Evolution of aerodynamic testing in F1 - Windtunnels

As a first part of the article series on the evolution of aerodynamic testing, we take a closer look at windtunnels.

The principle by which wind tunnels work was first described by Leonardo Da Vinci, whereby a body moving through a static fluid produces the same forces as a static body in a flowing fluid. Wind tunnels have been in use in Formula 1 since the 1970s when the importance of aerodynamics first became clear to teams. Throughout the 70’s and 80’s, into the early 90’s the majority of aerodynamic development in Formula 1 was performed at two academic institutions in the UK, Imperial College London and Southampton University; Peter Wright designed the ground-effect Lotus 79 at Imperial, while Adrian Newey among others favoured Southampton (Pininfarina and Fondmetal in Italy also got a lot of business).

Tired of sharing facilities and seeking a competitive advantage, through the 90’s there was an arms race with teams building larger and more specialized private wind tunnels. This allowed teams to schedule tests any day of the year with larger scale testing producing more accurate results. In addition to this, Imperial College and Southampton University could only accept models of up to ~40% scale; as we will come on to later in this article, the size of the model relative to the actual car is something to be carefully taken into account and so building tunnels capable of accepting full size vehicles (such as Sauber and Honda did) offers additional benefits to testing aerodynamic parts (although teams are now restricted by regulations to 60% scale models).

Wind tunnel basics

All the wind tunnels in Formula 1 are of a type called “closed return”, which means the air loops around a rectangular circuit in an enclosed system. There are several flow conditioners around the wind tunnel designed to aid the passage of air around the corners of the loop, as well as straighten the flow and break down the turbulence generated by the fans. As the air is recirculated around the wind tunnel it is heated by the fan motors, so large heat exchangers (similar to a car radiator) are used to cool the air and maintain a constant temperature - important for measurement certainty.

The most interesting part of a wind tunnel is its working (or test) section. This is the part of the tunnel were the model sits and measurements are performed. The wind tunnel cross section contracts leading into the working section to accelerate the air and create a more uniform airflow. The ground of the wind tunnel has a treadmill-like rolling road which matches the speed of the air and rotates the tyres. There is also an airflow phenomenon called the “boundary layer” (or 99% velocity gradient) which develops on any surface as air moves over it. It is the region of air near to the surface which is influenced by viscosity. The air almost ‘sticks’ to the surface and is slowed by friction. The boundary layer grows as it travels along the surface and can get relatively thick, especially in regions where surface pressure is increasing rapidly (adverse pressure gradient). If the ground plane was stationary, then the ground and car underbody boundary layers would merge in a way which is unrepresentative of the full-size car. Moving the ground means no ground boundary layer forms, which results in a more realistic simulation of underbody flow. Such is the magnitude of low pressure generated by Formula 1 underbodies that the rolling road belt has to be sucked to the ground to prevent it lifting up and stalling the car’s underbody. In some wind tunnels, like Sauber’s, the rolling road sits on a turntable which can yaw the road for cornering simulations.

Schematic of the Toyota Motorsport GmbH (TMG) wind tunnel working section, from Nakagawa et al (2016).

There is a fantastic video available on YouTube by Willem Toet explaining the Sauber F1 wind tunnel which we very much recommend watching. Below you'll find the first video in the series:

The following article hopes to give a slightly more generalized summary of wind tunnel methodologies, as well as an overview of how testing has developed over time.

Wind tunnel usage

Where teams used to run their wind tunnels for three shifts every day of the year, aerodynamic testing in Formula 1 is currently restricted by the FIA Aerodynamic Testing Restrictions (ATR), which are outlined in Appendix 8 of the FIA Formula 1 sporting regulations. The ATR are 7 pages long, so the following is just a summary:

The rules also limit the number of hours a week the wind tunnel can be run, teams are only allowed to occupy the wind tunnel for up to 60 hours, during which time they can only perform up to 68 individual test runs, or a maximum of 25 hours of “wind on” time (wind on time is defined as the length of time where the wind speed exceeds 15m/s), whichever limit is reached first. The time spent in the wind tunnel is then subtracted from the maximum CFD allowance (using the formula below), so teams must plan and balance their aerodynamic test programs efficiently - wind tunnel occupancy is filmed by FIA cameras so there is little to be gained by trying to cheat!

Where:


And for RCFD (Restricted CFD) Simulations Option A :


And for RCFD Simulations Option B :

Below is part of a log of wind tunnel activities from the Caterham F1 team, they were one of three Formula 1 teams using the wind tunnel at TMG in 2014, along with McLaren and Force India (Williams and Ferrari have also used the TMG facilities in the past). At that time the wind tunnel rules were slightly more lax than today; and over a 3 day period (16th-18th September 2014) the Caterham aerodynamicists occupied the wind tunnel for a total of 62 hours, with 83 individual runs completed over 12.5 hours of wind on time. It is possible this represents the total of their allotted testing time for that week with McLaren, Force India, or even Toyota’s LMP1 program taking over the wind tunnel on Friday 19th.

Wind tunnel models

A large part of the difference in wind tunnel testing over the past 20-30 years is the manufacturing methods used to create the wind tunnel models. In the 90’s the cars were handmade, predominantly using wood with some metal (aluminium) components. Through the 2000’s, a lot of carbon fibre was used which required similar lead times for the full-size car (i.e. mould manufacture, layup and curing time). Modern wind tunnel models are predominantly 3D printed directly from the CAD model used to manufacture the full-size car, albeit still with some structural aluminium components, namely the rear wing, front wing mainplane, and suspension members.

Willem Toet with the Benetton B193 model in the Southampton University wind tunnel circa 1993, from Toet “What did I do in Formula 1”, compared to the unraced Caterham CT07 in the TMG wind tunnel in 2014, from @KevTs/CaterhamF1.co.uk.

As mentioned earlier, larger model scales are desirable for improved accuracy, this is because of something called dynamic similarity, which is achieved when the surface flow features of the scale tests match the full-size car. Dynamic similarity is related to Reynolds number (Re), the ratio of speed () and length () to viscosity (),

,

simply put, if the model scale is 50% then the air speed must be 2x greater (assuming the same atmospheric conditions) to match the Reynolds number and produce the same aerodynamic profile. With 40% models, as was the limit in the Southampton and Imperial wind tunnels, the wind speed would have to be 2.5x faster for dynamic similarity. The problem is that faster wind speed requires ever greater power to move the mass of air around the wind tunnel circuit, fan power is already ~3MW, and that to achieve dynamic similarity with higher track speeds the wind speed starts to veer into transonic territory (air behaves differently when transitioning to supersonic speeds). As the RWTT limits teams to 60% scale models at 50m/s, the teams can only achieve dynamic similarity to the full-size car travelling at 108km/hr (67.5mi/hr). Teams try to design cars their cars to be insensitive to Reynolds number, i.e. the downforce coefficient varies little with speed, but dynamic similarity remains a source of some correlation issues between scale and full-size cars.

There are disadvantages to larger scale testing in F1, mainly material cost. The RWTT scale limit of 60% was imposed to try and maintain a level playing field between the richest and poorest teams.

Despite the obvious visual differences there are similarities, mainly the underlying aluminium spine onto which the bodywork and instrumentation are attached. The spine is also used to connect the model to the ceiling of the wind tunnel via the overhead strut which, as well as supporting the model, also shrouds the umbilical of electronic and pneumatic cables between the onboard instruments and the control room. The strut is aerofoil shaped to reduce its aerodynamic impact on the rear of the car, but it does create a wake which affects downstream surfaces and is one of the main disadvantages of wind tunnel testing compared to CFD. Other key similarities and differences between the eras are discussed below:

Model motion systems

The overhead support strut is also used to set the ride height and pitch (rake) of the car. In the 90’s these were the only degrees of freedom available to the aerodynamicists, whereas now the cars have multiple degrees of freedom. The models can be pitched (front & rear ride height), rolled (left and right ride height), and yawed to create aero-maps which include the effects of acceleration, braking and cornering, on top of determining the baseline downforce, drag, and aero-balance. The aerodynamicists can also steer the front wheels for a more accurate representation of cornering aerodynamics. The aero-maps are fed into driver-in-loop simulations, rather than steady state iterative lap time simulations, to help guide set-up choices ahead of a race weekend.

The nature of the ATR means teams also have to be cleverer to maximize the data collected from each FIA classified run. To this end teams have transitioned from fixed condition testing, where the model was driven to a particular attitude and data logged for ~20 seconds before moving to the next attitude, to continuous motion simulations in which data is logged as the model drives through a pre-programmed set of attitudes and steer angles. This model motion can be achieved by physically moving the model support strut using an overhead hexapod system, or via an internal 3-axis model motion system, similar to the commercial units shown below, with the vehicle ride height handled by the overhead strut. Both systems have positives and negatives and use of each can depend on the location of the main force/moment transducer (balance) either overhead, as in the TMG wind tunnel schematic above, or inside the car at the base of the support strut. It is preferred that model yaw be achieved from inside the model to keep the support strut aligned to the tunnel, minimizing disturbance.

Tyres

An obvious difference in the images above of wind tunnel testing in the past compared to now is the use of pneumatic tyres instead of solid metal, conical shaped (for camber) wheels. A stipulation in the contract for Formula 1’s official tyre supplier is that they must supply teams with special tyres for wind tunnel testing. The tyres are reduced in dimension to match model scale but are also designed deform in the same way as the full size tyres.

Another obvious difference is the way the tyres and wheels are mounted in the wind tunnel, today they are attached to the car via fully articulated suspension arms rather than mounted off the body using horizontal spars. The suspension members have to be free moving to prevent reactive forces from the tyre corrupting force measurements. Again, each method has advantages and disadvantages, but the on-body approach is preferred for improved aerodynamic accuracy, at the expense of absolute experimental repeatability.

One thing mounting the wheels to the body does not allow is the ability to vertically load the tyre to accurately deform the contact patch and sidewall profile. The correct geometric representation of the tyres under load and rotation is critical in order to accurately simulate the tyre contact patch squirt (the jet of air pushed out at high velocity from where the rotating tread meets the road) and overall shape of the tyre’s turbulent wake, so teams use a pusher (below) located inside the wheel to alter the sidewall shape, depending on the simulated vertical load.

References

Federation Internationale de l'Automobile. Formula One Sporting Regulations 2018; 2017 Dec [Accessed: 7/5/18];

Nakagawa M, Kallweit S, Michaux F, Hojo T. Typical velocity fields and vortical structures around a formula one car, based on experimental investigations using particle image velocimetry. SAE International Journal of Passenger Cars-Mechanical Systems. 2016 Apr 5;9(2016-01-1611):754-71.

Ogawa A, Yano S, Mashio S, Takiguchi t, Nakamura S, Shingai M. Development methodologies for formula one aerodynamics. Development Methodologies for Formula One Aerodynamics. Honda R&D Technical Review 2009, F1 Special (The Third Era Activities), pages 142{151, 2009.

Toet W. What did I do in Formula 1; 2015 Oct [Accessed: 7/5/18]

Other useful sources

Newey A. How to build a car;
Barlow J, Rae W, Pope A. Low-Speed Wind Tunnel Testing;
James Allen on F1. Insight: F1 wind tunnel technology reaches amazing new levels;
Mercedes AMG Petronas Motorsport. F1 wind tunnel model explained;
Racecar-Engineering. Manor MRT06: http://www.racecar-engineering.com/cars/manor-mrt06/ ;