Printing metal, and when it is worth it

Around 2013 the story was that everyone would print objects at home and manufacturing would decentralise. That did not happen and was never plausible. What did happen is that additive manufacturing became a normal industrial process with a well-understood envelope, sitting alongside casting, machining and moulding rather than replacing them.

The one economic rule

Conventional manufacturing has a high setup cost and a low cost per part. A mould costs tens of thousands and then each part costs cents. Additive has almost no setup cost and a high, nearly constant cost per part.

Therefore: additive wins at low volume, complex geometry and high part value. Conventional wins at everything else, and the crossover is usually somewhere between a hundred and a few thousand parts depending on size and material.

Every failed additive manufacturing business tried to beat injection moulding at volume. It cannot be done, and the arithmetic says so before anyone builds the factory.

Where it genuinely wins

Geometry that cannot be made otherwise. Internal cooling channels that follow a curved surface, lattice structures that are strong and light, heat exchangers with surface area impossible to machine. This is the real argument: not cheaper, but possible.

Consolidation. A fuel nozzle assembled from twenty machined parts, printed as one. Fewer joints, less weight, no assembly labour, fewer failure points. Aerospace has pursued this hard and it is where the largest documented savings are.

Individual fit. Every patient's bone is different. Hip cups, cranial plates, dental aligners and surgical guides are made per person, which is a volume of one by definition and therefore exactly the case additive was made for.

Tooling. Injection moulds with conformal cooling channels that follow the part shape cool faster and more evenly, cutting cycle time by double digits. The printed object is not the product; it makes the product.

Spare parts. A part for a machine built thirty years ago, where the tooling is gone. Printing one is cheaper than recreating a mould, and it removes an inventory holding cost.

The cost nobody quotes

The machine is not the expense. Metal powder must be handled in an inert atmosphere and is a respiratory and explosion hazard. Parts must be removed from a build plate, usually by wire cutting. Support structures must be machined away. Most metal parts need heat treatment to relieve stress, and many need hot isostatic pressing to close internal porosity. Surfaces that matter must be machined to tolerance. Then everything needs inspection, frequently computed tomography, because defects are internal and invisible.

Post-processing routinely exceeds half of total part cost. Any business case built from machine time and powder price is wrong by a factor of two.

The qualification problem

In aerospace and medicine, a part must be qualified, which means demonstrating that the process produces consistent properties. Additive processes are sensitive to machine, powder batch, build position and orientation. Qualifying a part is expensive, and it qualifies that part on that machine with that powder. Change any of them and much of the work repeats.

This is why adoption is fastest where part value is high enough to absorb qualification cost, and slow everywhere else.

What to watch

Watch in-process monitoring, which could turn qualification from a per-part exercise into a per-build one. Watch multi-laser machines, which attack the throughput limit. And watch whether any additive part reaches genuinely high volume in automotive, which would mean the crossover point moved.