Additive Manufacturing (3D Printing) in Production

Additive manufacturing (3D printing) builds a part layer by layer from a digital model, the opposite of subtractive machining, which cuts material away. ISO/ASTM 52900 groups every method into seven process families; in production the workhorses are material extrusion (FDM), powder bed fusion (SLS for polymers, DMLS for metals), and vat photopolymerization (SLA). Prototyping was the first act. Production is the one that changes how a plant sources parts.

For most of its history, 3D printing lived in the model shop: a way to hold a concept before committing to tooling. That is still real, but it undersells where additive earns its keep now, printing the jigs that hold your product, the fixtures your line runs on, the low-volume and geometrically impossible parts injection molding cannot touch, and the spare that used to take six weeks to source. This post walks the process families, sorts prototyping from production, covers tooling and spares, and lays out what quality control actually requires before a printed part is allowed on a machine. If your instinct is to compare it against cutting metal, hold that thought against CNC machining as you read.

What is additive manufacturing, and how is it different from machining?

Additive manufacturing is any process that produces a part by adding material, usually in layers, from 3D model data, rather than removing it from a solid block. That single difference cascades into everything else. Because you are building up rather than cutting away, geometry that a mill or lathe could never reach becomes routine: internal channels, lattice infills, one-piece assemblies that used to be five parts and a fastener.

The international definition and vocabulary come from ISO/ASTM 52900:2021 which classifies additive into seven process categories: material extrusion, powder bed fusion, vat photopolymerization, material jetting, binder jetting, directed energy deposition, and sheet lamination. Three of those do most of the production work; knowing which is which keeps a spec conversation from turning into a brand-name argument.

Which additive process should you use in production?

Pick the process by the job the part has to do, not by what printer is nearby. The three production families trade off speed, strength, material, and detail in predictable ways:

Two properties trip up teams new to production printing. The first is anisotropy: extruded parts are weaker between layers than along them, so print orientation is a design decision, not an afterthought. The second is post-processing support removal, heat treatment, machining of critical surfaces, curing. The print is often half the process. Budget for the other half or the part will fail qualification.

When does 3D printing move from prototyping to production?

It moves when the part has to survive real use, at real rates, with a repeatable spec, not just look right on a desk. The spectrum runs from concept models nobody touches to end-use parts that carry load in the field, and the requirements tighten sharply as you move along it.

For most plants the first real production use is not a glamorous end-use part, it is tooling: jigs, fixtures, assembly aids, gauges, and soft jaws. They are internal, low-risk, and pay back fast because a fixture that used to take a machinist two days now prints overnight for a few dollars of plastic. Once a team trusts printed tooling, moving to low-volume end-use parts is a smaller leap than it looked.

How does additive manufacturing change spare parts and tooling?

It converts physical inventory into digital inventory. Instead of stocking a shelf of rarely-used spares, or waiting weeks for a discontinued part from a vendor who may no longer make it, you keep the file and print on demand. Siemens Mobility, for example, uses additive to produce rail-industry spares, which removes most of the shipping and logistics time and keeps legacy equipment running past the point where parts are commercially available.

The plant-floor implications are concrete:

• Lead time collapses. A part that took six weeks to source can print in hours to days. For a line down waiting on one bracket, that is the difference between a shift lost and a shift saved, the same math that makes machine monitoring and fast response pay.
• Inventory shrinks. Slow-moving spares become files, not shelf space. You carry the risk digitally, not physically.
• Obsolescence stops ending equipment life. Discontinued and legacy parts can be reproduced from a scan or drawing, extending the life of machines the OEM abandoned.
• Design freedom pays. Consolidated assemblies and lightweight lattices reduce part counts and weight in ways that ripple into fabrication and assembly downstream.

How do you qualify a printed part for production?

Qualify the process first, then the part, then keep proving it stays in control. A printed part is only as trustworthy as the process that made it, and additive processes drift. Here is the disciplined path:

1. Define the part's duty and acceptance criteria. Loads, environment, tolerances, surface finish, and the tests that will prove them. If you cannot state pass/fail, you cannot qualify anything.
2. Lock the material and its traceability. Specific polymer or metal powder, supplier, lot, and storage conditions. Powder that absorbed moisture or was recycled too many times prints different parts.
3. Freeze the process parameters. Machine, layer height, orientation, laser power or nozzle temperature, supports, and post-processing. Changing any of them means re-qualifying, treat the parameter set like a controlled recipe.
4. Build a qualification lot and test it destructively. Mechanical testing, dimensional inspection, and for metals, porosity checks by CT or density. One good print proves nothing; a lot proves repeatability.
5. Add in-process monitoring where the risk warrants it. Melt-pool or thermal monitoring on metal builds, camera checks on polymer, and logged machine data so a failed build is traceable rather than mysterious.
6. Document the whole chain. File version, material lot, machine, parameters, operator, and test results, the same records logic as any quality control process, so an auditor sees a controlled process, not a hobby printer.
7. Monitor drift in production. Track first-pass yield, scrap, and build failures over time. Additive machines age; nozzles wear and lasers drift, and only a trend catches it before a bad lot ships.

What are the real limits of additive in production?

The honest limits are speed, cost per part at volume, material properties, and consistency. Additive wins on complexity and low volume; it loses to injection molding and machining the moment you need thousands of identical simple parts cheaply, molding still crushes it on unit cost at scale. Printed parts can carry porosity, anisotropy, and surface roughness that machined parts do not, and build-to-build consistency demands the process discipline above. Additive is a powerful tool in the mix, not a replacement for the mix.

Where does additive fit in a connected plant?

Additive is a production process like any other, which means it belongs inside the plant's data picture, not off in a side room. A print farm has uptime, scrap, queue, and quality records exactly like a bank of CNC machines, and the moment printed spares and fixtures feed the line, their status matters to production. The additive market reflects how mainstream this has become: industry analyses put the global additive manufacturing market in the tens of billions of dollars in 2026, growing at a double-digit CAGR, with metal powder bed fusion among the fastest-moving segments in aerospace, medical, and industrial applications.

That is where a connected operation pays off. When machine status, quality checks, and part traceability from the printers land in the same real-time picture as the rest of the floor, the substance behind smart factory technology and the Industrial IoT additive stops being a black box and becomes a managed capacity you can schedule, monitor, and audit. Harmony connects machine signals, quality data, and records across a plant into one live view without ripping out what you already run; you can see how that looked on a real floor in the CLS case study or walk the module map on the features section of our homepage.