NDT for Additive Manufacturing: Build an Inspection Evidence Plan

An inspection plan starts before the finished part reaches the laboratory

NDT for metal additive manufacturing should connect a specific part decision to the evidence that can support it. A build log, a computed tomography image and a material test report describe different aspects of the manufacturing outcome. Collecting all three does not automatically create a coherent qualification argument. Someone must explain what each record establishes, which questions remain unanswered and who is authorized to disposition the part.

For a U.S. manufacturer buying inspection services, the useful starting document is therefore an evidence plan, rather than a request for a generic scan. State the part's intended application, manufacturing route, relevant features and governing requirements. Then identify the decisions expected from the inspection supplier. This feature provides an original planning workflow and a hypothetical example. It does not specify examination settings, defect acceptance limits or a substitute for an approved manufacturing and inspection plan.

Separate monitoring, examination and release

Process monitoring observes information generated while manufacturing occurs. Finished-part examination evaluates the article in a defined condition. Product release combines the evidence required by the applicable quality and engineering process. Treat these as related activities with different responsibilities. A supplier may provide excellent monitoring software without being responsible for the final release decision, and an inspection laboratory may receive too little manufacturing context to interpret a requested comparison meaningfully.

NIST's Measurement Science for Additive Manufacturing program distinguishes measurement needs before, during and after manufacture. Its published program description includes process sensing, dimensional characterization and NDT as complementary research areas. That structure is useful context for organizing a project; it is not an endorsement of a particular production inspection route.

Create separate entries for the evidence generated at each stage. Identify the record owner, the physical article or region represented, and the authorized reviewer. If a proposal says that one stage will replace another, make that a distinct technical decision requiring documented justification. Do not allow the substitution to happen silently because a purchasing line item was removed or because an attractive monitoring dashboard appeared to show no anomalies.

Describe the manufacturing route precisely enough to be useful

The label metal additive manufacturing is too broad for an inspection request. A laboratory needs to understand the relevant manufacturing process, material, part geometry and condition. Supply the information approved for release by the manufacturer, with controlled drawing and specification references. Distinguish the process actually used from a planned alternative. A technical reviewer can then determine what further information is necessary before proposing an examination.

Record the sequence of subsequent operations as well. Separation from the build platform, heat treatment, machining, surface treatment and any other relevant processing may create different inspection opportunities or constraints. The engineering and quality owners should decide which stages require evidence. Purchasing should preserve those stages in the work scope so that an examination requested before one operation is not performed afterward merely because that schedule was easier.

Use a revision-controlled route summary rather than an informal narrative distributed across emails. When the route changes, identify the affected evidence and obtain a review of its continued applicability. An earlier result remains a record of the article's earlier condition. It should not be relabeled as evidence of a later condition without a justified connection between the two.

Turn the drawing into questions that inspection can answer

Work through the regions that matter to the intended use. For each, ask what condition needs to be detected, characterized or measured, and what decision follows from the result. Internal passages, transitions, thin sections and surface-connected features may present different questions. This exercise should be led by the responsible design, materials and NDE personnel; it is not an invitation to invent universal rejection criteria from a general article.

Keep geometric conformity, discontinuity evaluation and material-property evidence in separate columns. A dimensional match does not answer every question about internal integrity, while detection of an indication does not by itself determine whether the part is acceptable. Specify how the relevant requirements will be supplied to the examiner and who resolves a conflict or ambiguity. An inspection provider should not have to infer the intended engineering decision from the file name.

Document regions that cannot be examined as requested. A useful report describes the actual coverage and its limitations, including any areas deferred for another agreed approach. Avoid treating an incomplete examination as a full pass merely because the inspected regions produced no reportable findings. The evidence plan should make the unresolved area visible before the part reaches a release meeting.

Treat process signals as evidence with a defined relationship to the part

NIST's 2024 report on in-process monitoring and NDE for metal additive manufacturing reviews methods, instruments and knowledge gaps in the context of potential nuclear power plant applications. It discusses the challenge of making these technologies robust tools for qualification. It does not establish that a particular monitoring signal can universally replace final examination.

For an actual project, ask the monitoring supplier what the recorded output represents. Is it a measured signal, a processed feature, a statistical flag or a predicted material condition? Preserve those distinctions in the handover. If a model produces a predicted location, retain the coordinate convention and the transformation used to relate it to the finished part. A colored region without a defensible location relationship is difficult to investigate later.

Ask how missing observations, sensor interruptions and incomplete files are represented. An absent warning should not be indistinguishable from a period when no usable information was collected. Include a record-completeness check in the review package, separate from the technical interpretation of the signals. That gives the reviewer a way to distinguish a satisfactory observation from a gap in the observation itself.

Select finished-part NDT against the required evidence

An inspection method should be proposed against the actual application envelope. Ask the responsible NDE authority to explain how material, geometry, surface condition, access and the relevant discontinuity population affect the choice. The manufacturer and inspection supplier should agree what the proposed technique is intended to establish and what it will not address. A technique demonstrated on another part family may provide useful background without resolving the new application.

Industrial CT is particularly easy to oversimplify because the result can resemble a complete digital representation of the component. The useful question is whether the examination supports the required task in the relevant regions. NIST's 2019 study of CT acquisition parameters and AM defect detection investigated how acquisition choices affected image noise and detection results for a particular artifact. It supports asking application-specific questions, rather than relying on one advertised image specification.

For an outsourced service, request a technical proposal that explains the planned evidence and any necessary feasibility work. Keep that discussion separate from purchasing an instrument; our industrial CT system trial guide addresses that different decision. A service contract should make clear which investigation, examination, interpretation and reporting activities are included, and which require a separate authorization.

Define what a reported porosity value represents

A request for a porosity percentage needs a definition of the evaluated region and the analysis behind the number. NIST's 2017 investigation of cobalt-chrome AM pore structures explains that quantitative CT results depend on image segmentation. The study investigated several aspects of pore structure, rather than treating a single percentage as a complete description. Its reported experimental conditions are not acceptance limits for another manufacturer's parts.

For a purchasing brief, ask the technical reviewer to define the intended measurement and the supporting information needed to interpret it. Does the reported value cover the entire evaluated region or only a selected subvolume? How are the region boundaries recorded? Which analysis version produced the result? A comparison between suppliers should use compatible definitions, or clearly explain why the results answer different questions.

Retain the relevant spatial information when location or distribution matters to the engineering assessment. Two summary values can be numerically similar while describing different arrangements of detected features. Do not infer a mechanical property or a service-life conclusion from a porosity percentage without the required supporting engineering evidence. State exactly which measurement the inspection supplier is delivering and leave the broader disposition to the authorized process.

Choose representative trial items and explain their limits

A trial needs items selected for a reason. Identify which features, manufacturing variations and examination difficulties each item represents. A simple demonstration block can establish that a system produces data; it may not establish performance on the part features that drive the purchase. Have the technical owner document the connection between the trial population and the proposed production population.

Distinguish known reference information from assumptions. If a trial item's internal condition has been established through another examination or subsequent destructive work, retain the relevant records and their limitations. If the condition is only suspected, describe it that way. Agree how uncertain reference findings will be handled before comparing the supplier's results with an expected answer. Otherwise, disagreements can be hidden by changing the reference after the trial.

Witness coupons require the same discipline. Record their identity, relationship to the build, subsequent processing and intended use in the qualification plan. Do not turn a convenient coupon into a blanket statement about every production article. When the plan relies on a coupon to support a particular conclusion, the responsible reviewer should explain why that relationship is appropriate and what additional part-specific evidence remains necessary.

A hypothetical project: an internal-channel development part

Consider a fictional U.S. manufacturer evaluating a metal development part with an internal channel. The team wants to compare the intended passage geometry with the manufactured condition and investigate an unusual process-monitoring flag near a transition. These are two questions. The example contains no actual customer results, acceptance limits or recommendation to release the component for service.

The project engineer first identifies the part revision, build identifier, coordinate system and processing stage. The monitoring owner supplies the original flag record and explains how its location was calculated. The laboratory is asked to assess feasibility for the defined regions and to state what output can support each question. The design owner retains responsibility for determining what constitutes an acceptable passage and how any finding affects the development decision.

Suppose the supplier can deliver useful information about the passage but cannot support the requested evaluation at the flagged transition in the proposed configuration. The correct commercial response is to separate those outcomes. Record the completed task and the unresolved task, then ask the technical team to approve a suitable next step. Do not summarize the work as a successful scan of the whole part or quietly replace the difficult question with a more convenient measurement.

Preserve coordinates, identities and original records

Location relationships are a practical source of confusion when manufacturing and inspection systems describe the same component differently. Establish the authoritative part identity, drawing revision, units, orientation and reference features. Retain the transformations needed to connect manufacturing coordinates, examination coordinates and the reporting view. A reviewer should be able to locate the relevant region without reconstructing the author's intentions from a screenshot.

Define the handover before the work begins. Identify the original or native datasets required for review, the derived outputs, the report and the relationships among them. Include the software and configuration information needed to understand how the outputs were produced. The exact retention package should follow the applicable project requirements; a blanket instruction to retain everything can be as unhelpful as accepting only a presentation image.

Ask a receiving reviewer to open a representative package using the intended review environment. Confirm that identifiers, units, locations and exceptions survive the transfer. Our native inspection data handover guide and file manifest guide provide practical ways to organize this exercise. The objective is a usable record, not merely a folder that transferred without a network error.

Resolve the approval route for the actual U.S. application

Requirements depend on the product, customer and applicable program. For example, NASA-STD-6030 has a stated scope concerning additive manufacturing for NASA spaceflight hardware. Its existence does not make every commercial AM part subject to the same requirements, nor does following an isolated section establish approval for another industry. Use the actual contract and engineering requirements to identify the governing documents and their applicable revisions.

Create a responsibility record that names the design authority, manufacturing process owner, NDE technical authority, quality release owner and customer approval contact where applicable. Record which decisions each can make. This is especially useful when a machine supplier, a contract manufacturer and an independent laboratory each perform only part of the work. Clear responsibility prevents a gap from being mistaken for a decision that another organization must already have made.

Compare quotations on completed decisions and deliverables

An inexpensive scan can be poor value if the quotation excludes the interpretation, feasibility work or data package the project needs. Ask each bidder to separate the proposed tasks, assumptions, customer-supplied inputs and exclusions. Compare equivalent examination stages and regions. Treat unresolved feasibility as an open technical item rather than inserting a guessed price for work that has not been defined.

Use the free inspection quote comparison tool with actual supplier figures once the scopes are comparable. Include the agreed reporting, review and shipping responsibilities, and record which items remain unknown. The tool organizes commercial inputs; it does not determine whether a proposed examination is technically adequate. Keep the technical review alongside the cost comparison so the lowest numerical total does not silently become the preferred inspection plan.

Close the planning review with a concise decision record: what evidence is required, who will produce it, what must be resolved before work starts and who approves the final disposition. If the project is ready for supplier input, prepare an inspection request with the part location, manufacturing route, drawing references and intended deliverables. Share only information authorized for that recipient and make the scope specific enough for a useful technical response.

Sources and editorial scope

Primary sources checked October 31, 2026: NIST measurement science program, NIST IR 8538, published in 2024, NIST CT acquisition study, published in 2019, NIST pore-structure analysis, published in 2017, and NASA-STD-6030 scope and document record. The workflow and development-part example are NDT Connect editorial guidance, not a reported qualification study or an approved inspection procedure.