eRacing Magazine Vol 2. Issue 7 | Page 50

Complications of Computation

When Computational Fluid Dynamics (CFD) was first introduced into the motorsport industry it was clear that the potential of this method was huge. Teams and companies recognised this and immediately began investing in the technology which resulted in fast development rates; improving both computing power and accuracy. ‘We can now achieve a CFD case run in less than 4 hours between when we submit the case and get the results,’ highlights Franck Sanchez, Principal Aerodynamicist at Scuderia Toro Rosso; demonstrating just how far this technology has come.

However, there still remains fundamental issues within CFD calculations as these mathematical models can only simulate what the engineers have included in the coded equations, and not what hasn’t been included. Therefore, the accumulation of these assumptions and the errors associated with numerical analysis, results in an overall correlation delta between CFD and Wind Tunnels and thus reality. This is why even though the regulations have cut Wind Tunnel testing time by 70% since 2013 for Formula One teams, they still utilise this valuable time for testing parts. Every component that you see on an F1 car on track, has been validated in the Wind Tunnel as this is still the most accurate method.

Furthermore, although many automotive and motorsport projects are sold as ‘CFD’ development only, which illustrates the impressive advances in CFD, these projects still rely on a degree of Wind Tunnel testing. Take the new Jaguar XE, it is JLR’s (Jaguar Land Rover) most aerodynamically efficient saloon car yet, with an impressive drag coefficient (C_D) of only 0.26 counts.

‘The XE’s development process is virtually-lead, with no aerodynamic test properties,’ explains Adrian Gaylard, Technical Specialist in Computational aerodynamics at Jaguar Land Rover. ‘This proves that virtual development of an all

new automotive product is viable.

’This is somewhat true, however alongside the 1,200 CFD simulations and over eight million CPU (Central Processing Unit) hours, Wind Tunnel testing was still necessary to acquire surrogate data and used for final tuning. Therefore, this project was not entirely completed using CFD alone, and furthermore, the CFD codes used would have been based on Wind Tunnel data to start with for validation.

As Jan Jagrik, Chief Aerodynamicist at Skoda Auto summarises; ‘can you develop a proper CFD simulation without Wind Tunnel measurements? No. Before you can achieve an accurate result from CFD, you have to complete many Wind Tunnel experiments.’

A further example is the famous case of the Marussia Virgin Racing (now Manor Racing) VR01 which was the first Formula One car ever to be developed through CFD alone. In fact, throughout the team’s first two seasons, no Wind Tunnel tests were completed at all.

This radical approach was applied to try and accelerate Marussia’s progress up the grid as being a new team, they knew that they had to do something different to be in line for a podium finish. Although a brave move, it unfortunately didn’t quite pay off as they expected; during 2010, the team finished the Championship last and had to retire fifteen times. In 2011, they started no higher than the tenth row of the grid for the first six races. This disappointing start led them to part with their CFD technical partner, Wirth Research. Obviously, this is not purely down to CFD codes as teething problems are common for new teams, but it does illustrate that CFD alone was not sufficient enough yet to be the only development tool for a vehicle, even with the resources of an F1 team.

Wind Tunnel Woes

It is important to remember that Wind Tunnels are by no means perfect. Firstly, they are not fully representative of the real world either - a Wind Tunnel is itself a model, a simulation of the real world, just like CFD is. ‘I hear people say, ‘it’s measured’ as if that shows that it is the true value. With CFD we are very critical, however what is physically measured in a Wind Tunnel is still a simulation,’

highlights Professor Dr. -Ing Jochen Wiedermann from FKFS. ‘Factors such as Wind tunnel blockage, the width of the Wind Tunnel, whether the ground plane is moving or not, the type of boundary layer control, turbulence and the effects of instrumentation all result in a pressure gradient which is not present in real life especially in open jet Wind Tunnels – so you cannot just believe the results. Furthermore we are not using Wind Tunnels that are designed for automotive applications. The Automotive industry inherited Wind Tunnels from the Aeronautical industry around 70-80 years ago. Therefore, this is what these Wind Tunnels were designed for, yet we use them to optimise our cars with.’

An example of the consequences of these fundamental issues with Wind Tunnels can be found in the debut of the Drag Reduction Systems (DRS) in Formula One back in 2011. ‘These DRS systems were tuned in the Wind Tunnel, they went out on track and they didn’t work,’ explains Mark Gillan, Director of Motorsports Technology Group at MTS and who previously worked in Williams F1, Jaguar Racing, Toyota Motorsport GmbH and McLaren F1. ‘This was due to the fact that the intensity levels were too low in the tunnel, which meant that the separation point on the rear wing wasn’t properly reproduced. Therefore, we introduced artificial intensity into the tunnel and suddenly they had to be re-optimised and then they began working on track due to a better correlation.’

This is the key to achieving repeatable and accurate results – optimising the correlation of the Wind Tunnel with reality and this is achieved in many ways.

Images: Exa_ Jaguar XE simulation (1)(1) & Jaguar XE Flow structures

The new Jaguar XE was pitched as a CFD only project, however Wind Tunnels still played a part in the development process.’