Prototypage rapide : CNC vs impression 3D vs coulée sous vide (guide de décision 2026)
Choisir le bon procédé de prototypage selon quantité, fidélité et matériau : CNC vs impression 3D vs coulée d’uréthane comparés sur coût, délai, finition et matériaux — avec logigramme et points de bascule de coût.
Trois procédés dominent le prototypage rapide — l’usinage CNC, l’impression 3D et la coulée sous vide (uréthane) — et le moyen le plus rapide de gaspiller de l’argent est de mal choisir pour l’étape en cours. Ce guide organise le choix autour des trois variables décisives — quantité, fidélité et matériau — et montre comment un programme intelligent utilise les trois à mesure qu’il monte en échelle.
The quick answer (by quantity)
The three processes
- CNC machining — subtractive: cut the part from solid stock. Real production materials, tight tolerances, excellent finish. Our CNC service covers metals and plastics.
- 3D printing — additive: build the part layer by layer (SLA, SLS, MJF, FDM). Fastest for complex geometry, zero tooling. See rapid 3D printing.
- Urethane casting — replicate: make a master (usually printed or machined), pour a silicone mould, cast polyurethane copies under vacuum. Great finish and consistency for tens-to-hundreds. See urethane casting.
Master comparison table
| Factor | CNC machining | 3D printing | Urethane casting |
|---|---|---|---|
| Method | Subtractive | Additive | Replication (mould) |
| Sweet-spot quantity | 1–20+ | 1–10 | 20–500 |
| Tooling / setup | Fixturing only | None | Silicone mould (a few days) |
| Lead time (first part) | Days | Hours–1 day | About a week (mould + cast) |
| Materials | Real metals & plastics | Resins, nylons, some metal | PU resins mimicking ABS/PP/rubber |
| Tolerance | Best (±0.02 mm) | Moderate | Good (from master) |
| Surface finish | Excellent | Layer lines (post-proc) | Excellent (mould finish) |
| Strength / properties | Full material props | Anisotropic, weaker | Good, near-production feel |
| Cost per part (low qty) | Medium–high | Lowest for 1–few | Low once mould exists |
| Geometry freedom | Limited by tool access | Highest (freeform) | Follows the master |
Choose by quantity
Quantity is the strongest single signal. A single concept model? Print it. A handful of parts that must behave like the real thing? Machine them. Fifty to a few hundred presentation or pilot units? A silicone mould pays for itself fast in urethane casting. Past a few hundred repeat parts, the economics point to hard tooling (injection moulding) or production machining.
Decision flow
1. How many parts do you need now?
1–10 → lean 3D print or CNC. 20–500 → urethane casting is likely cheapest per part. 500+ → look at injection moulding / production CNC.
2. Must it be the real production material?
Yes (metal, specific engineering plastic, real strength) → CNC machining. No (form/fit/appearance) → printing or casting are cheaper.
3. Is the geometry impossible to machine?
Internal lattices, organic freeform, conformal channels → 3D printing is the only route.
4. Do you need many good-looking, consistent copies?
Presentation units, user testing, pilot run → urethane casting from a printed/machined master.
5. How tight are tolerance & finish?
Tight tolerance / fine finish → CNC. Moderate → printing (with post-processing) or casting from a good master.
6. What’s the timeline?
Need it tomorrow → 3D print. Days → CNC. About a week for a batch → urethane casting.
Choose by fidelity & material
Need real material / properties
- Functional test in aluminium, steel, or a specific engineering plastic → CNC.
- Full isotropic strength, real thread and bearing fits → CNC.
- Regulated material traceability → CNC.
Need form, fit or appearance
- Ergonomics, design review, packaging fit → 3D print or cast.
- Rubber-like or clear parts, overmoulds → urethane casting (PU resins mimic ABS/PP/rubber).
- Complex internal geometry a cutter can’t reach → 3D printing.
The cost crossover
Each process has a different cost curve as quantity rises, and where they cross is what determines the cheapest route for your batch:
- 3D printing has near-zero setup but a flat, relatively high per-part cost — cheapest at 1–few, but it never drops much with volume.
- CNC machining has modest setup (fixturing/programming) amortised over the run; per-part cost falls with quantity and it delivers real materials.
- Urethane casting has an upfront mould cost, after which each cast part is cheap — so it wins in the tens-to-hundreds range once the mould is amortised.
- The practical rule: printing for the first few, CNC when you need real parts or moderate volume, urethane casting once you need dozens of consistent copies, hard tooling beyond that.
For a deeper look at squeezing cost out of machined parts specifically, see our cost reduction guide.
Using all three across a program
The best NPI programs don’t pick one process — they move through them as the product matures. A typical path:
Concept: 3D printing
Print early iterations overnight to check form and ergonomics cheaply while the design is still moving.
Functional test: CNC machining
Machine a few parts in the real material to validate strength, fit, and tolerance before committing.
Pilot / user testing: urethane casting
Cast tens-to-hundreds of consistent, good-looking units for field trials, marketing and pilot builds.
Production: injection moulding or production CNC
Once volume and design are locked, transition to hard tooling or a production machining cell.
Prototype-to-production tips
Frequently asked questions
The questions teams ask most when choosing a prototyping route.
Foire aux questions
- It depends on quantity, fidelity and material. For 1–10 concept parts or complex geometry, 3D printing is fastest and cheapest. For 1–20 functional parts that must behave like production (real material, tight tolerance), CNC machining wins. For 20–500 consistent look-and-feel parts, urethane casting from a silicone mould is usually cheapest per part. Above a few hundred, move to injection moulding or production CNC.
- Once you need enough parts to amortise the silicone mould — typically from a few dozen upward. 3D printing has near-zero setup but a flat per-part cost that never drops much with volume; urethane casting has an upfront mould cost but very cheap parts afterward, so it wins in the tens-to-hundreds range with better finish and more consistent properties.
- When the prototype must behave like the real product. CNC parts are made from actual production materials (metals, specific engineering plastics) with full, isotropic strength, real thread and bearing fits, and tight tolerances — none of which a printed part reliably delivers. For functional testing, regulatory materials, or anything load-bearing, machine it.
- Polyurethane casting resins are formulated to mimic common production plastics — rigid ABS/PC-like, flexible PP-like, and rubber-like elastomers across a range of Shore hardnesses, plus clear and coloured options. It’s a strong way to get near-production look and feel in low volumes before committing to injection-mould tooling.
- Yes — and the best programs do. A typical path is: 3D-print early concepts, CNC-machine functional test parts in the real material, urethane-cast a pilot batch, then move to injection moulding or production CNC for volume. You can also split an assembly by requirement — machine the critical bracket, print the housing, cast the soft grips.
- 3D-printed parts can be ready in hours to a day. CNC-machined prototypes typically take a few days depending on complexity. Urethane-cast batches take about a week because a silicone mould has to be made first, after which casting copies is quick. Tell us your deadline and we’ll route the job to hit it.
CNC, 3D printing or urethane casting — which is best for prototyping?
When is urethane casting cheaper than 3D printing?
Why machine a prototype instead of printing it?
What materials can urethane casting mimic?
Can I use more than one process on the same project?
How fast can I get prototype parts?
Recommended services
Move from research to a quote in one click.
Articles liés
À propos de l'auteur
JLYPT Engineering Team
Rapid Prototyping & NPI Engineers
We run all three prototyping routes — CNC machining, 3D printing and urethane (vacuum) casting — and move customers between them as a program scales from first concept to pilot production. This guide is the selection logic we apply so you spend on the right process at each stage instead of over- or under-building a prototype.
Besoin d'un devis pour un projet similaire ?
Téléversez vos fichiers CAO — nos ingénieurs répondent sous 24 h.
Devis gratuit