Welding Faults: The Complete Guide to Types, Causes, Prevention, and Repair

A structural fabricator once told me the story of a Friday-afternoon handrail weld that looked perfect, passed a quick glance by everyone on the floor, and shipped. Three weeks later it came back cracked open at a jobsite, and the repair crew, the re-inspection, the crane time and the missed deadline cost the shop more than the entire job was worth. That is the quiet truth about welding faults: they rarely announce themselves when they are made — they announce themselves when the structure is already in service, and by then the bill is enormous.

The 2 A.M. Phone Call: Why Welding Faults Deserve Your Full Attention

A structural fabricator once told me about a Friday-afternoon handrail weld. It looked perfect, passed a quick glance on the floor, and shipped. Three weeks later it came back cracked open at a jobsite. The repair crew, the re-inspection, the crane time and the missed deadline cost the shop more than the job was worth. That is the quiet truth about welding faults: they rarely announce themselves when they are made. They announce themselves once the structure is in service, and by then the bill is enormous.

Nobody on that floor was careless in the way we imagine carelessness. They were rushed, under-trained on acceptance criteria, and working without a procedure that told them what “good” actually meant in measurable terms. That is how most welding faults happen: not through ignorance of how to pull an arc, but through gaps in the system around the arc.

Welding Faults Are Not a Cosmetic Problem

It is tempting to treat an ugly weld as a cosmetic issue and a pretty weld as a safe one. Both instincts are wrong. A welding fault is a structural problem first, because it changes how load travels through a joint. An undercut at the toe of a fillet weld reduces the effective throat, so the joint carries less force than the engineer calculated for it. A single cold crack in a load path can propagate under cyclic loading until the whole connection fails.

It is also a financial problem, a safety problem, and — less discussed but just as real — a health problem:

• Financial: rework labor, replacement consumables, re-run NDT, crane and mobilization time, liquidated damages. Prevention costs pennies compared to repair in the field.
• Safety: the U.S. Occupational Safety and Health Administration identifies both health hazards from welding, cutting and brazing — including exposures to metal fumes and ultraviolet radiation — and direct safety hazards such as burns, eye damage, electrical shock, cuts, and crushed toes and fingers (OSHA welding hazards and solutions). A weld that fails in service turns those theoretical risks into real ones.
• Health: the National Institute for Occupational Safety and Health states that welding fumes are composed of metals and that most fumes contain a small percentage of manganese (NIOSH welding topic page). Long-term exposure is an occupational health issue that the trade is still reckoning with.

Treat any one of those dimensions seriously and the case for getting faults right becomes overwhelming.

What This Guide Covers — and Who It Is For

This guide is built for the people who actually touch welding faults: shop welders and fitters who need to recognize what they are looking at, fabricators and shop owners who carry the financial risk, site supervisors who sign off on work, inspectors-in-training preparing for certification exams, and serious DIY welders building anything that has to hold a load. You do not need a metallurgy degree to use it. You need a willingness to look at a weld analytically instead of just glancing at it.

Here is what you will get, section by section:

1. Identification — what each major fault actually looks like, in the shop and under NDT, so you can name it correctly instead of guessing.
2. Root causes — the real mechanisms behind porosity, undercut, cracks and lack of fusion, organized through a systematic diagnosis framework rather than superstition.
3. Prevention — the procedures, habits and controls that stop faults before the arc is struck: qualified procedures, preheat discipline, consumable storage, fit-up tolerance and technique.
4. Inspection — which nondestructive test finds which fault, how acceptance criteria are written, and what a certified inspector is actually evaluating.
5. Repair — what to do when a fault already exists, who is authorized to approve it, and why “just capping over it” is one of the most dangerous shortcuts in the trade.

You Are Joining a Very Old Fight

If this problem feels personal, take some comfort: you are not the first generation to wrestle with it. The American Welding Society’s structural welding code, D1.1, was originally published in 1928 as the Code for Fusion Welding and Gas Cutting in Building Construction. Nearly a century later it has grown into D1.1/D1.1M:2025-AMD1, the worldwide reference standard for welding structural steel. AWS has since authored more than 350 standards covering welding practices and procedures — every one of them written, in some form, because a weld failed or nearly did.

That history matters practically, not just rhetorically. Every acceptance criterion in a modern code was bought with a lesson someone paid for. When a code says an undercut cannot exceed a certain depth, that number is not a bureaucratic preference. It is distilled experience.

It is also worth noting that the trade’s own workforce data points the same direction. The American Welding Society reports that 66% of the welding workforce supplements their formal education with AWS certifications — a signal that welders, inspectors and supervisors increasingly treat welding knowledge as something you keep validating, not something you finish learning in an apprenticeship. Structured education, and guides like this one, are how the knowledge gets passed on.

How to Read the Rest of This Guide

Read it front to back the first time if you can; the sections build on each other, and the prevention playbook lands harder once you have seen how faults actually behave. After that, use it as a reference. Come back to the fault you are staring at, find it in the identification sections, and work backward through cause, prevention, inspection and repair.

The fabricator with the handrail story did something else useful after that phone call. He stopped relying on eyesight and goodwill, wrote real procedures, and trained his crew to know the difference between a weld that looks fine and a weld that is fine. That is the transformation this guide is for. Welding faults are predictable, and when you treat welding as a controlled process rather than a craft practiced purely by feel, they are preventable. Let’s get into it.

Welding Fault vs. Welding Defect vs. Discontinuity: Why the Words You Use Matter

Two welders can look at the same bead and argue for ten minutes about whether it “passed.” Usually they are not disagreeing about what they see. They are disagreeing about what to call it. In inspection work, words are not decoration. A discontinuity, an imperfection and a defect sit at three different points on the same scale, and the difference between them decides whether a job ships, gets ground out, or gets rejected outright.

Start with the neutral word: discontinuity

A discontinuity is simply any break in the normal, expected properties or continuity of the weld or base metal. Porosity is a discontinuity. A toe crack is a discontinuity. So is a tiny arc strike. The word carries no verdict. It describes geometry and metallurgy, not acceptability. Every weld ever produced has some discontinuities; the trade’s job is to sort the harmless from the hazardous.

An imperfection takes it one step further: it is a measurable, characterizable deviation — you can put calipers, a gauge or a radiograph reading on it. The industry’s master分类 reference here is ISO 6520-1, which classifies fusion weld imperfections into five families:

• Geometric imperfections — undercut, overlap, underfill, excessive reinforcement, misalignment.
• Cracks — longitudinal, transverse, crater, toe, root and heat-affected-zone cracking.
• Cavities — gas pores, wormholes, blowholes, crater pipes.
• Solid impurities — slag inclusions, tungsten inclusions, oxides.
• Miscellaneous imperfections — arc strikes, spatter, grinding damage, start/stop mismatches.

A defect is the word with teeth. A defect exists only when a discontinuity or imperfection exceeds the acceptance limits of the governing standard. The same 4 mm pore is a tolerable imperfection in one project and a rejectable defect in the next. That is why experienced inspectors are careful with language: acceptance is never a property of the weld alone.

Why acceptance criteria are never universal

Different codes were written for different consequences of failure, so their limits differ in tone and tolerance:

• AWS D1.1/D1.1M:2025-AMD1 — structural steel; routinely referenced in project contracts, it defines how welders and welding procedures are qualified and how welds are inspected and accepted.
• ASME BPVC Section IX — pressure equipment; qualification-focused, with acceptance criteria largely deferred to the construction codes (Section I, VIII, B31 series).
• API 1104 — pipeline girth welds, with its own ring-gauge and gouge-and-repair philosophy.
• ISO 5817 and ISO 10042 — quality levels B, C and D for steel, aluminum and copper alloys, and for wrought and cast joints respectively.
• AWS D1.5 — bridge welds, stricter and more prescriptive than D1.1 because of fatigue loading.
• AWS D17.1 — aerospace fusion welding, where tolerances are measured in fractions of a millimeter.

The vocabulary itself is standardized, too: AWS A3.0M/A3.0 (current edition 2025) is the AWS standard for welding inspection and visual examination terminology, and it is the reference a Certified Welding Inspector leans on when a dispute over wording reaches the desk.

At the extreme end of acceptability sits fracture control. NASA-STD-5009C, Nondestructive Evaluation Requirements for Fracture Critical Metallic Components, issued August 3, 2023, defines NDE requirements where a quantitative demonstration of probability of detection (POD) is required. It exists to support NASA-STD-5019A fracture control requirements, which stipulate that all human-rated aerospace flight systems be subjected to fracture control procedures to preclude catastrophic failure — a useful reminder that somewhere, a “small imperfection” was once the difference between a mission and a tragedy.

The practical takeaway: before you judge a weld, name the code, find the clause and read the limit. One shop’s “good enough” is another shop’s rejected weldment, and the drawing always wins.

The Most Common Welding Faults You Will Actually See in the Shop

Every weld has a story to tell, and most of the trouble starts on the surface, right where a quick eye can catch it. The faults below are the ones that fill up rework tickets, hold up inspections, and send welders back to the grinder. Learn to spot each one the moment you strike the arc out, and you will already be ahead of most of the problems that cost shops real money.

Porosity: Gas Trapped in the Solidifying Pool

Porosity shows up as rounded cavities or pinholes in the weld face. Gas porosity looks like scattered blowholes, wormhole porosity reads as elongated, tunnel-like channels running along the weld, and surface porosity is the cluster of open pinholes you can see without a magnifier. All of it starts the same way: gas becomes trapped in the molten pool before it can escape.

The usual suspects are moisture, oil, paint, or rust on the base metal or consumables, shielding gas flow that is too low (or far too high, which pulls air in), drafts, and arc blow. Danger level: real. Porosity removes load-bearing cross-section, and the subsurface kind can sail through a visual check. Prevent it by grinding clean a couple of inches either side of the joint, storing low-hydrogen electrodes in a rod oven, checking hoses and flow meters, and welding away from open doors and fans.

Undercut: The Groove You Did Not Mean to Cut

Undercut is a groove melted into the base metal adjacent to the weld toe or weld root and left unfilled by weld metal. It looks like a thin notch or bite taken out of the plate right where the bead meets it. The classic causes are excessive current, travel speed that is too fast, and an incorrect electrode or gun angle. It matters more than it looks: an undercut is a built-in stress raiser, and fatigue cracks love to start exactly there. Drop the amperage, slow down, correct your angle, and favor stringer beads over wide weaves. Acceptance limits for undercut depth are written into the governing code, and AWS D1.1 is where most structural work in the United States gets judged.

Overlap, Lack of Fusion and Toe Inclusions

Overlap (cold lap) is molten weld metal that rolled over the base metal without bonding to it. It reads as a wrinkled, raised edge instead of a smooth tie-in. Low current, travel speed too slow, incorrect technique, or a dirty surface will do it. It is a fusion defect hiding in plain sight, and it quietly reduces the effective throat of the fillet.

Lack of fusion on the surface looks like a visible line or crevice between weld and base metal where the two never married. Inclusions at the toes are bits of slag or oxide trapped right at the edge of the bead, usually traceable to interpass cleaning that never happened or a weave that rolled slag under itself. Both are rejectable under most structural criteria, and both are preventable with heat, cleaning, and patience.

The Smaller Faults That Still Cost You

• Arc strikes on adjacent base metal: stray arcs outside the joint leave small, hard, quenched spots. On load-bearing steel they can crack and act as fatigue starters. Strike the arc only on the joint, or use a start plate. Grinding them out is the accepted fix.
• Spatter: those frozen droplets are mostly cosmetic, but they signal wrong parameters, a contaminated surface, or a bad gas situation, and they force extra grinding time. Fix the root cause rather than the symptoms.
• Underfill: the weld face sits below the required reinforcement height, leaving the joint thin. It usually means too little current, too much travel speed, or a joint that was never filled. The fix is honest: more passes.
• Excessive reinforcement: too much metal creates a sharp transition at the toe and a stress concentration. Cap the joint flush-ish, then dress it.
• Burn-through: a hole melted clean through the plate, common on thin material or a wide root opening. Slow down, reduce current, and tighten fit-up tolerance.

One more habit worth building: verify what the code actually permits before you argue with an inspector. Surface-breaking linear indications are judged far more strictly in some industries than others, and aerospace work follows NASA-STD-5009 for nondestructive evaluation acceptances. And while grinding and chipping these faults, remember that welding and cutting fumes are treated as a workplace hazard in their own right: NIOSH flags the eyes, skin, respiratory system and central nervous system as target organs for welding fumes. Clean the fault, and protect yourself while you do it.

Cracks, Distortion and the Faults That Hide Inside the Weld

Everything covered so far shares one trait: you can see it with your eyes, a flashlight, and a bit of common sense. The faults in this section are different. They live under the surface, in the heat-affected zone, along the root pass, or in the base plate itself, and they are the reason welding codes were written in the first place. A weld can look perfectly formed from the face and still be harboring a crack that will grow under load. That is why fracture control programs such as NASA-STD-5009 exist at all: critical structures depend on nondestructive evaluation to prove that crack-like flaws capable of initiating failure are not hiding in areas that matter.

Cracks: The One Fault You Never Argue With

Cracks come in two families, and knowing which one you are dealing with tells you how to prevent it.

• Hot cracks (solidification and liquation). These form while the metal is still hot, as the weld solidifies. Solidification cracking runs through the centerline of the bead and is encouraged by a concave profile, a deep and narrow joint, high travel speed, and sulfur-rich or otherwise poorly balanced chemistry in the base metal or filler. Liquation cracking sits at the fusion line or in the HAZ of aluminum and nickel alloys, where low-melting phases melt again under the heat of a nearby pass.
• Cold cracks (hydrogen-induced). The technical name is hydrogen-induced cold cracking, or HICC, and it most often appears in the heat-affected zone of hardenable steel. Unlike hot cracking, it does not happen while you are welding. It happens hours, sometimes days, later.

Hydrogen cracking is never a single-cause problem. It requires three factors acting at the same time, in the same place:

1. Hydrogen present in the weld, picked up from damp coatings, rust, oil, grease, or moisture-laden consumables.
2. A susceptible microstructure, typically a hard, martensitic HAZ that forms when hardenable steel cools too quickly without preheat.
3. Tensile stress, supplied by joint restraint, shrinkage, or the applied load of the structure itself.

Preheat, controlled interpass temperature, low-hydrogen consumables kept in a rod oven, slow cooling, and, where the code demands it, post-weld heat treatment are the standard countermeasures. Every one of them attacks one or more of the three factors.

Lamellar Tearing: When the Steel Itself Gives Up

Lamellar tearing is not really a weld fault; it is a base metal fault triggered by welding. Rolled steel plate can contain flat, plate-like inclusions parallel to the rolling plane. When a heavy T-joint or corner joint is welded under high restraint, the shrinkage strain acts through the plate thickness, pulling the steel apart along those internal planes. The result is a stepped, terrace-like crack beneath the weld, usually close to the fusion line, that visual inspection will never reveal.

The fix starts before the arc is struck: use through-thickness-tested Z-quality plate in highly restrained joints, specify lower restraint details, use weld-metal overlap or butterfly plates to spread the strain, and apply preheat to soften the response.

Internal Faults: Lack of Penetration, Porosity and Slag

• Lack of penetration. The weld simply did not reach the root. The face looks full; the root is untouched. Causes are classic: insufficient amperage, wrong electrode angle, too fast a travel speed, excessive root opening that consumes all your filler, or poor joint geometry. On a full-penetration joint this is a rejectable condition every time.
• Internal porosity. Gas pockets sealed inside the weld metal, caused by everything in the porosity discussion plus, crucially, poor shielding at the root of a joint where gas cannot circulate.
• Slag inclusions. Trapped flux or slag sitting inside the weld, usually from poor interpass cleaning, incorrect technique such as an arc that is too long or a wrong work angle, or a joint shape that physically cannot be cleaned and re-welded properly.

Distortion: The Fault That Costs You the Fit-Up

Distortion does not threaten structural strength directly, but it ruins dimensions, causes rework, and if left uncorrected can force so much forced fit-up that it creates the restraint that causes cracking. Weld metal shrinks as it solidifies and cools, and distortion is simply unbalanced shrinkage. The family includes:

• Longitudinal shrinkage along the weld length.
• Transverse shrinkage pulling the joint edges together.
• Angular distortion, where the plate folds along the weld.
• Buckling, where thin plate goes wavy from compressive residual stress.

The countermeasures are practical and old-fashioned: balance welds on both sides of a joint, use a deliberate sequence such as back-step or skip welding, fixture and clamp the assembly, alternate welds to spread heat, keep total heat input reasonable, and think about shrinkage before you tack, not after.

Note: Visual inspection alone will not protect you here. Where a fault can hide, the governing code will call for volumetric methods like radiography or ultrasonic testing. In the next section, we look at the full inspection toolkit and how acceptance criteria actually decide what passes.

The Root Cause Tree: Man, Machine, Material, Method and Medium

Every welder has been there: a fault shows up, and the shop starts guessing. Someone blames the wire, someone blames the new hire, someone blames the gas supplier. The guess that wins is usually the loudest one, and a week later the same fault comes back. Systematic diagnosis replaces that noise with a repeatable method. The most widely used framework in manufacturing problem-solving is the 5M tree — Man, Machine, Material, Method and Medium — and it maps onto welding faults almost perfectly.

Man: the person holding the gun

Human factors are the first branch, but “operator error” is a lazy stopping point. Dig deeper: is the welder formally trained and qualified for the process and position being run, or promoted into it? Is fatigue setting in on the fifth hour of overhead work? Is PPE compliance slipping — hood shade too dark, respirator missing — so the welder compensates by holding an inconsistent arc length? Even experienced hands drift when they are working around a defect they cannot see. Skill gaps, technique drift and comfort all live here, and the fix is usually qualification, coaching and rotation, not punishment.

Machine: the equipment doing the work

The second branch covers the power source and everything feeding it: amperage and voltage settings versus the WPS, wire feed speed calibration, a worn liner or contaminated contact tip, a clogged gas regulator, a cracked liner, a leaking hose, or a machine whose meters have drifted out of calibration. A regulator reading low flow at the gauge can still deliver gas that reaches the arc too slowly once hose length and fittings are factored in. Machine problems are cheap to fix and expensive to ignore, which is why a calibration and maintenance schedule belongs to this branch rather than to “method.”

Material: what is being welded

Third comes the metal itself and the consumables: base metal chemistry that differs from what the WPS assumes, mill scale, primer, galvanizing, oil or cutting-fluid residue, and damp or improperly stored electrodes and flux-cored wire. Low-hydrogen electrodes that skipped the rod oven are a classic material-side fault generator, because the moisture they pick up becomes hydrogen in the arc. Coatings and contamination belong here too — they do not care how good your technique is.

Method: how the work is being done

The fourth branch is the procedure in practice: joint design and bevel angle, fit-up and root gap control, travel speed, weave width, torch or work angle, preheat and interpass temperature, interpass cleaning, tack strategy, and whether the actual sequence on the floor matches the qualified WPS. Most method faults are really WPS-compliance faults — the procedure is written correctly, but the gap drifted, the bevel was cut short, or the preheat torch was skipped to keep schedule.

Medium: the environment around the arc

The fifth branch is the one shops underestimate. Drafts, humidity, ambient temperature, welding position, confined-space access and poor visibility all shape the weld. Environmental factors such as drafts are a leading cause of shielding-gas disruption and subsequent porosity in GMAW and FCAW. And ventilation deserves its own attention here: NIOSH guidance warns that you should not assume welding outdoors or in open areas provides adequate general ventilation, even when the welder uses proper positioning and natural drafts. Local exhaust systems should be positioned to draw fume and gases away from the welder and other workers in the area, and a workplace hazard assessment is a recommended step in NIOSH’s hierarchy of controls approach to welding exposure. OSHA’s welding, cutting and brazing hazard guidance similarly puts engineering controls, work practices and PPE at the center of controlling welding hazards (OSHA fact sheet).

A real diagnostic scenario: porosity after the move

Consider a shop that starts getting recurring porosity right after moving a welding cell next to a loading dock door. First they check Man: the welder was qualified, technique unchanged. Then Machine: wire feed, voltage and flow meter verified against the WPS, all in tolerance. Material: same heats, same wire, same batch of gas. Method: fit-up, travel speed and interpass cleaning audited, all compliant. That leaves Medium — and there it is. The dock door cycles open all day, and the cross-breeze from the traffic lane shears the shielding envelope just before it reaches the puddle. The fix is not a new welder or a new machine; it is a wind screen, repositioning the cell, or a local exhaust arrangement that draws fumes away from the welder and other workers instead of letting the draft fight the gas lens. The point of the tree is that you arrive at that conclusion with evidence in twenty minutes instead of after twenty wasted trials.

How Welding Faults Are Found: Visual Inspection, NDT and Code Acceptance Criteria

You can’t fix what you can’t see, and you can’t see what you don’t look for properly. Fault detection in welding isn’t one skill — it’s a layered toolkit, starting with the cheapest tool in the shop (your eyes) and escalating to machines that can find a void the size of a grain of rice inside an inch of steel.

VT: The First, Fastest Gate

Every inspection campaign starts with Visual Testing (VT), performed to the terminology defined in AWS A3.0 / AWS D1.1. VT is typically the first, fastest and most cost-effective inspection gate before any NDT is scheduled — roughly 80% of weld discontinuities are surface-visible, and a trained eye catches undercut, porosity, overlap, underfill and cracks with nothing more than a light, a mirror, a gauge and a magnifier.

But VT has hard limits. It cannot see a lack-of-penetration at the root of a groove weld, an internal slag pocket, or a toe crack hidden under a layer of spatter. That’s where Nondestructive Testing comes in.

The NDT Menu: What Each Method Finds and Misses

• Penetrant Testing (PT): A dye wicks into surface-breaking flaws, then a developer draws it back out as a bright indication. Excellent for cracks, porosity and lack of fusion on any non-porous material — stainless and aluminum included. It misses everything that doesn’t break the surface.
• Magnetic Particle Testing (MT): Magnetic field plus iron particles reveals surface and near-surface flaws (roughly 1–2 mm deep) in ferromagnetic materials. Faster and more sensitive than PT on carbon steel, but useless on stainless, aluminum or castings that won’t hold a field.
• Radiographic Testing (RT): X-ray or gamma radiation produces a film or digital image of the weld’s interior — porosity, slag, lack of fusion and root cracks show up as density differences. RT confirms internal flaws beautifully, but planar flaws (tight cracks, lack of fusion parallel to the beam) can slip through if the geometry isn’t right. AWS even created a Certified Radiographic Interpreter (CRI) credential specifically because reading weld radiographs is its own specialty.
• Ultrasonic Testing (UT): High-frequency sound waves bounce off internal discontinuities and return as indications. UT excels at planar flaws RT misses, works through thick sections, and poses no radiation hazard. Its blind spots: it needs a skilled operator, and geometry (corners, weld profiles) can mask real indications. Phased Array UT (PAUT)扫扫 the beam electronically, giving a visual cross-section and improving repeatability.
• Eddy Current Testing (ET): An induced current detects surface and near-surface variations in conductive materials. Common in tube and aerospace work; less common in structural fabrication.

Reading the Rules: Acceptance Criteria

Finding an indication is only half the job — the code decides whether it’s acceptable. Acceptance criteria are written as measurable limits: maximum allowable length, width and depth of an individual flaw, plus cumulative length per unit length (for example, total porosity allowed in any 12-inch run). Nothing about this is universal. A pore cluster rejectable under ASME may be perfectly legal under API 1104, which is why the governing code always travels with the drawing or WPS.

A Certified Welding Inspector (CWI) interprets those criteria for a living, verifying WPS qualification, monitoring parameters in the booth, performing visual examination, and signing off on what passes. AWS offers CWI endorsements for Structural Steel (D1.1), Structural Aluminum (D1.2), Bridge Welding (D1.5), Aerospace (D17.1), Pipeline (API 1104), Magnetic Particle Testing, Penetrant Testing, and ASME BPVC Section IX — plus CAWI and SCWI levels for inspectors at earlier career stages. The AWS D1.1:2025 edition notably lists refined acceptance criteria and expanded inspection guidance among its updates over the 2020 version.

The High End: Probability of Detection

At the aerospace and nuclear level, “did the inspector find it?” becomes a statistical question. Probability of Detection (POD) studies blind-test NDE systems against known flaws of varying sizes and produce a curve showing the flaw size at which detection becomes reliable (commonly the 90/95 threshold — 90% probability with 95% confidence). NASA-STD-5009C defines primary requirements for NDE in support of fracture control and requires exactly this kind of quantitative POD demonstration. It’s a useful mindset for any shop: an inspection method you’ve never validated is a guess, not a guarantee.

The takeaway: choose the method to match the fault, the material and the governing code — and treat visual inspection not as the cheap option, but as the foundation everything else stands on.

The Prevention Playbook: WPS, PQR, Preheat and the Habits That Kill Faults Before They Start

Every fault covered so far has something in common: it was preventable. Not by luck, and not by a welder with “great hands,” but by a controlled process that repeats the same result on Monday morning that it produced on Friday afternoon. That control starts on paper and ends in daily shop habits. Here is the playbook.

Start With a Qualified WPS Backed by a PQR

A Welding Procedure Specification is the recipe: process, filler classification, shielding gas, diameter, polarity, current range, voltage, travel speed, joint design, preheat, interpass temperature and cleaning requirements. It only means something if it is backed by a Procedure Qualification Record, the test data proving that the recipe actually produces sound welds in the thickness range you intend to run. AWS D1.1 establishes how welders and welding procedures are qualified, which welding processes are typical, and how welds are inspected and accepted, and AWS D1.1:2025 brings updated and clarified requirements to both procedure and welder qualification. Read the current edition rather than the shop binder that was photocopied a decade ago.

Qualification papers are only half the job. Welders need performance qualification and a continuity record, usually maintained through documented work in the qualified range within a six-month window. A welder who has been running aluminum MIG for a year and jumps back on structural carbon steel without requalification is a fault looking for a place to happen.

Preheat and Interpass Control: The Cheap Insurance Policy

Preheat slows cooling, lets hydrogen diffuse out of the joint, and reduces the thermal shock that invites cracking. Interpass temperature keeps that logic running through every pass instead of letting pass three quench pass two. Both are specified on the WPS, and both are measured, not estimated by holding a hand near the plate. On a restrained joint in heavier material, skipping preheat is the single fastest route to a hydrogen-induced crack in the heat-affected zone.

Consumables, Gas and Joint Prep: Where Quiet Faults Start

Damp low-hydrogen electrodes are a classic source of porosity and cracking; rod ovens and strictly observed holding times exist for exactly that reason. Shielding gas deserves the same discipline: correct flow rate, a hose of reasonable length, no leaks at the connections, no CO₂ hose coiled in a puddle, and cylinders changed before they run near empty. On the joint itself, bevel angle, root face, root opening and cleanliness are tolerances, and widening a gap because the fit-up “looked off” is how underfill and lack of fusion begin.

Interpass cleaning matters more than most welders admit. Wire-brushing or grinding each pass removes oxide films and slag that otherwise become the inclusion you find later on a radiograph. Travel speed and technique discipline follow: too fast produces undercut and cold lap, too slow produces excessive reinforcement and burn-through, and weaving beyond the width the procedure allows drives heat where it does not belong.

Fixtures, Tacks and the Daily Shop Rhythm

A well-designed fixture and a deliberate tacking sequence fight distortion before it starts; tack size, tack placement and weld sequencing are part of the method, not improvisation. Behind all of it sits the unglamorous maintenance calendar: calibration dates for ammeters, voltmeters and gas flow meters, scheduled machine service, routine wire liner and contact tip changes, and nozzle cleaning. Most shops only notice these items after a fault appears.

Keep the environment clean too. NIOSH recommends elimination and substitution controls such as using a less toxic welding type or consumable and keeping welding surfaces free of coatings, dirt and dust, and it advises cleaning with a HEPA-filter vacuum or wet methods instead of dry sweeping or compressed air. OSHA Fact Sheet 3647 frames fume and gas control as part of a compliant ventilation and protection program, and a clean, well-ventilated cell is also a more readable one, which makes faults easier to catch early. For the paperwork trail, see OSHA’s welding, cutting and brazing hazard solutions guidance.

Repair or Reject? How to Handle a Welding Fault You Have Already Made

After hours of hard work, the last thing any welder wants to hear is the words, “We have a reject.” However, when a welding fault is discovered, the actions that follow will be the difference between a safe, code compliant repair, or a catastrophic failure down the road. With this being said, the decision must always be repair or reject, never ignore.

Stop, Tag, and Document the Issue

The very first step to handling a welding fault is stop-work discipline. As soon as an inspector flags a rejectable condition, all work pertaining to that joint must be halted until the issue is properly evaluated. This prevents any further alterations from hiding the flaw or making it worse.

Once stopped, the weld must be clearly tagged as non-conforming and documented. Items that should be recorded include the joint location, the type of fault, the inspection method that found it, the date, and the initials of the individual who found it. These records not only keep your shop organized, but they also provide traceability for the repair down the road.

Excavating the Fault: Get Rid of Every Trace

One of the most dangerous mistakes a welder can make is simply “capping over” a defect. Cracks, in particular, will almost always propagate if the crack tip is left behind. Due to this, the entire flaw must be removed by grinding or gouging. For full-penetration joints, back-gouging is often required to ensure total removal from the root side as well.

However, visual removal isn’t enough. D1.1:2025-AMD1 compliant repairs require verification that the crack tip has been fully removed before rewelding. For this reason, penetrant testing is commonly used after grinding out a crack to verify the excavation is clean. This extra step prevents re-introducing a crack directly into the new weld deposit.

Preheat, PWHT, and the Repair Procedure

Not every repair is as simple as rewelding the cavity. Certain materials and joint thicknesses will require preheat and interpass temperature control to combat hydrogen induced cracking. In some instances, Post Weld Heat Treatment (PWHT) may be required to relieve residual stresses from the weldment. These parameters will be dictated by the governing code of the project.

Who Approves It and What Gets Inspected

Repair work is never a “self approval” task. Per the majority of fabrication codes, a repair procedure must be authorized by the engineer-of-record or their designated representative. Additionally, AWS Certified Welding Inspectors (CWIs) use the code as the standard benchmark to evaluate weld quality and determine the acceptability of a weld.

Once the weld has been repaired, re-inspection requirements go into full effect. The completed repair must pass the same NDT method that originally rejected it, ensuring that the fix is sound prior to the part being released. All of these results, along with the repair procedure used, must be updated into the project’s inspection and quality control records.

A Real Life Repair Scenario

Take for example, a shop that discovers a 1.75″ long crack at the toe of a structural fillet weld. Instead of capping it over, the crew immediately tagged the joint and halted production. A CWI referenced the project’s acceptance criteria under AWS D1.1:2025-AMD1, approving a documented repair. The crack was then ground out to a smooth excavation, verified to be free of any remnants through PT, preheated to the specified temperature, and re-welded utilizing the qualified WPS. After completion, the repair was re-inspected via PT and passed with zero indications. The joint was released with its updated traceable repair records, saving the project from a devastating in-service failure.

At the end of the day, welding faults will happen. However, with discipline, verification, and code compliance, a reject does not have to be a loss. It can be proof that your quality system is working exactly as it was designed to.

The True Cost of Welding Faults: Safety, Health, Money and Reputation

When a welder grinds out a bad bead, the visible cost is small: an hour of labor, a few inches of wire, a little shielding gas. What most shops underestimate is everything that grows around that defect. Welding faults behave less like isolated mistakes and more like stones dropped in a pond. The first ripple is rework. The ripples after that reach the schedule, the customer, the insurance premium and, in the worst cases, the people standing beside or beneath the finished structure.

Three Layers of Cost: Direct, Indirect and Catastrophic

It helps to sort fault costs into three tiers, because shops almost always calculate only the first one.

Direct costs are the ones a foreman can tally by lunchtime. Indirect costs are where budgets quietly die. A rejected assembly repaired on site means a crane booked again, a crew rescheduled, a lowboy dispatched, and possibly a rental yard charging for a second week. Expedited shipping to recover a missed date can erase the entire margin on a job. Studies in manufacturing consistently show that indirect rework costs run several times the direct repair cost, which is why experienced estimators treat a rejected weld as a schedule event rather than a welding event.

Catastrophic costs are the ones nobody recovers from. A cracked weld found in service can mean failure investigation, liability exposure, regulatory citations, and a reputation that takes years to rebuild. As OSHA makes clear, the hazards of the trade extend well beyond the arc: OSHA’s welding, cutting and brazing hazard resources cover both fire and explosion risks and the exposure risks that follow the fume.

The Health Dimension: What the Fault Costs a Body

There is a cost layer that never appears on a rework ticket, and it is paid in the lungs and nervous systems of the people doing the work. Welding fume is a complex mix of metal oxides, silicates and gases generated from the consumable, the coating and the base metal. The CDC/NIOSH welding topic page collects the current occupational health picture, and it is worth a slow read.

Manganese is the one to know. NIOSH reports that prolonged exposure to high manganese concentrations greater than 1 mg/m³ in air may lead to a Parkinsonian syndrome known as manganism. Less obvious is the effect at lower levels: some studies indicate that welders exposed to manganese below 0.2 mg/m³ perform more poorly on tests of brain function and motor skills, including changes in mood, short-term memory, reaction time and hand-eye coordination. That matters for fault prevention directly, because a welder whose reaction time and short-term memory are slipping is a welder who misses fit-up tolerance and interpass temperature.

The published limits themselves show how much disagreement exists in this area:

• NIOSH REL: 1 mg/m³ TWA and 3 mg/m³ STEL for manganese
• OSHA PEL: 5 mg/m³ ceiling, per the NIOSH Pocket Guide entry for manganese
• ACGIH TLV: 0.02 mg/m³ TWA for the respirable fraction

That is a 250-fold gap between a legal ceiling and the recommended occupational limit. Compliance with the PEL alone does not mean the exposure is benign, and shops that treat the OSHA number as a health target rather than a legal floor are fooling themselves.

Other exposures worth respecting: stainless and chromium-bearing alloys can generate hexavalent chromium, a confirmed carcinogen with its own OSHA standard. Welding on galvanized or plated steel can cause metal fume fever, with chills, fever and nausea appearing hours after exposure. Iron oxide inhaled with welding fume has been found in the lungs years later, even after removal from exposure, according to NIOSH-cited research — the damage does not reset when the welder clocks out.

And the newest entry: CDC and NIOSH researchers identified welder’s anthrax as a new, rare occupational disease, with seven patients reported from 1994 to 2020 and five of the seven dying. It comes from inhaling spores in contaminated clothing, hair or worksite dust rather than from the arc, which is exactly why it went unrecognized for decades.

Why Respect for Fumes Belongs in Fault Prevention

It is tempting to treat health as a separate conversation from weld quality. In practice they share a root cause: both are failures of control. A shop that lets shielding gas drift, skips preheat and welds by feel will also skip ventilation checks and respirator fit testing, because the underlying habit is the same — accepting an unmeasured process. The codes that govern weld quality and the standards that govern exposure both assume you are measuring something. The shops with the fewest faults and the healthiest crews are almost always the same shops: the ones that treat welding as a controlled process, tracked on paper, reviewed openly, and never waved through because the bead looked good on a Friday afternoon.

Quick-Reference Checklists, Common Mistakes and Your Next Step

You have read through porosity, undercut, cold cracks and lamellar tearing. Now let us turn all of it into something you can carry to the shop floor. The best welders I know do not rely on memory for quality; they rely on routines. Below are three working checklists, a fast-fix table for the mistakes that cause most rework, and a clear path for readers who want to go further.

The Pre-Weld Checklist

Most welding faults are born before the arc is struck. Run this list every morning and again whenever conditions change:

• Confirm the governing code and current WPS/PQR are on hand and match the joint in front of you.
• Verify base metal grade, thickness and heat number against the drawing, and remove mill scale, oil, paint or coating.
• Inspect fit-up: root opening, root face, joint angle and alignment within the WPS tolerance.
• Check consumables: low-hydrogen electrodes from a heated rod oven, clean wire properly spooled, full gas cylinders.
• Confirm shielding gas flow rate and hose condition; long hoses bleed pressure.
• Set required preheat and verify your thermometer or temp stick reads true.
• Machine check: calibrate current, clean or replace the contact tip, inspect the liner, set amperage, voltage and wire feed to the WPS range.
• Clear the area: no drafts blowing across the joint, adequate ventilation, dry floor, good lighting.
• Confirm position and access. If you cannot reach the root comfortably, reposition or fixturing comes first, not improvisation.
• PPE on and correct for the process: hood shade, fume protection, gloves, leathers.

The During-Weld Checklist

Quality is maintained in the middle of a pass, not inspected at the end of it:

• Strike the arc on the joint, never on adjacent base metal.
• Hold travel speed. Too fast causes undercut and lack of fusion; too slow causes overlap and burn-through.
• Watch the puddle, not the arc. Bead wetting at the toes tells you what the finished weld will look like.
• Clean between passes: chip the slag, wire brush, remove inclusions before the next pass covers them forever.
• Hold interpass temperature within the WPS range, especially on HSLA and chrome-moly steels.
• Follow the written travel sequence and skip pattern. Distortion is a sequencing problem before it is a metal problem.
• If anything looks wrong, stop. A five-minute stop is cheaper than a five-day repair.

The Post-Weld Visual Inspection Checklist

Do this before the inspector, the crane or the shipper gets there:

• Remove slag completely. You cannot inspect what you cannot see, and neither can anyone else.
• Check surface porosity, pinholes and wormholes along the entire bead, including starts and stops.
• Measure bead width, height and profile against the acceptance criteria. Excessive reinforcement is not free strength.
• Inspect toes and edges for undercut, overlap and cold lap.
• Check for cracks at every start, stop and tie-in. These are their favourite hiding places.
• Look for arc strikes, spatter and damage to adjacent base metal.
• Verify the weld is complete on every side the drawing requires, including the root of full-penetration joints.
• Mark, tag and document anything questionable before it moves to the next station.

Five Common Mistakes and Their One-Line Fixes

Your Next Step: Building a Career on Top of This Knowledge

If this guide changed how you see welds, the natural next move is to formalize what you now know. The American Welding Society offers credentials including Certified Welding Inspector, Certified Welding Supervisor, Certified Welding Educator, Certified Welding Engineer and Certified Robotic Arc Welding, spanning nine industry categories. AWS also lists 39 in-person or virtual welding courses and training resources to get you there.

On the technical side, spend real time inside the code that governs your work. AWS D1.1 touches virtually every role on a structural steel project: engineers, fabricators, contractors, inspectors and educators. Read its crack prevention, preheat and repair clauses alongside your own shop’s procedures, and you will start predicting faults before they happen.

Keep that same seriousness on the health side. NIOSH recommends that welders and metalworkers understand their occupational health risks, and OSHA’s Hazard Communication Standard requires employers to inform and train workers on hazards and protective measures. Reduce fume and gas exposure through the hierarchy of controls: elimination, substitution, engineering controls, administrative controls, then PPE, as described in OSHA’s welding, cutting and brazing guidance.

Welding faults are predictable. Nearly every crack, cavity and incomplete joint in this guide traced back to a machine setting, a stored consumable, a skipped preheat, a draft, or a decision made in a hurry. Treat welding as a controlled process with written rules, measured variables and honest inspection, and the Friday-afternoon weld that used to haunt you becomes just another weld that ships, stays sound, and never calls you at two in the morning.

Frequently Asked Questions

What is the most common welding fault?

Porosity and undercut are the two most frequently encountered welding faults in general fabrication. Porosity appears as gas pockets or pinholes caused by contamination, poor shielding or drafts, while undercut is a groove melted into the base metal beside the weld toe and left unfilled.

Why does my weld keep getting porosity?

Recurring porosity almost always traces back to one of four things: moisture, oil or rust on the base metal or consumables; a shielding gas problem such as wrong flow rate, an empty cylinder or a leaking hose; drafts disrupting the gas envelope; or arc blow from poor grounding. Fix the cause, not the symptom — capping over porosity does not solve it.

What is the difference between lack of fusion and lack of penetration?

Lack of fusion means the weld metal did not bond with the base metal or a previous pass, usually along the sidewall or root face. Lack of penetration means the weld did not reach the full thickness of the joint at the root. Both reduce load-bearing cross-section and are rejectable in structural work.

Are all weld cracks rejectable?

Yes. In structural codes such as AWS D1.1, any crack — hot, cold, crater or toe crack — is a rejectable discontinuity with no tolerance allowance. Cracks must be completely removed by grinding or machining, verified (commonly by penetrant testing), and the joint rewelded under a qualified procedure.

How much undercut is allowed in a structural weld?

It depends on the governing code. AWS D1.1 allows small amounts of undercut in static-loaded members (typically up to about 1/32 in or 0.8 mm) but is stricter or fully restrictive for tension members and cyclically loaded connections. Always check the acceptance criteria in the code named in the contract.

What NDT method best finds internal weld faults?

Radiographic testing and ultrasonic testing (including phased array UT) are the primary methods for internal flaws such as porosity, slag inclusions, lack of fusion and lack of penetration. Surface methods like penetrant and magnetic particle testing cannot see inside the weld.

Can welding faults be repaired instead of rejected?

Often yes, but only under the rules of the governing code and with approval from the welding inspector or engineer. The fault must be fully removed, the preparation verified, the repair welded to a qualified procedure, and the repair area re-inspected.

How do welding faults affect worker health and safety?

Faults themselves are a structural risk, but the process that produces them also exposes welders to metal fumes and gases. NIOSH links welding fume exposure to metal fume fever, potential lung cancer, and manganese-related neurological effects, and has identified welder’s anthrax as a rare but deadly occupational disease in metalworkers.