Visual Inspection of Welding: The Complete Inspector’s Guide

A fabricator calls you over to a finished column, points at a fillet weld, and asks the question that decides whether the piece ships: “Is that good enough?” No dye, no X-ray, no ultrasound — just your eyes, a flashlight, and a gauge. What you do in the next two minutes determines whether a defect gets caught in the shop or discovered later by someone far less happy to see it.

Visual Inspection of Welding: The Answer Up Front

A fabricator calls you over to a finished column, points at a fillet weld, and asks the question that decides whether the piece ships: “Is that good enough?” No dye, no X-ray, no ultrasound — just your eyes, a flashlight, and a gauge. What you do in the next two minutes decides whether a defect gets caught in the shop or discovered later by someone far less happy to see it.

The honest answer is that you cannot respond with an opinion. You respond with a process. Visual inspection of welding — abbreviated VT, the term the American Society for Nondestructive Testing uses for the method — is the systematic examination of a weld and the base metal around it using the eyes plus simple aids: light, magnification, gauges, mirrors, borescopes. The purpose never changes: judge whether the joint meets the applicable code, the drawing, and the workmanship requirements written into the contract. It is judgment backed by measurement, and it is recorded, not guessed.

Here is what surprises people new to quality control: this is not the low-tech option you settle for when the budget runs out. ASNT identifies visual testing as foundational to all nondestructive testing, because every other method still depends on someone’s eyes to interpret what a screen, a film, or a particle indication actually shows (ASNT, Visual Testing). It is the oldest examination method, the cheapest per inch of weld, and by a wide margin the most used. AWS’s own pocket handbook puts it plainly: visual inspection performed by properly trained inspectors discovers the vast majority of defects that would otherwise only be found later by far more expensive NDT methods (AWS Welding Handbook, Vol. 2). The dedicated tutorial document for the method is AWS B1.11M/B1.11, Guide for the Visual Examination of Welds, and VT is defined as a formal examination method in AWS B1.10M/B1.10.

Two things define it before we go further. First, VT is not a glance at the end. It happens in three phases: before welding, to verify joint preparation and fit-up; during welding, to catch problems while they can still be corrected; and after welding, for final acceptance. The first two phases are what save rework — final inspection can only reject what is already finished. Second, VT is documented. An undocumented verbal “looks fine to me” is not visual inspection; it is an assumption with no traceability when the question comes back six months later.

Who This Guide Is For

This guide is written for everyone who has to make or support an accept-or-reject call. Welders who want their work to pass the first time. Inspectors building or sharpening a workflow. Supervisors and engineers who need to know what their QC staff should be doing and documenting. Fabricators working alone or in small shops with no dedicated QC department. Even serious DIY fabricators who have realized that “it looks good” is not a standard.

What You Will Learn

We will start by defining exactly what visual inspection covers, from joint prep and root condition to spatter, arc strikes, and distortion, and draw the line between direct and remote (indirect) visual examination. You will get a practical rundown of the inspector’s kit — fillet weld gauges, undercut gauges, cam gauges, mirrors, and magnifiers that turn arguments into measurements — plus the lighting and vision requirements that quietly decide how many defects you actually see.

From there, the article builds a repeatable step-by-step workflow you can run on any joint, then moves into the defect catalogue: cracks, undercut, porosity, incomplete fusion, overlap, underfill, undersized fillets, mismatch, and how each one is judged. Because acceptance is never personal preference, we compare the governing rulebooks — AWS D1.1 for structural steel, AWS D1.6 for stainless, ASME BPVC Sections IX and V, API 1104 for pipeline, and ISO 5817 quality levels — and show why the same undercut can pass under one document and fail under another.

We will also be blunt about failure: the human and process mistakes that cause missed defects, from dirty welds and glare to trusted welders and end-of-shift rushing. We then compare VT honestly with PT, MT, UT, and RT, so you know when surface or volumetric NDT adds real value and when it is wasted money. Finally, we look at building a program that survives an audit — written procedures, qualified inspectors with documented eye exams, controlled gauges, hold points, weld maps, and repair feedback loops — plus a short list you can act on Monday morning.

Visual inspection of welding rewards people who treat it as a craft with rules rather than a formality. By the end of this guide, you will know the rules — and you will be able to answer that shop-floor question with evidence instead of a hunch.

What Visual Inspection of Welding Actually Covers

Most people hear “visual inspection” and picture someone glancing at a finished weld and nodding. That image is exactly why welds get sent back. Visual testing (VT, in NDT shorthand) is a systematic examination of the weld, the joint, and the surrounding base metal, measured against a written set of requirements. Its scope is far wider than the bead you can see, and it begins before anyone strikes an arc.

What Is Actually on the Inspection Table

Run through what a proper VT covers and the list is much longer than “the weld”:

• Joint preparation and fit-up: bevel angle, root opening, root face (land), alignment or mismatch between the two members, cleanliness of both faying surfaces, and the location, length, and quality of tack welds. Cleanliness discipline is a fit-up item as much as a welding one — the guide to Aluminum Cleaner for Welding: How to Prep Aluminum Right explains why oxide and oil have to come off before the arc, not after.
• Root condition: whether the root opening stayed consistent along the joint, whether backing is correctly fitted, visible burn-through or melt-through on open roots, and root reinforcement or suck-back where the back side is accessible.
• Bead sequence and interpass condition: pass order, whether each layer was cleaned before the next went down, and whether each bead wets properly into both toes. Bead placement is where most workmanship defects are born, as covered in Aluminum MIG Welding: The Complete Practical Guide.
• Final weld profile: contour, toe blend, undercut, overlap, convexity and concavity, reinforcement height, and cap width consistency.
• Weld size and length: fillet leg size, effective throat, specified length and continuity, and whether welds start and stop where the drawing says they should.
• Spatter and cleanup: heavy adhered spatter, slag, and flux residue that hide the surface and must be removed before final examination.
• Base-metal damage: arc strikes outside the weld, grinder gouges, tool marks, and damage to adjacent surfaces. A weld can meet every profile requirement and still be rejected because someone dragged an electrode across the plate two inches away.
• Distortion and dimensional conformance: angular change, shrinkage, bowing, and whether the assembly still holds its drawing dimensions. Distortion-prone assemblies, like the panel work described in Aluminum Boat Welding: The Complete Guide for Strong Hull Joints, often fail VT on geometry long before they fail on weld appearance.

Every one of those items is judged with the same basic kit: eyes, light, and a gauge. Fillet gauges, a square, a rule, and a written acceptance criteria do most of the work. That is why VT is simultaneously the cheapest test in the shop and the easiest one to do badly.

Direct vs. Indirect Visual Examination

ASNT distinguishes two categories of visual testing. Direct visual testing means the inspector’s eye has an unobstructed line of sight to the surface, aided if necessary by mirrors, lenses, or magnifiers (conventionally up to about 10×). In practice, direct VT is done close in: personnel standards commonly place the inspector within roughly 24 inches (600 mm) of the weld, at an angle of about 30 degrees or more to the surface, under adequate lighting. Indirect visual testing covers everything the eye cannot reach directly — pipe interiors, box sections, root passes behind structure — using borescopes, videoscopes, fiberscopes, mirrors at angles, or cameras. Knowing which one you are doing matters, because indirect examination usually requires its own written procedure, personnel training, and recorded image handling.

Discontinuity vs. Defect: The Two Words That Decide Everything

AWS A3.0M/A3.0 draws the line that underpins every accept/reject call you will ever make: a discontinuity is an interruption in the uniformity of the object, while a defect is a discontinuity that does not meet the required product specification. In other words, a discontinuity only becomes a defect when the governing document says it does.

This is also why saying “it’s just a discontinuity” is not automatically a defense, and why calling something a defect does not by itself dictate the repair. The code does both.

Three Phases: Before, During, and After Welding

AWS B1.11, Guide for Visual Inspection of Welds, organizes the fundamentals of visual examination into three phases, and the ordering is the whole lesson:

1. Prior to welding: verify base-metal and consumable identity, joint preparation, fit-up, root gap and bevel angle, cleanliness, tack welds, and preheat condition.
2. During welding (in-process): observe the root pass, bead placement, interpass cleaning, travel-speed and heat indications, and crater filling at every stop.
3. After welding: remove slag and spatter, examine the final profile, measure size, mark discontinuities, and record the disposition.

The reason those first two phases matter is blunt: final VT can only reject. Once the cap is on, a root gap that closed to zero, a contaminated joint, or an interpass layer that never got cleaned is hidden. In-process inspection turns rejections into corrections — you fix a fit-up gap in thirty seconds instead of cutting out a completed joint an hour later. Final inspection tells you what went wrong; prior and in-process inspection is what lets you stop it from going wrong at all.

One caution defines VT’s ceiling, and it belongs here rather than buried later: a weld that looks perfect on the surface can still hide lack of fusion or lack of penetration. A root may look fully fused from the face side while the back side is open; a cap may sit beautifully while the land between passes never melted. This is the structural limitation VT cannot overcome, and it is why codes pair visual examination with radiography or ultrasonic testing on full-joint-penetration welds in critical service. Visual inspection is the first filter, not the last word.

The Inspector’s Eyes, Light, and Tools

Every visual inspection comes down to three variables: what the inspector can actually see, the light falling on the weld, and the tools in their hand. Get any one of them wrong and you end up with a confident, documented, wrong answer. The encouraging part is that all three are cheap to fix, and one of them is free.

Vision Requirements: Qualify Your Eyes First

Before anyone judges a weld, they qualify their own eyes. Under AWS B5.1, Specification for the Qualification of Welding Inspectors, an inspector must read Jaeger Number 2 letters or smaller at a minimum of 12 in. (30.5 cm) in at least one eye, with or without correction, plus pass a color perception test and a contrast differentiation test that asks them to separate shades of gray.

Those two extra tests are not formalities. Heat tint bands, light surface porosity, and tight toe cracks register as subtle shifts in tone and gray value, not obvious shapes. An inspector with a color deficiency or weak contrast discrimination can look directly at a defect and genuinely see nothing.

The exact acuity number is not universal. Visual testing personnel qualification under ASNT SNT-TC-1A is governed by the employer’s written practice, which commonly specifies Jaeger J-1 or equivalent. So the number you must meet is the one your employer wrote down, not a national constant.

Welders are held to a different threshold, which is a useful reality check. New York City requires welder license applicants to pass near-vision acuity on Jaeger J2 at 12 in. or greater (NYC DOB licensing requirements). Seeing your own work at arm’s length is simply not the same skill as judging someone else’s weld against a code table.

Light: The Failure Point Nobody Blames

Poor lighting causes more missed defects than poor eyesight. You can pass your Jaeger chart and still miss a 1/32 in. undercut because the lamp was behind your shoulder.

The angle of light matters more than raw brightness. Light raking across the surface at a low angle throws undercut, inconsistent ripple, and undersized reinforcement into sharp relief; light aimed straight down your line of sight flattens all three into invisibility. Glare is the opposite problem: a freshly wire-brushed or ground surface bounces light straight into your eyes and camouflages the very profile you are trying to judge. Shield the lamp, kill the overhead fluorescents, or simply change your position instead of staring harder.

Color temperature distorts judgment too. Warm incandescent light hides blue-violet heat tint and makes clean stainless look dirty. A neutral or daylight-balanced LED, roughly 4000–5000 K, keeps tone differences honest. And where shadows fall — inside a groove root or the toe of a fillet — that is exactly where cracks hide. A penlight held from a second angle plus a small mirror opens those corners up.

The AWS visual examination guidance in AWS B1.11, Guide for Visual Inspection of Welds is built around these fundamentals: sufficient illumination, controlled viewing angle, clean surface, documented procedure.

Viewing Distance and Angle

ASNT guidance for visual testing recommends observing the area of interest from within about 24 in. (600 mm) and maintaining a viewing angle of no less than 30 degrees to the surface. Both numbers are easy to ignore and expensive to ignore. A shallow viewing angle foreshortens the weld profile; from five feet a seam looks like one continuous pretty bead, and from twelve inches you find the crater that was never filled. Distance also silently shrinks what you can resolve: pores smaller than about a millimetre simply stop existing at three feet.

The Physical Tool Kit

Opinions are free and worthless. A calibrated gauge is cheap, and it ends debates instantly. A $20 fillet gauge set resolves arguments that a hundred opinions cannot, because the throat and leg dimensions are either inside the drawing tolerance or they are not.

Tool	What it settles
Fillet weld gauge set (telescoping / Hi-Lo)	Leg length and effective throat on fillet welds
V-wedge and undercut gauge	Undercut depth in thousandths or fractions
Cam (AWS profile) gauge	Reinforcement height, profile angle, misalignment
Steel rule, tape, and bridge gauge	Weld length, continuity, size, dimensional conformance
Magnifier, 5x–10x	Surface porosity, hairline cracks, start-stop defects
Telescoping mirror and penlight	Root condition, back side, restricted corners
Borescope / videoscope	Indirect visual exam inside tubes, vessels, box sections
Calipers and pit gauge	Pit depth, remaining wall, groove dimensions
Weld map template, marker, camera	Locating, tagging, and proving what you found

Two practical notes on magnification: up to about 10x is plenty for acceptance work, and beyond that you start inspecting the metal’s grain rather than the weld’s workmanship. And on cleanliness — spatter, slag film, and grinding dust hide more defects than any code limit ever will, which is why surface prep comes before the first look. On aluminum work especially, proper aluminum cleaner for welding and deglossing before inspection changes what the eye can honestly report.

Documentation is part of the kit, not an afterthought: a tagged location on a weld map, a photo with a scale reference, a filled-out report form, and a gauge calibration record. Without those, an accurate observation becomes an unverifiable claim the moment it leaves the shop floor.

A Step-by-Step Visual Inspection Workflow

Knowing what a crack looks like is one thing; running the same examination the same way every time is another. The workflow below is the one experienced inspectors follow on structural, pipe, and pressure work. It runs from paperwork to report, and every step exists because skipping it has failed a weld somewhere.

Step 1: Read the Contract Documents and Applicable Code

Before you look at metal, establish which rulebook governs the job. Contract drawings and specifications outrank everything else, then the referenced code, then general workmanship expectations. On structural steel that usually means AWS D1.1, where Clause 8.9 addresses visual inspection and Table 8.1 lists the acceptance criteria you will actually apply. Note the edition printed on the drawings: the AWS D1.1:2025 edition reorganized clauses and refined inspection and acceptance criteria compared with the 2020 edition, so a habit of citing “Clause 8” from an old copy can put you on the wrong page.

Step 2: Confirm the WPS, Qualifications, and Materials

Verify that an approved welding procedure specification exists and that the joint you are about to inspect is covered by it. Confirm welder performance qualifications are current for the process, position, and material thickness. Watch the dates: under AWS D1.1, a welder’s performance qualification lapses if that welder has not used the process for roughly six months. Then check that base-metal heat numbers and filler classifications match the WPS. A selection reference such as our alloy steel welding electrodes guide helps here, because a sound weld made with the wrong consumable is still a nonconformance.

Step 3: Inspect Fit-Up and Joint Preparation

This is where most rework is prevented. Check root gap, bevel angle, land, alignment or mismatch, and cleanliness of the groove faces and the adjacent land. Look for oil, paint, mill scale, moisture, and rust. On aluminum, residual oxide and cutting fluids are the usual culprits, and a proper aluminum cleaning routine before assembly saves a lot of arguments later. Tack welds matter too: they should use the same process and filler, be sound, sit where they do not distort the joint or leave an arc strike in the finished weld, and be either incorporated or removed and dressed.

Step 4: Observe the Root Pass and Interpass Condition

If the joint is accessible, look at the root while it is being made and between passes. Excessive interpass temperature, heavy melt-through, slag trapped between beads, or a restart with cold lap cannot be judged fairly once the final cap covers them. This is the phase where inspection earns its keep: final VT can only reject what it finds, but in-process checks stop the defect from being buried.

Step 5: Prepare the Surface for Final VT

Visual testing is only as good as the surface it reads. Remove slag, flux residue, and spatter; wire-brush or grind only as the code permits, and never let paint or heavy grinding hide the weld toe. NRC inspection guidance notes that VT acceptance criteria prohibit cracks and lack of fusion while permitting only limited undercut, porosity, reinforcement variation, and undersized fillet welds — none of which you can judge accurately through a layer of slag.

Step 6: Scan Systematically, Then Measure

Work in one direction with a repeatable pattern: start at one end, scan the weld toe on both sides, then the crown, then the adjacent heat-affected zone. Overlap each pass by about half a viewing width so no inch is skipped. Rake the surface with light held at a low angle, change your viewing angle rather than trusting one glance, and get close enough for the magnification the procedure calls for.

Then measure instead of estimating:

• Fillet leg and throat with a weld gauge; overall length with a tape.
• Undercut depth with a V-wedge or undercut gauge.
• Reinforcement height and profile with a cam gauge.
• Porosity size and spacing with a pit gauge or magnifier.

A twenty-dollar gauge settles a dispute that a dozen opinions cannot, and it puts a number in the report.

Step 7: Mark, Record, and Disposition

Circle every discontinuity with a paint marker, number it, and sketch it on a weld map so the repair crew finds the same spot you did. Record location, type, size, and measured value against the governing acceptance table. Then make one of three calls: accept, reject, or repair, and document who made it, when, and under which code clause. Small shops can do this with a numbered form and a phone photo; the medium matters less than the fact that it exists. Unwritten verbal approvals are how traceability dies.

The final inspection report closes the loop: code edition, WPS and procedure references, welder identification, joint and weld map, instruments used, results, discontinuity records, disposition, and signatures. That packet proves the weld met requirements months after the paint went on, and it becomes the raw material for the metrics and nonconformance tracking discussed later in this guide.

Reading a Weld: Discontinuities and Acceptance Criteria

You have done the paperwork, confirmed the WPS, watched the fit-up, and cleaned the surface. Now comes the part every inspector actually enjoys: reading the weld itself. This is where vocabulary matters more than most people expect. A discontinuity is any interruption in the normal structure of the weld or base metal. A defect is a discontinuity that exceeds the acceptance limits of the governing code and must be repaired or the part rejected. The same porosity cluster can be a rejectable defect in one joint and an acceptable feature in the next, depending entirely on the specification in the contract. That is why the industry leans on AWS A3.0, Standard Welding Terms and Definitions, so that everyone on the job is naming the same thing the same way.

Cracks: The One Family With No Size Limit

Cracks come in several flavours, and you should be able to name each one on sight:

• Longitudinal cracks run along the weld axis, usually in the centreline or the face.
• Transverse cracks run across the bead, typically from restraint and shrinkage stress.
• Crater cracks sit in the depression left when the arc is pulled without filling the crater, often star-shaped.
• Toe cracks start at the weld toe where stress concentration is highest.
• Heat-affected zone cracks sit in the base metal just beside the weld, sometimes invisible without magnification or surface NDT.

Cause is usually some combination of hydrogen pickup, excessive restraint, a wrong travel speed or technique, or an unfilled crater. Disposition, however, is simple: cracks are prohibited outright under structural steel visual acceptance criteria, not limited to a size. The same is true of lack of fusion. Repair means grinding or gouging out the full length until sound metal is found, re-welding to an approved repair procedure, and re-inspecting — often with magnetic particle testing to prove the crack is gone.

The Core Catalogue at a Glance

AWS B1.11 and the AWS Welding Inspection Handbook treat surface porosity, undercut, incomplete fusion, incomplete joint penetration, overlap, and underfill as the core visual discontinuity categories. Add the dimensional and cosmetic issues below and you have the working list most inspections run on.

What you see	Likely cause	Usual disposition
Sharp linear indication (any direction)	Restraint, hydrogen, crater not filled	Reject, excavate, re-weld
Undercut notch at the weld toe	Too much current, wrong angle, fast travel	Accept if under code limit, else blend or repair
Scattered round pores or one deep blowhole	Contamination, damp consumable, wrong gas	Accept if within area limits; excavate and weld if not
Cold lap / bead sitting on unfused base metal	Low heat input, slow travel, arc too short	Reject (lack of fusion)
Fused face, unfused root visible from the back	Insufficient penetration, joint too tight	Repair if full penetration was required
Weld metal lapped over the toe	Low current, dragging technique	Blend out; reject if it hides fusion issues
Over-built bead, poor profile	Over-welding, slow weave	Accept or grind to profile per code
Weld below the plate surface at the toe	Too little filler, wrong joint geometry	Reject (underfill)
Leg length short of the drawing	Wrong settings, rush, welder judgement	Reject; add weld or reclassify size
Dark glassy pockets in a multi-pass bead	Incomplete cleaning between passes	Excavate and re-weld if surface-breaking
Pitted, torn spots outside the weld	Arc strikes during fit-up or adjustment	Grind smooth; reject if cracking is suspected
Plates not flush, joint bowed or twisted	Uneven bead sequence, skipped back-stepping	Re-straighten or re-fit; often no weld repair

Reading Each One on the Job

Undercut is the classic measured defect: a sharp groove melted into the base metal at the toe. AWS D1.1 visual acceptance criteria commonly allow about 1/32 in. (0.8 mm) of undercut in statically loaded connections, with noticeably tighter values for cyclically loaded structures — the current edition cited in the contract is the only one that counts. Small surface porosity is judged by size, spacing, and accumulated area; piping porosity is different, a deep channel with a crater at the surface that often signals continuing gas escape and is far more likely to be excavated. Incomplete fusion and incomplete joint penetration rarely announce themselves from the face, so you are looking for the tell-tale cold-lap edge, an unfused bevel face, or a root that never closed. Overlap and excessive convexity are technique problems showing up as profile problems, and most codes let you accept or grind them — but fat toes create stress raisers on loaded joints, so check the drawing before you wave it through. Underfill and undersized fillets are measured, not judged: a fillet weld gauge settles leg length and throat in seconds, and undersized length along the run counts just as much as undersized size. Excessive spatter is cosmetic in many specifications but must come off if it will interfere with coating or inspection; arc strikes are never cosmetic in hardenable steel because they can crack later. Cleanliness drives a lot of this — contaminated aluminum, for instance, shows porosity and oxide-driven lack of fusion almost immediately, which is why surface prep is covered so carefully in Aluminum MIG Welding: The Complete Practical Guide.

Finally, remember that a visual reject is rarely the end of the story. Expect a marked-up weld map, an excavate-and-repair instruction, possibly a root-cause note back to the WPS, and a full re-inspection of the completed repair. The visual acceptance tables in AWS D1.1 exist to make that loop predictable, not optional.

Codes and Standards: AWS, ASME, API, and ISO Compared

Nothing in visual inspection starts more arguments than one short question: “According to what?” A weld is never good or bad in the abstract. Every accept or reject call traces back to a written document, and the inspector’s first move is knowing which document has authority on this job, on this joint, today. Get that right and most code disputes dissolve before they start.

The Contract Always Comes First

Project documents sit at the top of the hierarchy: the drawing, the general specification, the purchase order, the owner’s requirements. Those documents usually name a code, and the code fills in everything the contract leaves unsaid. If a drawing calls for undercut not exceeding 0.5 mm on a cyclically loaded connection, that number governs even if the structural code would allow more. If the contract says nothing about weld profile, the invoked code’s workmanship clause becomes your acceptance criterion. This is why experienced inspectors read the paperwork before they look at the steel.

Keep the roles straight as well. OSHA’s welding rules are about fire prevention, ventilation, and personal protection rather than weld quality, so “OSHA says” rarely settles a bead-profile argument. One document that does shape the vocabulary is AWS A3.0, Standard Welding Terms and Definitions, the shared language behind every acceptance table in this business, including the discontinuity-versus-defect line you draw on each call.

Structural Steel: AWS D1.1 and the D1 Family

For structural steel in North America, everything runs through AWS D1.1/D1.1M. It is routinely referenced in project contracts, and it defines welder and procedure qualification, fabrication requirements, inspection methods, and acceptance criteria in one document. The AWS code page outlines its scope, and the 2025 edition preview shows how Clause 6 handles inspection, repair, and the acceptance tables for cracks, undercut, weld size, and profile. The wider D1 series applies parallel logic with metallurgy-specific rules to other materials: D1.6 for stainless steel, D1.2 for aluminum structures, where process variables from aluminum MIG welding interact directly with the acceptance table.

Qualification outside the D1 documents is shaped by AWS B2.1, Standard for Welding Procedure and Performance Qualification. A welder qualified for one process, position, and material grouping under one document is not automatically qualified under another. Consumable identity is part of that chain, since the WPS names a specific classification, as our guide to alloy steel welding electrodes shows through the strength and toughness letters on every stick rod.

Pressure Work: Section IX Qualifies, Section V Examines

ASME splits the job across two documents. BPVC Section IX handles qualification, covering WPSs, PQRs, and performance qualifications with its own material groups and variable rules that don’t map one-to-one onto AWS. The examination method lives in ASME BPVC Section V, Article 9, the visual examination article of the Boiler and Pressure Vessel Code. Article 9 is explicit where structural documents tend to be general: a minimum illumination of 500 lux (50 foot-candles), a maximum viewing distance of roughly 600 mm (24 in.) for unaided-eye work, magnification up to 6x, and defined terms for direct versus remote visual examination. Pressure work also commonly triggers mandatory volumetric examination alongside VT, which is exactly where VT’s role as a first filter rather than a stand-in for RT or UT becomes operational. For a plain overview of where visual testing sits among the NDT methods, the ASNT visual testing overview is a clean primer.

Pipelines: API 1104

API 1104 covers welding of pipelines and related facilities, and its acceptance criteria are written for field production: circumferential girth welds examined visually and radiographically, with porosity, undercut, and reinforcement limits tuned to weld geometry and pipe position rather than comfortable shop access. Repair provisions matter here too, including gouging technique and repair depth limits, territory our air arc welding and carbon arc gouging guide covers in practical detail.

International Work: ISO 5817 Levels B, C, and D

For European and international fabrication, ISO 5817 is the workmanship reference, and its structure is refreshingly simple: it defines weld quality levels B, C, and D, with B the most stringent and D the most permissive. The contract picks a level, and that choice becomes the acceptance table for cracks, porosity, undercut, and profile across the whole job. ISO 17637 describes the visual testing method itself, and ISO 9712 handles personnel certification in a scheme familiar to anyone who knows the ASNT ladder. Classification-society work, meaning offshore, ship, and marine fabrication, layers society rules on top of all of it, as our look at aluminum boat welding illustrates for hull joints.

One Undercut, Four Verdicts

This is the trap that catches people. Undercut at a weld toe is a stress riser in every rulebook, but the number that turns it into a rejection moves around. Common structural limits sit near 1/32 in. (0.8 mm), with tighter values for cyclically loaded structures; ISO level B judges the same geometry far more harshly than level D; a pressure code weighs undercut against profile and reinforcement; a pipeline spec consults its own table. Same weld, same flashlight, opposite answer. So “is this undercut acceptable?” is never a question for the group chat. Measure it with a gauge, then read the governing table.

Typical project	Governing document(s)	What drives the VT call
Building and bridge steel	AWS D1.1/D1.1M, contract-invoked	Clause 6 acceptance criteria: cracks, undercut, weld size, profile
Stainless process equipment	AWS D1.6 plus project spec	Profile, heat tint, cleanliness for corrosion service
Pressure vessels and piping	ASME BPVC Sec. IX and Sec. V Art. 9	Qualification records; Article 9 lighting, distance, and magnification rules
Pipeline girth welds	API 1104	Porosity, undercut, and reinforcement limits tied to radiography
Offshore and marine structures	Classification society rules plus AWS D1.1	Society-specific workmanship criteria and documented VT sign-off
European and export fabrication	ISO 5817 levels B, C, D	The chosen quality level as the acceptance table
Aerospace	AWS D17.1 and AMS specs	Near-zero crack tolerance and dense traceability records
Small custom fabrication	Contract plus OSHA 1910.252 for safety	Often unstated, so define workmanship criteria before the first weld

Who Is Qualified to Make the Call

Personnel rules differ as much as the acceptance tables. In the AWS world, the Certified Welding Inspector credential runs alongside AWS B5.1, Specification for the Qualification of Welding Inspectors, which sets the training and experience pathway for inspection roles. Regulated industries add layers: NRC material notes that acceptable qualification bases for visual inspection of AWS welds include a current or previous AWS Certified Welding Inspector certification, or an engineer or technician competent by training or experience. The NDT parallel is ASNT SNT-TC-1A and ANSI/ASNT CP-189, where VT Levels I through III are certified under an employer’s written practice, meaning the employer, not a blanket certificate, defines exactly what you are authorized to examine and report. If your shop has no written practice, that is the first document to draft.

The ladder keeps going up from there. Aerospace fabrication runs on AWS D17.1 and material-specific AMS specifications with tighter visual criteria and denser records than structural work ever demands; our aerospace welding guide maps that standard stack for anyone aiming at that industry. Whatever level you work at, keep a one-page summary in your inspection kit: which document governs this job, which clause carries the acceptance criteria, and who signs off when the drawing goes silent. That single page ends more arguments than any argument ever will.

Why Inspectors Miss Defects: Common Mistakes and Fixes

Here is the uncomfortable starting point: ASNT notes that visual testing depends largely on the vision and perception of the technician. In every other NDT method the instrument has a calibration record. In VT the instrument is a person, and people get tired, biased, rushed, and poorly lit. Most failed welds that reach the field were not hidden. They were visible, in plain sight, on a surface someone looked at the wrong way, at the wrong time, under the wrong light.

The good news is that nearly every miss traces back to a short list of predictable failure modes, and each one has a concrete fix.

Mistake	Why it fails	The fix	Supervisory control
Inspecting dirty, slag-covered, or painted welds	Oxide, spatter, and coatings mask profile and surface-breaking indications	Clean first; surface preparation to remove material that obscures defects is a recognized step in VT, and it costs real time — budget it	Make cleaning a documented hold point before VT sign-off
Glare, shadowing, or dim light	Specular highlights wash out undercut; flat light flattens profile	Rake the light across the surface at a low angle, then re-check from the other side	Provide handheld lights at each bay and verify lux at the workstation
Standing too far back or viewing at one shallow angle	Viewing angle changes the apparent size and location of a discontinuity	Examine from at least two or three angles and distances; ASNT recommends multiple viewing angles rather than a single glance	Require a stated scan pattern in the written procedure
Fatigue on long seams	Detection rates fall sharply as attention drains on repetitive bead	Inspect in defined segments with a break every length; alternate tasks	Cap continuous VT time and rotate inspectors
Rushing at end of shift, confirmation bias	You find what you expect, and closing time is the enemy of looking	Work backward from the end of the joint; keep a defect tally per weld	Discourage same-hour completion and sign-off pressure
Trusting a known-good welder	Reputation substitutes for evidence	Gauge it anyway; a $20 fillet gauge settles what a hundred opinions cannot	Sample every welder’s work regardless of history
Measuring weld size by eye	Estimation drifts, especially with convex beads and lap joints	Measure with a fillet gauge, V-wedge, or cam gauge and record the number	Issue calibrated gauges and track their control
Skipping the root and the back side	Incomplete penetration and poor root fusion rarely announce themselves from the face	Inspect both sides wherever access exists; use a mirror or videoscope	Put back-side examination on the traveler explicitly
Ignoring arc strikes away from the weld	Stray strikes are hard spots and potential cracks in the base metal	Scan the surrounding plate, not just the bead, and grind them out	Add adjacent-base-metal condition to the checklist
Verbal approvals	Nothing is traceable, and the argument is unwinnable later	Mark, date, and sign the map or report before leaving the joint	Require documented disposition for every hold point
Ticking VT after paint	Coating hides profile, toe geometry, and fine indications	VT before coating, always, per the applicable sequence in the contract documents	Sequence the traveler so coating cannot precede inspection

This is also why AWS B1.11 leans so hard on procedure rather than intuition: a written method, qualified people, and records turn a subjective glance into a defensible result. Housekeeping and work practices around the job also fall under rules like OSHA 1910.252, which is a useful reminder that inspection quality starts with an orderly workstation.

How Welders Can Make Their Own Welds Easier to Inspect

Inspection works both ways. A weld that is easy to examine usually passes, and the habits below take seconds:

• Clean as you go. Wire-brush each pass, chip slag completely, and wipe the toe line. Good prep habits start with fundamentals like choosing the right process and cleaning approach, and Aluminum Cleaner for Welding: How to Prep Aluminum Right shows how much surface condition changes what the eye can see.
• Keep a consistent bead sequence. Uniform, evenly spaced passes make profile judgment obvious and hide nothing. If you are still building that muscle, All Position Welding Rod Guide: Types, Uses & Tips covers rod choice for predictable out-of-position beads.
• Fill your craters. Crater cracks are the most avoidable reject there is: back up, reduce current, and fill the ripple rather than pulling straight out.
• Feather your restarts. Grind the stop, overlap by about an inch, and blend the tie-in so it reads as continuous metal, not a step.
• Self-gauge before you call the inspector. Run your own fillet gauge along the toe and check leg size. Catching it yourself keeps the weld in your hands instead of in the repair loop.
• Keep arc strikes off the plate. Strike on a start tab or the run-on plate, never on base metal you will have to explain later.

Missed defects are rarely mysterious. They are dirty surfaces, shallow viewing angles, fatigue, trust, and undocumented calls. Fix the light, slow the scan, gauge the size, write it down, and the same pair of eyes suddenly catches far more than it used to.

Visual Inspection vs PT, MT, UT, and RT: Limits and When to Escalate

Visual Testing (VT) is the filter every weld passes through first. It is the simplest form of examination and the only one that judges the weld as a whole, but it is not a catch-all. There is a point where the human eye reaches its physical limit, and Non-Destructive Testing (NDT) has to be brought in to interrogate what cannot be seen from the surface. The skill is knowing exactly when to escalate and when not to.

What VT Can and Cannot See

A trained inspector can assess fit-up, profile, reinforcement, cleanliness, and overall workmanship in a matter of minutes. Most of that judgement, though, is restricted to the weld surface and the adjacent base metal. Subsurface conditions, such as lack of fusion in the throat of a groove weld or lack of penetration down to the root of a full-penetration joint, are often completely invisible to even the most experienced eyes.

This is exactly why the industry relies on a broader family of methods. According to AWS B1.10M/B1.10, Guide for the Nondestructive Examination of Welds, visual, penetrant, magnetic particle, radiographic, ultrasonic, and other methods are all covered under one reference for the purpose of supplementing each other, not replacing one another.

The NDT Methods and Their Purpose

Each supplementary method has a very specific lane to run in. These are the ones most often called upon after a VT examination:

• Penetrant Testing (PT): As outlined by the ASNT Visual Testing overview, penetrant testing requires a clean, smooth surface and reveals surface-breaking indications only. It is excellent at finding fine, tight, surface-breaking cracks that can be near-impossible to spot with the naked eye, yet it complements rather than replaces visual examination.
• Magnetic Particle Testing (MT): Primarily used on ferromagnetic materials, MT locates surface and very shallow, near-surface discontinuities. Much like PT it is a surface-orientated method, but it can be far quicker on magnetic parts.
• Ultrasonic Testing (UT), including Phased Array: UT sends high-frequency sound waves into the weld, allowing it to interrogate the interior of the joint. Radiography and ultrasonics are able to detect lack of fusion and lack of penetration that VT cannot see. With phased array, an inspector can produce detailed cross-sectional views of the weld, making it invaluable for thick, high-integrity joints.
• Radiographic Testing (RT): Using X-ray or gamma radiation, RT creates an image of the internal structure of the weld. It is extremely effective on volumetric flaws such as gas pockets, slag entrapments, and certain fusion issues. Its use does come with strict safety controls, however, and it is far less forgiving when it comes to accessibility.

When to Escalate and When to Hold Off

VT should precede every other method. Not only is it the fastest and cheapest, but it prevents wasting expensive NDT on welds that fail on appearance alone. A heavily spattered, undercut fillet on a structural bracket may be rejected by VT before a single PT can is opened. The same principle holds true for precision work, such as aerospace welding, where a sound profile is scrutinised as closely as internal integrity.

Escalation to surface NDT, such as PT or MT, adds value when you are hunting fine, tight cracks or first-layer indications sitting on the edge of visibility, especially after grinding or dressing. Escalation to volumetric methods, such as UT or RT, is most often mandated for full joint penetration groove welds in critical service. That can range from pressure-retaining joints to load-bearing members, or even specialised work such as aluminum boat welding for vessels in demanding marine service. Material selection also plays a part, since procedures for alloy steel welding electrodes can require volumetric examination based on the governing code and service conditions.

That said, not every weld needs to be over-tested. Applying RT to a small, non-critical fillet weld is one of the most common budget-wasting traps. The opposite is far more dangerous: using VT as a substitute for required volumetric examination.

Practical Comparison

Method	What It Sees	Speed	Cost	Limitations
VT	Surface profile, size, fit-up, and obvious discontinuities	Very fast	Lowest	Cannot see subsurface or tight internal flaws
PT	Surface-breaking only	Fast	Low	Requires clean, smooth surface. No volumetric coverage
MT	Surface and very shallow near-surface on ferromagnetic metals	Fast	Low-Mid	Limited to magnetic materials
UT (including Phased Array)	Volumetric, including lack of fusion and lack of penetration	Mid	Mid-High	Requires skilled interpretation and good access to the weld face
RT	Volumetric, permanent image of internal indications	Slowest	Highest	Strict safety, access for source/film, less ideal for some geometries

Ultimately, the question is not “VT or NDT?”, it is “VT and the correct NDT for the risk”. For jobs involving air arc welding and carbon arc gouging repairs, or intricate repair work, VT will always be your first verification before committing to costly volumetric testing. With accessibility, surface finish requirements, and safety all taken into account, the most efficient inspection plan is the one that escalates only when VT runs out of answers.

Building a Visual Inspection Program That Holds Up

Individual skill gets you through one weld at a time. It does not get you through a third-party audit, a follow-up inspection, or the month your best inspector is on vacation. A program is what remains when the talented person leaves the room: written steps, qualified people, controlled tools, and records that tie a serial number to a welder, a date, and a disposition. This section is about building that system, whether you run a structural shop with a full QC department or a two-bay garage with no dedicated inspector at all.

Write the Procedure Before You Need It

A visual inspection procedure does not have to be fifty pages. It needs to answer four questions: what gets inspected, when, against which acceptance criteria, and how it gets recorded. Pull the acceptance criteria straight from the governing code and the contract drawings rather than paraphrasing them from memory, and reference the standard guide for visual weld examination, AWS B1.11, Guide for Visual Inspection of Welds. Name the governing document by edition, because criteria change between editions and an auditor will ask which one you used.

Qualify the People, and Keep Their Eye Records Current

Programs fail at the personnel layer more often than anywhere else. AWS B5.1, Specification for the Qualification of Welding Inspectors, defines what an inspector needs: documented experience, a written examination, demonstrated capability on real work, and proof of visual acuity. AWS builds on that with the Certified Welding Inspector (CWI) and Certified Associate Welding Inspector (CAWI) credentials, the practical baseline most employers ask for when they hire visual inspection personnel.

Not every shop uses AWS, though. ASNT’s visual testing method page points to the other common route: ASNT SNT-TC-1A is an employer-based recommended practice, meaning your own written practice defines the training, experience, and certification levels for VT personnel. Whatever you name the levels, the training records, examinations, and acuity checks must exist before the person signs an inspection line, not after the auditor asks. Re-test vision periodically, because near acuity, distance acuity, and color and contrast discrimination all drift with age, and log the results in the inspector’s certification file. For people starting down this path, a structured aluminium welding course builds the hands-on base that classroom certification cannot give you.

Control the Tools and the Light

Gauges need a verification or calibration schedule, a unique ID, and a storage place that is not the bottom of a toolbox. Log when each fillet gauge, undercut gauge, and magnifier was checked. Lighting deserves the same treatment: specify a minimum illuminance for inspection areas, position lamps so light rakes across the weld at an angle rather than shining straight down, and replace bulbs that have yellowed. Uncontrolled tools are how two inspectors arrive at opposite verdicts on the same bead.

Hold Points, Weld Maps, and Traceability

Put inspection hold points in the traveler so work physically stops until someone signs. Fit-up before the root pass, root and interpass where accessible, final examination before paint or coating. A weld map that assigns each weld a number, welder ID, date, and disposition is what lets a nonconformance trace back to a procedure, a consumable lot, or a person. Traceability is also where the aerospace welding world sets the standard every other sector should copy: nothing ships without a paper trail.

Close the Loop on Nonconformances

A rejected weld should trigger two actions: a repair, and a root-cause note. Was it undercut from travel speed, porosity from damp electrodes, underfill from an inconsistent weave? Feed that pattern back to the WPS or the welder’s training instead of repairing the same symptom every week. The repair itself needs re-inspection and its own record. If a defect keeps appearing on the same joint type, that is a procedure problem, not a people problem.

Track a Few Metrics That Actually Predict Trouble

You do not need a dashboard. Three numbers reviewed monthly tell you most of what is going on:

Metric	What it tells you
First-pass yield	Share of welds accepted on first VT with no repair. Falling yield means something upstream changed.
Repair rate by welder and joint type	Isolates training gaps from procedure gaps.
Defect Pareto	Most common defect types ranked, so you fix the top one or two causes first.

New Technology: Support, Not Substitution

AI vision cameras can flag surface indications in real time, weld monitoring systems log parameters alongside visual results, and drones put eyes on bridge and tank welds nobody can safely reach. All of that is genuinely useful for coverage and consistency. None of it removes the need for criteria, calibration, and accountable human judgment: the system flags, the qualified inspector decides, and a named person signs. Treat these tools the way you would treat a very diligent assistant with unreliable judgment on edge cases.

If You Have No Dedicated QC Staff

Small shops can still hold a real program with four habits: one page of written criteria taped inside the toolbox, a set of gauges bought once and looked after, a hold point before anything is painted, and a signed record for every job. Assign one person ownership of inspection, get them certified or at minimum trained against the employer written practice, and make cleaning and self-checking part of the welder’s own workflow. A program is not paperwork for its own sake. It is the reason the answer to “is that good enough?” never depends on who happens to be standing there.

Conclusion: The Cheapest Test You Cannot Afford to Skip

It is tempting to treat visual inspection as the low-tech option, the thing you settle for when the budget will not stretch to radiography. The framing is backwards. AWS B1.11 recommends visual testing (VT) as the primary method for any effective weld quality control program, and ASNT treats it as a distinct, fully developed NDT method in its own right. Most rejectable conditions in a welded joint live on the surface, which makes VT both the cheapest examination step and the highest-yield one.

The catch: it only works as a process, not a glance.

What the Whole Guide Comes Down To

Strip away the detail and the system is simple:

• A defined workflow. Documents, WPS, and qualifications first, then material identity, fit-up, root and interpass checks, surface cleaning, a systematic scan, measurement, documentation, and a written disposition. Skip the front end and final VT can only reject, never prevent.
• A small set of defect families. Cracks, undercut, porosity, incomplete fusion and penetration, overlap, profile problems, underfill, undersized fillets, spatter, arc strikes, surface slag, mismatch, and distortion. Learn their visual signatures and usual causes and you can call most in seconds.
• The right rulebook. AWS D1.1 for structural steel, ASME BPVC Section V Article 9 for the examination, API 1104 for pipe, ISO 5817 for quality levels. Contract documents outrank everything, and numeric limits come from the governing code, never memory.
• Documented acceptance criteria and qualified inspectors. This is the line that separates professional visual inspection from an opinion. A qualified eye working to a written criterion is evidence; the same eye working from habit is not.
• Enough light and enough time. Poor illumination, shallow viewing angles, dirty welds, and end-of-shift rushing cause more escapes than bad eyesight ever will.

Your Monday List

You do not need a department to start. You need five concrete moves:

1. Write a one-page VT procedure. Scope, criteria, tools, lighting, scan pattern, record format, disposition rules. One page is enough to make the process repeatable.
2. Buy the right gauges. A fillet gauge, an undercut gauge, a cam gauge, a tape, and a 5x to 10x magnifier. A twenty-dollar gauge ends arguments a hundred opinions cannot.
3. Fix the lighting. Adequate illuminance, raked across the surface at an angle rather than straight on, no glare. This is the cheapest quality improvement available to you.
4. Set hold points in the traveler. Fit-up, root pass, and final VT signed before the next operation covers the joint. Once it is welded shut, the root is gone.
5. Require documented eye exams. Near acuity, distance acuity, color perception, and contrast sensitivity, on record and renewed on schedule.

Run that list for a quarter and you will see what every mature shop sees: fewer repairs, fewer late surprises, fewer arguments at the paint stage.

A Skill, Not a Gift

If you weld, hear this directly: inspection is not a talent some people are handed and others denied. It is a learnable skill, built the way you built your hood time, through deliberate reps with feedback. Structured programs such as an Aluminium Welding Course: What You Learn, Costs & Careers show what formal training covers, and welders who take it consistently make welds easier to accept. Techniques help too: clean as you go, keep bead sequences consistent, fill craters, and blend restarts. A welder who inspects their own work before calling the inspector rarely loses a weld.

Small shops without QC staff can run every item on this list with a binder, a toolbox drawer, and a habit. The AWS guide to visual inspection exists precisely because this method scales down as well as up.

Treat visual inspection as the first and most productive filter in weld quality, give it light, time, and paperwork, and everything after it gets cheaper. That is not the low-tech compromise. It is the whole game.

Frequently Asked Questions

What is visual inspection of welding?

It is the systematic examination of a weld and surrounding base metal using the eyes plus simple aids such as good lighting, magnifiers, mirrors, borescopes, and weld gauges. The inspector compares what is visible against the applicable code, drawing, and workmanship requirements, then records the result. It is performed before, during, and after welding.

Can visual inspection detect lack of penetration or lack of fusion?

Generally no. Visual inspection only evaluates what is on or near the surface, so subsurface lack of fusion or incomplete joint penetration can pass a visual check. At the surface, incomplete fusion may appear as a visible line or unfused edge, but confirming internal soundness requires ultrasonic testing, radiography, or another volumetric method.

What vision standard must a welding inspector meet?

Under AWS B5.1, inspectors must read Jaeger Number 2 letters or smaller at a minimum of 12 inches in at least one eye, with or without correction, and pass color perception and shades-of-gray contrast tests. ASNT-based employer written practices often require Jaeger J-1 or equivalent, so the exact number comes from the governing document.

How much of a weld should be visually inspected?

Structural codes such as AWS D1.1 generally require visual inspection of all welds, while other specifications may define sampling percentages. The contract documents control. Even where sampling is permitted, many shops inspect 100 percent visually because it is fast and inexpensive compared with repair or field failure.

What is the difference between a discontinuity and a defect?

A discontinuity is any interruption in the uniformity of a weld or object. A defect is a discontinuity that fails the applicable acceptance criteria. The same undercut or porosity can be an acceptable discontinuity under one quality level and a rejectable defect under another, which is why the governing code must be identified first.

Do I need to clean a weld before visual inspection?

Yes. Slag, spatter, flux residue, oil, paint, and scale can hide cracks, undercut, and porosity, and ASNT recognizes surface preparation as a required preparatory step. Cleaning the weld and adjacent base metal is one of the most common reasons visual inspection misses a rejectable condition.

How much does visual inspection cost compared with other NDT?

Visual inspection is by far the least expensive examination method because it needs little equipment and no consumables or radiation controls. It is also the fastest. That economics is exactly why it should be used first to catch surface problems before paying for penetrant, magnetic particle, ultrasonic, or radiographic examination.

What tools do I need to start doing visual weld inspection?

At minimum: a strong, adjustable light, a fillet weld gauge set, an undercut or V-wedge gauge, a tape or rule, a magnifier, a mirror for restricted access, a marker, and a camera for records. Add a borescope for tubing and confined spaces, and calipers for dimensional checks. Calibration and condition of gauges matter as much as owning them.

Is visual inspection enough for structural welds?

It is necessary but sometimes not sufficient. AWS D1.1 requires visual inspection and often also requires nondestructive testing on certain joint types, particularly complete-joint-penetration groove welds. The project specification determines the required NDT extent; visual inspection alone is rarely accepted as the only examination for critical complete-joint-penetration welds.