The most employed method to collect aerodynamic data (above all during the development phase) and to isolate and improve cars aerodynamic performance is wind tunnel testing. This is a very complex field, which could take a series of article in itself just to describe its main aspects. We will try to give a very short overview of wind tunnels and wind tunnel testing.
As the name itself say, this kind of testing is done using a closed environment, a sort of tunnel, where the car is positioned and a powerful fan pushes air in. By using measurement devices while keeping the car in position, the most important metrics (downforce, drag, balance) are extracted for a certain wind speed.
Wind tunnel testing can be run in many different ways, using many different setups both for the car and the tunnel itself.
All wind tunnels have some common features, namely a fan, that moves the air inside the tunnel, a test section, where the car (or a scale model of it) is placed and all measurements are performed, a contraction section before the test one and a diffuser immediately after.
Another commonality to all tunnels is that the fan used to move the air will also tend to rotate it, creating turbulence. This movement will need to be killed, in order to have a straight flow hitting the car/model in the test section. This is why normally wind tunnels also have “anti-rotation” vanes to straighten the flow before it comes to the test section.
Beside these common elements, there are two main type of wind tunnels: open circuit ones and closed circuit ones.
Open circuit tunnels have a pretty simple architecture and are cheaper to build. They are called “open” because the air is taken from outside the tunnel and, after it has gone through it, is expelled again outside.
This kind of tunnels are not often used in the automotive and racing industry, because they are very expensive to run and require a lot of energy (the air and its energy are continuously lost).
The “closed circuit” type is much more expensive to build and require a much bigger area, but it is cheaper to run because the energy transferred to the air by the fan doesn’t get lost and is always reused. As a machine, such a tunnel is much more efficient and, consequently, it requires less energy to move the air at a certain speed.
This kind of tunnels are normally very big and they usually have the fan placed on the opposite side of the test section, as shown in the picture above. The fan itself normally requires a large power to be run (some Formula 1 tunnels have fun requiring a power of about 3 MW).
Temperature control is an important part of air cycle inside such a wind tunnel. Since the fluid always remains inside the tunnel and it receives energy from the fan, its temperature tends to rise. This is of course not desirable and air temperature needs to be controlled in order to have measurements repeatability. For this reason, normally there is also a unit able to cool the air down. Depending on the size of the tunnel, such a unit can also be very big.
Most of the time, wind tunnel testing makes use of scale vehicle models, instead of a real car, both for cost and practice reason: the development is often done when the car has not been built yet. Besides this, the larger the car/model, the bigger must be airspeed in order to test at useful speeds and the bigger are power and costs involved with running the wind tunnel itself.
On one side, it is desirable to use a bigger model or, ideally, a real car, because that helps to improve the accuracy. Using a bigger model requires a bigger test section (otherwise the results are affected by “blockage”, a term that refers to the control sections walls influencing the way air moves around the car/model itself) and a larger fan (power), in order to operate the tunnel at reasonable speeds.
Where the rules allow it, it would be ideal to run a real car in a tunnel able to handle such a big testing object: on one side this would improve accuracy, on the other side we would use exactly the same vehicle for wind tunnel testing and on the track.
However, there are very few wind tunnels in the world that can handle full-scale vehicles. Most of the times, proper engineering is very expensive: having ideal testing conditions in a wind tunnel, would mean a higher construction and operational costs, so a compromise must always be found. Many engineering companies running wind tunnels use scale models, sometimes because the rules mandated it but, often, also because this is the best compromise between costs and accuracy.
Test section design is very important, in order to obtain accurate results. First of all, this area must be big enough, to avoid disturbances during testing. We briefly mentioned blockage, which is what happens when the test section volume is too small compared to the car/model and the air flow gets disturbed.
In the picture above, showing a cylinder in a fluid flow, we see how the streamlines in a free flow situation (on the left side of the picture) bend their path when they meet the body on their way. On the right side, with the cylinder placed in closed volume (test section), we see how the streamlines running close to the chamber’s walls have their trajectory modified by the walls themselves. This changes the testing conditions, making them not comparable to a free stream situation anymore.
Aerodynamic engineers have developed coefficients to correct the results in cases where a blockage is not avoidable, but of course, the goal is always to have a test section big enough (in comparison to the tested model) to avoid similar phenomena.
A very important element is the design of a wind tunnel is the floor, seating below the model. To improve accuracy, above all with racecars with significant ground effects, in modern wind tunnels the floor moves in a similar manner to what the ground would do with respect to the car when travelling on a road. Most of the time wind tunnels also have a feature fore or aft of the moving floor to reduce boundary layer thickness, because it negatively affects results accuracy (above all when car’s ride heights are very small).
Another important aspect is how the car/model is constrained. This has an effect on results accuracy, but also often determines how the forces are measured. Among the many possible architectures, one, often employed with scale models, uses a pylon holding the vehicle from the top:
With this setup, forces are normally measured using a six components scale, able to assess forces in vertical, longitudinal and lateral directions and moments around the three axis. This allows the engineers to also derive the aerodynamic balance.
The pylon is indeed a disturbance for the airflow and, depending on its position, it can affect how the air moves around important components, for example, the rear wing.
An advantage of this setup is that it allows to easily control car/model position, with respect to the airflow and to the ground, setting, for example, predefined yaw or pitch angles.
The wheels themselves are extremely important elements in wind tunnel testing and, nowadays, they also are designed to behave as similarly as possible to the real ones (also in terms of deflection when the load is applied or when they are steered). In some cases, they are separate elements, held by beams lying more or less horizontally, as shown in the previous picture.
Sometimes the car/model is held from behind, but this setup seems to be less common in the motorsport industry.
The situation is slightly different in the few cases of tunnels that can handle actual cars, testing them at “track speeds”. An example of a similar tunnel is the Windshear one, in the US, also used by IMSA for its balance of performance tests.
In such a tunnel, also the way the vehicle is held in position needs to be different, as the image above shows and, hence, also the way forces are measured differs from tunnels where scale models are employed.
Downforce is measured directly under the wheels, thus having more accuracy and a direct reading of balance too. The model is held with tethers attached to the wheel area, which are also used for drag measurements.
The moving floor and the wind can reach a speed of about 290 kph. Windshear is recognized as one of the best wind tunnels in the world and is probably the only one of its kind made available for private customers.
Finally, another alternative to performing aerodynamic testing is to instrument the car with appropriate sensors and running it on a road, measuring the most important metrics using car’s data acquisition system.
We finally have road testing.
It is sometimes the easiest (if not the only) way that teams have to analyze their car aerodynamics and can be a useful approach when they receive no (or wrong) data from the manufacturer. The car must have sensors that allow a direct or indirect measurement of vertical load acting on the suspension system, see for example (ideally) strain gauges or (not optimal) suspension position sensors.
Moreover, a strategy to measure (and, ideally, control) ride heights must also be in place, for example using laser ride height sensors (or deriving ride heights from suspension position sensors, but this normally lead to a much lower accuracy). In any case, the effects of the wheels on the aerodynamics cannot be measured.
The two images above show respectively a strain gauge, as it would be mounted on top of a pushrod or at the end of a suspension link and a potentiometer mounted in parallel to a damper/spring unit, to measure its travel.
Road testing has the advantage of making use of the actual car, in running conditions. However, it lacks the repeatability typical of lab testing, like what can be done in a wind tunnel. Air parameters like temperatures and pressure are extremely important and determine directly air properties and, hence, define the magnitude of aerodynamic forces: in a road test, they fall completely out of team’s control.
Logged data also normally has a lot of “noise”, meaning that the readings vary continuously (just to say a few reasons, the road is never perfectly levelled, the car vibrates, there is sometimes wind blowing in a random direction, etc.).
The two most common tests that are performed in order to measure downforce, balance and drag are constant speed straight line testing and downcoasting.
The first one, consisting in driving the car at constant speed, allows isolating downforce and balance through the readings obtained by load (or suspension position) sensors. It is important to perform this test at a constant speed and on a straight line to avoid inertial effects (due to longitudinal or lateral acceleration) affecting the loads acting on each wheel.
The second one, downcosting, consists in bringing the car at a certain speed and then shifting to neutral (or pushing the clutch) and letting the car to freely decelerate. By analyzing logged data (in particular speed, time and deceleration), it is possible to isolate drag forces.
The downside of this method is that it doesn’t allow to isolate only the aerodynamic drag, as for how the car decelerates is a result of all the drag forces acting on it (aerodynamic drag, but also rolling resistance and friction in other components). Nonetheless, it can be a useful method to understand how big the change obtained with different setups (like, for example, different wing settings) could be.