Machine installation is not production launch

Acceptance of a laser powder-bed fusion system involves three questions. Does the machine meet vendor specification? Does it repeatedly produce the chosen material–parameter condition? Can the full chain deliver accepted parts at the required rate? Factory and site acceptance answer the first, capability studies the second and production ramp-up the third.

Confusing these levels is common. One successful calibration build does not prove monthly capacity, and a high nominal rate does not prove yield. Build the ramp plan backwards from part requirements.

1. Define the application boundary

Specify material, part family, critical characteristics, volume and acceptance level. Development tooling needs different evidence from aerospace or medical hardware. Also define excluded use: every alloy the machine can process is not automatically inside the initial qualified scope.

Create a matrix covering density, chemistry, tensile and fatigue performance, dimensions, surface, internal defects, cleanliness and traceability. Give each row a method, uncertainty, sampling plan, owner and decision rule.

2. FAT and SAT: prove machine functions

Factory Acceptance Testing checks configuration, safety, optics, gas, powder flow, software and contracted performance before shipment. Site Acceptance Testing repeats the relevant checks at the final site with local utilities, extraction, network and environment.

Do not reduce testing to green status lights. Retain raw data, calibration certificates, limits and deviation handling. On multi-laser machines test each field and overlap zone; for automated module exchange test repeated positioning, identity and containment.

3. Material and process baseline

Record powder supplier, lot, chemistry, particle size, morphology, oxygen and moisture. Document sieving, reuse and blending. Version optical calibration, parameters, slicer, firmware, gas flow, plate and maintenance state.

NIST describes qualification as collecting sufficient data to demonstrate that a material or process functions as expected. That means known relationships between critical variables, measurement uncertainty and results—not a single coupon.

4. Position and build-to-build capability

Plan several builds with coupons at edges, centre and laser overlap zones. Measure density, chemistry, microstructure, tensile properties, fatigue where necessary and dimensions. Variation, spatial dependence, build trends and measurement contribution matter alongside averages.

Coupon geometry should match the risk. A simple cylinder may not represent thick-to-thin transitions, enclosed channels or high thermal mass. Document the argument linking coupon and part.

5. Model capacity at finished-part level

Nominal cm³/h covers only part of exposure. Include data preparation, setup, recoating, cooling, exchange, depowdering, stress relief, separation, support removal, machining, finishing, NDT and release.

Model bottlenecks, not catalogues. If the printer restarts automatically but there is one depowdering station or furnace, work-in-progress may grow without faster delivery. Scenario-test module, operator, furnace and inspection capacity.

6. OEE for LPBF

OEE multiplies availability, performance and quality. In LPBF, availability should distinguish failures, cleaning, material changes, calibration and planned maintenance. Performance compares actual with standard cycle, while quality is the first-pass accepted share.

One OEE number can mislead. Add accepted parts per week, machine and total lead time, first-pass yield, rework, powder loss and downstream waiting. If only laser utilisation is rewarded, the system may overproduce uninspected work.

7. Powder, gas and safety

Closed powder supply can reduce exposure but does not remove risk assessment. Define inerting, extraction, cleaning, PPE, ignition control, waste and emergency response by material. Test failure cases such as module faults, filter changes and spills.

Powder mass balance is both production and quality data. Track virgin input, machine load, recovered fraction, samples, waste and unexplained loss. Losing lot identity can cost more than losing the powder itself.

8. Digital traceability

Connect order, CAD and build files, orientation, parameter version, machine and sensor logs, powder and gas lots, operator, post-processing and inspection. Use unique identifiers and controlled versions, not filenames. Monitoring data should drive acceptance only through validated decision rules.

9. Change control

Classify changes in advance: an equivalent spare, software update, laser replacement, new powder supplier, changed heat treatment or added machine need different evidence. Each change requires impact analysis, retesting, approval and rollback planning.

Do not assume a second machine is an automatic clone. Even with shared parameters, demonstrate machine equivalence for critical properties and the measurement chain.

10. Ramp gates

Use gates: installation complete; safe operation; machine capability; material–process baseline; representative part; pilot series; stable production. Give each a quantitative limit and closed deviation list. Increase volume only when the previous evidence is reproducible.

Editorial conclusion

LPBF ramp-up depends less on maximum nominal speed than on a demonstrated process window and a balanced chain. Automation creates value when more first-pass accepted parts leave the system, not merely when more builds begin. A strong acceptance plan is therefore technical, quality, logistics and data-management work at once.