Inspection and Metrology for Aerospace Parts
A part is accepted when every characteristic on the drawing has been measured with a method whose uncertainty is small relative to the tolerance, interpreted per the drawing standard (usually ASME Y14.5), and recorded. For new or changed parts, that record takes the form of an AS9102 first article inspection report.
Start with the drawing standard
Every geometric tolerance means what the referenced standard says it means. The current ASME edition is Y14.5-2018, reaffirmed in 2024 and placed on stabilized maintenance, and it replaced the 2009 edition. Many legacy drawings still reference Y14.5M-1994 or 2009, and some concepts changed between editions, so the inspector must read the title block and interpret to that edition, not to whatever is on the CMM software default.
Common interpretation problems that show up at first article:
- Position tolerances with material condition modifiers where the inspector reports only the RFS value and rejects good parts
- Profile of a surface applied to complex surfaces without a clear datum reference frame, leaving the alignment ambiguous
- Datum features that are not simulated the way the standard defines, for example a datum bore aligned with a least squares fit instead of the largest inscribed cylinder
- Unspecified edge breaks and surface texture notes that nobody measures
Resolve these before the first article, not after. A short call between the design engineer and the source inspector about datum setup saves weeks.
Choosing the measurement method
The right instrument is the cheapest one whose uncertainty is small compared to the tolerance. Calibration practice has long used a test uncertainty ratio of at least 4:1. ANSI/NCSL Z540.3 frames the real requirement as a probability of false accept no greater than 2 percent, and allows a 4:1 ratio when that probability cannot practically be estimated. Metrology sources caution that 4:1 alone is a crude control and can carry much higher risk in worst cases, so for tight tolerances use a calculated risk or a guard band.
| Characteristic | Typical methods | Watch for |
|---|---|---|
| Diameters and lengths, loose tolerance | Calipers, micrometers, height gauges | Operator technique, temperature of large parts |
| Bores, tight tolerance | Bore gauges, air gauges, CMM | Form error (taper, lobing) hidden by two point checks |
| Position, profile, complex datum frames | CMM, articulating arm, structured light | Alignment strategy, probe qualification, point density |
| Threads | Go and no-go gauges, thread wires, optical comparator | Gauge calibration, plating allowance |
| Surface texture | Contact profilometer | Filter and cutoff settings |
| Large structures and tooling | Laser tracker, photogrammetry | Thermal growth, line of sight |
CMM capability: what the spec sheet means
CMM accuracy is stated as a maximum permissible error for length measurement, tested per ISO 10360-2. The error usually takes the form A + L/K, where L is the measured length and A and K come from the manufacturer. One manufacturer example reads E0,MPE = 2.5 + 3L/1000 µm. The acceptance test measures five lengths, three times each, in seven positions through the volume, 105 measurements in all. The 2009 revision added an optional offset probe test at a default length of 150 mm.
That number is a best case under controlled temperature with a qualified probe. Actual task uncertainty on your part is larger and depends on the feature, the alignment, the number of points, the stylus, and the temperature of the part and machine. For tight work ask the source three questions: when was the CMM last verified to ISO 10360, what is the room temperature control, and how is the alignment built for my datum reference frame.
Machine tool positioning is a different test. ISO 230-2 covers accuracy and repeatability of positioning of individual NC axes and is not a measure of part accuracy. See the 5-axis machining page for machine capability topics.
Decision rules and guard bands
When a measured value sits close to a tolerance limit, measurement uncertainty decides whether you can claim conformance. ASME B89.7.3.1 provides terminology and the content a decision rule must address. Most shops apply simple acceptance, meaning inside the limit passes, which is reasonable when the uncertainty ratio is large. For tight tolerances, a guard banded rule shrinks the acceptance zone by an amount tied to the expanded uncertainty. One common formula from instrument makers following ILAC G8 is AL = √(TL² - U²), where TL is the tolerance limit and U the expanded uncertainty.
Put the decision rule in the quality clauses of the PO if it matters. Otherwise expect simple acceptance.
AS9102 first article inspection
AS9102 Revision C, released in 2023, defines the first article inspection report (FAIR). It requires three forms:
| Form | Content |
|---|---|
| Form 1, Part Number Accountability | Part identification, revision, sub-assemblies, reason for FAI, signatures |
| Form 2, Product Accountability | Materials, special processes and functional tests required by the design, with the certificates that prove them |
| Form 3, Characteristic Accountability | Every design characteristic, numbered to match a ballooned drawing, with actual results for variable characteristics |
Rev C renamed several Form 1 fields, removed the Form 2 signature, added a field for embedded software, and requires that the person reviewing the FAIR is not the same person approving it. Most rejected FAIRs fail because the three forms disagree with each other, for example a material cert on Form 2 for a different heat than the parts were made from.
Commonly cited triggers for a new or partial FAI include first production, a design change, a change of material source, process, tooling or location, and a gap in production of two years or more. Form 2 is where special process certs land, so the heat treatment, surface finishing and NDT records need to be complete before the FAIR can close.
Ongoing production and sampling
After first article, inspection is usually sampled unless the drawing flags key characteristics or the customer requires 100 percent inspection. Key characteristics are typically controlled with statistical process control, and the supplier should show capability data before reducing inspection. If your drawing does not identify key characteristics, consider adding them for the features that drive fit or function, because it changes how the supplier plans inspection and what data you receive.
Calibration records matter as much as the measurement. Instruments used for acceptance must be calibrated, traceable and within interval. A cert of conformance from a supplier is a statement that this system was followed, so auditing it occasionally is reasonable. For what to check at your own dock, see receiving inspection.
What to ask for in an inspection report
A useful dimensional report lets someone else repeat the measurement. Ask for the drawing revision and Y14.5 edition interpreted, the instrument and its calibration due date, the alignment and datum simulation used on the CMM, the number of points or scan density on critical features, the part temperature or room temperature at the time of measurement, and actual values rather than pass or fail for every variable characteristic. For CMM programs reused across lots, ask that the program revision be recorded too, so a later change in results can be traced to the part or to the program.
Get inspection or a FAIR done right
If you need CMM inspection, a full AS9102 first article package, or independent inspection of parts already made, send the drawing and the inspection scope. We source the work from qualified inspection providers, flow down the drawing standard, decision rule and report format, check the report against the drawing before release, and a person replies within one business day. Mutual NDA before files. If the drawing is export controlled, see controlled programs first. Back to the processes hub.
Questions
Which edition of ASME Y14.5 should I interpret to?
The edition on the drawing. ASME Y14.5-2018, reaffirmed in 2024, is current and replaced the 2009 edition, but many drawings still reference 2009 or Y14.5M-1994, and some rules changed between editions. If the drawing does not state an edition, ask the design authority. Interpreting a legacy drawing to the newest edition can turn a conforming part into a reject or the reverse.
Is a 4:1 test uncertainty ratio enough?
Often, but not always. ANSI/NCSL Z540.3 sets the real target as a false accept probability of no more than 2 percent and accepts a 4:1 ratio only when that probability cannot practically be estimated. For tolerances close to the capability of the measurement, metrology sources recommend calculating the risk or applying a guard band rather than relying on the ratio alone.
What does a CMM accuracy spec like 2.5 + 3L/1000 µm mean?
It is the maximum permissible length measurement error tested per ISO 10360-2, where L is the measured length in millimeters. At short lengths the error limit is about 2.5 µm and it grows with length. It describes the machine under controlled conditions with a qualified probe. The uncertainty of measuring your specific feature will be larger, depending on alignment, point count, stylus and temperature.
What changed in AS9102 Revision C?
Rev C, released in 2023, renamed several Form 1 fields, removed the Form 2 signature, added a field to capture embedded software, and requires that the person who reviews the first article report is not the person who approves it. The three form structure stays the same: part number accountability, product accountability for materials and special processes, and characteristic accountability for every drawing characteristic.
When do I need a new first article?
Commonly cited triggers are the first production run, a design change affecting the part, a change in material source, manufacturing process, tooling or location, and a production gap of two years or more. A partial FAI covering only the affected characteristics is often acceptable for a change. Your customer purchase terms may add triggers, so check them.
Related
Sources
- ASME: Y14.5 Dimensioning and Tolerancing
- ASME: B89.7.3.1 guideline for decision rules
- Micro Precision: ANSI/NCSL Z540.3 calibration requirements
- Morehouse: why a 4 to 1 TUR is not enough
- Tektronix: decision rule guide
- Mitutoyo: CMM accuracy and ISO 10360-2
- Mitutoyo: testing CMMs handout
- ISO 230-2:2014 positioning accuracy of NC axes
- Ideagen: AS9102 Rev C template and guide
- InspectionXpert: AS9102 first article inspection
- GNDCTL: how to fill out AS9102 forms
Ready to
source it?
A person replies within 1 business day. NDA first.
Request a quote