The weak point of a steel structure is not the material itself but the connections. Rolled steel sections from the mill have fairly consistent properties that can be traced back through mill certificates. Welds, on the other hand, are made on site and depend on the welder's skill, the weather and the welding position, which differ from joint to joint. Weld quality is therefore not uniform, and welds need to be inspected.

Depth of weld flaws versus how deep each method — VT, PT, MT, UT and RT — can see The left image is a cross-section of a single-V groove butt weld between two base-metal plates, divided into three depth zones by dashed lines. Zone one is the weld surface, with a surface-breaking crack and undercut. Zone two is shallow subsurface, with a small subsurface discontinuity. Zone three is the weld interior, with porosity, incomplete fusion along the groove face and slag inclusion. The right image is a bar chart showing how deep each method can see, using the same depth scale as the cross-section. VT (visual testing) and PT (penetrant testing) see only surface-breaking flaws. MT (magnetic particle testing) reaches shallow subsurface flaws and only works on ferromagnetic steel. UT (ultrasonic testing) and RT (radiographic testing) examine the full thickness. The bottom strip summarizes: always start with visual inspection; what the instruments find are discontinuities, which become defects only when they exceed the acceptance criteria; and choose the method by the depth of the flaw you need to find. How deep is the flaw? → Pick the right method Not to scale Single-V groove butt weld Depth each method sees Base metal Base metal Surface crack Undercut Subsurface Lack of fusion Porosity Slag incl. VT PT MT UT RT Surface Shallow Interior Weld VT visual · PT penetrant · MT mag. particle UT ultrasonic · RT radiography MT: ferromagnetic steel only Dashed line = depth zone limit 1 Always VT first Low cost, spots bad profile 2 Found ≠ always repair Check spec acceptance 3 Choose by depth Not by site convenience
Weld flaws can occur at the surface, just below it, or deep inside the weld. Each inspection method sees to a different depth, so the method must match what you need to know.

What Is Non-Destructive Weld Testing?

Non-destructive weld testing (Weld NDT) is the inspection of welds for discontinuities without cutting or damaging the part, after which the results are compared against the acceptance criteria stated in the project specifications.

Two terms need to be kept clearly apart: discontinuity and defect.

  • A discontinuity is what the instrument detects.
  • A defect is a discontinuity whose size or location exceeds the acceptance criteria and must be repaired.

So a weld that triggers an indication on the instrument is not necessarily unfit for service. The result must first be evaluated by a qualified inspector.

There are 6 discontinuities commonly found in structural steel welds: cracks, incomplete fusion, incomplete joint penetration, porosity, slag inclusion and undercut. Each lies at a different depth, so each calls for a different inspection method.

Why Inspect Structural Steel Welds?

Welds must be inspected because they transfer forces between members, and they are the part of a steel structure whose quality depends most on the welder's skill and site conditions. There are 3 main reasons.

  • Internal flaws cannot be seen with the naked eye — a weld with a smooth, neat surface may still hide porosity or incomplete fusion inside.
  • Cracks can grow under cyclic loading — in structures subjected to reversing loads, such as crane runways, bridges or machine bases, small cracks may grow with each load cycle.
  • It is a contract requirement — most structural steel contracts specify the scope of weld inspection, and the inspection results serve as evidence for handover of the work.
Weld quality control in three stages — before, during and after welding — with NDT in the final stage The timeline at the top divides welding into three stages, with the first arc strike separating 'before welding' from 'during welding'. A green band under the timeline covers the before and during stages, the period in which defects can be prevented. An orange band covers the after-welding stage, the period in which flaws are detected and must be repaired or the results confirmed. Panel one, before welding: review the welding procedure specification (WPS), verify welder qualifications, check joint preparation, and check electrodes and their baking. Its illustration shows a WPS document, a beveled joint with root opening, and an electrode oven. Panel two, during welding: control preheat temperature, control current and travel speed, clean between passes, and check the welding sequence. Its illustration shows an electrode over a multi-pass groove and a thermometer. Panel three, after welding: visual inspection first, surface inspection with PT or MT, internal inspection with UT or RT, and issue the inspection report. Its illustration shows a magnifying glass and a UT probe on the weld line. The box at the bottom summarizes that welds with well-controlled procedures from the start run into far fewer problems at the NDT stage. NDT is the last gate, not the only one Not to scale First arc Defects are prevented here Detect → repair / confirm 1 Before Review WPS Check welder quals. Check joint prep. Check electrodes & baking 2 During Control preheat temp. Control current & speed Interpass cleaning Check weld sequence 3 After Visual (VT) first Surface: PT or MT Internal: UT or RT Issue NDT report WPS Bake Bevel + root gap Multi-pass + preheat UT VT first, then instruments Welds with well-controlled procedures from the start have far fewer NDT problems
NDT can only check the end result. Real quality control starts before the first arc strike.

What Weld Testing Methods Are There?

There are 5 main methods, ordered from shallowest to deepest reach: VT and PT inspect the surface, MT inspects the surface and shallow subsurface, and UT and RT inspect the interior of the weld.

1. Visual Testing (VT)

VT is the visual inspection of welds by a qualified inspector, aided by measuring tools such as weld gauges and magnifiers. It must always come first, because it immediately reveals undercut, undersized weld legs, overlap and irregular weld profiles at the lowest cost. Its limitation is that it only sees what is on the surface.

2. Penetrant Testing (PT) — PT Testing of Welds

PT involves applying a penetrant to the cleaned weld surface and letting it seep into surface-breaking flaws. The surface is then wiped and a developer is applied to draw the penetrant back out as a visible colored indication. Reference guidelines are ASTM E165 and ASTM E1417.

PT works on almost any material with a non-porous surface, making it the main option for non-magnetic materials where MT cannot be used, such as stainless steel and aluminum. Its limitation is that it only detects flaws that are genuinely open to the surface; flaws sealed by grinding or covered by paint will not be detected.

How PT testing of welds is done — 5 main steps

  1. Pre-cleaning — remove weld spatter, rust, oil and paint from the weld and adjacent area, then let it dry completely, because dirt plugging the mouth of a flaw stops the penetrant from getting in.
  2. Apply penetrant and wait for the dwell time — spray or brush a visible dye or fluorescent penetrant over the whole weld, then leave it for the time specified by the standard and the manufacturer, typically about 5–30 minutes depending on temperature and the type of flaw being sought.
  3. Excess penetrant removal — wipe the surface first with a dry cloth, then with a cloth slightly dampened with cleaner. Never spray cleaner directly onto the surface, as it will wash the penetrant out of the flaws too.
  4. Apply developer and wait for the development time — spray a thin, even coat of white developer. It draws the penetrant trapped in flaws back to the surface as red indications (or fluorescent indications under UV light). Development time is generally at least 10 minutes.
  5. Inspect, evaluate and record — inspect under adequate lighting (or UV light for fluorescent penetrant), separating relevant indications such as cracks and surface-breaking lack of fusion from false indications. Evaluate against the acceptance criteria of AWS D1.1 or the project specification, photograph and record locations, then clean the surface after inspection (post-cleaning).

PT takes longer per location than MT because of the dwell and development times, but it needs no electricity or magnetizing equipment, which makes it convenient for inspecting welds at height or in confined spaces.

3. Magnetic Particle Testing (MT)

MT works by inducing a magnetic field in the part and dusting magnetic particles onto the surface. A flaw lying across the field creates flux leakage that pulls the particles into a visible line, while flaws parallel to the field barely show. Each location must therefore be tested at least twice, with the field direction rotated 90°. Reference guidelines are ASTM E709 and ASTM E1444.

MT only works on ferromagnetic materials such as carbon steel, which is already the main material in structural steelwork. Its advantages over PT are that it also reveals shallow subsurface flaws and is faster, since there is no dwell time.

4. Ultrasonic Testing (UT)

UT sends high-frequency sound waves through a probe into the steel and measures the signals reflected from boundaries or internal discontinuities. The reference guideline is ASTM E164.

UT's strengths are that results can be read immediately on site, it is sensitive to cracks and vertically oriented incomplete fusion, and it needs access from only one side. Its limitations are that results depend heavily on the operator's skill, the instrument must be calibrated against a reference block before use, and the surface must be smooth enough for good probe contact.

5. Radiographic Testing (RT)

RT passes radiation through the weld onto film or a digital imaging plate. Areas with porosity or inclusions show a different density from sound steel. Reference guidelines are ASTM E94 and ASTM E1032.

RT's strengths are that it produces image evidence that can be archived and re-examined, and it shows porosity and inclusions clearly. Its limitations are that it requires access to both sides of the part, a radiation control area must be cordoned off and nearby work stopped, so its safety and licensing requirements are much stricter than those of the other methods.

The two images below summarize these principles. The first is a 2D diagram running from the five PT steps to what MT, UT and RT can and cannot see. The second is a 3D model of a steel column-beam connection showing which method suits each type of weld.

2D diagram: the five PT steps for weld inspection and the principles of MT, UT and RT The top row is a weld cross-section with a surface-breaking crack, showing the five steps of penetrant testing (PT): clean the surface and let it dry; apply red penetrant and wait a dwell time of about 5 to 30 minutes; remove excess penetrant with a dry cloth and then a damp cloth, without spraying cleaner directly onto the surface; spray white developer and wait at least 10 minutes until the penetrant in the flaw bleeds back out as a red indication; and inspect, evaluate, record and clean the surface. The bottom row has three panels. MT, plan view: a magnetic yoke straddles the weld, and the magnetic field lines cross a crack running along the weld, so the particles gather into a line, while flaws parallel to the field lines are not seen, so two perpendicular directions must be tested. UT, butt-weld cross-section: an angle-beam probe sits on the plate beside the weld, the sound wave reflects off the back wall and strikes incomplete fusion at the groove face, and the reflected signal appears as a peak on the screen. RT: the radiation source is above, the film is below the part, porosity shows as dark spots on the film, and a crack whose plane is not parallel to the beam may not appear. How each method works (2D) Not to scale PT penetrant · 5 steps (surface-breaking) ASTM E165 / E1417 1 Clean Remove spatter, rust, oil; dry 2 Penetrant Dwell ~5–30 min 3 Wipe excess Dry → damp cloth, no spray 4 Developer Thin coat, wait ≥ 10 min 5 Inspect · record Real vs false indic. MT mag. particle (plan) Surf. + shallow · ASTM E709 / E1444 N S ✓ Flaw ⊥ field → particles line up ✕ Flaw ∥ field → not visible Test 2 perpendicular dirs. Ferromagnetic material only ? UT ultrasonic (section) Interior · ASTM E164 45°–70° Lack of fusion Echo Sensitive to planar flaws One-side access · instant read Calibrate on ref. block RT radiography (section) Interior · ASTM E94 / E1032 Source Film/plate: pore = dark spot ✓ Clear pores & inclusions ✕ May miss off-axis planar flaws 2-side access · radiation zone
2D diagram — Top row: the five PT steps — clean, apply penetrant and wait the dwell time, wipe off the excess without spraying cleaner directly, spray developer, then inspect and record. Bottom row: MT shows flaws lying across the field lines, so two directions must be tested; UT is sensitive to planar flaws and needs only one-side access; RT shows porosity clearly but may miss planar flaws not parallel to the beam. (Image: SPN Soil Engineering)
3D model: weld inspection points on a steel column-beam connection, and a three-question method selection chart A 3D view of an H-section steel column on a base plate, with an H-section steel beam connected to the column flange. The beam's top and bottom flanges are joined to the column flange by complete joint penetration butt welds. An inspector places a UT probe on the beam top flange beside the weld to examine the interior. The beam web is attached with fillet welds. At the column base, fillet welds run around the column, and an MT magnetic yoke is used to check for surface and shallow subsurface cracks. Every weld receives visual inspection as the baseline. The right-hand panel is a three-question method selection chart: how deep is the flaw of concern, is the material ferromagnetic, and what site access is available — with a note that the inspection scope and acceptance criteria must be stated in the contract documents, and only Level II or higher inspectors may evaluate results. 3D model Not to scale Beam flange: CJP butt weld Full tension → internal UT/RT UT probe on beam flange Sound into weld from one side Web: fillet weld Surface: VT + MT Column base: fillet all round MT yoke across weld VT: visually inspect every weld first Leg gauge, magnifier A-scan Pick a method: 3 questions 1 How deep is the flaw? Surface / shallow → VT + MT or PT Internal (butt weld) → UT or RT 2 Is it magnetic? Carbon steel → MT (faster) Stainless / aluminum → PT instead 3 What site access? One side / can't stop work → UT Two sides + radiation zone → RT (with UT) Scope % & acceptance Must be in contract (e.g. AWS D1.1) Level II+ evaluates Per ASNT SNT-TC-1A
3D model — Beam flanges with complete joint penetration butt welds carry full tension and are inspected internally with UT or RT, while the fillet welds at the beam web and column base are surface-inspected with VT plus MT. The right panel summarizes the three questions used to choose a method; the actual scope and acceptance criteria must come from the contract documents. (Image: SPN Soil Engineering)

Comparing the 5 Methods: How Deep Can Each See?

VT and PT inspect the surface only, MT reaches shallow subsurface flaws, and UT and RT reach the interior of the weld. The table below summarizes the capabilities and limitations of each method to help in defining the inspection scope.

← Scroll to see the full table →

MethodDetection depthAdvantagesLimitationsReference guidelines
VT Visual testingWeld surface onlyFast, low cost, applicable to every weldCannot see anything below the surfaceAWS D1.1
PT Penetrant testingSurface-breaking flawsWorks on non-magnetic materials; simple equipmentSurface must be very clean; sealed flaws go undetectedASTM E165 / E1417
MT Magnetic particle testingSurface and shallow subsurfaceFaster than PT; reveals shallow subsurface flawsFerromagnetic materials only; field must lie across the flawASTM E709 / E1444
UT Ultrasonic testingWeld interiorInstant results; one-side access; sensitive to cracksHighly operator-dependent; instrument calibration requiredASTM E164
RT Radiographic testingWeld interiorPermanent image record; clearly shows porosity and inclusionsRequires two-side access and a radiation exclusion zone; high costASTM E94 / E1032

Although UT and RT both reach the weld interior, they are not fully interchangeable. RT excels at volumetric flaws such as porosity and slag inclusions but may miss planar flaws such as cracks or incomplete fusion if the beam is not parallel to the plane of the flaw. UT is more sensitive to planar flaws, so work requiring high confidence often uses the two methods together.

How to Choose the Right Inspection Method

Choose by asking 3 questions in order: how deep is the flaw of concern, is the material ferromagnetic, and what site access is available. On real projects, the contract documents and referenced standards such as AWS D1.1 usually already specify the inspection method for each type of weld; these three questions help explain the reasoning behind those requirements.

  1. How deep is the flaw of concern? For undercut or surface cracks, VT combined with MT or PT is sufficient. But for complete joint penetration butt welds in tension, UT or RT is required, because the critical flaws lie inside the weld.
  2. Is the material ferromagnetic? Carbon steel can be tested with MT, which is usually faster, while stainless steel or aluminum must use PT instead.
  3. What site access is available? If only one side is accessible, or nearby work cannot be stopped to set up a radiation exclusion zone, UT is a more practical option than RT.
💡 In practice, several methods are usually combined

Visually inspect every weld as the baseline, then select the critical welds specified in the contract for further testing with a method that sees deeper. Inspecting with one method and drawing conclusions for the entire job usually leaves questions unanswered. The same principle applies to inspecting other materials; for an example of method selection for concrete, see NDT guide for non-destructive testing of concrete structures

For welds in spliced piles, which have their own particular conditions — frequent splices and limited site access — read the separate article How to inspect welded joints in spun piles and micropiles: 5 defects to check

What Percentage of Welds Must Be Inspected?

There is no universal figure that applies to every job. The inspection scope must be specified in the contract documents and project specifications. AWS D1.1, the most frequently referenced structural steel welding standard, leaves it to the contract documents to state which welds are inspected, by which method and in what proportion.

The specification should cover 4 items in full:

  1. Types of welds to be inspected
  2. Method used for each weld type
  3. Proportion or number of locations to be inspected
  4. Acceptance criteria used for evaluation

If any one of these is missing, disputes easily arise at handover, because each party will interpret it differently.

In practice, welds carrying full-section tension and structures under cyclic loading are usually assigned a higher inspection proportion than ordinary welds, because the consequences of a defect are more severe.

⚠️ Do not set inspection proportions from figures passed down by word of mouth

Follow the current edition of the standard referenced in the contract, together with the judgment of the project's engineer of record, because criteria and proportions vary with the type of weld and the nature of the loading. A figure that works on one project may not suit another.

What Qualifications Must a Weld Inspector Have?

Inspectors must be certified under the scheme specified by the project, such as ASNT SNT-TC-1A, and those evaluating results must be Level II or higher. Beyond the individual, the inspection method must be backed by a written procedure and the equipment must be calibrated. If any one of these is missing, the report cannot be relied on.

  • Inspector qualifications — the widely used guideline is ASNT SNT-TC-1A, which defines Levels I, II and III. Only Level II or higher inspectors may evaluate results; Level I may only perform inspections under supervision.
  • Written inspection procedure — there must be an approved procedure document specifying the inspection parameters before work starts, rather than each person inspecting by habit.
  • Equipment calibration — UT instruments must be calibrated against a reference block at the specified intervals, with evidence attached to the report.
  • Traceable reports — state the location of each weld inspected, the method, equipment, inspector, date and the acceptance criteria applied.

In the broader context of building inspection, weld testing is an engineering test separate from statutory building inspection. The difference between the two is explained in 9 types of building inspectors vs. structural testing: what's the difference? For cases where the load capacity of a completed structure must be verified with real loads, see Bridge load testing: verifying the load capacity of special structures

Summary

Choosing a weld inspection method for structural steel starts with the question of how deep the flaw of concern lies, not which method is best. VT and PT answer for the surface, MT for the surface and shallow subsurface, and only UT and RT can answer for the weld interior. Material type and site access conditions then make the final decision.

Just as important, the inspection scope and acceptance criteria must be written clearly into the contract from the outset, and inspectors must be qualified under the scheme the project specifies. Otherwise the inspection results become documents that cannot settle anything when a dispute arises.

⚠️ This article provides general information

The inspection method, scope and acceptance criteria for any specific project must be based on the edition of the standard specified in the contract and reviewed by that project's engineer of record. The figures and guidance in this article should not be applied directly to actual work.