Finishing internal channels in AM parts: solving the access problem

The single biggest reason to use metal additive manufacturing is geometry that no other process can make: topology-optimised hydraulic manifolds, turbine fuel nozzles with swirl passages, lattice heat exchangers, conformal cooling channels that follow a mould cavity. Every one of these is defined by its internal surfaces — and internal surfaces are exactly where conventional finishing fails.

A milling cutter, a grinding wheel, a blasting nozzle, a polishing mop, a shot-peen stream — all are line-of-sight tools. They finish what they can see and touch. The inside of a 4 mm cooling channel that loops through a part is invisible and untouchable to all of them. Yet that channel's surface is what determines whether the part works.

This guide is about the processes that can reach inside, why the as-built internal surface is so rough to begin with, and the design rules that decide whether internal finishing is even possible.

Why as-built internal surfaces are so rough

The external surface roughness of an LPBF part is typically Ra 5–15 µm. Internal channel surfaces are worse — Ra 10–35 µm — and the reason is geometric.

On a downward-facing internal surface (the "roof" of a horizontal channel), there is no solid material beneath the melt pool — only loose powder. The melt pool sinks slightly into that powder and pulls partially-melted particles with it. These particles fuse to the surface as adherent satellites. The result is a rough, particle-studded wall that is far coarser than a vertical or upward-facing surface.

This matters for three concrete reasons:

• Flow performance. A rough channel has a higher friction factor. For a cooling or hydraulic passage, that means higher pressure drop and lower flow coefficient (Cv) for the same pump. Published AFM studies on LPBF channels report flow-coefficient improvements of 15–40 % after internal finishing.
• Fatigue. Surface asperities are stress concentrators. For a pressurised internal passage, the rough wall is the crack-initiation site. Internal roughness is one reason as-built LPBF fatigue strength trails wrought.
• Cleanliness. Adherent partially-fused particles can break loose in service — unacceptable in a fuel system, a hydraulic valve, or a medical fluid path.

Before any of this can be addressed, the channel has to be empty.

Step zero: get the powder out

No internal finishing process works on a channel that still contains trapped powder. Depowdering is the prerequisite. For complex internal geometry this is not trivial — powder compacts and lightly sinters against hot walls during the build, and gravity alone will not clear a channel that loops back on itself.

The practical rules:

• Through-channels only. A blind internal channel — one with no second opening — cannot be reliably depowdered or finished. Design every internal passage with an inlet and an outlet.
• Minimum diameter ~1 mm for depowdering, larger for finishing. Below ~1 mm, powder bridges and compacts; below the practical finishing minimum (~0.5 mm for AFM media flow), no finishing medium will pass.
• Multi-axis agitation. Automated systems rotate and vibrate the part through a sequence of orientations so every channel drains in turn.

Only once the channel is verifiably clear — confirmed by CT scanning for critical parts — does finishing begin.

The three processes that can reach inside

1. Abrasive Flow Machining (AFM)

Abrasive flow machining forces a viscoelastic abrasive medium — abrasive grit suspended in a polymer carrier — through the channel under hydraulic pressure. The medium is a self-conforming lapping tool: it abrades wherever it flows, and it flows fastest where the channel is most restricted, so material removal concentrates exactly at the high spots.

AFM is the only general-purpose method for finishing arbitrary internal passages. It reduces internal Ra from ~15 µm to ~1 µm in 10–30 cycles for typical LPBF channels.

The catch is that AFM is owned, commercially, by a single dominant equipment maker, and its process parameters are not published. To plan a job — how many cycles, what grit, what pressure — you have to estimate. The AFM Cycle Planner models this logarithmic curve to sanity-check cycle count feasibility before committing to a trial.

AFM's hard limit is the same as depowdering's: the medium must be able to flow through. A through-passage with adequate inlet and outlet works; a blind pocket or a sub-0.5 mm capillary does not.

2. Electropolishing (wet, DLyte, and Hirtisation)

Electropolishing dissolves surface peaks electrochemically — no tool contact at all, which is what makes it geometry-agnostic. Three variants matter for AM:

• Wet electropolishing immerses the part in an electrolyte and passes current. It levels micro-roughness efficiently but struggles with the large adherent-particle asperities of an as-built surface, and current density is uneven deep inside a channel. The electropolishing estimator models time-to-target-Ra and flags when as-built Ra is too high to start from EP alone.
• Hirtisation is a pulsed electrochemical process developed specifically for AM internal geometry. It combines support removal and internal smoothing in one immersion sequence and reaches lattice interiors and channels that wet EP current distribution cannot.
• Dry electropolishing (DLyte) uses a solid granular electrolyte that conforms to complex external surfaces — strong for external finishing and lattices, less so for long internal channels.

Electropolishing removes material isotropically, so any close-tolerance internal dimension must be finished before electropolishing, not after.

3. Chemistry-assisted flow

For very fine internal passages where even AFM media will not pass, chemical and electrochemical flow processes (including the Hirtisation family above) become the only option. They trade speed and predictability for the ability to reach geometry nothing else can.

Choosing a process

In practice, internal finishing is often sequential: depowder → AFM to knock down the gross asperities → electropolish or Hirtisation for the final Ra. No single process does everything.

Design for internal finishing

The most important decisions are made in CAD, long before the part is finished:

• Every internal channel needs two openings. An inlet and an outlet so depowdering and AFM media can traverse the full path.
• Keep diameters above the finishing minimum. ~1 mm to depowder reliably, ~0.5 mm as the practical floor for AFM media flow. Capillaries below this cannot be finished — only printed clean.
• Avoid sharp internal bends. AFM media bridges at tight radii, leaving the apex unfinished. Generous bend radii finish more uniformly.
• Allow for material removal. Electropolishing and AFM both remove stock from the wall. If the channel has a flow-critical diameter, design it slightly undersize and open it up during finishing — or account for the removal in tolerancing.
• Provide a verification path. If the internal Ra has to be proven, the geometry must allow either a destructive section on a sister coupon or a CT scan with sufficient resolution.

These rules sit alongside the broader design-for-metal-AM ruleset and the general post-processing workflow — internal finishing is one branch of the larger post-build pipeline.

Verifying an internal finish

This is the honest limit of internal finishing: you usually cannot measure the result directly. A profilometer stylus cannot reach inside a 4 mm channel any more than a polishing tool can.

The options, in descending order of practicality:

• Destructive section of a sister coupon printed in the same build, same orientation, same parameters. The most reliable way to know the actual internal Ra — at the cost of the coupon.
• Flow testing — measure pressure drop or flow coefficient against the design intent. Indirect, but it tests the property you actually care about.
• CT scanning confirms the channel is clear and dimensionally correct, but CT voxel resolution (typically 5–150 µm) is usually too coarse to quantify Ra in the single-micron range. CT tells you the channel is clean and the right shape — not that it is smooth.

Plan the verification method at design time. If a programme requires a proven internal Ra, build the sister coupons into every build from the start.

The takeaway

Internal channels are the reason to use AM and the hardest thing to finish about an AM part. The processes that reach inside — abrasive flow machining, electropolishing, and Hirtisation — each have a distinct envelope, and none of them rescues a geometry that was designed without an exit path. The leverage is in CAD: a through-channel with adequate diameter, generous radii, and a verification coupon can be finished predictably; a blind sub-millimetre passage cannot be finished at all.

Start with the AFM cycle planner and the electropolishing estimator to scope feasibility, and use the post-processing reference to compare the full set of finishing routes against your material and geometry.

Sources: Abate et al. (2022), AFM of LPBF IN718 internal channels; Han & Fang (2019), fundamentals of electropolishing, Int. J. Machine Tools & Manufacture; published application data from Extrude Hone (Kennametal), GPainnova/DLyte, and Hirtenberger Engineered Surfaces. Parameter figures are planning-grade — validate against test coupons for production work.