all depend on the velocity profile that is no longer there.
Wheels are wrong. Stationary wheels shed a different wake than rotating ones. Wheel and wheelhouse aerodynamics can account for a significant share of total drag, and none of it is represented correctly.
The wake is wrong. Errors from the underbody and wheels convect downstream and change base pressure — the term that dominates drag on most road cars.

The result is a repeatable measurement of the wrong configuration. Repeatability without fidelity is exactly how programmes end up optimising a car that only exists in the tunnel.

2. The ground simulation ladder

Ground simulation is not a single feature but a series of steps, and modern facilities are built so they can be activated one at a time — which is also how you quantify what each one is worth. In the IVK tunnels (Institute of Internal Combustion Engines and Automotive Engineering, University of Stuttgart), for example, the sequence is documented as: no ground simulation, then boundary layer suction, then the centre belt, then wheel rotation with mini-belts.

Boundary layer suction (BLS). A scoop and distributed suction upstream remove the floor boundary layer before it reaches the car. Tangential blowing can be added to re-energise the near-wall flow. This alone recovers a large part of the error, and it is the cheapest step.

Centre belt. A moving belt under the vehicle centreline gives correct relative motion where the underbody flow lives.

Wheel rotation with mini-belts. Four small belts, one under each wheel, rotate the wheels while the car remains supported. Together with the centre belt this forms the 5-belt system , the configuration most full-scale production-car tunnels use today.

Single wide belt. One belt under the whole vehicle, with the car held by struts from above or by the wheels' hub restraints. It is the most faithful simulation and demands the most: high belt speed and power, cooling, air bearings, precise tension control and a balance system that works around the belt. Motorsport and dedicated aerodynamic development facilities favour it.

3. Blockage: the other half of the problem

A car in a test section displaces a significant fraction of the flow area. In a closed section, that constrains the streamlines and raises local velocity, overstating forces. In an open-jet section, the jet can expand, but a shear layer forms at the jet boundary and its instabilities can drive resonance that shifts the pressure distribution in the test section.

Neither behaviour is a defect to be eliminated — both are physics to be corrected. Automotive practice has converged on correction methods developed by Mercker, Wiedemann, Cooper and co-workers, which account for nozzle interference, jet expansion and deflection, collector interference and the static pressure gradient along the test section, and which have been shown to bring measured drag coefficients toward on-road values.

The practical implication is uncomfortable and important: a drag coefficient is not a property of a car alone. It is a property of the car, the tunnel, the ground simulation state and the correction method. Comparing a Cd from one facility with a Cd from another without knowing all four is a common and expensive mistake.

When quoting a figure, state the configuration. When receiving one, ask for it.

4. What else the moving ground unlocks

Once ground simulation exists, several other measurements become meaningful in the same facility:

Cooling drag and thermal management. Airflow through the front intake, heat exchanger stack and engine or battery compartment behaves realistically only with correct underbody flow — increasingly the dominant question on electric vehicles.
Lift and balance distribution. Front and rear lift, and their shift with ride height, are the numbers that determine high-speed stability.
Aeroacoustics. Wind noise assessment requires the flow field to be right and the tunnel's own noise floor to be low.
Soiling and water management. Where spray goes depends on wheel rotation.

5. Facility implications

If you are specifying or upgrading such a facility, the ground simulation system drives a surprising amount of the design:

Belt speed, power and cooling. Belts running at road speed for hours generate heat and demand precise tension and tracking control.
Balance integration. The force measurement system has to work around belts, wheel restraints and suction plena without contaminating the load path.
Suction plant. Boundary layer removal is a real air-handling system, not a slot.
Section geometry and correction strategy. Open-jet or closed section is a decision that determines which correction method you will live with and how comparable your data will be with others.
Scale. Reduced-scale tunnels (commonly around 1:2 to 1:4) deliver far more test hours per unit of energy and are widely used for development, with full-scale reserved for validation.

The general trade-offs of circuit architecture apply here as everywhere — see

closed-circuit vs open-circuit tunnels

and our

comparison of wind tunnel types

6. Five mistakes

Comparing uncorrected numbers. Two tunnels, two correction methods, one meaningless comparison.
Reporting Cd without the ground simulation state. "With BLS only" and "5-belt, wheels rotating" are different cars, aerodynamically.
Ignoring the pressure gradient. A static pressure gradient along the test section produces a horizontal buoyancy force that shows up directly in drag.
Testing at the wrong ride height. Aerodynamic loads change ride height, which changes aerodynamic loads. Static geometry is not the flying geometry.
Treating CFD correlation failures as CFD errors. Frequently the simulation is right and the tunnel configuration was not what the simulation assumed.

7. Building it in

Ground simulation is not an accessory; it is the part of an automotive tunnel that decides whether the data describes a road vehicle. Decide early which level you need — suction only, 5-belt, or single belt — because each implies different structure, power, cooling and balance design.

We design, manufacture and commission custom tunnels around the measurement programme, with ground simulation, balance and instrumentation integrated from the aerodynamic concept rather than added afterwards.

Tell us the vehicles and speeds you need to test , and we will return a concept: section type, ground simulation level, drive power and footprint.

Sources and notes

5-belt road simulation with boundary layer suction, tangential blowing and four wheel rotation units, and the step-by-step activation sequence (no ground simulation → boundary layer suction → centre belt → wheel rotation): "The New 5-Belt Road Simulation System of the IVK Wind Tunnels", SAE 2003-01-0429.
Open-jet correction practice, including nozzle interference, jet expansion and deflection, collector interference and static pressure gradient effects: "The necessity for boundary corrections in a standard practice for the open-jet wind tunnel testing of automobiles".
Shear layer instabilities and resonance in open-jet automotive test sections: Hu et al., "The low frequency pressure pulsation and control of the open-jet wind tunnel", Scientific Reports 12, 19090 (2022).
Effects of ground simulation and blockage ratio on predicted vehicle aerodynamics: MDPI Vehicles, 2(2), 18.
Reduced-scale ranges of roughly 1:2 to 1:4 reflect common industry practice and vary by facility.
Facility design guidance is TunnelTech engineering practice.