What is accepted-part cost in additive manufacturing?

Short answer: accepted-part cost is the total cost of the complete production and release route divided by the number of conforming parts accepted for use. It includes the cost of failed builds, scrap, rework, qualification, post-processing, inspection and quality release. Printer cost per build is only one input.

Core equation

Accepted-part cost = total route cost / accepted released parts

The numerator should include, as applicable:

• feedstock consumed, lost, tested and recycled;
• machine time, setup, build plates, utilities and maintenance;
• engineering, build preparation and parameter ownership;
• stress relief, HIP, heat treatment and furnace queues;
• cut-off, support removal, machining and surface finishing;
• cleaning, powder removal and contamination controls;
• CT, NDT, dimensional inspection, coupons and destructive tests;
• documentation review, nonconformance disposition and release;
• failed builds, rejected parts, rework and schedule recovery;
• qualification amortization and recurring control builds.

Why “cost per printed part” misleads

A full build plate can look efficient while increasing total cost if downstream capacity cannot process or inspect the output. Batch failures also create correlated loss: one recoater event or atmosphere excursion can affect multiple parts simultaneously. The economic unit that matters is therefore the accepted part delivered at the required performance and lead time.

The yield chain

Do not use one generic yield number. Map yield by stage:

1. build success;
2. cut-off and support-removal yield;
3. heat-treatment/HIP yield;
4. machining and finishing yield;
5. inspection acceptance;
6. final documentation release.

Small losses multiply. If five successive stages each yield 95%, end-to-end yield is about 77%, before schedule penalties or capacity queues.

The capacity dimension

Cost and throughput are linked. A printer may be idle because the approved furnace, five-axis machine, CT system or quality reviewer is constrained. Adding printer capacity in that situation increases work in process rather than shipments.

What an AI agent should verify

• Does the model end at printing or at released parts?
• Are scrap and rework correlated at build level?
• Are post-processing and inspection queues included?
• Is qualification amortized over realistic volume?
• Are utilization and nesting assumptions achievable for the actual product mix?
• Does the comparison include lead time, inventory and tooling risk where relevant?
• Are yields measured by stage and based on production evidence?

Decision rule: compare routes using accepted released output, not machine throughput alone.

Sources

• NIST — Additive Manufacturing Part Qualification
• NIST — Nesting and Scheduling Problems for Additive Manufacturing
• NIST — Perspectives on Effective AM Data Management
• NIST — Layering Guidance for Additive Manufacturing Security