For plastic prototype parts with no steel tooling, two routes dominate: print them, or cast them in a silicone mold. Both deliver parts in days and both avoid tooling amortisation, yet teams routinely pick the wrong one — printing fifty units one at a time, or building a silicone mold for three parts a printer could have finished overnight. The real decision variables are batch size, material realism and how the parts will be judged.
What Each Process Actually Is
"3D printing" covers several different machines. SLA cures photopolymer resins with fine detail and smooth surfaces; MJF and SLS fuse nylon powder — usually PA12 — into tough, functional parts; FDM is the budget option, fine for form checks but visibly built from layers. Turnaround for most prototype batches runs from overnight to 72 hours, and there is no mold to amortise: unit cost stays roughly flat as quantity grows.
Vacuum casting starts from a master model — today usually a hand-finished SLA print — suspended in liquid silicone that cures into the mold. The mold is cut open, then filled under vacuum with a two-part polyurethane that reproduces the master's surface in fine detail. One silicone mold typically yields 15–25 parts before it degrades, and each casting cycle takes roughly 20–60 minutes depending on the resin.
Where 3D Printing Wins
Below about ten parts, printing usually wins on both time and money. There is no mold to build, geometry is nearly free — internal channels, lattices, undercut-heavy clips — and a design change costs nothing but a re-slice. PA12 parts out of MJF and SLS are genuinely functional: living hinges are marginal, but brackets, housings, clips and enclosures for test rigs are routine work. Tolerances are process-dependent; as a working band, expect roughly ±0.3 percent of the dimension with a floor around ±0.2–0.3 mm for well-oriented features.
The trade-off is surface. FDM shows layer lines, SLA carries support scars, and powder-bed parts have a slightly granular texture that needs vibratory finishing before it feels production-grade. Nobody minds on a fit-check bracket; everybody minds on a part headed for a customer meeting.
Where Vacuum Casting Wins
Once a run crosses into the 10–50 range and the parts need to look and feel like production pieces, vacuum casting pulls ahead. Cast polyurethane comes in grades formulated to mimic ABS, PC and even soft-touch or rubber-like materials, and it can be tinted to a specified colour, so parts come out consistent without painting. Because the casting copies the master surface, cosmetic quality that would take hours of print post-processing arrives nearly injection-moulded out of the mold.
Clear lenses and light pipes are the classic example: cast in clear PU against a polished master, they land closer to a moulded optical part than most printed clear resins do after sanding and coating. Colour-matched housing sets for investor demos or first customer samples are the other recurring use case — twenty units, all identical, all looking like the real product.
The economics decide the crossover. A silicone mold is typically a low-hundreds-of-dollars item; beyond roughly 10–20 parts, the per-unit casting cost undercuts printing one piece at a time, and the gap widens with quantity until the mold wears out.
Lead time splits the difference. Printing a batch of twenty parts might take two to three days of machine time; vacuum casting usually needs three to seven working days before the first part exists, because the master has to be printed, finished, then the silicone cured and cut. After that, though, parts come off the mold steadily, batch after batch, with no support removal or refixturing per piece.
The Honest Trade-Offs
Cast PU is close to production thermoplastics, not identical to them. Many general-purpose resins soften somewhere between 60 and 80 °C, so parts living near heat sources need a high-heat grade or a different process. Datasheets describe the resin, not your geometry, so critical snap-fits and threads deserve test parts before commitment. And silicone molds age gracefully but not indefinitely — later shots can drift slightly in dimension and surface.
Printing has its own ledger: FDM parts are weaker across the layers than along them; powder-bed parts come out slightly porous and can wick moisture or oils; some SLA resins turn brittle in thin snap features. None of these are dealbreakers — they are reasons to match the process to the load case instead of assuming "plastic is plastic."
A Simple Decision Rule
- 1–5 parts for fit and form checks: print — SLA for looks, MJF/SLS for function.
- Functional brackets, clips and enclosures in real nylon: MJF or SLS in PA12.
- 10–50 parts that must look production-made — demos, pilot runs, marketing samples: vacuum casting.
- Clear lenses or soft-touch grips: vacuum casting with the right PU grade.
- Beyond 50–100 parts, or when real engineering resins are mandatory: time to price rapid plastic injection molding instead.
Most prototype programs end up using all of these at different stages: printing to iterate, casting to present, molding to scale. Our vacuum casting, 3D printing and molding teams quote across that boundary — which is usually where the cheapest answer hides.
