The mission of this project, sponsored by Hooligan Discs, is to apply a rigorous engineering design process to develop the “farthest flying disc on the market.” The project entails the creation of two distinct driver discs: one that conforms to the technical standards of the Professional Disc Golf Association (PDGA) for competitive use, and a second, unconstrained experimental disc designed for maximum possible distance. My role was the design and development of the PDGA regulations “constrained” disc.

Our team followed traditional requirements, ConOps, and basically that whole systems “V.” However, my goal always remained simple: I needed to make the farthest flying PDGA-legal disc on the market.

Design and Analysis

Initial exploration involved materials, manufacturing, and design trade studies. The aerodynamic theory for the disc golf driver disc was an interesting problem.

Modeling the disk in a stream tube helped me to understand key lift generation characteristics applying Bernoulli’s. Moreover, coupled with rotation, these observations allowed me to understand how to make the disc fly farther for a standard right-hand back hand throw.

The primary tool I used to measure aerodynamic performance of a design before manufacturing and flight test was ANSYS Fluent. These simulations were conducted at a translational velocity of 35 m/s, a rotational velocity of 119.15 rad/s, and SST k-ω turbulence modeling for accurate boundary layer modeling. The values for the velocities were determined based on average professional throw speeds from disc golf players’ forums. From the start, there was a clear understanding that CFD is simply a tool and would require validation, so a wind tunnel testing plan was created before leaning more heavily on CFD to improve designs. Temporarily, though, initial values were compared to similar studies on disc golf discs from research on similar geometry. My initial values for the disc’s C_L and C_D were within 7% of the previously referenced research.
Following the completion of the wind tunnel testing, there was a tight timeline to iterate the design for manufacturing and flight testing, so design studies were conducted based on feedback from players and flight test data.
Initial feedback focused on iteration 1’s over-stability due to its relatively flat flight plate. Disc stability falls into three categories: overstable, stable, and understable. For further distance, an understable disc is better at resisting the natural right-handed
backhand throw’s tendency to curve left for a longer distance. This behavior means, though, that an understable disc can still eventually fade to the left.

Hooligan’s current competitor to the market leader was the Yeet, and I was able to compare its flight path data to the market-leading Innova Destroyer.

So, to improve the design, naturally I first looked to the market leader—the Innova Destroyer.

I whipped out the dremel and calipers to find a few surprising observations. One path to achieving understability and better flight is to increase the “dome” of the disc; however, the destroyer did not utilize a high % of dome. Moreover, this helped point me in the correct direction of disc diameters to start at. After setting up a parametric model, I conducted a few studies in CFD.

The domage study helped me to confirm there is a drop off past a certain ratio of domed diameter to flat diameter of the disc. Moreover, the AOA study shows that the more-domed Iteration 5 disc clearly has better stall characteristics across commonly seen AOA ranges than a flat-top disc (Iteration 1).

The dome design changes were great, but the corresponding manufactured discs still fell 6ft short of the market leader. So, to make the disc more understable, I looked at increasing dome again and lowering the leading edge height of the disc. This is also where the parting line is located for injection molded discs.

This change ultimately led to the best aerodynamically performing disc in CFD and professional flight testing.

Build and Test

Wind tunnel testing data was awful and compliance in the model made it worse. I could have CNC disc out of aluminum or something stiffer for better results.

For prototype design, hand-feel and target weight were essential for better comparisons with the existing injection-molded market leaders.

I initially prototyped using Silk TPU 98A to best match disc material as PETG was rough to hold/throw for the flight test. We then proceeded to test iterations with the TAMU professional disc golf team (best in the nation) against market leading discs.

As my designs improved, the final designs were actually CNC’d out of injection molded blanks. I reached out to a company called ProtoFlyte that has dialed in the process for CNC with higher deflection materials.

Specifically, for testing, we used a weather station to record wind conditions alongside a range finder to measure disc performance. Another critical factor was player feedback and ratings of iterations through google forms and live flight test days.

The pink disc is actually the 4th fully tested iteration (6th CNC disc design iteration), and consistently flew the farthest. This disc averaged 30 ft more of flight than the market leader, and a later iteration I designed was great at rolling on the ground and went even further. These two farthest flying PDGA constrained discs were my final designs I delivered to our stakeholder: Hooligan Discs.

Since I was already working on my single element Camaro wing, I decided to iterate off of that for my final mechanisms project. I attempted to model a Drag Reduction System (DRS) used to reduce drag and increase straight-line speed through active aerodynamics. The requirement for this project, however, that proved most painful was that the mechanism needed to be spatial. The wing from side-view was truly just a planar mechanism (though cantilevered). So, I need to figure out out-of-plane-actuation.

I chose to use bevel gearing for a servo pinion to drive the arm that raises the second element flap.

By simplifying the model to two 4-bar linkages, I was able to figure out the input-output relation from the actuator to the flap opening. I then used MATLAB Simulink to model the actuation and output from my SolidWorks model. I simplified it as the pinion mating gave me a lot of trouble, and the pinion bevel simply gave a 1:2 ratio to the input I was using in MATLAB.

After 3D printing the model (all tolerances I used actually worked first try), this was the assembled result below. Because the mounting was PLA-CF, I did adjust the tolerance for the larger nozzle size I used on my printer, but otherwise the PLA parts behaved as my previous PLA prints have.

Though this oversimplified example ignores operation under high aerodynamic loads and is obviously not the solution I would use in my actual wing, I did learn and gain experience with actuation. I not only had to spec for the torque required, but the friction in the system itself was also a huge factor alongside gear alignment. I would also use some form of stopping device in the endplates to fix the “closed” position of the flap without relying on the servo holding the load. Below is the video of it in action.

After initial research on closed-wheel race car wings, such as GT3 wings, I selected a few design features to lock in: airfoil shape/sizing selection, number of elements, endplate shape/sizing, degrees of twist in the wing, and mounting.

I began with real-life testing on my car (Chevrolet Camaro). I first downloaded a track map of the Spa circuit to determine what speeds I would likely see from my car going into corners. From there, I selected 3 speeds to check for flow attachment on the rear windshield. While I drove the car, I had a friend use my camera to record the tufts on the rear windshield of my car.

Moreover, though I had a CFD model running, I wanted to try and find the true angle of twist required for my wing. The reason there is a twist in the wing is due to the downwash of the rear windshield, which increases the local angle of attack (AOA) of the wing. Thus, I would need to twist the middle portion of the wing to a lower AOA. Though, due to the relatively gradual slope of the rear windshield (especially compared to a vehicle like a Miata), I expected a small required twist.

To test this, I made the jig below.

From this jig, I then created a grid with sharpie on the Lexan endplates. After hot gluing this to my car and mounting my camera, I observed the difference in angle of attack between the outboard and inboard cardboard flaps (required some bending of the flaps to catch more air). This difference from the camera angle appeared to be approximately 4 degrees, which I reflected in my design.

Simulation

For my simulation set up, I used ANSYS Fluent. To start, however, I did basic airfoil analysis in XFOIL.

In my studies, MSHD had the highest possible downforce compared to the other profiles, but produced the most corresponding drag and had “peaky” performance. The last thing you want while driving is the car to be on the edge and lose the rear. I ended up choosing the CH-10 airfoil due to it having decent efficiency, smooth stall characteristics, and it overall performed well as a single element (my original goal was to make a single element wing).

From here, I set up a simulation with the car body and the wing together.

I first optimized vertical and longitudinal location of the wing while also making it rules-legal in GT3 terms and a few other race regulations. From there I optimized the wing angle of attack itself.

This computational study using the native ANSYS optimization function pointed me towards a 12.5 degree starting AOA for my wing for cornering.

The final area I started iteratively designing was the endplate. After optimizing sizing based off of convention and computational optimization, my final endplate sizing appeared as below.

Manufacturing

For manufacturing, I wanted to explore resin infusion. I did not get the chance to attempt it during our 1-year quick and tightly-budgeted design cycle in Formula SAE.

The first area I researched was tooling, but once I saw the costs to CNC molds, I quickly turned to 3D printing.

Though fiberglass molds from the 3D printed molds were an option, I did not plan on producing multiple wings from 1 mold, so budget-wise, a 3D printed mold would suffice.

From this general shape, I needed to prepare the surface for resin infusion. This preparation required SANDING, LOTS AND LOTS OF SANDING. Thought I was done after FSAE, NOPE. I documented my process and initial planning using Miro as shown below, but some work was subject to change as was my process.

I noted what went wrong in each attempt and updated the process accordingly.

For example, this first attempt left a pattern on the mold after I demolded the part. This leftover means the release was poor, and the part was sticking to the poorly prepared surface too much. This problem can be resolved through better sanding, clear coating (avoid orange peel), and a proper buff/polish.

I later added larger flanges to the molds in order to prevent vacuum and resin insertion creases from reaching the part. Additionally, the image above shows a much better-prepared surface, which corresponded to an easy release.

Though some of the wax transferred into the finish of the part, I was able to make it look a lot better with some sanding and clear coat. After completing these tests, my next steps are to print/assemble the big mold, space out resin insertion ports, and attempt the full span’s infusion. The mold will also serve as a jig for where to cut mounting insertion points and for where to bond the ribs (which have a carbon spar running through them).

Motivation

When I entered Texas A&M as a National Merit Scholar on a prestigious scholarship, I
expected the system to work with me a little. Instead, for two years, I was shut out of every research opportunity I pursued. My advisor eventually admitted, “the system has truly failed you.” That moment defined me. I stopped waiting for permission and started creating my own path.
I built my first research experience myself—entering the McLaren F1 CFD Challenge,
writing my own syllabus of work, and earning a perfect score. That self-motivated mindset led me into Formula SAE, where I went from sanding parts to leading an all-new aerodynamics sub team in less than a year. I shaped the design philosophy, ran simulations, manufactured composite parts with my own hands, and tuned the aerodynamics package to improve overall lap time by over 2 seconds.

Design & Analysis

Three things needed to come first before design: interpretation of the rules/identification of loopholes, justification of why the car needed aero (what does an aero kit do?, F=ma_c, etc.) and a full-team philosophy that provided a shared metric for improvement to the car in the form of reduced lap times.

As lead, I was responsible for setting reasonable goals/requirements. In a new 1-year design cycle team, research was essential to achieving the goal. Initial research targeted information that would help our aerodynamics sub team perform trade studies on components to include on the car. The resulting components selected were a high nose cone, multi-element front wing, undertray/diffuser, multi-element rear wing, and dacron body panels. I then referenced existing research for target C_L and C_D values for this kind of package.

As important as overall downforce is to the car, stability and drivability were core to our design philosophy, so the aerodynamic center of pressure (CoP) location was a target. For stability’s sake, I set a longitudinal CoP aft of the center of gravity (CG) along the car’s centerline. Moreover, from a side-view of the car, the CoP needed to be aft of the CG in order to provide stability from side-winds. Finally, CoP migration needed to be minimized to provide a consistent driver feel/confidence, which was a large reason I opted for an undertray as a centralized aerodynamic device.

The system I had to develop to constantly justify aerodynamic changes was running CFD on designs to find aerodynamic values, then taking those inputs alongside car weight, powertrain, and suspension data to run lap time simulations. Hard to argue with our overall argument of the change making the car faster.

How did I set up simulations? How did I arrive at a target speed? Well, the first iterations of designs were designed with a target speed of 33 mph for cornering a hairpin-like radius as the rules for endurance recommend. Then, I worked with suspension and testing car AIM data to find the average cornering speed of an old track with a real FSAE car and back-calculated the average cornering radius for endurance to use for ‘25. Weighting these by prominence on track provided an average speed of 36 mph, which produced the following results.

Moreover, running a Yamaha R6 engine allowed us to not worry about drag for endurance and autocross (our prioritized events as a full team). Regardless, I did estimate the bhp absorbed by our endurance set up. This was the equation I used from McBeath’s book on Competition Car Downforce:

Working with a teammate, I managed to run simulations for straight-line and sensitivity studies. These sensitivity studies (dynamic conditions: pitch, roll, yaw) were essential as this was the school’s first ever ground effect FSAE IC car. Undertrays are notoriously sensitive to dynamic conditions, so I interfaced a lot with suspension to figure out max roll, pitch, yaw (we even ended up using stiffer springs than we designed for as we got faster lap times).

Overall, simulations were run with a target y+, factoring in our use of wall function because the supercomputer was down, so we had to reduce computational demand. Moreover, I validated our simulation methodology initially using Ahmed Body studies. We did not gain access to a sizeable wind tunnel to thoroughly validate our CFD. So, I modeled the existing Ahmed Body (similar to automotive design) and compared our CFD results to real life existing wind tunnel data for that bluff body. We fell within 7% while being able to run simulations on our laptops and found a happy medium for iteration there. Additionally, I selected k-omega SST turbulence modeling (lots of boundary layers and their interactions).

Though I was responsible for the aero package working well as an ecosystem, I also had to design and manufacture my component: the undertray/diffuser.

Key design points for my undertray:

• Steeper inlet angle than diffuser exit
• Create pressure gradient (high front, low pressure rear) to accelerate air underneath the car
• Bernoulli’s effect
• Maximize diffuser exit length for flow attachment
• Encouraged side-tunnels over purely underbody
• Shorter throat is less sensitive to changes in ride height/dynamic conditions
• Lip/floor edge houses vortex that helps seal the flow of the tunnels
• Rear wing (if you have “beam-wing” style lower elements) can “extend” the diffuser and allow for a higher exit angle while helping keep flow attached and low pressure at the exit
• An open venturi system differs from a quasi-1D tube in the way that it is dominated by vortices. Flow is energized and stays attached through the control of vortices
• Inlet in-board of tire wash
• Do suspension members require fairings?
  • Tufts showed minimal turbulence ahead of the floor
• Strakes help to maintain vortices in corners
• Experimented Diffuser angle 11-13°, 13° had peak downforce and minimal increase in drag

My iterative design involved computationally determining inlet and outlet angles for the diffuser, then slowly modifying the design to perform better in dynamic conditions. The greatest improvement to the design came from introducing more vorticity with strakes. Below are some CFD results to better understand the flow:

Manufacturing

I led almost all layups, tracking durations and procedures through a “layup sheet” on my iPad. This process was helpful in not repeating mistakes and improving on every layup.

For my Undertray:

• 6 layer wet layup (overbuilt – hindsight is 20/20)
  • Conducted bend tests on test pieces of different # layers to determine optimal rigidity
  • Vacuum bag failed to properly cover complex contours
  • 3D-printed, sanded, bondo’d mold
  • A mold was 3D-printed out of ASA to create the 2 diffuser tunnels.
  • 6 layers of 3k carbon weave were then wet laid over the mold and a flat center area.
• Tunnel insides were repaired and finished with flash tape, epoxy, and 2-part clear coat for smooth wetted surfaces
• The underbody was fastened to the chassis using DZUS fasteners (flush, minimal aero impact), and the tunnels required rods to prevent flexing under load.
• Strakes were 3D-printed and attached with adhesive
• Rods used 10-32 bolts with nylock nuts
  • I designed a jig to align and weld the chassis tabs
  • Slot-head bottom bolt to use the same tool to take off DZUS and bolt, removing diffuser fully

Testing

Testing involved a lot of back and forth with suspension to find the optimal ride height for overall vehicle dynamics and underbody performance (lap time trumps all). Moreover, I did not model the suspension rods in our CFD simulations and wanted to check their influence on flow entering the inlet. Thus, I ran some tufting as shown below. At endurance/autocross speeds, our GoPro footage showed no problems in that area.

I also tuned the front wing and rear wing for overall better aero balance and faster lap times. Upwash from the front wing decreases rear wing performance, and this effect grows under braking, leading to a less predictable car. I took driver feedback and lap times to mitigate aerodynamic balance problems.

(We also did some CFD validation with coast down testing and skid pads at varying radii, but the skid pad results were not usable).

Final Notes

There is definitely more depth to a lot of this that I am happy to have conversations about, but I truly could not fit the past year’s worth of work in a portfolio post. More importantly, however, I learned firsthand the difference that a good team makes. I could not have done it without y’all aero boys: Nick, Kevin, Mauri, Liam, and Sid. I remember waking up every morning coughing up blood for 2 months straight, but you guys helped remind me I had a reason to show up (eventually realized my apartment unit had bad mold poisoning). Thank you for a season of memories and delivering when life called on you.

“To do something well is so worthwhile that to die trying to do it better cannot be foolhardy. It would be a waste of life to do nothing with one’s ability, for I feel that life is measured in achievement, not in years alone.”

Bruce McLaren