Quality, inspection and non-destructive testing
📍 Where we are: Part IV · Making It Real — Chapter 13 of 15. The bracket has been printed and post-processed (Post-processing). This chapter asks the next question: is it actually good, and how can we tell without breaking it?
Imagine buying a loaf of bread that you are not allowed to cut open. You can weigh it, measure it, look at the crust, tap it, and maybe take an X-ray. You can also bake a small test roll from the same dough and cut that one open. Inspecting an AM part works the same way. Some checks look at the outside, some see inside without damage, and some are done on sacrificial test pieces built alongside the real part. No single check tells the whole story, so you combine them.
What this chapter covers
We list what can go wrong inside and outside a printed metal part, then walk through the main checks: dimensional metrology, surface roughness, density, computed tomography (CT), other non-destructive testing (NDT), and witness coupons with mechanical tests. We finish with the idea that matters most for a designer: every method has blind spots, so the inspection plan is part of the design. Most examples are about metal laser powder bed fusion (see Laser powder bed fusion), because that is where inspection is hardest and most regulated.
What can go wrong
Think of defects as belonging to three groups.
- Internal voids. Lack of fusion means regions that did not melt together properly, often with irregular, sharp-edged shapes. Gas or keyhole pores are rounder voids left by trapped gas or an unstable melt pool (the small pool of molten metal under the laser). In laser powder bed fusion, insufficient energy can leave unfused scan tracks, and incomplete wetting and balling can cause pores [1]. In an X-ray tomography study of electron beam melted Ti-6Al-4V, most pores were small spherical gas pores concentrated in the hatched infill region, while rarer irregular pores, attributed to lack of fusion, lay mostly in the contour region [2].
- Cracks and residual stress. Cracks can form during build or heat treatment. Residual stress, discussed in Post-processing, shows up as distortion.
- Geometry and surface. The part may be out of tolerance, warped, or rough. As-built metal surfaces are markedly rougher than machined ones, with values that vary widely by process, material and orientation [3], where Ra is the average height deviation of the surface profile.
Not every defect matters equally. A review that used X-ray tomography to study how defects affect mechanical properties reports that the smallest pores are harmless, and that tomography can help separate critical defects from harmless ones [4]. Size, shape and, above all, distance from the surface decide how dangerous a defect is.
Checking the outside
Dimensional metrology
Metrology is the science of measurement. A coordinate measuring machine (CMM, a probe that touches points on the part) or an optical 3D scanner compares the real part with the CAD model. Machined faces are checked to drawing tolerance; as-built faces are checked more loosely. To judge a machine rather than a part, ISO/ASTM 52902 describes test artefacts, benchmark shapes built to assess a system's geometric capability, and says what to measure on them [5].
Surface roughness
A profilometer drags a fine stylus, or an optical instrument scans, across the surface to report Ra and related numbers. Roughness matters for fatigue and for sealing faces. Studies use different measurement techniques and ways of characterising the surface, and a review of surface texture metrology for metal AM calls for better methods and reporting [3], so compare numbers with care.
Checking the inside
Density (Archimedes)
Archimedes' principle gives density by weighing the part in air and then submerged in liquid; the difference gives its volume. ASTM B962 is a test method for powder-metallurgy products, with a companion method, B311, for materials under 2 % porosity [6]. These were written for powder-metallurgy parts, so applying them to AM parts is an extension; check which method your specification names. Density is cheap and fast, but it is only an average: it cannot say where a pore is, and one large near-surface pore can hide in an otherwise dense part. NIST work on porosity monitoring built reference samples and measured their porosity with several techniques before correlating ultrasonic signals with it [7].
Computed tomography (CT)
CT takes many X-ray images from different angles and reconstructs a 3D map of the inside. A 2018 review calls micro-CT an established method for AM parts, especially for dimensions and porosity analysis, and discusses practical limits [1]. An earlier review names limited resolution for porosity and high cost and slow speed as the main barriers to wider adoption [8]. Two limits matter. First, resolution: small witness specimens can be imaged at higher resolution than a large complex part, so smaller defects can be seen in them. Second, cost and data: for big parts, resolution is limited by computing resources, and the review cites cost and lack of standards as the main obstacles to wider use [1]. CT can also check internal channels that no probe can reach, which is a major reason it is used on AM parts.
Each defect type has one or two methods that find it well; none finds everything, so an inspection plan combines several.
Original diagram by the authors, created with AI assistance.
Other non-destructive methods
ASTM E3166 is a guide to non-destructive examination of metal AM aerospace parts after build. It covers CT, eddy current (a probe that senses surface and near-surface flaws through electrical effects), optical metrology, penetrant testing (dye that seeps into surface cracks), process compensated resonance testing, radiography, infrared thermography and ultrasonic testing [9]. It applies to as-built, intermediately processed and finished parts. It sets no accept/reject criteria and does not cover in-process monitoring during the build [9].
One interaction worth knowing: HIP closes internal pores but not those with surface-connected ligaments [10]. So inspection after HIP looks for a different set of flaws than inspection before it.
Witness coupons and mechanical tests
NDT can find flaws but cannot measure strength. For that you break something. A witness coupon is a small test piece built in the same job as the part, with the same material and, ideally, the same orientation and heat treatment. It is then tensile or fatigue tested. NASA-STD-6030 builds witness tests into statistical process control against a reference distribution [11].
Where and how a specimen was built matters. ISO/ASTM 52909 gives guidelines for reporting orientation and location dependence of mechanical properties, and asks for part location (as coordinates in the build volume), orientation and scan strategy to be reported [12]. Properties of an AM metal can differ with direction, so a coupon built flat does not automatically represent a feature built upright; a widely cited review documents this anisotropy and heterogeneity of microstructure and properties in metal AM [13].
For the bracket, a plausible (illustrative, not a specification) plan is: scan the machined faces with a CMM, measure density on a coupon, tensile test two coupons from the build, and CT a first-article part. For a non-critical bracket that might be enough; for a flight bracket it would grow.
Why it matters
AM can produce hidden voids and unusual surfaces that conventional inspection was not designed for. If you cannot inspect a feature, you must either design it to be inspectable, prove the process so well that inspection is a confirmation, or both. That is the bridge to qualification, covered next.
In the real world
Inspection scales with consequence. A hobby bracket needs a calliper. A spacecraft bracket needs documented CT, coupons and traceable records. Inspection costs time and money, so the plan should be set during design, when a drain hole, a flat reference face or a thicker wall can make inspection cheap.
Key terms
- Metrology — the science and practice of measurement, including checking part dimensions.
- CMM (coordinate measuring machine) — a machine that touches points on a part to measure its geometry.
- Ra — the average height deviation of a surface profile, a common roughness number.
- Lack of fusion — a void where material did not melt together properly.
- Keyhole pore — a round void caused by an unstable, deep melt pool trapping gas.
- Archimedes density — measuring density by weighing a part in air and in liquid.
- Computed tomography (CT) — X-ray imaging from many angles, reconstructed into a 3D map of the part's inside.
- Non-destructive testing (NDT) — inspection methods that do not damage the part.
- Witness coupon — a small test piece built alongside the real part to be broken or sectioned in its place.
- Test artefact — a benchmark shape built to assess an AM system's geometric capability.
Where this leads
Checks only mean something against agreed rules. Standards, qualification and certification shows who writes those rules and how a part earns trust.