Skip to main content

Qualification and certification: the central bottleneck

📍 Where we are: Part V · Industrial Reality — Chapter 12 of 19. Parts I-IV covered the physics, modeling, sensing and new materials and systems; this chapter asks why so few of those advances reach certified, flying or implanted hardware.

The simple version

Imagine a bakery whose cakes taste slightly different depending on the oven, the day and the flour batch. For a birthday that does not matter. Now suppose each cake must be guaranteed never to collapse under a load, and a regulator wants proof. You would have to bake and break an enormous number of cakes, and if you change the oven's thermostat even a little, the regulator asks you to start the proof over. That is qualification of additively manufactured (AM) parts: the cost of proving the cake is safe, again and again.

What this chapter covers​

We define what "qualification" and "certification" mean, then follow the five-part structure of this book: the problem, the current state of the art (NASA, FAA, EASA, FDA and ISO/ASTM routes), why it stays hard, what is still open, and what it does to industry. Claims are tagged by evidence level: lab-scale (shown in research), production (shown in qualified series use) or contested (experts disagree).

Problem​

Qualification is the evidence that a process, machine, material or part meets its requirements; certification is the regulator's or customer's formal acceptance of that evidence. For conventional metals, a long-established apparatus exists: wrought and cast alloys have published, statistically derived design allowables (for example "A-basis" and "B-basis" values, which bound the strength that a stated fraction of material exceeds with stated confidence), maintained in handbooks such as MMPDS (the Metallic Materials Properties Development and Standardization handbook).

AM breaks the assumption behind those handbooks: the material is created at the same time as the part. Properties depend on powder lot, machine, laser calibration, scan strategy, build orientation, location on the plate, geometry and post-processing [1] [2] [3] [4]. NIST describes current qualification as "extensive empirical testing consisting of thousands of individual tests", costing millions of dollars and taking several years, with minor process changes requiring complete re-qualification [5] (page updated March 2025; it states the burden qualitatively and gives no per-program figures).

State of the art​

Aerospace​

  • NASA-STD-6030 (baseline approved 21 April 2021) is the most explicit single framework. It contains 115 numbered "AM requirements", classifies parts by consequence of failure (Class A, B, C), and builds on an AM Control Plan, a Qualified Material Process, and a Material Properties Suite whose allowables are substantiated through statistical process control (SPC) of the ongoing production [6]. It covers laser powder bed fusion (L-PBF) and wire or blown-powder directed energy deposition (DED) for metals, and several polymer processes; "adaptive" processes whose heat input changes during the build, such as electron-beam powder bed fusion, are outside its scope [6]. A 2026 NASA lessons-learned presentation advises engaging NASA AM experts early and documenting tailored requirements in the control plan [7]. Evidence level: production use in NASA programs; I did not verify the number of programs.
  • FAA published Advisory Circular AC 33.15-3 (23 June 2023) as an acceptable, but not the only, means of showing compliance with 14 CFR 33.15 for powder-bed-fusion engine parts [8].
  • EASA Certification Memorandum CM-S-008, Issue 04, states that effort should be proportionate to novelty, criticality and complexity, and adds guidance for low-criticality applications [9]. (One industry document of December 2025 cites a different issue number; use EASA's own text as authoritative.)
  • Industry guidance: the Aerospace Industries Association recommends that statistically based material and process data exist at certification, with a "building-block" test pyramid from coupons to components [10].
  • Demonstrated in production: GE reported in April 2015 that its printed cobalt-chrome T25 sensor housing for the GE90 had become the first 3D-printed part certified by the FAA to fly in its commercial engines [11], and GE Aerospace now states that more than 10 additively made parts across the GE90, LEAP, GEnx and GE9X are FAA-approved [12]. Both are statements by the manufacturer about its own products; we did not find an FAA statement of a part count, and we give none.

Medical​

FDA's guidance (final, December 2017) is non-binding and divides expectations into design/manufacturing considerations (quality system) and device testing considerations for premarket submissions, including patient-matched devices; it recommends the Pre-Submission process when unclear [13]. It qualifies a device, not a material or printer in general. Evidence level: production (many AM implants are cleared), through case-by-case submissions.

Cross-sector standards​

ISO/ASTM 52920:2023 sets qualification requirements for AM processes and production sites, on top of a quality system such as ISO 9001, EN 9100 or ISO 13485; qualification may be part-specific or process-generic, and changes (hardware, firmware, feedstock, relocation) are evaluated individually and trigger re-qualification only if the process or part is impacted [14]. ISO/ASTM TS 52930:2021 defines installation, operation and performance qualification (IQ/OQ/PQ) for laser powder bed fusion machines [15]. A review of the state of standardization argued early that standards for feedstock, process and testing are a prerequisite for qualification [16]. A 2025 AFRL/NIAR guidance document supplements the AMS 7003 and 7032 specifications with minimum key process variables for machine operational qualification: scan-field calibration, shielding-gas flow, oxygen content, laser power and beam spot [17].

Pressure equipment and nuclear​

ASME code qualification is under way rather than finished: an Argonne plan addresses laser powder bed fusion 316 stainless steel for high-temperature nuclear (Section III, Division 5) service, where only 316H is currently qualified for Class A construction [18]. Evidence level: lab-scale and in-progress.

Why it is hard​

  1. The material is made with the part. Allowables from coupons may not transfer to a real geometry, where thermal history and defect distributions differ [1].
  2. Process sensitivity and variability. Build-to-build and machine-to-machine variation make it difficult to claim that two machines produce one material; reviews list this as a core gap [2].
  3. Fatigue and fracture dominate. Static strength is easy to reproduce; fatigue life depends on rare surface and subsurface defects, so the lower tail of the distribution, not the mean, governs design [1].
  4. Inspection limits. Computed tomography is often the only way to see internal defects, but is constrained by part size, density and resolution [2].
  5. Economics of change. If every parameter change risks re-qualification [5], engineers freeze processes, so improvement itself becomes expensive.

Qualification pyramid from material and process through machine to part and component, with the gates NASA, FAA, EASA, FDA and ISO/ASTM apply. The qualification stack: feedstock, machine (IQ/OQ/PQ), process, material allowables and the part each carry their own evidence, and a change at any lower tier can reopen the tiers above it. Original diagram by the authors, created with AI assistance.

Open questions​

  • Qualify the machine or the part? NASA-STD-6030 qualifies a material process and monitors it statistically; ISO/ASTM 52920 allows part-specific or process-generic routes; machine-focused documents define IQ/OQ/PQ [6] [14] [15]. Whether machine-level qualification can substitute for per-part testing is contested: it requires machine-to-machine equivalence that has not been demonstrated generally.
  • Can monitoring replace testing? Using in-situ data (Chapter 6) and models (Chapter 4) to reduce coupons is a stated aim in industry guidance [10] and in NIST's measurement-science review for real-time control of powder bed fusion [19], but I found no regulator accepting monitoring alone as the sole basis for a flight-critical part. Evidence level: lab-scale.
  • Minimum change that triggers re-qualification. Standards say the impact must be evaluated [14], yet no agreed, quantitative rule exists in the sources reviewed.
  • Common test artifacts and data sharing. Proposals for a common L-PBF qualification test artifact exist, but I did not find an adopted, cross-vendor one in the sources reviewed.
  • Pathway from OEM qualification to civil certification for repair and spare parts, which EASA's memorandum addresses only in part [9].

Industry impact​

Qualification cost and time push adoption toward high-value, low-volume or hard-to-make parts where the business case absorbs it, and toward large organizations that can fund test programs. The same cost favors consolidating many parts into one (a design freedom reviewed in [20]) so a single qualification pays off repeatedly. Smaller suppliers face an expensive entry, a theme taken up in the digital thread chapter and in the cost chapter. Statements about the dollar or time cost of an individual program are quoted in sources only qualitatively here; no verified program-level figure is given.

Why it matters​

A technology that cannot be certified cannot carry safety-critical loads, and most of AM's high-value promise lies there. Qualification, not printing, sets the pace of adoption.

In the real world​

An engine maker that wants a printed bracket must freeze the powder specification, machine settings and post-processing, qualify the process under a regime such as AC 33.15-3 or its own design-approval system, and re-open that work when the machine's laser module is replaced. A medical device maker follows the FDA guidance route per device, with patient-matched variants handled through defined design envelopes.

Key terms​

  • Qualification — evidence that a process, machine, material or part meets stated requirements.
  • Certification — formal acceptance of that evidence by a regulator or customer.
  • Design allowable — a statistically derived strength value used in design, such as A-basis or B-basis.
  • Statistical process control (SPC) — ongoing statistical monitoring used to show a process stays within its qualified distribution.
  • IQ/OQ/PQ — installation, operational and performance qualification of a machine.
  • Building-block approach — a test pyramid progressing from coupons to elements, subcomponents and full components.
  • Witness coupon — a test specimen built alongside parts to check the build.
  • Re-qualification — repeating some or all qualification after a process change.
  • Patient-matched device — a medical device designed to fit an individual patient's anatomy.
  • Part criticality class — a classification by consequence of failure that sets the rigor required.

Where this leads​

The evidence qualification needs is data, and data must be traceable across design, build and inspection. That is the subject of Digital thread, data and the supply chain.