Powder that never comes out

An internal cavity with no drainage path fills with metal powder during the build and keeps it. The part comes off the plate looking correct and weighing more than it should. The canonical case is a hollow titanium implant that shipped 30% overweight because nobody put drain holes in it. An entire depowdering-equipment industry exists because this happens often enough to sell machines against.

It gets worse after the build, not better. Heat treatment can sinter loose powder in an undrained channel into a solid plug, and at that point the channel is gone permanently — there is no post-process that recovers it. For a buried cooling channel in a mould insert, that means a new insert.

Two different problems that look the same in CAD

“Internal channel” covers two failures with different fixes, and the screen separates them:

• Not drainable. The cavity is sealed — nothing gets out. Fix: drain holes.
• Drainable but not connected. Each void has an exit, but they do not form a continuous circuit. The powder comes out and the channel is still useless, because nothing can flow through it. If the part is meant to carry coolant or propellant, this is a dead circuit — a functional failure that a depowdering check would pass.

Four gates cover this ground: trapped_powder (enclosed cavity volume with no path out), internal_channels (the passage network itself), flow_path_continuity (whether the network actually connects inlet to outlet), and channel_wall_margin (how much material stands between a channel and the nearest surface).

The rule, and why it is two holes

The bureau-canon mitigation the screen reports is ≥2 drainage holes, ≥2 mm, on the lowest faces of each cavity. Two rather than one because a single hole gives powder nowhere to go as it drains — you need an air path in as well as a powder path out. On the lowest faces because gravity is doing the work, and which face is lowest depends on orientation, which means the drainage question and the orientation question are the same question.

The cost of the fix is a small geometry edit; the trade-off is that the holes may need plugging afterward. The screen states both rather than presenting the mitigation as free.

Honesty about resolution

The same discipline runs the other way. An earlier version of the passage gate counted 2,437 passages on a part that has six — voxel dust read as geometry. The fix that removed the dust then over-corrected and began discarding real buried passages. Both errors were found in-house and both are fixed; they are recorded here rather than quietly patched, because a screening tool that only ever reports its hits is asking to be trusted rather than earning it. The published blind-protocol runs against NIST and EASI-STRESS benchmark data — including their misses — are in the validation receipts.

Why it is worth checking before the file leaves your desk

Bureaus see this constantly — “most of the designs we receive to be metal printed are unprintable” is the founder of one saying so publicly. And the contract terms are uniform across the industry: the customer is solely responsible for design correctness. Manufacturability feedback that arrives at quote time arrives after the risk is already yours. This check belongs before the upload.

Check a part

Every gate named on this page runs automatically on an uploaded STL or STEP at amprintanalyzer.com — free, no sign-up, and the geometry is never modified. Findings are pinned to the model and each one is labeled by its evidence class: measured (a direct geometric fact), screening (a dimensionless index that localizes risk), or estimate (a planning number with its assumptions stated).

If the screen disagrees with something you have actually measured on a printed part, that is the useful case — say so through the feedback strip. Disagreements are the calibration data.