Choosing a 3D-printing process—and knowing when not to print

01

Write the problem before naming a process

Describe function, quantity, geometry, envelope, interfaces, loads, stiffness, mass, tolerance, surface condition, material environment, temperature, chemicals, ultraviolet exposure, fire behavior, fatigue, service life, repair, appearance, schedule, budget, and consequence of failure. Rank each item as mandatory, target, or optional. Separate prototype needs from production needs; a form-and-fit model and a flight component may share geometry but not evidence.

Include organizational constraints: available operators, ventilation, powder or solvent controls, electrical service, inert gas, furnace capacity, post-processing, inspection, software, cybersecurity, and waste management. ISO/ASTM 52910 provides general design considerations across additive processes but does not supply process-specific material data or a universal solution ².

• What value must manufacturing create?
• Which requirements are measurable?
• What happens if the item fails?
• Which capabilities and controls already exist?
• What operations must a supplier perform outside the printer?

02

Screen material and process together

Do not choose a printer and then hope its material behaves like the required engineering material. Begin with the material state needed in service, then identify processes that can produce and verify it. A polymer name does not guarantee formulation or properties; an alloy name does not capture porosity, microstructure, heat treatment, surface, or orientation. A cementitious mixture needs pumpability, buildability, curing, reinforcement strategy, durability, and code evidence.

Use the standardized process families to identify the feedstock and joining mechanism ¹. Then ask what supports, atmosphere, washing, depowdering, debinding, sintering, curing, heat treatment, machining, infiltration, coating, or assembly are required. Reject routes that cannot provide safe material handling, clean internal geometry, or meet the service environment.

03

Test geometry against capability

Map wall thicknesses, channels, holes, gaps, overhangs, support contacts, lattice features, datum surfaces, and machining access. Compare the part envelope with the validated work volume, not only the advertised maximum. Consider thermal mass, nesting, recoating or peel forces, collision clearance, residual-material escape, and the ability to remove the object safely.

Capability should come from measured artifacts and representative parts. NIST’s test artifact connects specific measured errors with machine or process performance and can support periodic checks after maintenance ³. Supplier capability matrices should state machine, material, parameter range, orientation, sample count, measurement method, and date. “Minimum feature” without those conditions is a clue, not a design rule.

• Fits the validated build envelope with access margin
• Supports or anchors can be removed
• Powder, resin, or support media can escape
• Critical surfaces can be finished and measured
• Representative geometry has been demonstrated

04

Compare quality, rate, and economics honestly

Build time is only one component of lead time. Add preparation, queue, setup, heating, cooling, cleaning, support removal, thermal processing, machining, inspection, rework, and documentation. Calculate material yield using support, purge, unsalvageable powder, failed setup, and downstream removal. Include labor and facility controls. A fast print followed by two days of cooling and finishing may not be the fastest route.

For one custom shape, additive tooling avoidance may dominate. At high repeat volume, molding or forming may repay dedicated tooling. Binder-based processes may offer productive build stages while shifting control into debinding and sintering. DED may rapidly create near-net metal but depend on machining. Evaluate first-pass yield, throughput at the bottleneck, and accepted-part cost—not theoretical deposition rate.

05

Use hybrid manufacturing deliberately

Hybrid routes assign each requirement to the process that handles it best. Print a near-net shape and machine datums, holes, sealing faces, or bearing seats. Binder-jet a sand mold and cast the final alloy. Print a polymer pattern for investment casting. Scan a damaged component, add material by DED, machine to final geometry, and inspect. Print modular segments and assemble them where transport or build-envelope limits dominate.

ORNL describes hybrid directed-energy-deposition workflows in which additive and subtractive operations may be interleaved, creating planning challenges around access and coordinate control ⁴. Hybrid does not mean improvised. Define datum transfer, stock allowance, fixturing, thermal state, contamination control, intermediate inspection, and the order in which features become inaccessible.

• Add for near-net geometry, internal routes, repair, or customization.
• Cut for critical interfaces, finish, and datum control.
• Form or cast for repeatable bulk shape when tooling is justified.
• Assemble for repairability, access, modularity, and material separation.
• Use human craft where judgment and finish create value.

06

Gate the decision by consequence

For each candidate route, list failure modes and the evidence that would control or detect them. Low-risk visual models may need dimensional spot checks. Fixtures may need load cases, proof tests, wear review, and replacement limits. Pressure, medical, structural, transportation, electrical, fire-safety, or lifting applications require applicable engineering, codes, material control, validated processes, inspection, documentation, and authorized acceptance.

NIST’s qualification work emphasizes that complex internal features, pores, rough surfaces, anisotropy, gradients, residual stress, and post-processing challenge acceptance ⁵. If the required evidence cannot be produced economically or safely, redesign, use a mature conventional route, or lower the application claim. A decision not to print can be the most informed additive-manufacturing decision.

• Select the best-supported route, not the most novel.
• Pilot uncertain features before committing to the full part.
• Define acceptance and change control before production.
• Revisit selection when quantity, requirements, suppliers, or technology change.