Acceptance Criteria for Critical Parts

For defense, aerospace, and other regulated applications, “acceptance criteria” is not a formality—it is the technical and contractual definition of what “good” looks like for a critical metal part. Done well, acceptance criteria aligns design intent, process capability, inspection methods, and supplier documentation into a single, auditable pass/fail framework. Done poorly, it creates late-stage nonconformances, disputed rejections, schedule slips, and rework that can exceed the cost of the part.

This article provides a practical method to set acceptance criteria manufacturing teams can actually execute—across additive manufacturing (AM) and traditional workflows—while staying compatible with common quality systems and regulated supply chain requirements (e.g., AS9100, NADCAP, ITAR, DFARS). The focus is on metal parts where failure carries mission, safety, or program risk: load-bearing brackets, housings, rotating hardware, structural fittings, fluid manifolds, and parts with pressure boundaries or high-cycle fatigue exposure.

Defining CTQs

Acceptance criteria should be built from Critical-to-Quality (CTQ) characteristics. CTQs are the features, properties, and attributes that directly affect fit, function, reliability, safety, or regulatory compliance. The key is to avoid treating every dimension or note as “critical” (which drives cost and slows production) while ensuring the real failure drivers are explicitly controlled.

Step 1: Start with the functional failure modes, not the drawing. A reliable CTQ list typically comes from a design review that includes engineering, quality, manufacturing, and (for defense/aero) program leadership. Use your internal risk method—FMEA, hazard analysis, or a tailored risk matrix—to identify what would cause:

• Loss of structural integrity (yield/ultimate, fatigue, fracture toughness)
• Leakage (pressure boundary, porosity, sealing surfaces)
• Loss of alignment (bearing seats, datum schemes, positional tolerances)
• Thermal/creep deformation (hot sections, engine-adjacent hardware)
• Corrosion-induced failure (galvanic couples, salt fog exposure, stress corrosion)

Step 2: Translate failure drivers into measurable requirements. CTQs must be measurable with defined methods and limits. Typical CTQ categories for critical metal parts include:

1) Dimensional and geometric CTQs
• Datum structure and feature relationships (true position, concentricity, runout)
• Interfaces: mounting pads, bolt patterns, bearing fits, spline geometry
• Surface form: flatness, cylindricity, perpendicularity where load transfer matters

2) Material property CTQs
• Chemistry and heat/lot conformance (material spec, melt/heat number)
• Mechanical properties (tensile, yield, elongation, hardness) at relevant temperatures
• Microstructure constraints (grain size, phases, carbide/nitride control, alpha case for Ti)

3) Defect and discontinuity CTQs
• Internal porosity and lack of fusion risk for PBF (DMLS/SLM) parts
• Inclusions, shrinkage, cracking risk in cast or wrought sources
• Surface-connected discontinuities (cracks, laps) particularly after machining

4) Post-processing CTQs
• Heat treat / aging condition
• Hot Isostatic Pressing (HIP) parameters and verification for AM or PM-HIP parts
• Residual stress relief for PBF parts prior to machining
• Coatings and surface treatments (anodize, passivation, plating), including NADCAP process controls when applicable

Step 3: Define CTQ ownership across the manufacturing route. The acceptance criteria must reflect the real workflow. For example, a typical critical metal AM route might be:

Powder qualification → PBF build (DMLS/SLM) → stress relief → support removal → HIP (if specified) → rough machining → NDE (as applicable) → finish 5-axis CNC machining → surface finishing/coating → final inspection → documentation pack.

Each CTQ should be assigned to the operation that most effectively controls it. Internal porosity is controlled primarily by build parameters, powder condition, and HIP (if used). Final true position is controlled by machining strategy, fixturing, datum transfer, and CMM verification. Acceptance criteria that ignore where capability actually resides create “inspection-only control,” which is both expensive and unreliable.

Step 4: Set limits based on capability and risk. For critical parts, do not assume “tightest possible” is best. Limits should come from a combination of:

• Functional need (what the part requires to work safely)
• Process capability (what your qualified route can repeatedly achieve)
• Inspection capability (what you can measure with acceptable uncertainty)

Where history is limited (new AM geometry, new alloy, first-time supplier), consider phased criteria: prototype criteria with enhanced inspection and learning, then production criteria after process capability is demonstrated.

Inspection plans

Acceptance criteria only works when paired with an inspection plan that is explicit, feasible, and aligned with your quality system. In regulated programs, this plan becomes the backbone of the inspection record and the basis for objective evidence in your certification package.

Step 1: Create an inspection matrix tied to CTQs. For each CTQ, define:

• Requirement (tolerance/limit and drawing/spec reference)
• Inspection method (CMM, CT scanning, ultrasonic, penetrant, profilometer, hardness, tensile coupons)
• Stage (incoming, in-process, final, post-HIP, post-machining, post-coating)
• Frequency (100%, per-lot, first article only, periodic)
• Acceptance rule (pass/fail criteria, rework allowance, disposition path)

Step 2: Use the right metrology for the geometry and risk. Common approaches for critical metal parts include:

• CMM (Coordinate Measuring Machine) for datum-based GD&T verification, positional tolerances, and form on machined features. Ensure probe access and measurement strategy is planned early—especially for deep pockets or complex AM manifolds.

• CT scanning (industrial computed tomography) for internal geometry verification and internal defect characterization in complex AM parts (e.g., lattice structures, internal channels). CT is powerful but must be specified carefully: voxel size, scan energy, part orientation, and defect sizing method drive repeatability.

• Surface roughness and texture measurements using profilometry or optical methods. For PBF parts, clarify whether acceptance is based on as-built surfaces, post-processed surfaces (machined, shot peened, abrasive flow), and which parameter (Ra, Rz, Sa) applies.

• Mechanical testing via witness coupons or test bars tied to the build/lot. For AM, define coupon orientation, location on the build plate, and post-processing equivalence (stress relief, HIP, heat treat). Ensure acceptance criteria states which data is required for release (e.g., tensile at room temperature, hardness, density).

Step 3: Stage inspections to prevent expensive late discovery. A practical plan usually includes:

• Incoming inspection: verify material certs, heat/lot traceability, powder batch controls (AM), and any special process certifications required.
• In-process inspection: after rough machining, check datums and critical stock conditions; after HIP/heat treat, verify hardness or key property proxies as allowed.
• Final inspection: full dimensional verification of CTQs, final NDE as required, final visual, and documentation completion.

For high-value critical parts, add a “hold point” before irreversible steps (e.g., before coating, before final assembly) so that nonconformances are captured when rework is still possible.

Step 4: Plan for special processes and accreditation needs. If your workflow includes NADCAP-controlled processes (e.g., heat treating, NDT, chemical processing, coatings), acceptance criteria should reference the approved process and the objective evidence required (furnace charts, test results, operator certifications, process certs). Even when NADCAP is not contractually required, aligning to those controls reduces risk and supplier variation.

Sampling

Sampling is where engineering intent, risk posture, and procurement realities meet. For critical parts, sampling decisions should be defensible, repeatable, and tied to CTQs—not a generic “inspect 10%” rule. Over-sampling raises cost and cycle time; under-sampling increases escape risk and program exposure.

Step 1: Separate 100% checks from sampled checks. A practical split:

Typically 100% for:
• Safety-critical dimensions and interfaces (fits, sealing surfaces, bearing seats)
• Serialization, marking, and traceability fields
• Visual criteria where defects are obvious and inspection is quick (burrs, sharp edges, damage)

Typically sampled for:
• Non-critical dimensions with proven capability
• Some material properties when verified by lot-based testing (hardness, chemistry via cert)
• Certain NDE where lot-based sampling is acceptable and contractually permitted

Step 2: Tie sampling level to CTQ risk and process maturity. Consider three maturity states:

• New part / new supplier / new AM parameter set: treat as high-risk. Use enhanced inspection (often near-100% CTQ verification) until capability is established.
• Stable production with proven capability: reduce sampling where statistical evidence supports it.
• Any change event: after a change in powder lot strategy, machine, recoater, build parameters, HIP cycle, heat treat furnace, CNC program revision, or post-processing supplier, temporarily increase sampling and re-validate key CTQs.

Step 3: Use lot definitions that match the manufacturing reality. In additive manufacturing, “lot” can be ambiguous. Define it explicitly in acceptance language and inspection planning, for example:

• A build lot: all parts built in the same machine run under the same parameter set and powder condition controls.
• A post-processing lot: all parts HIP’d in the same cycle or heat treated in the same furnace run.
• A machining lot: all parts machined under the same setup and program revision (often less meaningful than the build/HT lot for properties but meaningful for dimensional capability).

Ensure witness coupons (for mechanical testing or density) are tied to the correct lot definition. Misaligned lot definitions are a common cause of disputes during audits and source inspections.

Step 4: Choose an acceptance rule that matches how you will disposition nonconformances. Sampling is not only “how many to inspect,” but what happens if you find a failure. Your plan should specify:

• Containment (stop-ship, quarantine lot, notify customer)
• Additional inspection (expand sampling or 100% screen)
• Root cause and corrective action (8D, SCAR response timeframe)
• Rework/repair rules (what can be re-machined, blended, welded, or reprocessed—and what cannot)

For critical programs, make sure this aligns with your AS9100 nonconformance control and corrective action process, and with any customer-specific flowdowns.

NDE requirements

Non-Destructive Evaluation (NDE) is where critical-part acceptance criteria often become either extremely robust—or unworkable. The goal is not “NDE everything,” but to select methods that can actually detect the defect mechanisms relevant to your part, material, and process route.

Step 1: Match NDE method to defect type and geometry. Common pairings include:

• Dye penetrant (PT): effective for surface-breaking cracks in non-porous materials; useful after machining of high-stress features. Not suitable for internal defects and can be limited on rough as-built PBF surfaces unless properly prepared.

• Magnetic particle (MT): effective for surface/subsurface defects in ferromagnetic alloys (many steels). Not applicable to titanium and many nickel alloys.

• Ultrasonic testing (UT): effective for internal defects in many wrought products; can be challenging for complex shapes, thin walls, and rough AM surfaces without proper couplant and scan planning.

• Radiography (RT): good for certain internal defects; may struggle with complex AM geometries, varying section thickness, and interpretation variability.

• CT scanning: uniquely suited to complex internal channels and intricate AM geometries. Also useful for porosity characterization, but acceptance criteria must define defect sizing, reporting thresholds, and what constitutes a rejectable indication.

Step 2: Be specific about what “acceptable” means for indications. A frequent problem is specifying NDE without defining acceptance thresholds. Your criteria should address, as applicable:

• Indication type (pore, lack of fusion, crack-like, inclusion)
• Size limits (maximum equivalent diameter, length, or projected area)
• Density/spacing limits (e.g., “no more than X indications per unit volume” for CT-based porosity criteria)
• Location sensitivity (tighter near high-stress regions, threads, sealing surfaces, or pressure boundaries; more permissive in low-stress bulk regions if justified)

For AM parts, also consider whether HIP is required to close internal porosity. If HIP is part of the route, define whether NDE is required pre-HIP (to baseline process) and/or post-HIP (for final acceptance). Post-HIP CT can confirm residual defect populations, but be aware that HIP may alter defect morphology, and resolution limits can affect detectability.

Step 3: Control NDE execution as a process, not an inspection event. For critical hardware, include requirements for:

• Procedure qualification (written NDE procedure aligned to the part family and material)
• Technician qualification (certifications per your required scheme and customer flowdowns)
• Calibration (reference standards, sensitivity checks, CT system calibration and artifact controls)
• Data retention (CT volumes, radiographic images, UT scans, reports retained per contract)

If you require NADCAP-accredited NDE, state it clearly in the contract language and in the acceptance criteria section of the PO/RFQ to avoid later disputes.

Step 4: Tie NDE to machining and final condition. For parts that are finish-machined after HIP or heat treat, define when NDE occurs relative to machining. A common best practice is:

• NDE after critical machining steps that could expose subsurface discontinuities or introduce cracks (e.g., aggressive hog-outs, sharp internal corners).
• Final NDE after all machining that affects high-stress regions and before coatings that may mask indications.

Documentation

Documentation is often treated as “paperwork,” but for regulated manufacturing it is part of the product. Acceptance criteria should explicitly require the documentation needed to prove conformance, ensure traceability, and support audits or field investigations.

Step 1: Define the minimum acceptance data package (cert pack). For critical metal parts, a typical documentation set includes:

• Certificate of Conformance (CoC): stating compliance to the drawing, specifications, revision levels, and PO requirements.
• Material certifications: mill test reports with chemistry and mechanical properties; for AM, powder certifications and powder handling records as required.
• Traceability records: heat/lot traceability, serialization mapping, traveler/router history.
• Special process certifications: HIP cycle record, heat treat charts, coating certs, passivation/anodize certs, shot peen certs, etc.
• Inspection reports: dimensional reports (CMM outputs with datum references), NDE reports with acceptance results, surface finish reports, hardness records.
• First Article Inspection (FAI): for aerospace-style programs, an AS9102-format FAI is commonly expected for first production, after major changes, or per contract.

Step 2: Control configuration and revision alignment. Many nonconformances are documentation-driven: wrong drawing revision, superseded spec, missing flowdown, or mismatched acceptance criteria between PO and drawing. Implement a simple rule in your acceptance criteria: no release without revision verification. This includes:

• Drawing revision and ECO status
• Spec revision and customer addenda
• Approved process plan/router revision
• CNC program revision and verification record

Step 3: Ensure traceability for AM + HIP and PM-HIP workflows. For additive and PM-HIP (powder metallurgy + HIP) parts, traceability must cover more than “heat number.” Practical requirements to include in acceptance criteria:

• Powder batch ID and any reuse/reclaim rules (blend ratios, max reuse cycles, sieve records).
• Build ID (machine, parameter set, build file revision, operator, date/time).
• Post-processing link (stress relief cycle, HIP cycle ID, heat treat lot, machining job number).
• Part-to-build mapping (which serial number came from which location/build plate, especially if properties vary spatially).

This level of traceability is frequently necessary to support root cause investigations and to comply with customer or government quality expectations.

Step 4: Address regulated data handling (ITAR/DFARS) in documentation. If technical data is export-controlled (ITAR) or subject to DFARS cybersecurity requirements, acceptance criteria and procurement language should specify:

• Controlled unclassified information (CUI) handling rules
• Where data may be stored and who may access it
• Required markings on travelers and cert packs
• Flowdown to sub-tier processors (HIP, NDE, coatings)

These requirements are not “nice to have”—they determine whether a supplier is eligible to touch the program at all.

Contract language

Even the best-engineered acceptance criteria fails if it is not contractually clear. Procurement teams need language that converts engineering intent into enforceable requirements without ambiguity. Engineers need language that avoids unintended constraints that drive cost or lead time without improving quality.

Step 1: Put acceptance criteria in the RFQ/PO package, not only in email. Critical requirements should be in controlled documents: PO notes, quality clauses, and referenced specifications. If acceptance criteria changes during negotiation, ensure the final PO reflects it. Verbal agreements are not auditable and are difficult to enforce during disputes.

Step 2: Use a structured “Conformance Requirements” section. A practical template for contract language includes:

• Product definition: drawing number, revision, applicable specs, approved deviations.
• Manufacturing route constraints: e.g., “PBF (DMLS/SLM) using qualified parameter set X,” “HIP required per defined cycle window,” “finish machine on 5-axis CNC to final datums.” Only constrain what matters to quality or qualification status.
• CTQ list: either referenced via drawing key characteristics or provided as an attachment. For complex programs, include a CTQ matrix with verification method and frequency.
• Inspection and NDE requirements: method, timing, acceptance thresholds, and reporting requirements. Specify if NDE must be performed by a NADCAP-accredited facility and whether raw data must be delivered (e.g., CT volumes).
• Sampling and lot definition: explicit lot boundaries (build lot, HIP lot), witness coupon requirements, and escalation rules if nonconformance is detected.
• Documentation package: required certs, FAI requirements, traceability expectations, retention period, and delivery format.

Step 3: Define nonconformance and rework rules before parts are built. For critical parts, contract language should answer:

• What deviations require customer approval? (repair welding, blend-out limits, re-heat treat, re-HIP, alternative NDE method)
• What constitutes scrap? (cracks in CTQ regions, missing traceability, unapproved parameter changes)
• Who pays for rework and reinspections? (especially when acceptance criteria is ambiguous)

This reduces adversarial dispositions and keeps schedule risk under control.

Step 4: Require process change notification and configuration control. Many quality escapes happen after “minor” changes. For AM and HIP-driven routes, require suppliers to notify and obtain approval for changes such as:

• Machine change (new PBF machine or major rebuild)
• Parameter set change (laser power, scan strategy, layer thickness)
• Powder handling change (reuse rules, new powder vendor, sieve practice)
• HIP/heat treat cycle change (temperature, pressure, soak time, cooling rate)
• Sub-tier supplier change (NDE lab, coating house, machining subcontractor)

Pair this with a defined re-qualification trigger—often an updated FAI, additional mechanical testing, and/or expanded NDE for the first lot after change.

Step 5: Align acceptance criteria with supplier qualification and audits. Acceptance criteria works best when suppliers are qualified to meet it. In your sourcing process, verify:

• Quality system alignment (AS9100 or equivalent; controlled nonconformance and corrective action)
• Special process controls (HIP and heat treat process control, NADCAP status where required)
• Metrology capability (CMM capacity, CT scanning access, measurement uncertainty knowledge)
• Traceability discipline (material control, serialization, traveler integrity)

When suppliers understand your acceptance criteria and can demonstrate capability, you reduce the need for overly conservative inspection burdens and shorten the path from prototype to stable production.

Putting it together: a practical checklist
Before releasing an RFQ or approving a first production lot, confirm you can answer these questions clearly and consistently:

• What are the CTQs, and why do they matter?
• How will each CTQ be verified (method, timing, frequency)?
• What is the sampling plan and lot definition, especially for AM + HIP?
• What NDE is required, with explicit acceptance thresholds and reporting?
• What documentation must ship with the parts to prove conformance and traceability?
• What contract language prevents ambiguity around rework, change control, and sub-tier processing?

Clear acceptance criteria protects both the buyer and the supplier: it reduces disputes, improves first-pass yield, and ensures critical metal parts are released based on objective evidence—not assumptions.