FSAE 2025

The creation of the full-car CFD model began with learning the fundamentals of STAR-CCM+ and the underlying logic of computational fluid dynamics. This involved understanding how CFD simulates fluid flow around objects and how to interpret results effectively. To build proficiency, I initially worked on setting up simple CFD simulations using existing models, focusing on essential parameters such as boundary box definitions, mesh sizing, and layering techniques. These exercises allowed me to grasp the impact of mesh density and refinement layers on simulation accuracy and computational efficiency. Through this iterative process, I developed the skills necessary to structure and execute reliable CFD studies.

As the project transitioned to full-car CFD simulations, the first step was to create a simplified suspension and wheel assembly to balance computational efficiency with accuracy. This step was driven by the powertrain subteam’s need to evaluate the feasibility of mounting radiators on the sides of the car, which required detailed airflow visualization around the suspension and side aerodynamic elements. The simplified suspension geometry was constructed using X, Y, and Z coordinates generated in Optimum Kinematics, defining the suspension points in a 3D space. These coordinates were integrated into SolidWorks using its equations functionality, allowing the model to be easily updated whenever the suspension team refined their layout.

Alongside the suspension assembly, simplified wheel models were developed to incorporate a moving floor and rotating wheels in the CFD model. Using the highly detailed wheel design from previous iterations would have significantly increased meshing complexity and computational time, so the simplified wheels were designed to preserve key aerodynamic effects without overwhelming simulation resources. 

To ensure the accuracy of our aerodynamic simulations, I utilized a benchmark validation approach using the Ahmed body, a well-documented test geometry in computational fluid dynamics (CFD). This validation process was critical for verifying that our CFD methodology, including turbulence modeling, mesh resolution, and solver settings, would provide accurate aerodynamic predictions before applying it to the complex geometry of the FSAE vehicle. 

Initial simulations showed a 10.3% error in drag coefficient (Cd) and a 9.95% error in lift coefficient (Cl) compared to experimental data. This discrepancy indicated the need for further mesh refinement and adjustments to turbulence modeling parameters. After increasing the number of prism layers to 15 and setting the mesh growth rate to slow, the final results achieved a 0.667% error for Cd and 4.75% error for Cl. These results demonstrated that our CFD model could accurately predict aerodynamic performance within an acceptable margin of error. 

In addition to the Ahmed Body study, we validated our simulation settings using the previous year's vehicle. A full-car CFD model was developed based on the prior year's CAD data, and coast-down tests were conducted with the physical vehicle. After completing the tests and processing the data, the experimentally measured drag coefficient (Cd) was found to be 1.20, while the simulated Cd was 1.13. This resulted in a percentage error of 5.83%, which was considered acceptable for validation purposes. 

Building upon the initial full-car CFD model, I continued to refine the simulation to enhance both accuracy and performance insights. This iterative process involved improving the fidelity of the model and collaborating closely with team members by providing feedback derived from CFD results, which helped guide design decisions across the aerodynamic package. 

To increase simulation realism, additional components such as dampers and a radiator fan with a turbulent flow outlet were integrated, better replicating real-world flow behavior. A major update included the introduction of side aerodynamic elements to increase overall downforce. This addition required redesigns of the front wing, nosecone, and rear wing to maintain the center of pressure (CoP) behind the center of gravity (CoG), ensuring aerodynamic balance and predictable handling. 

I also conducted attitude sensitivity studies to explore how yaw, roll, and pitch affected downforce and drag. Based on these findings, I suggested targeted changes to endplates and adjusted the ride heights of the front wing and side elements to reduce sensitivity to vehicle attitude changes. As a result of these refinements, the car achieved a 37.4% increase in total downforce and a 20.1% improvement in the lift-to-drag (L/D) ratio, significantly enhancing the overall aerodynamic efficiency of the vehicle. 

Beyond my responsibilities in CFD development, I was heavily involved in the hands-on aspects of the vehicle’s manufacturing process. I led the complete fabrication of the nose cone, beginning with the generation of a sliced STL shell of the CAD model, which was sent to a sponsor for 3D printing as a male mold. The manufacturing workflow included bonding the printed segments, applying body filler (Bondo), and sanding to achieve a smooth surface finish. This was followed by a wet hand layup for the initial composite cure, and subsequent post-processing to ensure surface quality and dimensional accuracy.

 In addition to the nose cone, I contributed to the production of various aerodynamic components. Our team employed XPS foam cores for airfoil structures and used Divinycell and Nomex cores in areas requiring increased panel rigidity. 

I also took ownership of the vehicle’s livery design, collaborating with a local wrap shop for vinyl application and personally placing each sponsor decal to ensure alignment with our branding and partnership commitments. 

During the testing and tuning phase, our efforts were primarily focused on two key areas: aerodynamic validation and dynamic performance tuning. To validate the CFD simulations, we conducted coast-down tests to experimentally determine the vehicle’s drag coefficient (CdA). The results showed strong correlation with simulation data, with only a 2.08% deviation—real-world CdA measured at 1.329 compared to the simulated value of 1.302. Unfortunately, due to time constraints and integration issues with our load cell electronics, we were unable to validate the lift coefficient (ClA) prior to competition. However, yarn tuft testing was employed to qualitatively assess flow attachment across aerodynamic surfaces. The yarn behavior closely matched the streamline predictions from CFD, indicating good agreement in flow behavior.

During dynamic testing, we observed higher-than-expected roll and pitch rates, which led to ground clearance issues—particularly with the front wing and side elements. The front wing, mounted on adjustable brackets, was easily repositioned. However, the side elements lacked built-in adjustability, requiring physical modifications to the endplate geometry and repositioning of the floor section. To ensure these changes did not compromise aerodynamic balance—especially the location of the center of pressure (CoP)—I re-ran CFD simulations to validate performance retention.

In addition to mechanical adjustments, we also tuned the aerodynamic setup based on driver feedback. To address understeer, we shifted the CoP slightly forward, improving front-end responsiveness. While not quantitatively measured, driver feedback confirmed that the aerodynamic package provided noticeable downforce through all corners and contributed to high-speed stability.

While our competition outcome did not meet expectations, there were several key successes and valuable lessons learned. On the positive side, our vehicle successfully passed mechanical, electrical, and EV active technical inspections—validating that our aerodynamic package was fully compliant with all regulations. During the design review, our aerodynamic package was well received by the judges, earning 13 out of 15 available points. The judge noted that our design was among the strongest in our review queue, which was a significant validation of our development process.

Unfortunately, an issue encountered during rain tech prevented us from participating in any dynamic events. This was particularly disappointing given the strong performance the car demonstrated during testing, and our confidence in achieving a top-10 finish. Reflecting on the development timeline, one key improvement would be accelerating the transition to full-car CFD earlier in the design cycle. Earlier integration would have allowed for more impactful design iterations and better optimization of the aerodynamic package.

Additionally, I would prioritize the development of advanced cornering simulations and initiate yaw, roll, and pitch sensitivity studies much earlier. While straight-line simulations are less resource-intensive and easier to execute, they are not representative of real-world racing conditions, where the vehicle is rarely traveling in a straight line. A more robust focus on vehicle attitude sensitivity would have enabled us to design a package that maintained performance across all phases of cornering, ultimately improving consistency and competitiveness on track.