Radiographic testing (RT) is the most audit-visible form of nondestructive examination in structural steel fabrication. The radiograph is a permanent record — it goes into the project file, it is reviewed by the engineer of record and the owner's inspector, and it is the primary evidence that the weld meets the acceptance criteria of the governing code.
AWS D1.1:2025 Chapter 9 (Inspection) establishes the RT acceptance criteria for structural welds. Understanding those criteria is not optional for a CWI or QC manager who signs off on RT reports. This article covers what the code requires, how the criteria differ between static and cyclic load applications, and what an RT report must contain to survive a quality audit.
Why RT acceptance criteria matter beyond weld quality
Many fabricators treat RT as a binary pass/fail: did the lab call it good? If yes, file it. That approach works until an audit, an owner dispute, or a warranty claim forces someone to look at the actual radiographic report.
At that point, questions arise: Was the IQI correct for the base metal thickness? Was film density within the acceptable range? Did the interpreter look at the right weld length? Was the report signed by a qualified RT Level II or Level III?
A weld called "accept" by an RT lab that used incorrect technique, incorrect IQI, or outside-spec film density is not an accepted weld under AWS D1.1:2025 — regardless of what the report says. The CWI responsible for the project must understand enough RT procedure to know when to push back on a report.
Automatic rejection: cracks, incomplete fusion, incomplete joint penetration
AWS D1.1:2025 establishes a category of discontinuities that are unconditionally rejectable — no size exemption, no engineering fitness-for-purpose allowance at the code level:
Cracks: Any crack visible on the radiograph is a rejection. Cracks are typically identified as linear indications with irregular edges and often a branching pattern. A cold lap or hot tear in the weld deposit counts. A toe crack in the HAZ counts. The word "any" means what it says.
Incomplete fusion (IF): Lack of fusion between weld passes or between the weld and the base metal fusion face is rejectable. On a radiograph, IF typically shows as a thin, straight, dark line following the weld bead boundary or the root face. It can be confused with slag in some orientations.
Incomplete joint penetration (IJP): In CJP groove welds where full penetration is required, incomplete penetration is rejectable. In PJP groove welds where the detail is designed for a specific throat size, the radiograph must show that penetration to the effective throat was achieved.
For a thorough review of CJP vs PJP design differences and their qualification requirements, see CJP vs PJP groove weld WPS requirements.
Porosity: size and aggregate limits
Porosity is the most common discontinuity visible on structural weld radiographs. AWS D1.1:2025 limits porosity in two ways:
Individual pore size: A single spherical or elongated pore cannot exceed 3/8 in (10 mm) in its greatest dimension. This is the maximum regardless of weld size or base metal thickness.
Aggregate area: The total area of porosity within any 12-in (300-mm) length of weld is limited as a function of weld area. The code specifies the maximum aggregate porosity as a fraction of the projected weld area in that length. Thin base metal sections have tighter limits because the pore-to-weld cross-section ratio is higher at a given absolute pore size.
Elongated (piping) porosity and wormholes: These are treated more strictly than spherical porosity because their elongated geometry can act as stress risers. AWS D1.1:2025 places length limits on piping porosity independent of the aggregate area limits.
Cyclically loaded structures: For fatigue-sensitive applications, the porosity limits tighten. The code recognizes that small pores acceptable in a statically loaded column may be unacceptable in a transverse groove weld at a high-stress location in a crane girder or bridge girder.
Slag inclusions and elongated indications
Slag inclusions appear on the radiograph as irregular, dark, often elongated areas, typically with less defined edges than porosity. AWS D1.1:2025 limits slag inclusions by length and by separation distance between adjacent inclusions:
- Individual slag inclusion length is limited as a function of base metal thickness (typically 2/3 of the thinner base metal thickness, but not to exceed a maximum absolute length).
- Separation between adjacent inclusions must exceed a minimum distance (typically 6× the length of the longer inclusion) for the inclusions to be evaluated separately rather than as an aggregate indication.
- Aggregate slag in any 12-in (300-mm) length is also limited.
Parallel linear indications (sometimes called "tramline" slag or slag lines along weld bead boundaries) are treated as a single elongated indication for length-limit evaluation purposes.
Image quality requirements: IQI, film density, and geometry
The acceptance criteria for the weld apply only if the radiograph meets image quality requirements. AWS D1.1:2025 specifies:
IQI (Image Quality Indicator / penetrameter): Wire-type or plaque-type IQI must be placed on the source side of the joint (the side facing the X-ray source or gamma source). The essential wire or essential hole must be clearly visible on the radiograph. If the IQI is not visible or is placed incorrectly, the radiograph does not meet D1.1 technique requirements and cannot be used for acceptance/rejection decisions.
Film density: Acceptable film density under most RT procedures cited by D1.1 is in the range of 1.8 to 4.0 (optical density) for the area of interest. Underexposed films (density below 1.8) lack the contrast to detect fine linear indications. Overexposed films (density above 4.0) are too dark to read on a standard viewer.
Source-to-film distance and geometric unsharpness: These affect the sharpness of the indication edges. Geometric unsharpness must not exceed the limits of the applicable RT procedure; typically Ug ≤ 0.020 in (0.5 mm) for structural applications. An RT report that does not document source-to-film distance and source size cannot be verified for geometric unsharpness compliance.
What the RT report must include for D1.1 audit compliance
An RT report that survives a quality audit contains:
- Welder/weld identification: Joint ID or weld map reference, WPS number used, welder ID
- Technique data: Source type (X-ray kV or gamma source type and activity), source-to-film distance, film type and brand, IQI type and designation
- Film density readings: At the IQI and in the weld area, for each exposure
- Indication mapping: Location and size of each indication found, referenced to a weld coordinate system (typically distance from one end)
- Acceptance/rejection decision: Per the applicable code criteria (state the code and edition — AWS D1.1:2025), for each indication
- Interpreter qualification: Name, certification level (ASNT Level II or Level III per SNT-TC-1A or ACCP), and signature
Reports that are missing technique data cannot be verified for image quality compliance. Reports without interpreter certification referenced are insufficient for most owner quality requirements.
For an overview of how RT fits into the full NDE documentation package — including UT, MT, and PT — see NDE method selection for structural welds.
RT vs UT: when does D1.1 allow substitution?
AWS D1.1:2025 allows either RT or UT (ultrasonic testing) for volumetric examination of groove welds, subject to agreement with the engineer and owner. For thicker base metal (above approximately 3/8 in / 10 mm depending on the application), UT is often preferred because:
- UT is faster on thick sections where multiple RT exposures would be required
- UT can detect planar discontinuities (cracks, lack of fusion) that are oriented unfavorably for RT detection
- UT does not require radiation safety control areas
RT has one advantage UT cannot replicate: the radiograph is a spatial record of the weld interior that any qualified interpreter can read years later. UT is a real-time examination — the report documents what was found, but the actual scan is not retained as a visual record unless digital phased-array UT (PAUT) is used with full waveform storage.
For high-accountability work (bridges, hospitals, critical infrastructure), many owners require both RT and UT for full-penetration groove welds in primary members. Know your project specification before assuming either method alone is sufficient.
Connecting RT results to your WPS and audit packet
An RT rejection on a production weld is not just a quality event — it is a potential WPS event. If multiple RT rejections are concentrated in one welder's joints, or in joints welded at a particular ambient temperature or with a specific heat lot of filler metal, the failure pattern may indicate a WPS essential variable is being violated in production.
For example: a cluster of porosity rejections at the root pass on joints welded during a cold-weather week might indicate the preheat specified in the WPS was not being maintained — see cold weather welding requirements under AWS D1.1:2025. Or a string of slag inclusions in SAW joints might indicate flux depth or travel speed drifted outside the WPS range.
Tracking RT rejection patterns against WPS parameters, welder IDs, and production conditions is the QC manager's job. A software-based WPS management system with integrated audit-packet export makes it practical to correlate NDE reports with the WPS and PQR records without manually cross-referencing paper files.
Rule library based on AWS D1.1:2025; verify against your governing edition.