Most additive failures are process-selection failures wearing a disguise. The part warped, the surface disappointed, the unit cost exploded — because the part was on the wrong process before the first layer went down. Here's the filter we teach engineering teams.
The seven families in one table
| Process | Mechanism | Best for | Watch out |
|---|---|---|---|
| FDM / FFF | Melted filament extrusion | Fixtures, jigs, functional prototypes, enclosures | Layer lines, weak Z axis |
| SLA / DLP | UV-cured liquid resin | Fine detail, patterns, visual and fluidic parts | Brittleness, UV aging |
| SLS | Laser-fused polymer powder | Support-free functional nylon parts | Cost, grainy surface |
| MJF | Agent + IR fused powder | Production polymer volumes | Dimensional spread, gray-only |
| Metal LPBF | Laser-melted metal powder | Complex high-value metal (aero, medical, tooling) | Cost, supports, mandatory machining |
| Binder jetting | Glued powder, then sintered | Higher-volume metal parts | Sinter shrink and distortion |
| DED / WAAM | Welded wire/powder deposition | Large near-net shapes, repair, cladding | Coarse — always finish-machined |
The four filters, in order
- Material, non-negotiables first. Temperature, chemical exposure, certification (biocomp, flame, food). If the part must be PEEK or 17-4PH, most of the table just disappeared — good, that's the filter working.
- Geometry. Internal channels and consolidated assemblies point to powder beds (SLS/MJF/LPBF), which need no supports (polymer) or tolerate complexity (metal). Large and chunky points to FDM or DED. Fine detail points to SLA.
- Quantity and unit economics. Additive has no tooling, so it wins at 1–1,000 and usually loses to molding at 10,000+. Run the crossover: mold cost ÷ (molded-unit savings) = break-even quantity. Between those bands, MJF and binder jetting stretch additive furthest.
- Finish and tolerance. Decide what surfaces matter, then check the process can hold them — or budget the machining/finishing step honestly (see our tolerance guide).
Design rule
Reject processes with specific reasons — 'SLS is out because the part exceeds the build volume and needs a certified ESD material' — and the shortlist defends itself in the design review.
Fast heuristics that hold up
- Fixture or jig, needed this week → FDM, almost always.
- Snap-together consumer-grade housing, 500 units → MJF.
- Master pattern, dental, or microfluidic detail → SLA/DLP.
- Ducting, brackets with internal lattice, one-piece mechanisms → SLS (polymer) or LPBF (metal, if the value justifies it).
- Meter-scale metal or a worn shaft to rebuild → DED.
This is the exact framework in our free AM Foundations course — seven process families, nine decision filters, and a worked shortlist exercise. Free with an account, no card.
Take Foundations free Next: the DfAM checklist