Method

Both the new and old rear wing models were designed using SolidWorks whilst simulations were set up using Starccm+ to demonstrate and evaluate the aerodynamics around them. To optimise efficiency, both wings were made symmetrical along the right plane allowing for only half the wing to need to be tested and consequently lower computational load.

Figure 11 – SF71H, Ferrari’s 2018 F1 car [51]

Figure 13 - F1-75, Ferrari’s 2022 F1 car [52]

Figure 12 – CAD and CFD design of the 2018 rear wing

Figure 14 – CAD and CFD design of the 2022 rear wing

The design of the old rear wing takes its inspiration from the Ferrari SF71H, their 2018 racing car (Figure 11). To lower computational costs, the rear wing design was simplified (Figure 12), eliminating non-essential features such as gurney flaps and slits.  

The new rear wing design draws inspiration from the 2022 Ferrari Formula one car, the F1-75 (Figure 13). This design features a prominently curved structure with no sharp edges, where the endplates curve inwards to connect to the main beam. Gurney flaps and slits have been completely removed as the new rear wing has been designed much simpler. This wing design (Figure 14) has been designed symmetrical to ease computational load.

Construction of the 2018 Rear Wing

The main beam is constructed using loft and 3D-sketch techniques to integrate the two different heights (Figure 15). This approach enables the wing to naturally curve towards the middle, encouraging an upwards lift of airflow. The endplates are straight and encase the main beams with two sections being removed from the top right and bottom left of the design. 

Construction of the main beams for the 2018 rear wing

Construction of the 2022 Rear Wing

The 2022 rear wing endplates consist of two distinct components. The top half is straight and connects the endplate to the mainplane, this section is also elevated at the back to aid in the connection of the upper shorter beam. The second section of the wing is angled to encourage inward airflow with an angle of 68.2 degrees. 

The main beam of the 2022 rear wing closely resembles the design used in 2018. Constructed using loft techniques, the beam achieves a spoon-like structure that is more pronounced compared to the 2018 model and the wings extended length causes many modifications (Figure 16). 

Figure 16 – Construction of main beam for the 2018 rear wing

Simulation Set-up

Geometry and Boundary set up:

To optimise computational efficiency, a block is constructed to encompass half of the symmetrical wing therefore halving the computational requirements. boundary layers are subsequently defined for each wall, with the front and rear surfaces designated as the inlet and outlet, respectfully (Figure 17, on the left). The inlet is configured as a velocity inlet with initial condition set at 45ms-1 for cornering speed [53] and 100ms-1 for straight-line speed without their drag reduction system open [54]. The outlet is specified as a pressure outlet, with the pressure set at 1 atmospheric pressure or 101325 Pa. All other walls are set to slip conditions to eliminate any friction except the dividing wall which was made into a symmetry plane. This symmetry condition ensures that the simulation is mirrored onto the other half of the wing as well as allowing for a cross section to be tested as it splits the wing in half.  

Figure 17 – Block boundary layer surrounding the wing.

Mesh:

The mesh configuration has been applied to capture the finer details of the wing’s geometry, ensuring that the small curves and the distinct spoon-like structure are accurately represented. For both wing designs, a combination of meshing techniques has been consisting of polyhedral mesher, surface re-mesher, prism layer mesher, auto surface repair.  

Figure 18 - The mesh design for the old rear wing design

Figure 19 – The mesh design for the new rear wing design

The two mesh configurations (Figures 18 and 19) were designed with a base size of 0.04m, target surface size of 10, surface curvature value of 360 and 5 prism layers. Adjusting the number of prism layers improves the accuracy of the data collected around the wing’s symmetry plane (where the wing meets the boundary layer), providing more detailed insight into all results based around the wings camber such as pressure distribution and velocity vectors. These modifications to the default mesh controls ensure that all the small curves and intricate features to be accurately represented. The validation of mesh is evaluated using five checks: 

 

  • Cell Quality – ratio of ca cell’s longest length to the shortest length. The ideal aspect ratio is 1. The smaller it is, the higher quality of an element it has. This allows for an optimal range of between 0-1. [55]

 

  • Skewness Angle – normalised distance between a line that connects two adjacent cell centroids and the distance from that line to the shared face’s centre. The ideal skewness angle is 0 with an acceptable range of 0-60. [55]

 

  • Face Validity – For better quality cells, the face normal must point away from the attached cell centroid. A face validity of 1 means that they are properly pointing away from the centroid. [56]

 

  • Chevron Quality – Chevron cells are pairs of slender cells that share a common face whose angle is such that the line joining the cell centres does not go through the common face. The ideal chevron indicator value is zero. [57]

 

  • Volume Change – ratio of the volumes of the adjacent cells. The ideal ratio is one. The smaller it is, the higher quality of an element is. An acceptable range for this ratio is between -1 and +1. [55]

 

Although not all values may perfectly align with the ideal metrics, the results fall within limits for running the simulation (Figure 20 and 21). 

 

Figure 20 – Mesh validation for the old rear wing design

Figure 21 – Mesh validation for the new rear wing design

Table 1 – Mesh validation results

Since all mesh validation results fall within their optimum ranges (Table 1), the mesh is deemed both satisfactory and sufficient for the simulation. For the skewness angle of the new rear wing design, all cells predominantly remain below 30 with all cells in the boundary layers valued under 35. 

Simulation Physics

This simulation is conducted in three dimensions for the purpose of realism. Other properties of the flow were determined using the Mach number:

Ma = Mach number 

  • = Airflow velocity (ms-1): 45ms-1 and 100ms-1
  • = Speed of sound (ms-1): 343ms-1

By applying equation 1, we can determine the Mach number for both 45ms-1 and 100 ms-1 inlet velocities. Using the speed of sound, 343ms-1, the Mach number for inlet velocity of 45ms-1 will be 0.131 and for inlet velocity of 100ms-1, a Mach number of 0.291 is calculated. Since both of these values fall below the threshold of Mach 0.3, an assumption can be made that density change is negligible and therefore the flow remains incompressible. 

Reynolds number for a rear wing typically exceeds one million, placing it firmly within the turbulent flow regime. Turbulent flow results in higher downforce and lower drag compared to laminar flow. The behaviour of the airflow for these rear wing designs can be determined using Reynolds equation:

Re = Reynolds number 

𝜌 = Fluid density (kgm-3): Density of air is 1.229kgm-3.[56]

Ma = Mach number 

  • = Airflow velocity (ms-1): 45ms-1 and 100ms-1
  • = Speed of sound (ms-1): 343ms-1

 

By applying equation 1, we can determine the Mach number for both 45ms-1 and 100 ms-1 inlet velocities. Using the speed of sound, 343ms-1, the Mach number for inlet velocity of 45ms-1 will be 0.131 and for inlet velocity of 100ms-1, a Mach number of 0.291 is calculated. Since both of these values fall below the threshold of Mach 0.3, an assumption can be made that density change is negligible and therefore the flow remains incompressible. 

Reynolds number for a rear wing typically exceeds one million, placing it firmly within the turbulent flow regime. Turbulent flow results in higher downforce and lower drag compared to laminar flow. The behaviour of the airflow for these rear wing designs can be determined using Reynolds equation:

Re = Reynolds number 

𝜌 = Fluid density (kgm-3): Density of air is 1.229kgm-3.[56]

V = Fluid Velocity (ms-1): The two inlet velocities being 45ms-1 and 100ms-1.

L = Chord length (m): Old rear wing 0.4 m whilst new rear wings is 0.46 m.

𝜇 = Fluid viscosity (kgm-1s-1): Viscosity of air is 1.73 x10-5 kgm-1s-1. [56]

Applying equation 2, The Reynolds number for the old rear wing design is calculated as 1.3 x106 at an inlet velocity of 45ms-1 and 3.0 x106 at 100ms-1. For the new rear wing design, which features a longer chord length, the Reynolds number is 1.6 x106 at 45ms-1 and 3.5 x106 at 100ms-1. Since the Reynolds number for all configurations exceeds one million, the flow regime is comfortably classified as turbulent. This ensures that the simulation captures all aerodynamic behaviour, such as vortices, accurately under the most realistic conditions. 

The airflow in the simulation is conducted as coupled. Although this method requires more processing time, it is well-suited to handle turbulent models and complex geometries. For similar reasons, K-Epsilon turbulence has been selected. This two-equation model type means in addition to the standard conservation equations, it solves turbulence kinetic energy (k) and turbulent dissipation rate (epsilon). These two equations will be important when determining state of the airflow left in the wake of both rear wing designs, providing valuable insight into the aerodynamic behaviour and performance of the wings. 

A Downforce-Drag single plot (values presented in Table 2) was generated to evaluate the results and assess simulation convergence through iterations of the residual plots. This Downforce-Drag plot was processed until a constant gradient for both forces was achieved. Residual plots provide a graphical representation of the simulation’s convergence by tracking discrepancies in the governing equations across iterations, providing insights into solution stability and accuracy.[57] These plots are computed at every iteration, with the objective of minimising the residuals to a small value, indicating that the solution has converged. Each wing design and respective inlet velocity was processed until all values were at least below 1 (Table 3).  Notably, the new wing design exhibited faster convergence, requiring fewer iterations. This efficiency is attributed to its simplified structure, which reduced computational complexity. 

The results collected in table 2 align closely with theoretical expectations, conforming that the new rear wing generates significantly less downforce. This consistency between simulation outcome and established aerodynamic principles strengthens the reliability of the findings and provides confidence in their validity.