Additive manufacturing lets you put material only where the load flows and leave air everywhere else — the promise behind every lightweighting story. But lattices are not a free lunch. They add design and simulation effort, they can trap powder or resin, and on some processes they are impossible to clean out. This guide covers when lattices actually pay off, the main lattice families, how they differ from slicer infill, the topology-optimization workflow, and which processes can build and clean them.
When lattices pay off
Reach for a lattice when mass, energy, or surface area is the point:
- Mass-critical parts — aerospace brackets, drone frames, motorsport — where every gram carries a real cost.
- Energy absorption — crash structures, padding, and cushioning tune stiffness and collapse behaviour through cell design.
- Thermal and flow — heat exchangers and filters exploit the enormous surface area a lattice packs into a volume.
- Medical implants — controlled porosity encourages bone in-growth.
It is rarely worth the effort on a cheap FDM bracket that just needs to be a bit lighter; there, slicer infill does the job. Lattices earn their keep when the performance gain justifies the design, simulation, and post-processing cost.
The three lattice families
Lattices fall into three broad families, each with a different manufacturability and performance profile.
| Family | Examples | Character | Strength and notes |
|---|---|---|---|
| Strut / beam | BCC, FCC, octet truss | Nodes joined by straight struts | Predictable and easy to tune; stress concentrates at the nodes |
| TPMS (surface) | Gyroid, Schwarz-P, diamond | Smooth continuous surfaces, no sharp nodes | Excellent stiffness-to-weight and fatigue; great for flow and heat |
| Stochastic | Voronoi, foam | Randomised cells | Organic, roughly isotropic; harder to simulate |
Every strut and wall must clear your process's minimum feature size with margin. Sub-minimum struts print thin, weak, or not at all — and on powder or resin processes, a lattice too dense to clean out is worse than no lattice.
Infill is not a lattice
"Just turn up the infill" is not the same as designing a lattice.
- Slicer infill (gyroid, grid, honeycomb) is a 2.5D pattern extruded vertically through the part. It lives in the slicer, not the CAD model, and it is uniform — you cannot steer it to follow load paths.
- A true lattice is 3D geometry in the model itself. You control cell type, size, and grading — thickening struts where stress is high, thinning them elsewhere — and you can simulate it.
For a functional prototype, infill is fine and free (see the infill density guide for where it tops out). For a qualified, load-following lightweight part, you design the lattice into the model, and the base material sets the ceiling on what it can carry (see the materials comparison).
The topology optimization workflow
Topology optimization and lattices are complementary: topopt tells you where material must be, lattices fill the resulting envelope efficiently. A typical loop:
- Define the design space — the maximum envelope the part may occupy.
- Fix keep-out and interface regions — bolt holes, bearing seats, and mating faces the optimizer cannot touch.
- Apply real load cases — every force, constraint, and safety factor the part actually sees.
- Optimize for a target — minimum mass at a stiffness target, or maximum stiffness at a mass target.
- Interpret the result — the raw output is organic and messy; rebuild it as clean, manufacturable geometry and drop a lattice in where a solid is not needed.
- Validate with FEA and re-run. The first result is never the last.
Tools like nTop, Altair Inspire, Autodesk Fusion, and Ansys handle the optimize-and-latticize mechanics; the engineering judgment in the keep-out, load, and interpretation steps is where the value sits.
Process fit and escaping powder or resin
A lattice you cannot clean is scrap. Process choice decides whether a lattice is even buildable.
| Process | Lattice fit | Why |
|---|---|---|
| SLS / MJF (powder) | Excellent | Surrounding powder supports the struts; no internal supports needed |
| Metal LPBF | Good | Fine struts possible, but needs supports, self-supporting angles, and powder removal |
| SLA / DLP (resin) | Limited | Struts print well, but uncured resin must drain or it cures and traps |
| FDM | Poor | Internal supports cannot be removed; only open, vertical cells work |
On metal LPBF, keep struts self-supporting — generally steeper than about 30–45° from horizontal — or the down-facing struts sag and roughen, and note that strut diameters below roughly 0.3–0.5 mm get unreliable. Match the part to the right process early with the process selection guide.
Design powder and resin escape routes from the start. Give every enclosed lattice volume at least a couple of drain holes — roughly 3–5 mm or larger — at the low points in the build and cleaning orientation. Trapped powder adds dead weight and is a genuine safety and quality issue; trapped resin cures into a solid slug that defeats the whole point.
Lightweighting judgment — process fit, lattice sizing, and reading a topology result — is covered end to end in the Foundations course. Learn the reasoning, not just the buttons.
Start with Foundations See Pro pricingLattice checklist
- Confirm the part is actually mass-, energy-, or surface-critical before latticing.
- Pick a family to match the job — TPMS for stiffness and flow, struts for tunable, stochastic for organic.
- Keep every strut and wall above the process minimum feature, with margin.
- Choose a process that can build and clean it — powder-bed first, FDM last.
- Add escape holes at the low points for powder or resin.
- Validate the design-model lattice with FEA, then a printed test coupon.