Dimensional Inspection for Critical Parts

In defense, aerospace, and other regulated manufacturing environments, dimensional inspection is not a paperwork exercise—it is a risk-control step that protects downstream assembly, performance, and certification. When a part is additively manufactured (AM) by powder bed fusion (PBF) such as DMLS / SLM, densified via Hot Isostatic Pressing (HIP) or PM-HIP, and then finished by CNC machining (often 5-axis), the geometry you receive is the cumulative result of multiple processes, each with its own variation mechanisms. A robust inspection plan verifies that the final, delivered configuration meets the drawing, the model-based definition (MBD), and any program-specific requirements under AS9100 quality systems and (when applicable) ITAR/DFARS-controlled workflows.

This article provides practical, procurement-ready basics for the main inspection modalities used on critical parts—CMM, optical scanning, CT / X-ray, and ultrasonic NDE—and shows how to choose the right method, how acceptance is typically documented, and how to specify dimensional inspection services clearly on RFQs.

CMM vs scanning

Coordinate Measuring Machines (CMMs) and 3D scanning both compare the manufactured part to the nominal definition, but they do it in very different ways and with different strengths. Understanding those differences helps engineering and procurement avoid two common failure modes: (1) paying for a high-end method that doesn’t reduce risk for the tolerance structure, or (2) using a fast method that can’t defend acceptance on tight features.

CMM inspection uses a calibrated machine and probe system (touch-trigger or scanning probe) to measure discrete features or continuous surfaces in a controlled coordinate system. A CMM excels when you need traceable, defensible measurements on critical features such as datums, true position, bores, and flatness/parallelism relationships. For aerospace drawings with GD&T (ASME Y14.5) and well-defined datum schemes, CMM data aligns naturally with the intent of the print. CMM is also often the default expectation for First Article Inspection (FAI) evidence because it provides repeatable feature-by-feature results that can be tied back to datums.

3D scanning (structured light, laser scanning, or photogrammetry-based systems) captures a dense point cloud rapidly and produces a mesh for comparison to CAD. Scanning is highly effective for freeform or organically contoured surfaces—common in AM brackets, manifolds, impellers, and conformal structures—where defining dozens of discrete CMM features is inefficient. Scanning also shines for part-to-part comparisons, shrink/warp studies, and process tuning (e.g., characterizing distortion before and after HIP or stress relief).

Key practical considerations engineers should address:

• Measurement traceability and uncertainty: A well-maintained CMM in a temperature-controlled metrology lab typically supports lower measurement uncertainty for tight tolerances. Scanning accuracy depends strongly on system calibration, surface condition, spray/coating usage, and how the scan is registered to datums. For acceptance of tight tolerance features, confirm the scanning system’s achievable uncertainty is comfortably below the tolerance band.

• Datum strategy: CMM workflows naturally establish and report datums. Scanning can be datum-based as well, but the method of alignment (best-fit vs datum alignment) matters. Best-fit alignment is useful for process characterization, but it can hide datum-related issues that will matter in assembly. For procurement, explicitly call out datum alignment required if the data will support acceptance.

• Surface condition: Highly reflective or rough AM surfaces can challenge optical scanners. Post-processing (bead blast, machining, tumbling) changes surface texture and may change measured edges if not defined consistently. For critical edges, it’s common to inspect after the final machining operation and final surface finishing called out on the drawing.

• Feature accessibility: CMM probes need line-of-sight and physical access; deep pockets and narrow channels can be difficult. Scanners also need optical access. When internal passages or hidden features must be verified, CT is often the only practical method.

CT and X-ray overview

X-ray radiography and computed tomography (CT scanning) are non-destructive examination (NDE) methods that use X-rays to reveal internal structure. They are particularly valuable for AM and PM-HIP components because internal porosity, lack-of-fusion, inclusions, and unmachined internal features can be mission-critical yet invisible from the outside.

2D X-ray radiography produces projection images. It is often used for welds, castings, and some AM parts when the goal is to detect gross internal discontinuities. It can be fast and cost-effective, but interpretation depends on part thickness, material, and the orientation of features relative to the beam.

CT scanning produces a 3D volumetric dataset from many radiographic projections. With CT, you can do more than “find flaws”—you can measure internal dimensions, check wall thickness, verify internal lattices, and confirm the as-built position of internal channels relative to external datums. This makes CT uniquely useful for complex PBF components such as heat exchangers, manifolds, turbine components, and optimized structures that rely on internal geometry for performance.

Practical CT notes for regulated manufacturing:

• Dimensional metrology vs defect detection: CT can be used for (a) dimensional CT where the emphasis is geometric measurement, or (b) NDE CT where the emphasis is discontinuity detection and characterization. These are related but not identical. Dimensional CT requires tighter control of scale calibration, artifact correction, and feature extraction methods.

• Material and thickness limits: Dense alloys (e.g., Inconel 718, cobalt chrome) and thick sections attenuate X-rays strongly, increasing scan time and potentially reducing resolution. Titanium alloys (e.g., Ti-6Al-4V) are generally more CT-friendly than high-density superalloys at the same thickness, but geometry and required resolution drive feasibility.

• Resolution and detectability: A CT system’s voxel size (often discussed as “resolution”) is not the same as detectability for a given defect type. Acceptance should be based on a defined inspection plan: what must be detected (porosity size? crack-like indications?), at what location, and with what confidence. For procurement, avoid vague requirements like “CT scan part” and instead specify objective criteria (e.g., minimum detectable pore size in a defined zone, or dimensional CT comparison of internal channel diameters to CAD).

• Data deliverables: CT outputs can include a pass/fail report, annotated slice images, defect tables, 3D renderings, and sometimes the volumetric dataset. For ITAR-controlled programs, clarify whether datasets can be shared and how they will be stored and transferred under controlled access.

• Relationship to HIP: HIP is used to close internal porosity in castings and AM parts and is common in aerospace AM workflows. Even when HIP is performed, CT or X-ray may still be required to verify that critical zones meet acceptance for residual porosity or to confirm internal geometry after densification and any subsequent heat treatment.

Ultrasonic basics

Ultrasonic testing (UT) is an NDE technique that introduces high-frequency sound into a part and analyzes reflections to detect internal discontinuities and to evaluate material properties. UT is widely used in aerospace for forgings, billets, and critical structural parts because it can inspect deeper sections without the radiation and heavy shielding requirements of X-ray methods.

Where UT fits well in critical-part workflows:

• Detection of planar flaws: UT is particularly sensitive to crack-like or planar discontinuities oriented favorably to the sound path. For some defect types, UT can be more effective than radiography.

• Thick-section inspection: For thick metallic parts where CT becomes impractical or cost-prohibitive, UT can provide a viable inspection route. That said, geometry complexity and access can limit UT effectiveness, especially on highly contoured AM parts.

• Process control and incoming material verification: UT is often used for incoming inspection of raw material (e.g., titanium or nickel alloy bar/forging stock) to ensure it meets internal quality requirements before machining. In a hybrid AM + machining program, UT may be applied to the substrate, build plate, or feedstock-related items as part of a broader material control strategy.

Important limitations and real-world cautions:

• Coupling and surface conditions: UT requires good coupling (gel/water) and stable contact. Very rough AM surfaces can degrade coupling and signal quality unless surfaces are machined or prepared.

• Geometry and interpretation: Complex geometry causes reflections that can be misinterpreted as indications. Qualified technicians, qualified procedures, and (for critical hardware) controlled acceptance criteria are essential. In aerospace contexts, UT is commonly performed under a Nadcap-accredited NDE program when required by customer flowdowns.

• Not a direct dimensional tool: UT is fundamentally an NDE method, not a dimensional metrology method. It can support thickness measurement in some scenarios, but if the requirement is strict GD&T conformance, a CMM, scanning, or dimensional CT approach is more appropriate.

When each method is used

Most successful defense and aerospace suppliers do not pick a single inspection method; they build an inspection stack that matches the tolerance risk, feature accessibility, and certification requirements. Below is a practical way to select methods based on real manufacturing workflows, including additive + HIP + machining.

1) Start from the drawing and the function. Identify critical-to-quality (CTQ) features: datum structure, mating interfaces, sealing surfaces, bores, threads, and any features tied to performance (flow, fatigue, balance). For AM parts, also identify internal features that cannot be probed (channels, lattice zones) and zones prone to defects (overhangs, thick-to-thin transitions).

2) Map the manufacturing route. A typical route might be: PBF build → stress relief → support removal → HIP (if required) → heat treatment → rough machining → finish machining → surface finishing → final inspection. Each step can change geometry. For example, HIP can produce small dimensional shifts; machining establishes final datums and tight features. This is why final acceptance is usually tied to the post-machined condition.

3) Choose the primary dimensional method for external features. If the part has tight GD&T on accessible features, CMM is often the primary acceptance tool. If the part has complex freeform surfaces with moderate tolerances, 3D scanning may be the most efficient, especially for profile tolerances or surface deviation maps—provided alignment and uncertainty are appropriate.

4) Add internal verification where needed. Use CT when internal geometry must be measured (wall thickness, channel diameter, internal feature position) or when internal defect detection is required in complex AM geometry. Use X-ray radiography when you need a faster screening method for internal discontinuities and the geometry/material supports meaningful interpretation. Use UT for thick-section discontinuity detection, incoming material verification, or when radiography/CT are impractical.

5) Match method to tolerance and risk. For very tight tolerances, the measurement system capability matters as much as the nominal accuracy. A practical rule in high-consequence manufacturing is to ensure the measurement method’s uncertainty is small relative to the tolerance (so acceptance decisions aren’t dominated by measurement noise). If you are buying to aerospace tolerances, ask the supplier how measurement uncertainty is controlled (calibration, environmental controls, fixturing, alignment strategy, and software traceability).

6) Use in-process checks strategically. Waiting until the end to discover nonconformance is expensive. Many programs use: (a) in-process CMM checks after rough machining to validate datum establishment; (b) scanning to check overall distortion trends and machining stock; and (c) targeted CT on first builds or process changes to validate internal geometry prior to full-rate production.

7) Consider compliance and accreditation flowdowns. If the customer requires Nadcap for NDE or special processes, the inspection method must be performed under the accredited scope. Similarly, an AS9100 system typically expects controlled procedures, calibrated equipment, and traceable records. For ITAR or DFARS programs, ensure data handling, access control, and record retention are compatible with contract requirements.

Reporting and acceptance

Inspection that can’t be audited or tied back to requirements is not very useful in regulated supply chains. For critical parts, acceptance is usually based on objective evidence packaged in a documentation set that supports internal QA and customer review.

Common reporting elements for dimensional inspection services:

• CMM report: Typically includes the datum reference frame used, the feature IDs measured, nominal values, actual values, deviations, and pass/fail against tolerance. For GD&T, reports often include true position, profile, flatness, perpendicularity, and related callouts. Ensure the report clearly states the revision of the drawing/model used and the calibration status of the equipment.

• Scan deviation report: Usually includes color maps of deviation to CAD, specific cross-sections, and quantitative summaries (e.g., max/min deviation in defined zones). For acceptance, the report should state the alignment method (datum-based vs best-fit), filtering/smoothing settings (if any), and the inspection software version or procedure reference when traceability matters.

• CT / X-ray NDE report: May include the technique (energy, exposure settings, voxel size for CT, number of projections), sensitivity/quality indicators used (when applicable), interpretation criteria, and results. If the requirement includes internal dimensional verification, the report should specify how measurements were extracted from the volume and how scale was calibrated.

• UT report: Should include the procedure and acceptance criteria, transducer details, scan coverage, calibration blocks/standards used, indications found (location, amplitude, characterization), and disposition.

Acceptance in aerospace and defense often ties to additional quality documentation:

• First Article Inspection (FAI): Many programs require an AS9102-style FAI package for first production or after significant changes. The inspection plan should be able to produce the characteristic-by-characteristic evidence needed for FAI, not just a single “pass” statement.

• Material traceability and CoC: For critical alloys and controlled programs, the inspection report should be supported by material certifications, lot/heat traceability, and a Certificate of Conformance (CoC) referencing the applicable drawing revision and purchase order requirements.

• Record control: Under AS9100 and defense contracts, retention, configuration control, and controlled access to inspection data matter. If you require raw point clouds, CT volumes, or native CMM program outputs, specify it—otherwise many suppliers will only provide PDF summaries.

• Nonconformance handling: A mature supplier will document nonconformances, perform root cause analysis as required, and control rework/repair steps. For AM parts, rework may be limited; for machined parts, rework may affect datums or surface integrity. Acceptance should clearly state what rework is allowed and how it must be re-inspected.

How to specify inspection on RFQs

RFQs frequently fail at inspection definition. The buyer assumes “standard inspection,” while the supplier prices a minimal approach that may not satisfy program needs. For critical parts, write inspection requirements the same way you write manufacturing requirements: specific, testable, and tied to deliverables.

Step-by-step RFQ approach that works in aerospace/defense supply chains:

1) State the governing definition. Identify whether acceptance is to a 2D drawing, a 3D model (MBD), and which revision controls. If you use a digital thread, specify the data format and any PMI requirements.

2) Identify CTQs and inspection method expectations. Do not require one method for every feature; instead, specify what must be proven and allow the supplier to propose the method, with exceptions for internal geometry or mandated NDE. For example: “Critical datum features and positional tolerances shall be verified by CMM” or “Internal channels shall be verified via dimensional CT.”

3) Define deliverables and reporting format. Call out required reports (CMM feature report, scan deviation maps, CT/UT reports), whether pass/fail is sufficient, and whether you need raw data. Also specify whether the inspection results must be included in a manufacturing/quality documentation pack.

4) Specify standards and accreditation flowdowns. If Nadcap is required for NDE, state it explicitly. If AS9102 FAI is required, state it explicitly. If ITAR/DFARS controlled information is involved, require compliant handling and define any restrictions on data transfer.

5) Clarify sampling and frequency. Prototype, first article, and production runs often use different inspection intensities. A common pattern is 100% inspection of CTQs on first article, followed by reduced sampling in production based on capability and risk. Define whether every part needs full reporting or if periodic CMM/CT is acceptable.

6) Address fixturing and datum simulation. If the drawing requires functional gaging or datum simulation (e.g., assembly constraints), specify it. For thin-walled AM parts, clamping can distort parts; metrology fixturing should mimic functional constraints without inducing measurement error.

Example RFQ language you can adapt (keep it consistent with your internal quality system and customer flowdowns):

• Dimensional inspection: “Supplier shall provide dimensional inspection services demonstrating conformance to drawing/model revision __. Datum features A/B/C and all GD&T position/profile characteristics marked as CTQ shall be verified using a calibrated CMM (or equivalent metrology method with documented measurement uncertainty). Provide a characteristic-by-characteristic report with nominal, actual, deviation, and pass/fail.”

• Scanning for freeform surfaces: “Provide 3D scan-to-CAD deviation report for surfaces __ with datum-based alignment to A/B/C. Include color map, section views at __ locations, and quantitative max/min deviation in defined zones. Identify scanner type and inspection procedure used.”

• CT for internal geometry: “Provide dimensional CT verification of internal channels __ including minimum wall thickness and channel diameter at __ stations. Report voxel size (or equivalent), calibration method, and measurement extraction approach. Provide annotated images and measurement table.”

• NDE requirement: “Perform NDE per customer-approved procedure; if Nadcap accreditation is required, NDE shall be performed within supplier’s Nadcap scope. Provide NDE report and technician qualification evidence upon request.”

• Documentation pack: “Deliver CoC, material traceability documentation, heat/lot data, process certifications (HIP/heat treat where applicable), and inspection reports with shipment. Records shall be retained for __ years.”

• Change control: “Supplier shall notify buyer prior to changes in AM build parameters, HIP cycle, heat treat, machining strategy, or inspection method that could affect form/fit/function. Changes may require a delta FAI.”

Finally, procurement teams should evaluate inspection not only as a line item cost, but as a capability indicator. A supplier that can explain their metrology plan, measurement uncertainty, data controls, and compliance posture (AS9100, ITAR/DFARS, Nadcap as applicable) is typically better positioned to deliver consistent, certifiable hardware—especially when AM, HIP, and precision machining are combined in one controlled workflow.