Additive at volume

Mass production and 3D printing: what step are we now?

This page is for engineers and buyers deciding whether a part should be printed at volume or machined. We walk through six steps: screening the geometry, fixing a process, setting the print parameters, checking tolerance, and switching to CNC where the print stops paying off. Read it before you commit a tool to either route.

Get Instant Quote Talk to an Engineer

6 screening stepsPrint vs. CNC dataTolerance ±0.005 mmNo minimum order

Mass production and 3D printing compared with CNC machining

On this page

1
What you should know before choosing a route
2
Mass production and 3D printing: how the two routes differ
3
Screen the geometry before you pick a process
4
Fix the process and the material batch
5
Check tolerance, finish and inspection at volume
6
Decide when to hand the part back to CNC
7
Six steps to move a part into volume production
8
When to print and when to machine

Key takeaways

What you should know before choosing a route

Print wins on lattice and internal channelsShapes a cutter cannot reach are the only reason to accept lower surface finish.

Mass production and 3D printing need a fixed processA print run is only repeatable after the machine, layer height and resin batch are locked.

CNC takes over where tolerance tightensBelow ±0.05 mm on a mating face, milling usually costs less per good part.

Hybrid runs are normalPrint the lattice, machine the bore. One part can use both routes.

Step 1

Mass production and 3D printing: how the two routes differ

Additive manufacturing builds a part by adding material layer by layer. Subtractive machining starts from a billet and removes material. That single difference drives every downstream decision: surface finish, internal geometry, per-part cost curve and inspection method.

In mass production and 3D printing, the printer is not competing with a mill on the same shape. It competes on shapes a mill cannot reach. Internal cooling channels, lattice infill and organic brackets are the classic cases. If your part is a simple prismatic block, printing it is usually the slower and more expensive choice.

The cost curve also moves in opposite directions. Machining has high setup cost and low per-part cost, so it gets cheaper as volume rises. Printing has low setup cost and a flatter per-part cost, so it stays competitive at low volume and in high-mix production. Around a few thousand units a year, the two curves often cross.

We run both routes in the same shop. That matters because most real programs do not pick one. A bracket may be printed in PA or PEEK and then have its mounting bores reamed on a 5-axis mill to hold ±0.005 mm. The engineer's job is to decide which feature goes to which process, not which process wins overall.

Step 2

Screen the geometry before you pick a process

Start with a feature-by-feature list, not a whole-part verdict. Mark every face that mates with another part, every bore that carries a bearing or a pin, and every surface that seals. Those are the tolerance-critical features. Everything else is cosmetic or structural.

Tolerance-critical faces usually belong to CNC. A bearing bore at Ø20 H7 needs a roundness and size control that most polymer printers cannot hold repeatably. A printed bore can be printed undersize and reamed later, but that adds a second operation and a fixture. Count that cost before you assume printing is cheaper.

Next, look at wall thickness and unsupported spans. FDM parts below 1.2 mm wall often warp or print with gaps. SLA and DLP hold thinner walls but get brittle. If a design needs a 0.5 mm living hinge, material choice matters more than the machine. PEEK and PA12 tolerate flex; standard resins crack.

Finally, list the internal features. A helical channel that a 6 mm cutter cannot follow is a print feature. A cross-drilled oil passage at 45° may still be machined with an angle head. Be honest about which features are truly unreachable, because every printed feature carries a finish penalty.

1
Mating facesSend to CNC unless the print can be reamed after.
2
Lattice and internal channelsPrint these; no cutter reaches them.
3
Thin walls under 1 mmCheck material ductility before committing.

Step 3

Fix the process and the material batch

A print run becomes production only when the process is frozen. That means one machine model, one layer height, one scan strategy and one material lot. If the resin or powder lot changes, tensile strength and shrinkage move. Re-qualify the part when the lot changes.

For photopolymer processes, layer height sets both finish and build time. At 50 μm, a DLP or SLA part comes off the platform with visible layer lines but good detail. At 100 μm, the build is roughly twice as fast and the surface is coarser. Pick the layer height from the drawing callout, not from the schedule.

For powder bed fusion, laser power, scan speed and hatch spacing control density. A part built at 90 percent density will fail a pressure test even if the outside looks solid. Ask for the density figure and the inspection method before you approve the process sheet.

Thermal history matters too. Large PA12 parts warp if they cool unevenly in the build chamber. Orientation on the platform can decide whether a 300 mm bracket stays flat. We usually nest tall parts vertically and add sacrificial ribs, then cut them off in a secondary operation.

Step 4

Check tolerance, finish and inspection at volume

Printers hold different tolerances on different axes. In the XY plane, a well-tuned SLA machine may hold ±0.1 mm. Along Z, layer stacking pushes that to ±0.2 mm or worse. Design your datum so critical features sit in the tighter plane, and note the axis on the drawing.

Surface finish follows the same logic. As-printed polymer typically sits around Ra 3.2–6.3 μm. Bead blasting brings it to roughly Ra 1.6–3.2 μm. If the drawing calls for Ra 0.8–1.6 μm, plan a finishing pass or a switch to machined metal.

At volume, inspection is where printed parts lose time. A 10,000-part run cannot be checked by eye. Build a first-article inspection report, then move to sampling with a defined AQL. For medical or automotive programs, keep traceability from material lot to finished part.

We inspect every shipment before it leaves the floor. Raw material certificates, in-process checks and a final dimensional report are standard. If your program needs full dimensional layout on every part, say so at quoting, because that changes the fixture and the cycle time.

Step 5

Decide when to hand the part back to CNC

Printing stops paying off when the tolerance band tightens, when the material must be metal, or when the annual volume grows past the printer's economic range. None of those are failures of additive. They are simply the point where the other process becomes cheaper per good part.

Metal printing is the clearest case. A DMLS or metal binder jet part at 200 g can cost several times a machined aluminium equivalent once you add support removal, heat treat and finish machining. That premium is justified only when the geometry is impossible to mill or the alloy is hard to cut.

Even then, the printed metal part usually needs CNC work on its interfaces. We see this constantly: a printed titanium bracket with machined bores and a faced mounting pad. The print supplies the shape, the mill supplies the tolerance.

The practical rule is to re-quote every 6 to 12 months. Machine and material prices move. A part that was cheaper to print last year may be cheaper to mill now, especially if the design has been simplified since the first run.

Step by step

Six steps to move a part into volume production

Work through these in order. Skipping step 2 is the most common cause of a failed first run.

1
List every tolerance-critical featureMark mating faces, bearing bores and sealing surfaces on the drawing. Note the datum and the tolerance band for each. Anything tighter than ±0.05 mm goes on the CNC list by default.
2
Separate print features from machine featuresHighlight internal channels, lattices and undercuts a cutter cannot reach. If fewer than about 20 percent of features are print-only, question whether the whole part should be printed.
3
Freeze the process sheetLock machine model, layer height (50–100 μm for SLA/DLP), scan strategy and material lot. Record the density target for powder processes, typically 99 percent or above for functional parts.
4
Set orientation and supportsPlace critical faces away from support contact. Support marks on a sealing face will cost you a finishing operation. For tall parts, orient vertically and add sacrificial ribs to control warp.
5
Build the first article and measure itMeasure the tolerance-critical features, not just the outside envelope. Compare XY and Z results separately. If Z is out by more than ±0.2 mm, adjust the process before scaling up.
6
Add a secondary CNC operation where neededReam bores, face mounting pads and cut threads on a mill after printing. Hold ±0.005 mm on those features and leave the printed surfaces as-built.

Route selection

When to print and when to machine

Use this as a first pass. Confirm with a DFM review before committing.

Factor	3D printing	CNC machining	Pick this when
Geometry	Lattice, internal channels	Prismatic, reachable faces	Cutters cannot reach the feature
Tolerance	±0.1 mm XY, ±0.2 mm Z	±0.005 mm	Mating face is tighter than ±0.05 mm
Surface finish	Ra 3.2–6.3 μm as-printed	Ra 0.8–1.6 μm typical	Drawing calls for Ra 1.6 μm or finer
Volume	Low to mid, high mix	Mid to high volume	Annual demand is in the thousands
Material	PA, PEEK, resin, some metals	Aluminium, steel, titanium, brass	Load path needs metal
Setup cost	Low, no tooling	Higher, fixture and program	Design is still changing
Per-part cost	Flat across volume	Drops as volume rises	Order quantity is above a few thousand

Print the shape, machine the tolerance

If a feature has to hold ±0.005 mm or seal against another part, plan on CNC for that feature. Print the rest, then re-quote the split every 6 to 12 months as prices move.

Get Instant Quote Talk to an Engineer

FAQs

Questions engineers ask about printing at volume

Can 3D printing replace CNC machining for production parts?

For some parts, yes. Lattice structures, internal cooling channels and patient-specific devices are printed in volume today because no cutter can make them.

For most load-bearing metal parts with tight bores, no. The printed shape is useful, but the tolerance-critical features still go to a mill. Plan for both processes on the same drawing.

What tolerance can I realistically expect from a printed part?

On a well-tuned polymer printer, ±0.1 mm in the XY plane is achievable. Along the build direction, expect ±0.2 mm or worse because layer stacking accumulates error.

If your drawing needs ±0.005 mm, that feature belongs on a CNC. Print undersize and machine the feature to final size in a second operation.

How do I keep a print run repeatable over thousands of parts?

Freeze the process sheet and the material lot. Re-qualify when either changes. Measure the first article on the tolerance-critical features and keep the report.

Move to sampling inspection with a defined AQL once the process is stable. Keep traceability from material lot to finished part for regulated programs.

When does metal 3D printing make sense for production?

When the geometry is impossible to mill, when the alloy is difficult to cut, or when the part count is low and the design is still moving.

Remember that most metal printed parts still need CNC finishing on their interfaces. Budget for support removal, heat treat and a machining pass.

Do I need a minimum order quantity for a printed run?

No. We run from a single prototype up to 10,000-part runs on the same floor, so the process choice does not force a batch size on you.

What changes with volume is the process sheet. Small runs can tolerate manual support removal; large runs need fixtures and a written inspection plan.