What the French 3D Printing Conference in Lyon Reveals About Industrial AM

• 1
  How the French 3D Printing Conference in Lyon Frames the AM Value Chain
• 2
  Why Layer-by-Layer Builds Create Directional Properties
• 3
  Where Additive Manufacturing Stops Being the Right Answer
• 4
  Post-Processing and the Machined Interface on Printed Parts
• 5
  Five Checks Before You Route a Part to a Printer
• 6
  Additive and Subtractive Routes Compared Feature by Feature

Every industrial AM event now groups its floor the same way: machines, materials, software, post-processing, metrology. That layout is not marketing. It mirrors the order in which a printed part actually gets made, and the order in which it can fail. Walk the aisles in that sequence and the technology stops looking like a single process. It becomes a chain of separate decisions, each with its own physics and its own cost driver.

The French 3D Printing Conference in Lyon makes that chain visible because the exhibitors sit next to the service bureaus and the machine shops that finish their output. Powder bed fusion, binder jetting, material extrusion, and directed energy deposition all appear within a few meters of each other. The differences between them matter more than the similarities. A process that wins on geometry freedom often loses on surface finish, and the reverse is just as common.

The useful question is not which process is most advanced. It is which process removes the fewest constraints for this part, at this quantity, with this tolerance. A bracket with internal channels and no sealing face may never need a second operation. A manifold with a gasket groove will need one, because no as-built AM surface holds a seal reliably.

That single distinction, sealed face or not, decides more routing than any machine specification sheet. It is also the distinction most often skipped when a team gets excited about a new printer.

Metal AM melts or sinters material in discrete layers. Each layer cools against the one below it, and the thermal history leaves a grain structure that is not uniform in all directions. Tensile strength, ductility, and fatigue life can differ between the build direction and the in-plane direction. This is anisotropy, and it does not disappear after a stress relief cycle. It changes, but it stays.

In laser powder bed fusion of Ti-6Al-4V, the as-built microstructure is typically acicular martensite. Hot isostatic pressing and annealing convert it toward a lamellar or bimodal structure with better ductility. The numbers improve. The directionality remains, because the prior beta grain boundaries are still columnar along the build axis. Design allowables must reflect that, not the polished datasheet coupon.

Residual stress is the second consequence. Fast cooling and repeated thermal cycles lock stress into the part. Thin walls and long unsupported spans distort when the part is cut from the plate. Thick sections can crack during the build. Orientation on the plate is therefore a design variable, not a scheduling detail. Rotating a part 45° can turn a scrapped build into a repeatable one.

For polymer processes the story is simpler but not trivial. Fused deposition parts are weakest along the Z axis because the bond between adjacent roads is a partial weld. A printed clip loaded perpendicular to the layer plane may fail at a fraction of the in-plane strength. Load it in-plane and it behaves far better.

• 1
  Build direction sets strengthSpecify the load axis on the drawing so the build can be oriented to match it.
• 2
  Heat treatment changes, not erases, anisotropyHIP and anneal improve ductility; columnar grain boundaries persist.
• 3
  Residual stress drives distortionOrientation, support strategy, and stress relief decide whether thin walls stay flat.

AM earns its place when geometry is the hard part. Conformal cooling channels, lattice structures, organic brackets, and consolidated assemblies all use the process well. Each of those cases removes an operation or a joint that would otherwise cost more than the print. The value is in the eliminated step, not in the printing itself.

AM loses its place when tolerance is the hard part. An as-built metal surface on a powder bed machine lands around Ra 8–12 μm on a good day. A mating bore, a bearing seat, or a hydraulic sealing face needs far better. Those features get machined, and once you are machining them you have to plan stock allowance, datum strategy, and workholding for a part that may not have a flat face to grip.

Quantity is the other boundary. At one to fifty units, tooling cost dominates and AM wins on almost any complex part. Past a few thousand units, the per-part economics of casting or machining usually overtake it, unless the geometry truly cannot be made another way. The crossover point moves with part size, material price, and how much post-processing the design requires.

There is also a materials boundary. Aluminum and titanium print well. Copper and some high-strength steels are harder. Magnesium and certain refractory alloys are harder still. If the part must be Inconel or 17-4PH, both routes are available and the decision comes back to geometry and volume.

• 1
  Choose AM when geometry is the constraintInternal channels, lattices, and part consolidation are the strongest cases.
• 2
  Choose machining when tolerance is the constraintSealing faces, bearing bores, and datum surfaces need controlled cutting.
• 3
  Quantity flips the economicsLow volume favors AM; high volume usually favors casting or machining.

Most production metal AM parts are hybrid parts. They are printed near-net and then finished on a CNC machine. That means the print is not the final operation, and the drawing has to say so. Which surfaces are as-built, which are machined, and how much stock is left on the machined surfaces are all decisions that belong on the print, not in a conversation after the build.

A practical allowance for a machined face on a printed metal part is 0.5–1.0 mm per side. That covers distortion from residual stress and gives the machinist something to indicate from. Below 0.3 mm, a warped build can leave you with an unmachinable surface. Above 2 mm, you are paying print time for material you are going to cut away.

Datums are the hard part. A printed blank often has no reliable flat face, no true bore, and no square corner. The first machining operation has to create a datum before it can cut anything useful. That usually means a soft jaw setup or a fixture built from the CAD model, and it adds a setup that a casting would not need. Plan for it in the quote.

Surface finish also splits. A bead blast or tumble gets the as-built surfaces to a uniform appearance. Functional surfaces get milled or turned to Ra 0.8–1.6 μm, and sealing faces may need finer. On a 5-axis machine, the machined features can be reached in one setup, which keeps the print-to-machined alignment tight.

The alignment question matters most. If a printed internal channel has to meet a machined port, the tolerance between the two features is a stack-up across two processes. Keeping the machined side to ±0.005 mm helps, but the printed side brings its own variation. Position the interface where a small offset does no harm, or allow a slip fit and seal with an O-ring instead of a metal-to-metal face.