Manufactured Racecar Model

The assignment fixes the deliverable and the division of labor between processes. The car body is cast in silicone from a mold I designed and 3D printed. The wheels are injection molded from plastic into a metal mold machined from my own CAM program. The cockpit is 3D printed. The axles are turned metal rod, held in the body by printed tubes that get overmolded into the casting.

The brief suggested a 1920s Grand Prix car because the body shapes are simple: a profile whose height falls away from the centerline toward the sides releases from a two-part mold cleanly and prints without supports. I picked a Formula 1 car, which is the opposite of that, and took on the redesign work that came with it.

The body mold is a two part split with the parting line running along the car's centerline, dowel pins for alignment, and an injection port at one end. Getting there took several rounds of edits to the car CAD itself: wall thicknesses raised where silicone would not fill, features pulled back where they would lock into the cavity, and geometry simplified where the F1 shape fought the draft the mold needed.

Both halves were printed in PLA and clamped together for injection. Printed tubes for the axles sit in the cavity before the pour so the silicone forms around them.

The cockpit is negative space in the casting, so a core fills that volume in the mold and is pulled after the silicone sets. The printed seat and dashboard then drop into the cavity it leaves, a snug fit backed up with glue.

This is where the F1 choice cost me. At this scale the steering wheel came out thin enough that it snapped off the print every time, and the cockpit itself was smaller than any of the three suggested cars would have given me.

The wheels are the machining half of the project. Each mold half starts as a 1.875″ × 1.875″ × 0.5″ block of 6061 aluminum, chosen because it takes heat from molten plastic and cuts easily with HSS tooling. Aluminum also means the die survives repeated shots, which mattered more than I expected.

The CAM program runs nine operations: an adaptive clear, four scallop and parallel finishing passes to bring out the tire and rim contours, then a center drill and a #30 drill for the injection hole. Tools step down from a 1/8″ end mill to 1/16″ for the tight radii. It was cut on a Tormach 770M at 10,000 rpm, well inside the machine's travel for a part this size.

This was the hardest part of the project and the least forgiving. The press has to be aligned and operated quickly, and it shifts as you orient it, so a shot that misses the injection hole or goes in too slowly produces a short, unusable wheel. I missed the hole, injected too slowly, and the hole itself turned out too small. It took more than ten trials to get four clean wheels in PLA.

The fix was mechanical, not procedural: I drilled a larger hole through the mold so more plastic could enter per shot. After that the process was straightforward.

The axles were turned on a lathe to length and diameter for a snug fit in the over-molded tubes with the wheels running true. Assembly is where the tolerances between four different processes meet: a cast part, a molded part, a turned part, and a printed part all have to agree.

The model came together and rolls.

The course’s final report asked how the same model would be produced industrially rather than in a shop, which reframes every choice made above. For the body, sand casting in A356 aluminum beats die casting at a quantity of one: a printed resin pattern and a packed sand mold cost a fraction of a machined die, and the loose tolerance sand gives is irrelevant for a display model. Pouring at 730 °C keeps flow and fill ahead of solidification.

The gating geometry follows from Bernoulli and mass continuity. A 0.15 m pouring height gives 1.716 m/s at the sprue exit and a flow rate of 1.72 × 10⁻⁴ m³/s, and the Reynolds number of 18,747 sits between laminar and turbulent, which means pouring has to stay constant or air gets entrained. Matching area to velocity down the sprue sets a 6.45˚ taper so the stream stays in contact with the walls.

The same treatment applied to the wheel mold: cutting 6061 at 10,000 rpm and 60 in/min with a 10 mm two-flute cutter gives a 0.076 mm undeformed chip, a 28.3˚ shear angle, and roughly 26 kW of cutting power at full width, numbers that decide whether a process is realistic before a machine is ever switched on.

Most of what went wrong traces back to one decision: an F1 car has thin, small features compared to the 1920s bodies the brief proposed, and every one of those features became a manufacturing problem downstream. The cockpit was cramped, the wheels were small, the steering wheel would not survive printing, and the wheel mold needed rework to fill.

What went well was the response to those problems. The CAD was revised when a feature would not cast, the mold was redimensioned when it would not fill, and the injection hole was opened up when the press could not keep up. More design iterations before committing to manufacture would have caught several of them earlier, which is the part worth carrying forward.