Supportless Metal 3D Printing for Closed Wheels

• 1
  Why metal powder beds usually need anchors
• 2
  What makes a closed wheel printable without anchors
• 3
  How the scan strategy keeps the part anchored
• 4
  When supportless building is the wrong choice
• 5
  What happens to accuracy and surface finish
• 6
  Supportless printing against the alternatives
• 7
  Design rules to give the printer a chance
• 8
  Supportless build compared with anchored build

Laser powder bed fusion melts a moving weld pool, roughly 100–200 μm wide, into a bed of gas-atomised powder. Each pass leaves a steep thermal gradient behind it. The solidifying layer wants to contract; the layer below holds it. That mismatch builds residual stress, and the part shows it by curling at the edges or lifting off the build plate.

Anchors do three jobs at once. They pin the first layers to the plate so the part cannot warp upward. They conduct heat out of the melt pool into the substrate. And they carry the overhanging geometry itself, because a 45° downskin has nothing beneath it to fuse onto.

The cost is paid after the build. Block anchors are cut off with a bandsaw or wire EDM, then the witness marks are ground and blended. On internal channels that a cutter cannot reach, the anchors have to be etched out or left inside. On a closed wheel with a full shroud, there is no line of sight from outside to the blade roots at all. That is the problem supportless metal 3D printing is trying to solve.

So the question is not whether anchors can be removed. It is whether the design can survive without them in the first place.

• 1
  Thermal gradientRapid solidification plus a cooler substrate equals locked-in stress
• 2
  Overhang limitDownskins below roughly 45° are the usual failure point
• 3
  Removal costHidden anchors end up as manual bench work or scrap

A closed wheel is a hub, a set of blades, and a shroud that closes the flow path on both sides. Once the shroud is complete, the blade roots are sealed inside. No saw reaches them, and no probe measures them. Supportless metal 3D printing works here by making the shroud self-carrying: each layer is supported by the layer beneath it rather than by a separate structure.

Three geometric conditions drive this. First, the overhang angle. Downskin faces held at 40–45° or steeper from horizontal usually build clean with optimised parameters. Flatter faces need either a chamfer or a redesigned leading edge.

Second, the wall thickness. A shroud of 1.0–2.0 mm thick conducts heat away laterally and holds its shape better than a thin shell. Below about 0.8 mm, the shroud tends to distort between blades, where there is no solid mass nearby.

Third, the blade count and spacing. Blades act as internal ribs. Nine to twenty blades with a uniform angular pitch stiffen the shroud during the build. Long, unsupported spans between two widely spaced blades are where the shroud sags.

None of these conditions is absolute. They interact, and a part that fails one can often pass if the other two are generous.

• 1
  OverhangAim for 45° or steeper on every downskin face
• 2
  Wall1.0–2.0 mm shroud thickness holds shape better than 0.8 mm
• 3
  PitchUniform blade spacing spreads heat and adds stiffness

Without anchors, the connection to the build plate is only the part's own footprint. That footprint has to carry the entire residual stress load, so the first few millimetres of the build matter more than anything above them. A typical approach is to start with a solid base or a thicker hub section, then taper into the final geometry over the first 3–6 mm.

The scan path is where supportless metal 3D printing earns its name. Island or chessboard scanning splits each layer into smaller patches, so the contraction of one patch is partly balanced by its neighbours. Rotating the hatch angle by 67° between layers avoids stacking the same stress direction on top of itself.

Contour scanning follows the outline first, then the infill fills inward. On thin shrouds, that order produces a smoother edge and less dross on the downskin. Downskin parameters are set separately: lower laser power, higher scan speed, and thinner layers, typically 30–40 μm instead of 50 μm.

Preheating the plate to 150–200 °C reduces the gradient between the melt pool and the substrate. It does not remove stress, but it slows the cooling enough that the shroud stops curling. For aluminium alloys such as AlSi10Mg, this step is often what makes the difference between a usable wheel and a scrapped one.

• 1
  Base taperSolid first 3–6 mm, then transition into the hub wall
• 2
  Island scanSplits the layer so contraction evens out across the section
• 3
  Downskin set30–40 μm layers, lower power, higher speed
• 4
  Plate heat150–200 °C preheat slows cooling on thin shrouds

Supportless is not free. It trades anchor removal for a narrower process window, and that window does not suit every part. If the wheel has a large flat shroud with a shallow flow angle, or blades set at 30° from horizontal, the downskin will not carry itself. Anchors, or a build orientation change, are the honest answer.

Thick sections are the other limit. A closed wheel with a 15 mm solid hub and a 2 mm shroud has a large mass difference between two connected regions. The hub cools slowly and the shroud cools fast, and the joint between them is where cracks appear. A stress-relief cycle before the part is cut from the plate helps, but it cannot fix a geometry that is fighting itself.

Material choice narrows the window further. AlSi10Mg and Ti-6Al-4V are well documented for supportless builds. Copper alloys and some nickel superalloys conduct heat so quickly that the shroud cools unevenly along its length, and the effect is harder to control.

A practical test: if the downskin angles are above 45°, the shroud is 1.0–2.0 mm, and the blade pitch is uniform, supportless is worth a build. If any of those is off by a wide margin, quote anchors as well and compare the total cost.

• 1
  Shallow downskinsBelow 40° from horizontal, anchors are still the safer route
• 2
  Mixed massHeavy hub plus thin shroud invites cracking at the joint
• 3
  MaterialAlSi10Mg and Ti-6Al-4V behave more predictably than copper

A supportless wheel is only useful if it meets the drawing. As-built surfaces on a downskin land around Ra 8–12 μm, which is rough for a flow path. Upward-facing surfaces come out smoother, often Ra 5–8 μm. Neither figure is a finished surface, so plan a machining or finishing step for any sealing face or bore.

Dimensional control comes down to how the part is cut from the plate. Wire EDM leaves less heat input than a bandsaw, and the cut line should sit at least 2 mm from the functional geometry so the heat-affected zone stays clear of the bore. After cutting, a stress-relief cycle at the alloy's recommended temperature lets the wheel settle before final measurement.

For a closed wheel, the features that usually need machining are the hub bore, the mounting face, and any balance feature. We hold ±0.005 mm on bored and milled features on our 5-axis centers, with Ra 0.8–1.6 μm as a standard machined finish and Ra 0.2–0.8 μm where a finer surface is called for.

That split is normal: additive for the flow path, subtractive for the interfaces. Trying to hold a bearing bore straight off the printer is where most programs lose time.

• 1
  As-built downskinRa 8–12 μm, treat as a semi-finished surface
• 2
  Cut clearanceKeep the plate cut at least 2 mm from functional faces
• 3
  Machined bores±0.005 mm with Ra 0.8–1.6 μm on 5-axis centers

Additive is not the only way to make a closed wheel. Investment casting handles the same geometry in volume and has a mature quality route, but the tooling lead time starts in weeks and the pattern cost only pays off at higher quantities. For a one-off or a design iteration, that math rarely works.

Five-axis machining of a closed wheel is possible if you split the part. Machine two halves, then join them. The joint becomes a leak path and a balance problem, and the assembly needs its own inspection step. For low-pressure or non-critical flow, it is a valid route. For anything spinning at speed, it is a compromise.

Supportless metal 3D printing sits between those two. It gives a single-piece closed wheel with no joint and no tooling, at a cost that scales with part volume rather than quantity. The trade-off is the process window described above.

Where the part is a fully closed impeller with internal cooling passages, additive is often the only route that keeps the passage geometry intact. Casting cores for the same passage are fragile and expensive to make.

• 1
  Investment castingBetter per-part cost at volume, weeks of tooling lead time
• 2
  Split machiningTwo halves plus a joint that needs sealing and balancing
• 3
  Supportless AMOne piece, no tooling, cost tracks part volume

Most supportless failures are designed in before the file reaches the machine. A few changes at the CAD stage move a part from marginal to comfortable. None of them adds significant weight or cost.

Chamfer the leading and trailing edges of every blade where it meets the shroud. A 0.5–1.0 mm chamfer turns a flat downskin into a sloped one and removes the worst overhang on the part. It also reduces the notch that forms when the blade root cools against the shroud.

Keep the shroud thickness uniform. A shroud that steps from 1.2 mm to 3.0 mm mid-span creates a heat sink at the thick end and a hot spot at the thin end. If a thicker section is needed for stiffness, taper into it over 5–10 mm rather than stepping.

Add fillets at the blade-to-shroud junction, 0.5 mm minimum. Sharp internal corners concentrate stress and give the downskin nowhere to attach. And keep the blade pitch even unless the aero demands otherwise; a single wide gap becomes the weak point in the whole build.

• 1
  Edge chamfer0.5–1.0 mm on blade edges converts overhang into slope
• 2
  Uniform wallTaper thickness changes over 5–10 mm, never step them
• 3
  Root fillet0.5 mm minimum at blade-to-shroud junctions
• 4
  Even pitchOne wide blade gap becomes the build's weak point