Welding Faults and Defects: Identification, Causes, Prevention, and Repair for Safer Joints

On a bridge retrofit job, a small undercut missed in visual inspection can become a fatigue crack years later. The same applies in shops and field work: knowing welding faults and defects and catching them early is what keeps welds safe, compliant, and economical.

Introduction to Welding Faults and Defects

Last winter, a mid-size fabrication shop finished 40 structural column splices for a warehouse expansion. The welds looked fine from three feet away. Then a third-party inspector gauged a few and found undercut along the toes of nearly a third of the horizontal joints. Not catastrophic, but rejectable under the governing code, ground out and rewelded at the shop’s own cost. Nobody got hurt. Everybody got an education.

That is the reality of welding faults and defects: they rarely announce themselves, and they cost more the longer they hide. This guide helps you identify, prevent and repair them correctly the first time.Weld Discontinuities vs Defects: Standards, Terminology, and Acceptance Criteria

Ask ten welders whether a small porosity cluster is a “defect” and you will get ten different answers. That confusion changes whether a bridge girder ships or gets gouged out and rewelded. The difference between a discontinuity and a defect is not severity alone; it is what the governing code says about that specific imperfection in that specific joint.

Discontinuity vs. Defect: The AWS A3.0 Definition

The welding world’s rulebook for vocabulary is AWS A3.0, Standard Welding Terms and Definitions. Its logic is simple and worth memorizing:

• Discontinuity: any interruption in the normal physical structure or configuration of a weld. Cracks, porosity, undercut, and inclusions are all discontinuities by definition.
• Defect: a discontinuity that exceeds the acceptance criteria of the applicable code or specification. In other words, a rejectable discontinuity.

The AWS Pocket Guide for Visual Examination of Welds makes the point: virtually every real weld contains some discontinuities. Acceptable porosity in a fillet weld is a discontinuity. That same cluster in a CJP groove weld on a seismic column is a defect. The weld metal did not change; the acceptance criteria did.

How AWS D1.1 Decides What Is Acceptable

For structural steel in the United States, the controlling document is AWS D1.1, Structural Welding Code – Steel (currently the 2025 edition). Its visual acceptance criteria live in Table 6.1, organized around one critical question: Is the weld subject to static loading or cyclic loading?

That distinction matters more than beginners realize. A discontinuity perfectly acceptable in a dead-load-only bracket can be rejectable in a fatigue-sensitive crane girder, because cyclic stress turns small stress raisers into crack initiators over time. Under AWS D1.1, limits on undercut, porosity, and underfill tighten considerably when cyclic loading is specified, which is exactly why undercut limits vary so much between jobs.

Then there is the absolute rule: per AWS D1.1, any crack is unacceptable regardless of size, shape, or location — surface-breaking, internal, anywhere. There is no such thing as an acceptable hairline crack. That is why crack prevention and hydrogen control sit at the center of every serious welding procedure, from a routine pipe welding procedure to heavy structural work.

Table 6.1 is also not the only source of acceptance criteria. Fabricators working to API 1104 (pipeline), ASME B31.3 (process piping), ASME Section VIII (pressure vessels), or AWS D1.5 (bridges) will find different limits for the same discontinuity. The qualified AWS D1.1 preview from AWS Publications is a good starting point for understanding how these requirements are structured.

Size, Location, and Orientation: Why Context Decides Everything

Three attributes turn a neutral observation into an engineering judgment:

• Size: a 1/16 in. (1.5 mm) deep undercut may pass under one criterion and fail the next. Depth, width, and length are all measured, usually with a weld gauge.
• Location: porosity at the toe of a fatigue-loaded fillet weld is far more serious than the same porosity buried mid-face of a static bracket. Discontinuities near the root of a CJP groove weld can be catastrophic, because the root carries the highest stress.
• Orientation: a crack running transverse to the weld axis sits under tensile stress and can open and propagate. Slag stringers aligned lengthwise along the weld are far less damaging than transverse ones.

This is also why elongated porosity matters more than scattered round porosity: elongated voids act like notches, and notches are where fatigue cracks start.

Who Decides: The Inspector and the CWI

Acceptance is only as good as the person applying it. The AWS Certified Welding Inspector (CWI) program, governed by AWS QC1, plus companion standards like AWS B5.1 for inspector qualification and personnel schemes such as ASNT SNT-TC-1A, ASNT CP-189, and ISO 9712 for NDT technicians, exist to make sure “acceptable” means the same thing on every shift and every job site.

A CWI does not use judgment alone. They work from the drawing, the contract documents, the applicable code, and the approved AWS inspection documents and AWS welding inspection handbooks. When a welder asks, “Is this porosity okay?” the honest inspector answer is a question back: “Static or cyclic? Which code? What does the WPS say?”

Why This Section Changes How You Weld

Once you internalize that acceptance is a code-driven judgment, your priorities shift. You stop chasing “perfect-looking” welds and start targeting the discontinuities codes actually reject and that actually cause failures in service — cracks, lack of fusion, and poorly controlled profile geometry. Those are the subjects of the next several sections.

Geometric and Profile Welding Faults

While internal flaws can be hidden beneath the surface, the shape of a weld can tell you just as much about its integrity. Geometric and profile faults are the most visible issues a welder will encounter, yet they are also some of the most common to overlook. These discontinuities directly impact fatigue resistance, joint strength and overall compliance. Catching them early through visual examination (VT) and simple gauges will always be far cheaper than a repair.

Undercut

Undercut is a groove melted into the base metal adjacent to the weld toe or root that is left unfilled. This creates a sharp notch, which can drastically reduce fatigue resistance, especially in cyclically loaded structures. Per AWS D1.1:2025, limited undercut may be acceptable depending on static or cyclic loading as outlined in Table 6.1. Excessive undercut, however, is rejectable in most structural applications.

It is often caused by excessive travel speed, an incorrect work angle, or running too high of an amperage for the electrode. For those learning the trade, proper technique plays a larger role than many realize. Guides such as 3/32 E7018 amperage, polarity and how to run it, or 5/32 E7018 amperage, polarity and technique can help dial in these parameters. Prevention starts with control, and if undercut is present, a qualified repair procedure with careful blending will be required. For a more in-depth breakdown, refer to welding defects undercut: causes, acceptance limits, and how to fix it.

Overlap (Cold Lap)

Overlap, often referred to as cold lap, occurs when weld metal rolls over the base metal without properly fusing to it. Unlike excess reinforcement, the issue is fusion, not volume. This creates a hidden toe that acts as a stress riser, weakening the joint without adding strength. It is most prevalent when travel speed is too slow, the electrode angle is incorrect, or insufficient heat is directed to the toes. Techniques covered in resources like push or pull stick welding: which technique is right and why can drastically mitigate this fault.

Underfill

Underfill is the opposite of excess reinforcement. It is a depression in the weld face or root that leaves the joint below the required thickness. This directly reduces the cross-sectional area of the weld, lowering its load-carrying capacity. Common causes include improper weaving, inadequate filler metal deposition, or a lack of attention to joint fit-up. Careful pacing and consistent passes are the keys to prevention.

Excess Reinforcement and Poor Weld Contour

While a small amount of reinforcement can be beneficial, too much can create sharp transitions that promote fatigue cracks. An irregular or “humped” contour can also trap moisture or make downstream processes, such as coatings, far more difficult. Excessive convexity in fillet welds is especially problematic. On the contrary, a highly concave face can mask an undersized throat. As noted by the American Welding Society, profile control is just as critical as penetration for many code applications.

Concave, Convex Fillets and Insufficient Throat

The effective throat is the most important dimension for a fillet weld’s strength. A convex fillet may look large, however, an excessive convexity does not always equate to a larger effective throat and can be rejectable. A concave fillet can be aesthetically pleasing, but may be severely undersized if deposited carelessly.

Insufficient throat is a rejectable fault that compromises shear strength, and it often stems from low amperage, poor joint preparation, or incorrect travel speed. These issues are magnified in out-of-position work, such as overhead stick welding. Before striking an arc, factors such as how to set up a MIG welder and your chosen electrodes in welding should be verified for the job at hand.

Detection, Prevention, and Repair

The first line of defense for all of these faults is consistent VT, paired with proper weld gauges. For high-risk work, AWS inspection resources highlight the importance of measuring size, location and orientation for acceptability. Prevention revolves around a qualified procedure, sound joint prep, stable parameters and operator awareness. With that said, safety must always remain paramount, and proper welding PPE should be in use for any repair.

When a geometric fault is rejectable, remove the affected metal by grinding or gouging until sound metal is reached. Never simply “stack” over it. Reweld in accordance with a qualified procedure, inspect and verify compliance. The goal isn’t to cover imperfections, it’s to build a profile that can withstand the demands of its service. That is what separates a passable weld, from a safe one.

Fusion-Related Welding Faults and Defects

Fusion is the whole point of welding: two separate pieces of metal becoming one. When that bond is incomplete, the result is a fusion-related discontinuity, and it is one of the nastiest families of faults in the trade for a simple reason — the bead often looks perfectly acceptable on the surface while a sharp, planar gap hides underneath. A welder can lay a smooth, uniform-looking pass and still have left a joint that will not carry its design load.

Incomplete fusion, fusion face failure, and incomplete joint penetration (IJP) are fusion discontinuities that can be rejectable depending on the code and acceptance criteria applied to the job. Understanding how they form, how to see them, and how to stop them is what separates a routine welder from someone trusted on structural and pressure work.

The Main Fusion Faults at a Glance

IJP and lack of root fusion deserve special attention because they are penetration problems that behave like fusion problems in service: the joint cross-section is simply smaller than the engineer intended, and the unfused notch at the root becomes a ready-made crack starter.

What Causes Incomplete Fusion

Almost every LOF root cause traces back to one of three things: not enough heat, wrong arc direction, or a dirty and badly prepared joint.

• Heat input. Amps that are too low, travel speed that is too fast, or an electrode diameter that is too large for the position leaves the base metal below its melting point. Settings matter more than most beginners realize, and they should be validated through the MIG welder setup process before production starts. Running a small 3/32 E7018 too cold is a classic recipe for unfused sidewalls; if you are unsure of the range, check the 3/32 E7018 amperage guide.
• Work angle and travel angle. The arc has to be pointed at the joint face, not at the bead you just ran. Dwell time at the sidewall is what melts the edge back. Push and drag produce different penetration profiles, so follow the technique matched to your process, as explained in push vs pull welding technique.
• Joint preparation and fit-up. Too narrow a root opening, too small a bevel angle, excessive root face, or a heavy land all limit the arc’s access to the joint corner. In GMAW, a stick-out that is too long pulls the arc off the sidewall and compounds the problem in out-of-position work.
• Contamination and cleaning. Mill scale, rust, oil, and — on aluminum — the stubborn oxide film that TIG sheet work demands you remove will block fusion no matter how hot the arc is.
• Procedural drift. Skipping interpass cleaning, running beads too close together without washing the edges, or weaving too wide without enough dwell all create interbead lack of fusion. Flux-cored processes hide this well because the slag bridges gaps that the metal never filled.

How Fusion Faults Are Detected

Visual examination (VT) is the first line of defense, and the AWS Pocket Guide for Visual Examination of Welds is the compact reference inspectors carry for discontinuity causes and remedies. A trained eye catches telltale signs: an unfused crease along the sidewall, a bead sitting proud without wetting into the bevel, cold and rounded root edges, or slag that was never fully removed between passes.

VT has a hard limit, though — subsurface fusion flaws are invisible. Radiographic testing shows lack of penetration as a dark line down the centerline, while ultrasonic testing is exceptionally sensitive to planar flaws sitting parallel to the beam. In practice, UT is often used post-modification to detect IJP and trapped slag in CJP welds, which is why repair and rework on structural members and qualified pipe welding procedures so often call for it. Magnetic particle and dye penetrant testing cover surface-breaking fusion flaws when the part is ferritic or non-porous.

Prevention: Build Fusion In, Don’t Inspect It Later

The cheapest fusion fault is the one that never happens. Anchor the work on a qualified WPS and PQR, verify joint dimensions with a bevel gauge before tacking, set amperage from the procedure rather than from habit, and keep travel speed slow enough that the arc is actually melting the joint face. Add interpass cleaning as a non-negotiable step, use the correct polarity and contact tip-to-work distance, and make sure the welder can physically reach the joint root — fixtures, positioners, and joint design changes are legitimate fixes, not luxuries.

Finally, remember that fusion faults do not just fail inspections; they fail structures. A joint that looks perfect but is not bonded is a hidden notch waiting for cyclic load, and that is where safe practice comes full circle — wear your welding PPE during grinding and gouging rework, because chasing fusion defects out of finished welds is where welders take their most avoidable burns and eye injuries.

Cracks: Types, Causes, Hydrogen, and Criticality

If undercut is a warning, a crack is a verdict. Among all welding faults and defects, cracks are the one discontinuity codes refuse to argue about. AWS D1.1 states it plainly: any crack is unacceptable, regardless of its size or location. Not “evaluate it,” not “monitor it.” Remove it and reweld.

The reason is geometry. A pore or a slag pocket has rounded edges that blunt stress. A crack has sharp tips, and a sharp tip concentrates stress so efficiently that it becomes a propagation site under fatigue or overload. That is exactly how a hairline at a weld toe, missed during a rushed visual check, turns into a running crack years after the job was signed off. For the vocabulary inspectors use to describe what they see, the AWS Pocket Guide for Visual Examination of Welds is the standard field reference, and AWS has published clear practical coverage of how weld cracks actually initiate and get found.

Hot Cracks vs. Cold Cracks

Hot cracks form while the metal is still hot, during solidification or just below it. Solidification cracking runs down the centerline of a bead when the shrinking pool pulls apart and no liquid metal is left to heal the tear. The usual triggers are an unfilled crater, a long thin bead, travel speed too fast for the puddle, or a joint that is deep and narrow.

Cold cracks, also called delayed cracks, are harder to deal with because they can appear hours or even days after welding, sometimes after the joint has already passed visual inspection. They need three conditions present at the same time, often called the hydrogen cracking triangle:

• Hydrogen in solution, picked up from rust, mill scale, paint, oil, or moisture in a damp or coated electrode.
• A susceptible microstructure, usually hard, low-ductility martensite formed in the heat-affected zone of rapidly cooled steel.
• Tensile stress, from shrinkage restraint in a heavy joint that is fixtured and cannot move.

Take away any one leg of that triangle and the risk drops sharply. Nearly every crack-prevention rule in the book comes back to it.

A Field Map of Crack Locations

Underbead and HAZ Cracking: The Hydrogen Problem

Underbead cracking sits beside the weld in the heat-affected zone, often hidden beneath the bead and invisible without magnetic particle or penetrant testing. It reliably needs all three ingredients at once: hydrogen in solution, a low-ductility microstructure such as martensite, and high residual stress. Steels with higher carbon or hardenability are the most vulnerable, and thick sections are worst because they cool fastest. A weld toe damaged by undercut behaves much the same way: it is a built-in notch that gives a crack somewhere to start.

Lamellar Tearing: Cracking Below the Weld

Lamellar tearing is not a weld-metal crack. It happens in the base metal beneath a heavy fillet or butt weld, stepping through the plate thickness in terraces. It needs three things as well: through-thickness weakness from flat sulfide or silicate inclusions rolled into the plate, strong shrinkage restraint, and a geometry that loads the short dimension of the plate. Heavy T-joints, box sections, and corner details in thick plate are the classic cases. The fixes are specify through-thickness (“Z-quality”) steel, butter the joint with a soft, ductile weld metal before the heavy passes, and detail the weld to avoid triaxial restraint.

Prevention: Preheat, Low-H Consumables, and PWHT

• Preheat and interpass control. Preheat slows the cooling rate, keeps the HAZ from going fully martensitic, and gives hydrogen time to diffuse out before stress peaks. Holding interpass temperature matters as much as starting preheat.
• Low-hydrogen consumables, stored dry. E7018 electrodes belong in a rod oven, not on a damp bench. Correct current settings, as covered in this 3/32 E7018 amperage guide, keep the arc stable and the slag release clean.
• Control heat input and bead shape. Deep, narrow beads lock in centerline stress. A wider, shallower bead with a lower depth-to-width ratio cracks far less.
• Backfill craters. Hold the arc over the puddle, overlap the last few millimeters, or use the crater function on the machine.
• Follow the qualified procedure. Preheat, interpass, heat input, and any post-weld heat treatment are written into the WPS for a reason. Pipe welding procedures show how tightly this is controlled on critical joints.
• Reduce restraint. Balanced beads, sensible weld sequencing, and avoiding heavy fixturing on joints that must be allowed to shrink.

Post-weld heat treatment, where the procedure calls for it, serves two purposes at once: it relieves residual stress and helps drive out diffusible hydrogen before it can do damage. In practice, the cheapest crack is the one that never forms. A little preheat, a dry electrode, a filled crater, and a sensible bead profile cost minutes; a crack found after installation costs a cut-out, an NDT campaign, and a repaired joint that will never be as good as the original.

Short Note: Because cracks can be hidden, delayed, and stress-driven, critical work should never rely on a visual check alone. Pair the right low-hydrogen electrode selection with NDT when the code, the service loading, or plain common sense calls for it. A rejected crack before shipping is a grind and a re-run. A crack in service is a failure investigation.

Gas Entrapment, Inclusions, and Other Volume Discontinuities

Cracks dominate the horror stories, but volumetric discontinuities are what inspectors run into most days. Porosity and inclusions hide inside the weld where the eye cannot reach, which is exactly why they are so easy to ship past a quick visual check. The AWS pocket guide for visual examination lists them as the classic gas and slag entrapment discontinuities, and each one traces back to a controllable variable: shielding, cleanliness, or arc technique.

Porosity: Gas Trapped as the Pool Freezes

Porosity forms when gas dissolved in the molten metal comes out of solution faster than it can float to the surface. If the pool solidifies first, the bubbles are locked in as spherical cavities or elongated voids. Shop welders meet it in several shapes:

• Scattered porosity: pinholes spread evenly through a bead, usually from moisture, dirty base metal, or a gas mixture that has gone off-spec.
• Aligned or linear porosity: a string of pores running along the weld centerline or the fusion line, often tracing a shielding interruption or an arc that wandered off the joint.
• Elongated porosity: teardrop-shaped voids that show the direction the gas was travelling when the metal froze.
• Piping or wormhole porosity: directional tunnels, common when a contaminant gasses violently or when travel speed outruns gas escape.
• Surface porosity and blowholes: visible craters on or just below the face, frequently caused by oil, rust, paint, mill scale, or a drag on shielding gas coverage.

The usual suspects are broken shielding from wind or a bumped flow meter, an empty or wrong cylinder, excessive flow that creates turbulence, wrong stickout, moisture in a flux coating, and welding over grease or cutting residue. Getting the machine dialed in first removes a surprising share of the problem, which is why a solid MIG welder setup routine matters more than any troubleshooting trick later.

Slag, Tungsten, and Oxide Inclusions

Slag inclusions are flux remnants trapped between passes or under a bead. They come from skipping interpass cleaning, using too little heat to float the slag out, a weave that traps it against the sidewall, or a groove too narrow for the rod. In multipass welds they show up as the dreaded “wagon tracks,” parallel slag lines at the toes of adjacent beads. Technique and cleaning prevent them, and practical flux core wire welding tips focus heavily on brush-out discipline between passes.

Tungsten inclusions belong to GTAW. They occur when the electrode dips into the pool or the arc gap collapses on a contaminated or balled tip. The fix is simple: stop, grind or replace the electrode, and restart. Sheet-metal work makes this more likely, and the settings and habits in the complete TIG welding sheet metal guide cover how to avoid it.

Oxide inclusions are an aluminium problem above all. Aluminium oxide melts near 3,740°F while the base metal melts near 1,220°F, so an unremoved oxide film simply floats as a seam. Wire-brush with a dedicated stainless brush, remove oil, and keep the shielding gas covering the puddle.

Detection, Fatigue Impact, and Repair

Surface-breaking pores and inclusions show up in visual examination; subsurface ones require UT or RT, with penetrant or magnetic particle testing used to confirm open indications after grinding. The structural concern is fatigue: a rounded pore is a stress raiser, and aligned porosity sitting perpendicular to the tensile axis acts like a notch. Discontinuity interpretation fundamentals are well covered in industry discontinuity references and university extension training material.

Repair means grinding or gouging back to sound metal, brushing clean, verifying removal, and rewelding under a qualified procedure such as those used on critical pipe welding procedures. Choose the right consumable dry and stored correctly, since hydrogen and moisture feed both porosity and cracking.

Process-Specific Welding Faults and Practical Examples

Every experienced welder learns this the hard way: fault patterns are not random. Each process has its own signature defects and its own failure “personality.” Switch from stick to flux core without learning those signatures and you will chase the same problems for weeks. Here is what you are most likely to meet with SMAW, GMAW/FCAW, GTAW, and SAW, with operator error separated from procedure issues.

SMAW (Stick Welding): Arc Strikes, Craters, and Slag

Shielded metal arc welding is forgiving on equipment and brutal on technique. The most underappreciated SMAW fault is the arc strike: an accidental arc initiated outside the intended weld zone, usually when a welder “scratches” the rod to start. Each arc strike creates a tiny, rapidly cooled re-melted area with its own heat-affected zone. AWS treats arc strikes as undesirable surface discontinuities because that local rapid cooling can harden the spot and initiate cracking, especially on restrained or high-carbon steels. The fix is discipline: start the arc on the joint or on a run-on tab, never on the base metal. If arc strikes occur, grind them smooth and blend, then examine the area, per the AWS Pocket Guide for Visual Examination of Welds.

Craters are the other classic SMAW end-of-weld fault. An unfilled crater is a shrinkage pipe and a ready-made crack starter, particularly in aluminium. AWS D1.1 requires craters to be filled, with the narrow exception of intermittent fillet welds ending outside their effective length. Back up slightly at the end of the bead, or use a crater-pour technique. Getting your 3/32 E7018 amperage settings right reduces spatter and makes crater control far easier, because the arc stays predictable at termination.

GMAW and FCAW: Burn-Through, Wormholes, and Penetration

Wire processes fail differently. On thin sheet, burn-through dominates, caused by excess heat, slow travel, or a wire feed that is too low. Our burn-through fixes apply the same logic here: reduce amperage, increase travel speed, and widen the heat sink with pads or tack spacing. In FCAW, wormhole porosity (elongated, tail-like gas pockets) usually traces back to moisture in the flux, excessive wind disrupting shielding, or too steep a work angle. The practical countermeasures are covered in our flux core wire welding tips, and they matter on jobsite work where wind is the enemy of shielding gas.

Lack of penetration in groove welds is the classic GMAW procedure fault. Short-circuit transfer in particular can run cold at the root, leaving a gap under the joint. The cause is almost always parameters, not the welder: correct voltage, wire size, stick-out, and travel speed per a qualified procedure. A proper MIG welder setup eliminates most of this fault before the arc is struck.

GTAW: Contamination, Tungsten Inclusions, and Craters

TIG welds are clean and unforgiving. Tungsten inclusions occur when the electrode touches the weld pool or the joint, shedding hard particles into the bead. The remedy is a clean cut on the electrode, correct gas coverage, and a smaller diameter electrode when access is tight. Oxide and dirt inclusions come from poor cleaning of the base metal, especially on aluminium. Craters and crater cracks are common again, and on aluminium, filling the crater with a slight post-flow dwell is essential. Our TIG welding sheet metal guide walks through the settings and common mistakes that produce these exact defects.

SAW: Wagon Tracks, Slag Lines, and High-Speed Cracking

Submerged arc welding looks flawless from the operator’s chair and hides a lot underneath the flux. Wagon tracks are parallel slag lines left along the weld toes when flux fails to drain properly, usually from excessive travel speed, wrong flux chemistry, or voltage that is too low. They are caught by visual inspection and often rejected because they act as crack initiation sites. Slag inclusions at interbead stops and excessive reinforcement from wrong flux depth are also common.

Operator Error or Procedure Issue? A Practical Test

Ask one diagnostic question: does the fault follow the welders, or the job? If three welders produce the same undercut or lack of penetration on the same joint, the WPS and joint design are at fault. Check heat input, preheat, travel angle, and groove geometry before blaming the crew. If the fault follows one welder, it is training and technique: travel angle, work angle, arc length, and travel speed. Even specialized processes carry their own fault patterns; aluminothermic rail welding brings entirely different contamination and porosity risks, yet the same inspection logic applies.

Finally, every grinding, gouging, and rework step is hazard work too. Fumes, UV, and hot spatter demand proper welding PPE, and wire work in particular requires careful attention to MIG welding safety precautions. Knowing which process generates which fault lets you inspect smarter, fix faster, and weld safer.

Prevention, Repair, and Inspection: Practical Tips and Solutions

Every welder eventually meets the same uncomfortable moment: a finished joint that looks fine from three feet away, but reveals undercut, a crack, or a lack of fusion once someone looks closer. The good news is that the overwhelming majority of welding faults are preventable. Prevention, honest inspection, and disciplined repair are three habits that separate reliable shops from ones that keep grinding out the same defect shift after shift.

Prevention Starts Before the Arc Strikes

The cheapest defect to fix is the one you never create. Sound fabrication begins with a qualified WPS (Welding Procedure Specification) backed by a PQR (Procedure Qualification Record). The WPS tells the welder what actually matters: process, filler classification, polarity, heat input range, preheat, interpass temperature, joint geometry, and welding position. Following it is not paperwork for its own sake; it is the practical definition of how cracking and fusion problems are controlled. For a detailed walkthrough, see our guide on pipe welding procedures covering WPS, positions, passes, and inspection.

Cracking prevention boils down to four controllable variables, as reflected in AWS guidance on weld cracks:

• Heat input: Amperage, voltage, and travel speed must stay inside the WPS window. Too much heat widens the HAZ and deepens residual stress; too little heat invites cold lap and lack of fusion.
• Hydrogen: Use low-hydrogen consumables, store and re-bake them properly, and keep base metal free of oil, moisture, paint, and rust.
• Residual stress: Manage weld sequence, avoid excessive restraint, use stitch or back-step patterns, and control fit-up gaps that force welders to “bridge” heavy joints.
• Joint design and fit-up: Excessive root openings, mismatched edges, and high depth-to-width ratios all raise the risk of cracking and poor fusion.

Day-to-day technique matters just as much as paperwork. Clean the joint immediately before welding, control travel angle, dial in correct parameters, and verify settings with reliable equipment. Our resources on setting up a MIG welder step by step, 3/32 E7018 amperage and polarity, and flux core wire welding tips show how proper parameter control directly reduces porosity, spatter, and fusion faults. Sound equipment helps too; bad welding cable connectors cause resistance heating and unstable arcs that quietly degrade weld quality.

Inspection: VT First, Then the Right NDT

Visual inspection (VT) is the primary inspection method, and for good reason: studies and AWS training material consistently show that visual examination finds the largest share of rejectable discontinuities, from undercut and overlap to cracks and profile problems, often before any instrument is picked up. AWS publishes an excellent Pocket Guide for Visual Examination of Welds covering discontinuity causes and remedies that every inspector and welder should keep handy.

Good VT habits include proper lighting, removing slag and spatter first, using weld gauges, and checking profile against the acceptance criteria of the governing code, such as AWS D1.1/D1.1M. When VT alone cannot see inside the weld, NDT fills the gap:

Repair Done Right

Repairs fail most often because the original defect was never fully removed. Welding over a crack is not a repair; it merely buries the crack tip, where it keeps growing. The correct sequence is simple but strict: remove the defect completely by grinding or arc-air gouging, verify removal by MT or PT where cracks were involved, clean the cavity thoroughly, confirm sound base metal, and reweld following a qualified repair procedure with the required preheat and controls. If you are fixing profile faults such as undercut, our piece on undercut causes, acceptance limits, and fixes covers proper blending and re-welding.

Every repair step involves hot metal, arcs, and grinding sparks, so welding PPE and fire-watch discipline are non-negotiable. OSHA’s welding, cutting, and brazing standards, including 1910.252 and 1910.254, set the baseline requirements for hot work, ventilation, and fire protection.

Prevention beats repair every time it is budgeted for. When shops invest in qualified procedures, disciplined parameter control, and honest visual inspection backed by the right NDT, they do not just pass audits; they send safer joints into service and spend far less time fixing them later.

Safety, Health Hazards, and Regulatory Compliance

Behind every weld fault lies a very real human cost. A rejected joint means rework, but a missed crack or a porosity-filled root can lead to injury, structural failure, or environmental harm. For this reason, quality and safety are not two separate programs on a job site. They are one and the same mindset, and the precautions taken to prevent defects often double as life-saving measures.

The Air Welders Breathe: Fumes and Gases

Welding fume is far from a single substance. Its makeup changes based on the base metal, coatings, consumables, and welding process. Some of its most concerning components are worth paying extra attention to:

• Hexavalent chromium (Cr(VI)) — Released when welding stainless steel and chromium-bearing alloys, Cr(VI) is a known carcinogen. It is regulated by OSHA under 29 CFR 1910.1026, with a permissible exposure limit of just 5 µg/m³ as an 8-hour time-weighted average.
• Manganese — Common in steel fume and linked to neurological effects with chronic overexposure.
• Carbon monoxide (CO) — Forms when the arc breaks down shielding gases and certain coatings.
• Ozone and nitrogen oxides — Created when UV radiation from the arc reacts with the surrounding air.
• Inert asphyxiants — Gases like argon, helium, and CO₂ displace oxygen. Argon is heavier than air, so it can pool in pits, trenches, and tank bottoms, creating silent, oxygen-deficient atmospheres.

To combat these hazards, OSHA’s welding, cutting, and brazing standards require ventilation to keep exposures below established limits. These rules, outlined in 1910.252 for general industry and 1910.254 for arc welding, follow a strict hierarchy. As reinforced by OSHA’s welding safety guidance and 1910.252(b)(2), natural ventilation may be enough for outdoor work or large, open shops. However, enclosed or poorly ventilated spaces demand local exhaust ventilation at the arc, and when that is impractical, air-supplied respirators must be used.

Confined Spaces: Where the Risk Multiplies

Tanks, vessels, pipe runs, ship compartments, and pits are some of the deadliest work environments for welders. Under 1910.252(b)(2)(iii), welding, cutting, and brazing in confined spaces must be done with adequate ventilation to prevent the accumulation of toxic fumes and oxygen deficiency. This calls for continuous air monitoring, ventilation ducting that reaches the deepest point, and a trained attendant outside the space. Even gas-shielded processes can be risky, which is why practices outlined in MIG welding safety precautions are crucial when working near openings or in low-lying areas.

Radiation, Burns, and Electric Shock

The arc emits intense ultraviolet and infrared radiation. UV can cause arc flash, essentially a sunburn of the cornea and skin, while IR radiates heat through gaps in clothing. Proper filter lens shades, leathers, and full skin coverage are non-negotiable. For shade charts and gear specifics, check out the complete guide to welding PPE and what welders actually wear. Thermal burns remain the most common welding injury, and their prevention is covered in this guide to weld burns.

Electric shock must also be taken seriously at any amperage. Wet gloves, damp floors, cramped positions, and poorly maintained equipment all increase the risk. Insulated gloves, dry work areas, properly connected ground clamps, and intact cable connectors and leads are all critical to staying safe.

Fire, Explosion, and Cylinder Safety

OSHA’s 1910.252(a) fire protection rules are the backbone of hot-work safety. Key requirements include:

• Remove or protect combustibles within 35 feet of the work. If removal is impossible, shield them with fire-resistant blankets and check for ignition on the far side.
• Have fire extinguishers on hand and know how to use them before striking an arc.
• Assign a fire watcher whenever conditions could lead to anything beyond a minor fire, and keep that watch for at least 30 minutes after welding operations stop to catch smoldering ignition.
• Never weld on containers that held flammable materials until they have been cleaned, vented, and tested.
• Treat oxygen and fuel gas cylinders separately: keep caps on, secure them upright, keep regulators clean, and never allow oil or grease on fittings.

Some jobs carry even more specific hazards. Gas line welding introduces purging and inerting concerns, while thermite rail welding involves molten metal at extreme temperatures in open-air public corridors.

Regulatory Framework and Training

OSHA’s Hazard Communication standard requires employers to train workers on the specific Safety Data Sheet (SDS) for every consumable and coating they encounter. Beyond compliance, good habits go a long way. For instance, a correct MIG setup can reduce spatter and fume, proper flux-core technique helps maintain shielding, and following qualified pipe welding procedures limits rework, which in turn reduces overall exposure.

From undercut acting as a fatigue initiator to a hydrogen crack in a pressure boundary, the faults covered in this guide directly translate to safety consequences. For further reading, the AWS safety and health resources and NIOSH welding fume guidance are excellent tools for building a job-specific safety plan.

Conclusion: Building Quality into Every Weld

When it comes to welding faults and defects, the smallest issue is almost always the most deceptive. A slightly uneven toe, an elongated pore, or a root pass that falls just short can seem harmless in the moment. However, left uncorrected, these minor discontinuities can become stress raisers, leak paths, or crack initiators. Caught early, they only take a few minutes of rework. That gap between a quick grind-and-reweld and a structural failure is the reason to build quality into every weld from the start, rather than trying to inspect it out later.

The Principles That Hold Up Under Pressure

Prevention beats repair, every time. Frontloading your work with a qualified welding procedure, proper preheat, controlled interpass temperature, and clean fit-up will always cost far less than gouging out a rejected weld and rewelding it under production pressure. This is exactly the mindset behind the AWS D1.1 structural welding code and its companion standards: qualify the procedure, verify the welder, and verify the weld.

Know your acceptance criteria before you strike an arc. Not every discontinuity is a defect. What is rejectable on a critical offshore member may be perfectly acceptable on a less critical joint. The code decides, not guesswork. By understanding undercut acceptance limits actually work, you can meet the spec without wasting time chasing harmless marks.

Catch faults while they’re cheap. Visual examination will always be the highest value inspection tool at your disposal. As AWS’s Pocket Guide for Visual Examination of Welds notes, a trained eye can spot the root causes of most discontinuities between passes. A few seconds of inspection at that stage prevents far more rework than any final check ever will.

Treat cracks as absolute. Above all else, no crack is acceptable, at any size, in any code, or in any position. Complete removal back to sound metal, confirmation that the crack is gone, and rewelding under a qualified procedure are the only acceptable paths. As AWS’s inspection guidance on weld cracks points out, repairing only the visible tip leaves the rest of the crack in the HAZ, waiting to propagate.

Control hydrogen and restraint together. Preheat, low-hydrogen consumables and their proper handling, moderate heat input, correct peening rules, and post-weld heat treatment where specified all work as one system. Skipping any of these can undo the rest of your crack prevention efforts.

Safety is part of quality. A distracted, uncomfortable, or unprotected welder will not produce consistent, sound welds. Proper PPE, fume control, and process-specific safety practices protect people first, and weld quality as a direct result. OSHA’s welding, cutting, and brazing standards, including 1910.252 and 1910.254, reflect these realities. Even the most common injury matters, so knowing how to treat and prevent weld burns should be basic trade literacy.

A Practical Pre-Weld to Post-Weld Checklist

1. Procedure: Weld to a current, qualified WPS with a supporting PQR. Confirm positions, heat input range, and preheat requirements.
2. Consumables: Use the correct low-hydrogen classification, store and handle per specification, and re-dry as required.
3. Fit-up and prep: Verify bevel, root opening, land, cleanliness, and tack quality. Remove mill scale and moisture from the joint.
4. Parameters: Set amperage, voltage, and travel speed to the procedure. From dialing in 3/32 E7018 rod amperage, to fine-tuning 5/32 E7018 settings, or learning how to set up a MIG welder, correct settings matter.
5. Technique: Maintain proper travel and work angles, control arc length, and remember that push or pull changes your fusion profile and gas coverage.
6. Between passes: Remove all slag, inspect visually, and control interpass temperature.
7. Craters and terminations: Fill or back up every crater. These are common starting points for cracks.
8. Post-weld VT: Check profile, undercut, overlap, porosity, and cracks before moving the assembly. Use gauges, not guesses.
9. NDT where required: Select PT, MT, UT, or RT based on joint criticality and the discontinuities you’re checking for.
10. Repair discipline: Remove defects completely, clean thoroughly, re-inspect the excavation, and reweld strictly per the qualified repair procedure.

A People-First Closing Thought

At its core, welding quality isn’t about codes, gauges, or film. It’s about the people who will ride the bridge, occupy the building, or rely on that pipe years after the crew has gone home. Every preheat check, every crater filled, and every discontinuity corrected is a decision made on their behalf. Know your acceptance criteria, respect the metallurgy, inspect honestly, repair completely, and send your crew home as safe and whole as they arrived. That is the standard worth carrying into every job: build quality into every pass.

Frequently Asked Questions

What is the difference between a welding fault and a welding defect?

A discontinuity (fault) is an interruption in material structure. A defect is a discontinuity that renders the weld unacceptable per applicable codes/standards (size/location/orientation exceed acceptance). All defects are discontinuities; not all discontinuities are defects.

Why are cracks considered the most severe welding defect?

Cracks have sharp tips and act as strong stress concentrators; they propagate under stress and are primary causes of catastrophic failure. AWS D1.1 requires any crack to be removed regardless of size/location.

What causes porosity in welds and how can it be prevented?

Porosity results from gas entrapment during solidification due to contamination, inadequate shielding, moisture, or improper technique. Prevention includes clean surfaces, proper shielding gas flow/coverage, dry consumables, and correct parameters.

How can undercut and overlap be avoided?

Undercut/overlap often stem from excessive travel speed, improper work angle, excessive current, or poor technique. Control parameters, maintain correct angles, ensure adequate filler metal deposition, and follow qualified WPS.

When is preheat/interpass temperature needed to prevent cracks?

Preheat/interpass helps reduce cooling rates and hydrogen diffusion, critical when welding hardenable steels, thick sections, or restrained joints. Requirements are specified in qualified WPS per code/material considerations.

What is incomplete fusion (LOF) and how is it detected?

LOF means weld metal did not fuse to base metal or adjoining beads. Causes include inadequate heat input, improper angles, poor access. VT may detect surface LOF; subsurface LOF often requires UT/RT depending on joint.

What safety controls are required when welding in confined spaces?

Confined spaces require adequate ventilation to prevent toxic gas accumulation/O2 deficiency, hazard assessment, and may require respiratory protection. OSHA 1910.252/1910.146 provisions apply; fire/explosion and air monitoring considerations are critical.

Can you weld over a crack to repair it?

No. The crack must be completely removed by grinding/gouging down to sound metal, followed by cleaning and rewelding using a qualified procedure. Simply welding over does not remove the defect and can lead to further failure.

What are common sources of welding fume hazards?

Fumes/gases vary by process/material: Cr(VI) from stainless/chromate coatings, Mn from many consumables, CO from combustion, O2 displacement by inert gases. Controls include local exhaust ventilation, positioning, substitution, and respiratory protection when needed.

How does AWS D1.1 determine if a discontinuity is acceptable?

AWS D1.1 provides acceptance criteria (e.g. Table 6.1 for visual) based on loading (static vs cyclic), discontinuity type, size, length, spacing, location. Cracks are never acceptable; other discontinuities have specific limits.