Design for Metal AM: the practical rules engineers actually need

Most DfAM guides give you a list of rules without explaining why they exist. This article does the opposite: for each rule, you get the underlying physics, the typical numbers, and the engineering judgement call you'll need to make.

The focus is metal powder bed fusion — LPBF and EBM — because that's where the constraints are tightest and the consequences of ignoring them are most expensive.

Why metal AM has different design rules

Metal LPBF builds a part by melting 20–60 µm layers of powder with a laser. Every layer is deposited on top of either solid material or loose powder. Solid material conducts heat away efficiently; loose powder does not. This asymmetry drives almost every DfAM rule.

When a layer is deposited over a down-facing surface (a surface that overhangs loose powder), the melt pool has poor thermal support. It cools more slowly, resulting in:

• A rougher surface (Ra 15–40 µm vs. 8–15 µm on up-facing surfaces)
• A larger melt pool with more tendency to sag or collapse
• Potential lack-of-fusion porosity in the first few re-solidified layers

The critical question for every design feature is: does this surface have adequate support beneath it?

Rule 1 — Self-supporting angles

The number: LPBF can print surfaces down to 45° from horizontal without support structures in most alloys. Some materials and machines can push to 35–40°, but 45° is the safe conservative threshold.

The physics: At 45°, each successive layer only overhangs by one layer thickness (20–60 µm) beyond the layer below. The partially melted powder beneath provides just enough thermal conduction to produce an acceptable surface. Below 45°, each layer overhangs by more than it extends into previously melted material, and the surface quality degrades rapidly.

EBM exception: EBM builds in a preheated powder bed (600–900°C). The sintered powder provides real structural support. EBM can routinely self-support down to 20–25° from horizontal without support structures.

Practical action: In CAD, identify all down-facing surfaces early. Reorient the part to bring critical surfaces above 45°, or chamfer/fillet transitions to eliminate near-horizontal faces.

Rule 2 — Minimum wall thickness

The number: Minimum wall thickness depends on the layer thickness and powder particle size. Conservative guidelines for LPBF:

Below the minimum, you risk:

1. Print failure — thin walls can curl or delaminate due to residual thermal stress
2. Dimensional instability — wall may be printed but varies ±0.1 mm, making it unreliable
3. Post-processing damage — support removal, blasting, and machining can destroy walls below 0.4 mm

Wall aspect ratio matters too. A 0.5 mm wall that is 50 mm tall (aspect ratio 100:1) is much riskier than the same wall at 5 mm tall. High-aspect-ratio walls accumulate thermal stress over many layers and are prone to cracking. As a rule of thumb, keep wall aspect ratio below 40:1 without additional stiffening ribs or support.

Rule 3 — Internal channels and conformal cooling

Internal channels are where AM delivers unique value — you can create cooling circuits, flow passages, and structural lattices that are impossible to machine. But channels have constraints too.

Self-supporting cross-sections:

The teardrop modification is the standard solution for circular channels: convert the top arc of a circle to a pointed apex. The pointed top is self-supporting regardless of channel diameter. This change costs nothing in CAD and eliminates internal support that would be impossible to remove post-build.

Minimum internal channel diameter: 0.8–1.0 mm for reliable powder removal. Smaller channels can be printed but depowdering becomes critical — use vibration, compressed gas, and/or ultrasonic cleaning. Channels below 0.5 mm are very difficult to guarantee powder-free.

Surface roughness inside channels: As-built internal surface Ra is typically 15–30 µm — much rougher than external surfaces. For heat exchangers or high-flow hydraulic circuits, specify abrasive flow machining (AFM) or electrochemical polishing to achieve Ra <5 µm internally.

Rule 4 — Support structure minimisation

Supports are not free. They:

• Add build time and material cost (typically 10–30% of build volume)
• Require post-processing labour to remove (often the most expensive manual step)
• Leave witness marks on the surface they touch
• Can be impossible to remove from internal features

Strategy 1 — Orient to eliminate supports. Before generating any support, ask: is there an orientation where all overhangs are self-supporting? Rotate the part systematically. A few degrees of orientation change can eliminate an entire support structure. Use the orientation advisor to find the minimum build height and support volume combination.

Strategy 2 — Design functional overhangs at 45°. Chamfer all horizontal shelves to at least 45°. Replace flat horizontal bases with tapered feet or standoffs with angled undersides.

Strategy 3 — Add self-supporting features. Replace a flat-bottomed pocket with a diamond or gothic-arch cross-section — same function, no support needed.

Strategy 4 — Accept support only where removal is accessible. When supports are unavoidable, ensure they contact accessible external surfaces. Position support contact points near edges and holes where tools can reach. Specify a contact distance of 0.1–0.2 mm (not zero) so supports snap off cleanly.

Strategy 5 — Use support-free lattice fill. In large hollow sections, use a sparse lattice infill (gyroid, octet truss) instead of support structures. The lattice supports the walls during printing and may add value as a weight-saving structural element.

Rule 5 — Build orientation strategy

The choice of build orientation is the most consequential design decision in metal AM. It affects:

1. Surface finish — critical surfaces should be up-facing
2. Mechanical properties — tensile strength and fatigue are typically 5–15% lower in the Z (build) direction for LPBF; EBM is less anisotropic
3. Support volume and removability
4. Build height and cost — build time scales with height, not volume
5. Residual stress and distortion — long, thin features aligned horizontally accumulate more distortion than vertical ones

The general hierarchy:

For fatigue-critical parts: LPBF fatigue strength in the Z direction is typically 10–20% lower than XY. If you have cyclic loading, orient the critical cross-section into the XY plane. Post-machining the functional surface will recover 80–90% of this anisotropy penalty regardless of orientation.

For thick prismatic parts: Consider orientating at 15–30° off-horizontal (a "tilted" orientation). This turns all flat horizontal faces into 15–30° faces that are self-supporting, eliminating large area support structures, at the cost of some additional build height.

Rule 6 — Lattice and topology optimisation integration

Topology optimisation (TO) tools remove material from lightly loaded regions; lattice design replaces solid regions with open-cell structures. Both are enabled by AM. A few practical notes:

Minimum printable strut diameter: 0.3–0.5 mm for LPBF (material-dependent). Thinner struts fail to build reliably or have significant deviation from nominal. Design lattice unit cells with struts ≥ 0.5 mm for production parts.

Minimum lattice cell size: ~3–5× the strut diameter for structural lattices. Below this, the space is too small for powder removal and the inter-strut surfaces merge unpredictably.

Topology optimisation + DfAM: Raw TO outputs often have near-horizontal faces, thin features, and non-45° overhangs. Always post-process TO geometry through a DfAM filter before building. Smooth the organic surfaces, reinforce thin connections, and verify all overhangs comply with the self-supporting angle rule.

Graded lattices: LPBF enables spatially graded lattice density — denser near load-bearing surfaces, sparser internally. This is the correct application of lattice design for structural parts; uniform lattice fill is rarely optimal.

Practical checklist before submitting for build

Before sending a design to the machine:

• All overhangs ≥ 45° from horizontal, or explicitly supported
• Minimum wall thickness ≥ 0.4 mm (0.5 mm for aluminium)
• Internal channels use teardrop or self-supporting cross-section
• All internal features accessible for powder removal (drain holes added if needed)
• Build orientation confirmed: primary load axis in XY, critical surfaces up-facing
• Support contact surfaces are accessible for tool removal
• Wall aspect ratio < 40:1 (or ribs/support added)
• Lattice struts ≥ 0.5 mm diameter with ≥ 1.5 mm cell spacing

This list is not exhaustive — every part needs engineering judgement. But running through it before file handover catches the 80% of issues that arise in new AM designs.

Further reading

• DfAM checklist tool — interactive checklist with process-specific rules for LPBF, EBM, SLS, and FDM
• Orientation advisor — find optimal build orientation for a given bounding box
• Support volume estimator — estimate support volume and cost impact before committing to an orientation