NDT welding inspection is the examination of a weld for flaws without damaging it, using five main methods: visual (VT), liquid penetrant (PT), magnetic particle (MT), ultrasonic (UT) and radiographic (RT) testing. The governing code sets which method applies; AWS D1.1, for example, requires visual inspection of all welds and has UT procedures for groove welds 5/16 in to 8 in thick.
No single method finds every flaw. This guide covers which methods to combine, under which code, performed by whom.
What does NDT mean in welding?
In welding, NDT (non-destructive testing, also called NDE or non-destructive examination) means checking a finished or in-process weld for discontinuities without cutting, bending or breaking it. The weld stays in service, so every production joint can be tested, not just a sample.
Destructive testing is still part of welding, but in a different place. Bend tests, tensile tests and macro-etches are used to qualify a welding procedure (the PQR behind a WPS) and to qualify welders. NDT is then used on production welds to confirm the qualified procedure was followed and produced a sound joint.
| Destructive testing | Non-destructive testing (NDT) | |
|---|---|---|
| Typical use | Procedure and welder qualification | Production welds, repairs, in-service checks |
| Sample | Coupon or test plate, destroyed | The actual weld, left intact |
| What it proves | Strength, ductility, fusion of the qualified procedure | Absence of rejectable discontinuities in this weld |
| Coverage | One coupon represents the procedure | Can cover 100% of welds, or a specified sample |
It also helps to separate two words that are often confused. A discontinuity is any interruption in the weld, such as a pore or a small slag line. A defect is a discontinuity that exceeds the code's acceptance criteria and must be repaired. NDT finds discontinuities; the code decides which are defects. The main flaw types are covered in EMA Industries' guide to weld defect types.
Key takeaway: NDT in welding checks the real production weld; destructive tests qualify the procedure that made it.
Which NDT welding inspection method finds which defect?
Each NDT welding inspection method has a reach (surface or volumetric), a material range and a flaw shape it detects best. VT, PT and MT are surface methods. UT and RT are volumetric. The table below compares the five main methods, with the ultrasonic variants listed separately.
| Method | What it finds best | Reach | Materials | Main limits |
|---|---|---|---|---|
| VT (visual) | Surface cracks, undercut, overlap, surface porosity, profile, size, craters | Surface only | All | Needs access, light and a clean surface; finds nothing below the surface |
| PT (liquid penetrant) | Fine surface-breaking cracks, pinholes, laps | Surface-breaking only | Any non-porous material, including stainless steel and aluminum | Flaw must be open to the surface; rough or contaminated welds give false indications; dwell time slows it down |
| MT (magnetic particle) | Surface and near-surface cracks, lack of fusion at the toe or root | Surface and shallow subsurface | Ferromagnetic only (carbon and low-alloy steel) | Not usable on austenitic stainless or aluminum; flaws parallel to the field are missed, so two field directions are needed |
| UT (conventional pulse-echo) | Planar flaws: lack of fusion, lack of penetration, cracks; through-thickness location and length | Volumetric | Most metals; best in fine-grained steel | Couplant and scanning access needed; coarse-grained austenitic welds scatter sound; results depend heavily on operator skill |
| PAUT (phased array) | Same flaws as UT, imaged from multiple angles at once, with a recorded scan | Volumetric | As UT | Needs a validated scan plan and setup; more equipment and data review time |
| TOFD (time-of-flight diffraction) | Through-wall height of planar flaws, from diffracted tip signals | Volumetric | As UT, mainly thicker butt welds | Dead zones near the scan surface and the back wall, so it is normally paired with pulse-echo or PAUT |
| RT (radiography, film or digital) | Volumetric flaws: porosity, slag inclusions, incomplete penetration, burn-through | Volumetric | All, limited by thickness and source energy | Planar flaws not aligned with the beam (such as lack of sidewall fusion) can be missed; radiation exclusion zone; usually needs access to both sides |
In short, planar flaws are best found by UT and rounded flaws by RT. That is why EMA Industries' articles on lack of fusion in welds and porosity in welding point to different primary methods.
Eddy current testing (ET) and alternating current field measurement (ACFM) also detect surface cracks in welds, often through coatings. They are used mostly for in-service crack checks, not as the main production method on new structural welds.
Key takeaway: Pair a surface method with a volumetric method, and pick the volumetric method by the flaw shape you most need to find.
How does each method work on a weld?
Each weld NDT method uses a different physical effect, and that effect explains its limits.
Visual testing (VT)
Visual testing is the first and most widely required weld examination. The inspector checks weld size and profile with fillet gauges and weld gauges, looks for cracks, undercut, overlap, arc strikes and surface porosity, and checks that the joint matches the drawing. VT is done before, during and after welding, not only at the end. More detail is in EMA Industries' guide to visual testing.
Liquid penetrant testing (PT)
PT draws a colored or fluorescent dye into surface-breaking openings by capillary action. After a dwell time, the excess is removed and a developer pulls the trapped dye back out to show the flaw. PT works on any non-porous material, which makes it the usual surface method for stainless steel and aluminum welds, but it cannot see anything that is not open to the surface.
Magnetic particle testing (MT)
MT magnetizes the weld, typically with a yoke, and applies fine iron particles. A crack at or just below the surface creates a leakage field that collects the particles. The weld is tested in two directions roughly 90° apart, because a flaw lying parallel to the field gives little or no indication. See EMA Industries' page on magnetic particle examination.
Ultrasonic testing (UT, PAUT, TOFD)
Conventional weld UT uses an angle-beam (shear-wave) probe, commonly 45°, 60° or 70°, that sends sound through the plate and into the weld from the side. Reflectors return echoes whose timing gives their position. PAUT steers and focuses the beam electronically and records the scan; TOFD uses a separate transmitter and receiver and measures the arrival time of signals diffracted from flaw tips. ASME-specific technique is covered in EMA Industries' article on ASME weld ultrasonic inspection, and the imaging method in phased array ultrasonic testing.
Radiographic testing (RT)
RT passes X-rays or gamma rays through the weld onto film or a digital detector. Thinner or less dense areas, such as pores and slag, show as darker spots. RT gives a permanent image, but it needs a radiation exclusion zone, and its sensitivity to a crack depends on how the crack lines up with the beam.
Key takeaway: The physics sets the limit: PT needs an opening, MT needs magnetism, UT needs a reflecting surface, and RT needs a density change along the beam.
Which code governs NDT on your weld?
The governing code depends on what is being built, and it is named in the contract documents. In North America, five code families cover most weld NDT. A key distinction runs through all of them: some documents tell you how to examine, others tell you what to accept.
| Work | Governing code | How it handles NDT |
|---|---|---|
| Structural steel buildings and general steel structures | AWS D1.1/D1.1M, Structural Welding Code – Steel (2025 edition is current) | Contains its own inspection clause: visual inspection of all welds, UT, RT, MT and PT procedures, and acceptance criteria. UT acceptance differs for statically and cyclically loaded connections |
| Highway and railroad bridges | AASHTO/AWS D1.5M/D1.5, Bridge Welding Code (2025 edition) | Applies to bridges designed to AASHTO or AREMA requirements; stricter than D1.1 in several areas, with a separate clause for fracture-critical members |
| Pressure vessels and boilers | ASME BPVC Section V (examination methods) with Section VIII Division 1 (acceptance) | Section V says how to perform RT, UT, PT, MT and VT; Section VIII sets which welds are examined and the acceptance standards |
| Process piping | ASME B31.3, with Section V methods | Sets examination extent (for example, random RT or UT of at least 5% of circumferential butt welds in Normal Fluid Service) and its own acceptance criteria table |
| Oil and gas pipelines | API Std 1104, Welding of Pipelines and Related Facilities | Contains its own NDT procedures and acceptance standards (Section 9); US federal pipeline rules cite it directly |
The B31.3 5% figure comes from paragraph 341.4.1(b)(1), as quoted in a 2021 PHMSA interpretation letter (PI-21-0009). The code also requires that the random sample include the work of each welder. Check the edition your project uses, because fluid service category changes the extent.
Examination standards vs acceptance criteria
ASME Section V is an examination standard. It does not, on its own, tell you whether a weld passes. A UT indication found with a Section V Article 4 technique on a B31.3 line is judged against B31.3; the same technique on a Section VIII vessel is judged against Section VIII. EMA Industries' explainer on ultrasonic testing acceptance criteria standards covers this split in more detail.
AWS D1.1 and API 1104 are different: each contains both its own procedures and its own acceptance criteria. For D1.1 visual and UT acceptance, including the indication rating d = a − b − c and the static vs cyclic qualifier, see AWS D1.1 weld inspection criteria.
Where API 1104 sits in US law
For gas transmission and distribution pipelines, the federal rule at 49 CFR 192.241 (see also the eCFR text) ties weld acceptability to Section 9 or Appendix A of API Std 1104. Its first paragraph also sets a qualification bar for the visual inspector:
"Visual inspection of welding must be conducted by an individual qualified by appropriate training and experience." (49 CFR 192.241(a))
The same rule states that Appendix A of API Std 1104 may not be used to accept cracks. Hazardous liquid pipelines have a parallel provision in 49 CFR 195.228.
Standards the original version of this article mislabeled
Two documents often appear in generic NDT articles, including an earlier version of this one. Neither is a governing code for US structural work.
| Document | What it actually is | What it is not |
|---|---|---|
| API RP 577, Welding Processes, Inspection, and Metallurgy (3rd edition, 2020) | A recommended practice giving API authorized inspectors guidance on welding inspection, welding processes, welder qualification and welding metallurgy for refinery and chemical plant equipment, piping and pipelines | Not a "recommended practice for NDT" and not a source of acceptance criteria. It supports API 510, API 570 and API 653 inspectors |
| ISO 23277:2015, Non-destructive testing of welds – Penetrant testing – Acceptance levels | Acceptance levels for surface-breaking indications in metallic welds found by PT, used with the ISO 3452 test methods and ISO 5817 quality levels | Not a PT procedure. It is the ISO route, and is rarely the governing document on a US D1.1 or ASME job |
Key takeaway: Find the governing code in the contract first; then use its examination method and its acceptance criteria, not a mix from different codes.
Who is qualified to perform and accept weld NDT?
Two different people are usually involved in NDT welding inspection: an NDT technician, qualified by method, and a welding inspector, often an AWS Certified Welding Inspector (CWI). Their qualifications come from different systems.
NDT technicians: SNT-TC-1A and employer certification
In the US, most weld NDT technicians are qualified under ASNT's Recommended Practice No. SNT-TC-1A. The current edition is 2024, with an addendum effective May 2025. SNT-TC-1A is a recommended practice, not a law: the employer writes a written practice, trains and examines the technician, and certifies them at Level I, II or III in each method. ASNT states this plainly:
"SNT-TC-1A-based certificates are employer-based certification programs for which ASNT does not hold the responsibility for certification." (ASNT, Explaining Employer-Based Certification Programs)
So a technician's certification belongs to the employer and normally lapses when they change jobs. ANSI/ASNT CP-189 is a stricter consensus standard used the same way, and ASNT's own ACCP and ISO 9712 schemes are central (third-party) certifications. More on the levels is in EMA Industries' guide to SNT-TC-1A requirements.
| Level | Typical role on weld NDT |
|---|---|
| Level I | Performs specific calibrations and tests under written instructions and supervision |
| Level II | Sets up and calibrates equipment, performs the test, and interprets and evaluates results against the code |
| Level III | Writes and approves procedures and techniques, administers the written practice, and examines and certifies Level I and II personnel |
AWS D1.1 has long required non-visual NDT personnel to be qualified to SNT-TC-1A or CP-189, with UT technicians also passing D1.1-specific practical examinations. Confirm the exact wording in the edition named in your contract.
Welding inspectors: the CWI role
The welding inspector checks the whole welding operation: materials, WPS compliance, welder qualifications, fit-up, preheat, visual inspection and final acceptance against the code. AWS certifies CWIs under AWS QC1, with qualification requirements in AWS B5.1:2025. The CWI examination has three parts (A, B and C), covering fundamentals, practical skills and an open-book code exam.
A CWI credential does not by itself qualify someone to perform UT or RT. The CWI accepts or rejects welds visually and reviews NDT reports; the method-qualified NDT technician produces those reports.
Key takeaway: SNT-TC-1A is employer certification by method; a CWI is an AWS third-party credential for welding inspection; most jobs need both.
Worked example: choosing methods for a groove weld
Choosing NDT for a weld means matching the code, the material, the joint and the flaws the process is prone to. Here is the reasoning for one common joint.
The weld: a complete joint penetration (CJP) groove weld splicing two 1 in (25 mm) carbon steel flange plates, single-bevel with backing, made by FCAW. The building is designed to AWS D1.1, the connection is statically loaded, and the contract documents call for UT of CJP groove welds.
Step 1: Visual inspection, 100%. D1.1 requires visual inspection of all welds. The CWI checks fit-up and root opening before welding, interpass cleaning during welding, and reinforcement, undercut, cracks and surface porosity after cooling.
Step 2: Surface method, if specified. The base metal is carbon steel, so MT is the natural choice; it is faster than PT and finds slightly subsurface cracks at the toes. PT would only be preferred on non-magnetic material.
Step 3: Volumetric method. At 1 in, the plate is inside D1.1's UT range of 5/16 in to 8 in. FCAW and single-bevel joints are prone to lack of sidewall fusion on the bevel face, which is a planar flaw. UT is the better choice than RT here, because RT can miss a fusion-face flaw lying at an angle to the beam.
Step 4: Check coverage with geometry. For a 70° probe on 1 in plate, the sound path to the bottom of the plate (first leg) is t / cos θ = 1 / 0.342 ≈ 2.92 in. The surface distance for one full skip is 2 × t × tan θ = 2 × 1 × 2.747 ≈ 5.49 in. The scanning surface therefore needs roughly 5.5 in of clean, spatter-free access beside the weld on each scanned side, which has to be agreed with the fabricator before painting or erection.
Step 5: Evaluate to the right table. Indications are rated with the D1.1 formula d = a − b − c and compared against the statically loaded acceptance table for 1 in thickness, not the cyclic one. The probe angle and scanning pattern come from the code's procedure tables for that thickness.
| Decision | Choice for this weld | Why |
|---|---|---|
| Surface examination | VT 100%, plus MT if specified | Carbon steel is ferromagnetic |
| Volumetric examination | Angle-beam UT | 1 in is within D1.1 UT range; main risk is planar lack of fusion |
| RT? | Not preferred | Planar fusion flaws can be missed; exclusion zone on an active site |
| Acceptance table | Statically loaded | Connection is statically loaded per design |
The same joint in a different job: if the joint were a 304 stainless steel butt weld in B31.3 process piping, the answers would change. MT would be ruled out because austenitic stainless is non-magnetic, so PT becomes the surface method. Random RT or UT would apply to at least 5% of circumferential butt welds in Normal Fluid Service, and the coarse-grained austenitic weld metal would make conventional UT harder, often favoring RT or a qualified PAUT technique.
Key takeaway: Code, material and the process's typical flaw decide the methods; geometry decides whether the chosen method can physically reach the weld.
What can't NDT find?
Every NDT method has blind spots, and an engineer relying on a report needs to know them. Clean NDT results mean no rejectable indication was found by that method, under that procedure, over the area actually scanned.
- Surface methods see only the surface. VT, PT and MT will not find buried slag, internal porosity or root lack of fusion in a CJP weld.
- Orientation matters. MT misses flaws parallel to its field; RT can miss planar flaws not aligned with the beam; UT can under-call a crack angled away from the probe.
- Tight cracks. Very tight, closed cracks can hold little penetrant and reflect weakly, especially in compressed regions.
- Dead zones. TOFD has blind areas near the scan surface and the back wall; contact UT has a near-surface dead zone under the probe.
- Material. Coarse-grained austenitic and dissimilar-metal welds scatter and bend ultrasound; MT is impossible on non-ferromagnetic metals.
- Access and geometry. Fillet welds, T-joints, nozzle welds and welds next to stiffeners may not allow full UT coverage or a usable RT shot.
- Sampling. Random or partial examination (such as 5% random RT) tests a sample. It controls quality statistically; it does not prove every weld is sound.
- Hydrogen cracking delay. Hydrogen-induced cracks can form hours after welding, so some codes and specifications set a delay before final NDT on high-strength steels. Check the governing code for the required hold time.
NDT also does not measure toughness, hardness or strength; those come from procedure qualification and specified mechanical tests.
How EMA Industries applies this
EMA Industries performs AWS D1.1 structural weld UT, MT and PT, and provides NDT Level III services and WPS/PQR support. The team is based in Southern California and mobilizes nationwide. On a weld inspection job, the scope starts with the governing code and acceptance table, then the method, coverage and access are agreed before welding is complete. See EMA Industries' ultrasonic testing services or contact EMA Industries to scope a weld inspection.
Frequently asked questions
What is the most common NDT method for welds?
Visual testing is the most common, because codes such as AWS D1.1 require visual inspection of all welds. For volumetric checks on steel structures, angle-beam ultrasonic testing is now more common than radiography, because it finds planar flaws such as lack of fusion and needs no radiation exclusion zone. Pressure piping still uses radiography widely.
Is ultrasonic testing better than radiography for welds?
It depends on the flaw. UT is generally better at finding planar flaws such as cracks and lack of fusion, and it can measure their depth. RT is generally better at showing volumetric flaws such as porosity and slag, and it leaves a permanent image. Many specifications allow either one, and critical welds sometimes get both.
Do MT and PT find the same defects?
Both find surface-breaking cracks, but they are not interchangeable. MT also detects flaws slightly below the surface and works only on ferromagnetic steel. PT detects only flaws open to the surface but works on any non-porous material, including stainless steel and aluminum.
Who certifies an NDT technician under SNT-TC-1A?
The employer does. Under SNT-TC-1A, the employer's written practice defines training, experience and examinations, and a Level III administers it. ASNT does not issue SNT-TC-1A certificates; its central certifications are ACCP and ASNT Level III.
Does a Certified Welding Inspector perform ultrasonic testing?
Not on the strength of the CWI credential alone. A CWI is certified by AWS for welding inspection, which includes visual inspection and review of NDT results. Performing UT, RT, MT or PT requires separate qualification in that method, usually under the employer's SNT-TC-1A or CP-189 program.
Sources
- American Welding Society, AWS Announces Release of D1.1/D1.1M:2025, Structural Welding Code – Steel, April 2025
- American Welding Society, New Undercut Tolerances and Treatments in the 2025 D1.5, Bridge Welding Code, 2025
- American Welding Society, Certified Welding Inspector (CWI), accessed October 2026 (lists QC1:2016-AMD1 and B5.1:2025)
- ASNT, Explaining Employer-Based Certification Programs
- ASNT, Recommended Practice No. SNT-TC-1A (2024), with Addendum
- 49 CFR 192.241, Inspection and test of welds: Cornell LII and eCFR
- 49 CFR 195.228, Welds and welding inspection: standards of acceptability: Cornell LII
- PHMSA, Interpretation PI-21-0009, October 2021 (quotes ASME B31.3 para. 341.4.1(b)(1))
- ASME BPV Committee on Nondestructive Examination (BPV V), committee scope
- API RP 577, Welding Processes, Inspection, and Metallurgy, 3rd edition, October 2020 (publisher listing via SAI Global)
- ISO 23277:2015, Non-destructive testing of welds – Penetrant testing – Acceptance levels (listing via Standards Council of Canada)

