Bad Weld Examples: Common Defects, Causes, Fixes, and How to Spot Them

What Counts as a Bad Weld? Separating Discontinuities From Defects

You pull the trigger, run a clean-looking bead, chip the slag, and then your inspector points at a pinhole you never noticed and condemns the whole joint. That moment, repeated on thousands of jobs, is exactly why learning to recognize bad welds matters more than learning to make pretty ones. A bad weld is not defined by how it looks in a photo—it is defined by whether it meets the standard it was built to.

Put simply, a bad weld is any weld that fails to satisfy its specified acceptance criteria. That benchmark may be a structural code like AWS D1.1, a customer workmanship standard, or a fitness-for-purpose evaluation signed off by an engineer. The same bead can be perfectly acceptable on a garden gate and rejected on a bridge. Context decides.

A code-compliant weld and a fit-for-purpose weld are not always the same thing. Engineers may accept a discontinuity when analysis shows the structure will still perform safely.

Discontinuity vs. Defect: The Distinction That Matters

Welding codes draw a sharp line between two words that get used interchangeably on the shop floor. AWS D1.1/D1.1M:2020 defines a discontinuity as an interruption of the typical structure of a material—such as a lack of homogeneity—and states plainly that a discontinuity is not necessarily a defect.

A defect, by contrast, is a discontinuity (or group of discontinuities) that by nature or accumulated effect renders a part unable to meet the minimum applicable acceptance standards. One tiny pore is a discontinuity. Stack thirty of them in a pressure boundary and that accumulated effect turns the weld into a defect.

The stakes are higher than a rejected part. The UK Health and Safety Executive studied 47 real incidents where weld quality contributed to failure in its report RR1215, “When welding goes wrong: learning from past failures.” Crack-like or volumetric flaws showed up in 49% of cases and poor or undersized weld geometry in 45%—often in joints that had already passed a quick visual once-over. Standards exist to remove that guesswork: AWS published its first fusion welding code (Code 1 Part A) in 1928, and AWS D1.1 is now in its 24th edition.

These bad weld examples are usually grouped by where the flaw lives. Across this guide we will work through:

• Surface and profile problems—undercut, overlap, and excessive convexity
• Internal fusion and penetration issues—lack of fusion and incomplete penetration
• Porosity and trapped gas
• Slag and tungsten inclusions
• Cracking—hot, cold, and everything between

Here is the uncomfortable truth. A smooth, evenly rippled cap can still be a reject. A cold lap lurking under the toes, a shallow root, or a buried crack all hide beneath a bead that photographs beautifully. Appearance is a proxy for quality, never a guarantee—which is why visual inspection alone cannot clear a critical joint. The machine side matters too: worn consumables like a degraded contact tip quietly seed defects before the arc even starts.

For a process-specific visual companion, see our guide to good and bad MIG welds, which lays healthy and defective GMAW beads side by side. Building the eye to tell them apart is a core welding practice skill—and it starts long before your first inspection.

Treat the pages ahead as a field guide: the more bad weld examples you can name on sight, the faster you catch them before an inspector does.

The Main Types of Welding Defects: A Visual Field Guide

Think of this as a memorizable taxonomy you can run through in your head at the bench. ESAB’s welding defects guide groups the usual suspects into lack of fusion, porosity, cracks, slag inclusions, undercut, spatter, and distortion. Here, each defect gets a short description, a typical location, and a quick “what it looks like” cue — the sort of thing you can mentally tick off as you walk a finished weld. Building the eye for this takes deliberate repetition, which is why structured welding practice beats random bench time. Naming a defect correctly is half the repair, because each one points to a different root cause.

Porosity

AWS A3.0 defines porosity as cavity-type discontinuities formed by gas entrapment during solidification. Its forms include aligned or linear porosity, cluster porosity, piping porosity (wormhole), and elongated porosity. Piping porosity has a length greater than its width and lies roughly perpendicular to the weld face, while elongated piping porosity typically hides in the face reinforcement of a groove weld or the convex face of a fillet weld. The gas sources vary — moisture, mill scale, paint, rust, and shielding problems — but a worn contact tip or poor gas coverage will leave these holes fast. They rarely announce themselves before a bend test or a radiograph; the surface often looks perfectly fine.

The Visual Checklist

Cracks

Cracks are the family that fails without warning. Hot cracks (solidification cracking) open while the metal is still hot, often down the centreline of a deep, narrow bead. Hydrogen-assisted cold cracks appear hours after cooling, usually in the heat-affected zone or at the toe — sometimes overnight, which is why a part that passed inspection on Friday can crack by Monday. Learn the six positions: longitudinal, transverse, crater, toe, root, and heat-affected-zone cracks. Because tight cracks hide, use magnification and dye penetrant; a bright developer line is often the only visible clue. And never “burn one out” and move on, since it keeps growing under load. For a process-specific version of this list, see our guide to good and bad MIG welds and our overlap welding defect breakdown.

Where Defects Cluster by Joint Type

Defects are not random. AWS Inspection Trends notes that in large single-pass fillet welds the biggest problems are incomplete fusion at the root and slag inclusions, while groove welds suffer incomplete fusion, rope-like weld beads, and slag inclusions. In plain terms, fillet welds lean toward fusion and slag; groove welds add rope-like beads and inclusions on top. Once you know where a joint type tends to fail, you can aim your inspection there first instead of eyeballing the whole part. These tendencies shift with position, wire, and technique, so treat them as starting points rather than rules. Track your own scrap this way and the patterns appear fast.

Bad Weld Examples You Can See: Surface and Profile Defects

Most of the “bad weld examples” people share online are visible ones. You don’t need X-ray vision to spot them—just a good light, a fillet gauge, and the willingness to look past a pretty bead cap. Surface and profile defects matter because they concentrate stress, thin the cross-section, and give corrosion and fatigue a place to start.

Undercut: The Sharp Notch at the Toe

Undercut is a groove melted into the base metal along the weld toe or root, left unfilled by filler metal. It reduces the base metal cross-section and creates a sharp notch—a classic fatigue crack initiation site. Root undercut is worse because it sits on the loaded side of a joint and is easy to miss from above. Run too much current, hold too long an arc, or travel at a steep angle and you’ll carve it in seconds. The fix is posture and heat, not speed alone; our guide to welding practice covers arc-length control in detail.

Overlap and Rollover: The Most Confused Profile Defect

Overlap (also called rollover) is filler metal that rolls over the toe without fusing to the base metal beneath it, creating a mechanical notch. Beginners routinely mix these two up. Undercut removes metal below the surface line; overlap adds metal that never bonded. The practical tell: overlap has no fusion line at its underside, while undercut is a visible groove. We go deeper in our overlap welding defect article.

Both are bad for the same reason: a sharp, unfused notch sits exactly where bending stress peaks.

Convexity, Concavity, and Underfill

Excessive convexity—the tall, “ropey” bead—looks strong but hides cold lap at the toes and wastes filler. Excessive concavity (a hollow face) and underfill (a groove below the surface) both shrink the throat, which is where a fillet weld actually carries load. A slightly convex bead is fine; a ropey one usually means travel is too slow or voltage too low.

Leg Length, Bead Uniformity, and Wandering Beads

A fillet weld’s strength lives in its leg size. Undersized legs and a wrong leg-length ratio (unequal legs on a joint that needs symmetry) quietly cut capacity. Non-uniform width and wandering, crooked beads signal unstable travel speed or poor joint tracking—common in semi-automatic work and a big reason shops move to robotic submerged arc welding for repetitive runs.

Spatter, Arc Strikes, and Craters

Spatter on stainless isn’t just ugly; it’s a corrosion site and a cleanup cost. Stray arc strikes and arc burns leave hardened, crack-prone zones wherever the electrode touched down. Deep craters at the end of a bead are shrinkage voids waiting to crack—never stop and pull out at full heat.

Surface-Breaking Porosity and Pinholes

Pinholes and surface-breaking porosity are the ones inspectors love to condemn. AWS D1.1 Table 6.1 gives the visual acceptance criteria, with specific limits on surface-breaking piping porosity because it can significantly reduce the weld cross-section. One quirk worth knowing: the code has no general visual requirement for all surface porosity—if appearance matters, the owner and contractor must agree on limits. That’s why a written spec beats a verbal “make it look good.”

How to Measure What You See

Inspectors verify leg size with a fillet weld gauge, and gauges can check both minimum and maximum sizes per AWS D1.1. Undercut depth gets measured against the code limit, not eyeballed. If you’re learning to judge your own beads, spend a shift running practice plates and gauging every one—see our welding practice guide.

Bad Weld Examples You Can’t See: Internal Fusion, Penetration, and Inclusion Defects

The most dangerous bad weld is the one that looks perfect. A smooth, evenly rippled cap can sit directly over a crack, a cold lap, or a pocket of trapped slag — and none of it shows until the joint fails under load. That is exactly why visual inspection alone is never enough on critical work, and why inspectors lean on volumetric methods to look inside the metal.

Incomplete Fusion

Incomplete fusion (IF) means the filler metal never formed a metallurgical bond with the base metal or a previous pass. It hides in three favorite spots: the sidewalls of a groove, the boundary between passes (interpass), and the root. Common causes are travel speed too fast, current too low, a poor electrode angle, or simply failing to aim the arc into the corner where the wall meets the previous bead. A disciplined welding practice routine prevents most of it.

Incomplete Joint Penetration

Incomplete joint penetration (IP) is the classic hidden defect in groove welds. On a single-sided weld, the joint never fuses all the way through the root.

Double-groove welds fail at the root for a specific reason: the welder didn’t back gouge deep enough, or the back-gouge geometry had too low a depth-to-width aspect ratio to give the arc adequate access. Too shallow and too narrow, and the electrode simply can’t reach the bottom — so you cap a gap that was never closed.

Slag and Flux Inclusions

Slag inclusions are islands of solidified flux that never floated out. They show up as isolated pockets, long linear stringers, or the parallel “wagon tracks” that trace a weaving pattern. In FCAW and SAW, molten flux can be mechanically entrapped when gun angle, nozzle condition, or travel speed is off — a big reason small consumable problems like a worn contact tip matter more than they look. AWS D1.1 treats slag inclusions as a specific concern alongside cracks, undercut, and porosity in Table 6.1.

Internal Porosity, Tungsten, and Copper Inclusions

Internal porosity is gas that solidified inside the deposit instead of escaping — harmless in small amounts, rejectable when concentrated. TIG welds can carry tungsten inclusions when the electrode dips into the puddle, and copper inclusions can appear when a contact tip or backing strip melts into the weld. Every one of these hides beneath an intact surface, which is why wire-fed processes demand careful setup and upkeep. The mechanics are well documented in general welding defect references.

Internal Cracking

Internal cracks are the most insidious of all. Hydrogen-induced cracking can appear hours after the weld cools; solidification cracks form as the puddle freezes; lamellar tearing pulls apart through the plate thickness. Because they are tight and planar, they open and close under load instead of presenting an easy pocket for an inspection beam to find.

Detection limits decide everything here. The FHWA technical advisory on fracture-critical members recommends both RT and UT for quality-control inspection of groove-weld butt splices, with UT used for owner quality-assurance auditing. That dual approach exists precisely because AWS D1.1 Section 6 requires UT for many cyclically loaded and fracture-critical applications where radiography alone would miss planar defects.

High-deposition processes can industrialize a hidden defect. Automate the arc with robotic submerged arc welding and you can mass-produce sound welds — or reproduce the same incomplete fusion on every part once a single parameter drifts. That is why automated cells still demand scheduled verification, not blind trust.

Same Defect, Different Process: Bad MIG, Flux-Core, TIG, and Stick Welds

Porosity looks like porosity in a photo. But the story behind it—and how you fix it—changes completely depending on which torch you’re holding. Two welders can produce identical pinholes with opposite mistakes: one running voltage too high, the other too low. That’s the trap with bad weld examples: the defect class tells you what failed, but the process tells you why. Read the bead like a signature, and you stop chasing symptoms.

GMAW/MIG: Cold Lap, Burn-Through, and Travel-Angle Traps

MIG is fast and forgiving, which is exactly why its defects hide. Cold lap and overlap—where molten metal rolls onto the base plate without fusing—often come from a travel angle that pushes the puddle instead of washing into the toe. You’ll see the classic overlap welding defect as a smooth, rounded edge that looks fine until it peels under load. Burn-through on thin sheet is the opposite failure: too much voltage, too slow a travel, or a gap you didn’t manage. Lack of fusion usually means voltage too low, letting the arc ride on top instead of biting in. Spatter, worm tracks, and a ropey, convex bead profile trace back to voltage/wire-feed mismatch or a worn contact tip. For a visual baseline, our guide to good and bad MIG welds pairs nicely with this section.

FCAW/Flux-Core: Slag, Wagon Tracks, and Heat Starvation

Flux-core brings its own signatures. Slag inclusion—trapped flux under a bead—happens when you drag too fast. “Wagon tracks” (two parallel lines of trapped slag along the toes) signal insufficient heat and cold tie-in. Incomplete fusion is the same story: not enough current for the thickness. Then there’s porosity, which in some FCAW setups points to too-short electrode extension—increase stick-out, amperage drops, and the pinholes usually vanish. If elongated porosity persists, slightly reduce wire feed speed. Polarity matters too; most gasless flux-core runs electrode-negative (DCEN), and the wrong flux-core polarity kills penetration and bead shape. Check your gasless MIG setup before blaming technique—par for the course in welding defect guides.

GTAW/TIG: Tungsten Inclusions, Arc Wander, and Crater Cracks

TIG rewards cleanliness and punishes shortcuts. Tungsten inclusions appear when the electrode dips into the puddle or when current is excessive for the electrode size—a small ball of grey metal frozen in the bead. Contamination from dirty filler, oil, or mill scale causes ugly floaters and porosity. Arc wander usually means wrong polarity or a fouled electrode; get your polarity right (DCEN for most steel, per our AC or DC for TIG steel breakdown) and the arc stabilizes. On thin material, crater cracks form if you snap off the arc instead of tapering the current. Slow the taper, add a touch of filler as you fade out, and the crater usually closes clean.

SMAW/Stick: Slag Entrapment, Arc Strikes, and Wet Electrodes

Stick welds fail in the field more often than in the booth. Slag entrapment comes from poor restart technique or running before the previous pass cools, and flux-related inclusions show up as glassy pockets at the toes when you whip too far ahead of the puddle. Arc strikes—those little scars outside the joint—create hard, brittle spots that can crack. Porosity here often means one thing: moisture. Damp E7018 or a wet rod low in hydrogen control will off-gas straight into the puddle, and hydrogen embrittlement often follows. Consistency in E7018 amperage and polarity helps, but dry storage is non-negotiable.

The fix always starts with polarity and settings: deliberate practice, changing one variable at a time, beats guessing.

Why Bad Welds Happen: Root Causes From Fit-Up to Technique

Bad welds rarely come from a single mistake. They’re usually the end result of small problems stacking up: a gap nobody clamped, a gas flow rate nobody checked, a contact tip that should have been retired weeks ago. To stop defects for good, you have to move from symptom to cause. Here are the root causes, grouped by where they live.

Material and Joint Preparation

Everything starts before the arc strikes. Mill scale, rust, oil, paint, and moisture all drag hydrogen, carbon, and porosity into the pool, so cleaning is a welding step, not a chore. Poor fit-up is just as damaging:

• Excessive gaps and high-low mismatch force you to bridge air instead of metal.
• Too-small root openings and rough bevels choke penetration.
• Parts designed without fabrication in mind — no torch access, no root room — almost guarantee repairs.

TWI reports that welder skill, weld location or accessibility, and poor fit-up prior to welding top the list of factors affecting repair rates, and poor fit-up is consistently named among the biggest contributors to weld repair rates in fabrication surveys.

Process Parameters and Consumables

Amperage, voltage, wire feed speed, travel speed, polarity, heat input, and interpass temperature all interact. Too much current burns through; too little leaves cold lap. Polarity quietly controls penetration and arc behaviour — a detail worth learning in our guide to DCEN and DCEP explained. Heat input and interpass temperature are quiet culprits too: run too hot between passes and you coarsen the grain and invite cracking. Consumables are chemistry: wet electrodes, the wrong filler, contaminated flux, or the wrong shielding gas change the weld from the inside out, no matter how steady your hand is.

Technique, Equipment, and Environment

Technique is where experience shows — travel angle, work angle, arc length, weave, stringer manipulation, tie-in, and crater fill. Skip the crater fill and you leave a crack starter at the end of every bead. Equipment failure and environment overlap constantly: a worn contact tip, liner drag, gas leaks, a bad ground, or a dirty nozzle cause erratic arcs, and our contact tip maintenance guide covers the usual suspects. Meanwhile, drafts blow shielding gas away, humidity adds hydrogen, and cold steel changes cooling rates and hardness.

Personnel: Skill and Qualification

Untrained or unqualified welders — or welders qualified only in the wrong position — are a root cause in their own right. A welder who’s flawless flat may struggle overhead, so position qualification matters. Skill is built through repetition and feedback, not assumed; see our guide to building real welding skills.

What the HSE Data Says

UK HSE research report RR1215 found the most prevalent causes were inadequate welder supervision (26%), insufficient percentage coverage of inspection (26%), and joints not designed for ease of fabrication. It identified the most common failure modes as fatigue and ductile or brittle failure due to overload of the welds.

Fix the Cause, Not the Symptom

AWS Inspection Trends advises determining the root cause of an unacceptable weld rather than simply judging the welder, starting by checking whether your parameters fall within the ranges listed on the WPS. Replacing a rejected weld without asking why it failed nearly guarantees the same defect next time. Working out whether the trail leads to fit-up, a parameter, or a person is the whole skill — and it’s the same patience that separates good and bad MIG welds. Record what you find, so the next job starts with the fix built in rather than discovered.

The Cost and Consequences of Bad Welds: Money, Safety, and Liability

A bad weld is never a small problem. It’s a bill that shows up three times: once when you repair it, once when the schedule slips, and once when someone gets hurt or the client walks away. Understanding that arithmetic is what turns “pretty enough” into “fit for service.”

The Repair Math Nobody Budgets For

Repair welding is brutally inefficient compared with making the weld correctly the first time. Research cited by the American Institute of Steel Construction (AISC) shows Volkov estimated the labor cost of repairing a 100–200 mm defective region at 1.5–3.5 times the labor cost of making 1 m of the original weld — and that multiplier climbs to 5–10x for particularly difficult welds.

The pain doesn’t stop there. AISC reports a 20–30% chance that any single repair weld will itself require re-repair. That’s how repair cascades start: grind it out, weld it, inspect it, fail it, repeat. Each cycle burns labor, consumables, and time, and each heat cycle risks distorting the very joint you’re trying to save.

Scale that across a project and the numbers get ugly. TWI reports average weld repair rates of 1–3% in oil and gas and power fabrication, with peak rates up to 25% and as high as 55% logged at specific locations. On a 10,000-inch pipeline, even a 3% rate means hundreds of inches reworked.

Then come the hidden line items: re-inspection and NDT (often double-charged when a repair forces a fresh radiograph or ultrasonic scan), scrap on parts that can’t be salvaged, re-qualification of procedures and welders whose coupons failed, and the schedule delay that pushes every downstream trade.

When the Weld Fails for Real

Repair cost is the best case. The worst case is structural failure. The UK Health and Safety Executive’s RR1215 study reviewed weld failure cases and found that in 49% of them the result was complete collapse of the structure, 21% led to injuries, and 11% led to one or more fatalities. Nearly half of those failures didn’t just crack — they brought the structure down.

The Human Cost: Fume, Manganese, and Chromium

Not every bad weld hurts by failing. The process that makes it hurts the person making it. CDC/NIOSH notes that prolonged exposure to welding fume may cause lung damage and cancers of the lung, larynx, and urinary tract, and that prolonged manganese exposure can cause Parkinson’s-like symptoms. Welding stainless and coated steels generates hexavalent chromium; OSHA’s Permissible Exposure Limit for Cr(VI) is just 5 µg/m³ as an 8-hour time-weighted average — a tiny fraction of a grain of sand per cubic meter. Ventilation, fume extraction, and respiratory protection aren’t extras; they’re part of the job, as covered in our guide to planning welding facilities.

The Business Blowback

Failures also wreck reputations. Lost revenue from downtime, warranty repairs out of your own pocket, liquidated damages for late delivery, liability lawsuits, and the quiet death of future work when a client stops calling — all trace back to a joint that wasn’t right. Learning to tell good and bad MIG welds apart on the bench is far cheaper than learning it in a courtroom.

Note: The cheapest weld you’ll ever make is the one done correctly the first time.

How to Inspect for Bad Welds: Visual Checks, Gauges, and NDT

Inspection isn’t a single event at the end of the job — it’s a chain of checks that starts before the first arc and continues long after the last bead. Skip a link and you hand the next person a defect they can’t un-see. Here’s how that chain works in practice.

Before the Arc: Pre-Weld and In-Process Checks

Start with paperwork, not metal. Confirm the Welding Procedure Specification (WPS) matches the joint, material, thickness, and position, and that the welder holds a valid qualification for that scope. Then verify reality against the document: joint prep and bevel angles, root gap, cleanliness, and fit-up. Poor fit-up is one of the quietest causes of lack of penetration and overlap, so measure it rather than eyeball it.

Check tacks — cracks and undercut hiding in a tack will survive into the finished weld. During welding, spot-check parameters: amperage, voltage, wire feed, travel speed, and gas flow. A wandering contact tip or drifting settings can wreck a run that looked fine ten minutes earlier. The point of these checks isn’t just catching bad welds — it’s catching a bad process before it produces a hundred of them.

Visual Inspection: The Cheapest, Most Powerful Tool

Visual testing (VT) finds more real defects than any other method, at almost no cost. But it only works if you do it properly. Clean the weld first — slag, spatter, and oxide hide the very features you’re hunting. Then get serious about lighting: use a good white light and a magnifier, and inspect at the correct angle so you’re seeing surface shadows, not glare. A head torch and a mirror let you reach the hidden side of a fillet.

For sizing, use gauges. A fillet weld gauge measures leg size and throat; an undercut gauge measures depth; a reinforcement gauge checks cap height. Judge results against the applicable code. Under AWS D1.1, Section 6 covers inspection and Table 6.1 lists the visual acceptance criteria, with Tables 6.2 and 6.3 giving ultrasonic acceptance-rejection criteria for statically and cyclically loaded nontubular connections. Always confirm you’re working from the correct edition for the project.

Nondestructive Testing: PT, MT, UT, and RT

When visual inspection isn’t enough, or the code demands more, NDT takes over. Each method sees differently.

• Liquid penetrant (PT) — finds surface-breaking defects like cracks and porosity. Per the FHWA, PT will not detect subsurface defects, so clean surfaces are essential.
• Magnetic particle (MT) — suits surface and near-surface defects in ferromagnetic steel; MT is only useful for defects in or very close to the surface. In the field, use a yoke: the FHWA warns that copper-prod MT can deposit copper into grain boundaries at the contact point and introduce surface cracks.
• Ultrasonic testing (UT) — best for planar internal defects like lack of fusion and tight cracks.
• Radiography (RT) — reveals internal volumetric defects such as slag inclusions, porosity, and lack of penetration, but may miss tight planar cracks that UT catches. RT and UT are complementary, not rivals.

One structural rule matters more than any technique: under typical quality-control structures, welders and operators are not permitted to inspect their own production welds for acceptance. Independence is the point.

Acceptance Criteria and Fitness-for-Purpose

Inspection only means something against a standard. Code compliance asks, “Does this meet D1.1 or the specified standard?” Fitness-for-purpose asks, “Is this safe and adequate for how it will actually be loaded?” The two overlap but aren’t identical — a defect permissible under strict code may still be unacceptable in a fatigue-critical joint, and vice versa.

Inspection is also quality control on your procedures and operators. Defect records are data: the same flaw three times in a row tells you when a procedure needs revising or a welder needs re-qualification. That’s how rework and repair stops being a treadmill. Catch defects like overlap early, and you fix a habit instead of a part.

Fixing Bad Welds and Preventing Them: Repairs, Training, and Practical Tips

Fixing a bad weld is not about hiding it with a hot pass. It is about removing the defect completely, then rebuilding the joint to a qualified procedure. Get that sequence wrong and you create a second, often worse, defect on top of the first.

How to Repair a Defective Weld

Start by locating and defining the defect. A visual check rarely tells the whole story, so confirm length, depth, and type with a method suited to the job — penetrant testing (PT) or magnetic particle testing (MT) for surface-breaking flaws, ultrasonic or radiographic testing for buried ones. Mark the extent clearly so you excavate everything, not just the visible tip.

Then follow a disciplined sequence:

1. Excavate fully. Grind, gouge (carbon-arc or plasma), or machine the defect out to sound metal. Shallow, feathered dishes tend to trap slag and lack of fusion.
2. Use a proper excavation profile. Aim for a width-to-depth ratio that lets the torch reach the root and fill cleanly — typically a U-shaped prep with no sharp corners.
3. Verify removal. Re-test the excavated area, usually with PT or MT, before any wire touches it.
4. Re-weld to a qualified procedure. A repair WPS may need different parameters than the original.
5. Re-inspect. Treat the repair as a new weld and run the same acceptance criteria.

Why Repair Welds Can Be Riskier Than the Original

Repairs concentrate heat in a small, already-restrained area — that raises real risks.

• Heat-input and cooling control. Too little heat input or too fast a cool produces hard, brittle martensite in the heat-affected zone (HAZ). AISC’s The Effect of Repair Welds on Service Performance notes repair welds generally perform satisfactorily, but documents failures caused directly by defective repairs — including HAZ embrittlement from martensite formation when heat input and cooling rate were poorly controlled.
• Restraint and cracking. A repair in a stiff joint can crack as it cools; preheat, interpass control, and sometimes peening help.
• Repeated repairs. Each pass degrades toughness and grain structure. Good practice limits repair cycles.

Prevention Beats Repair

AWS is blunt: it is less costly to prevent defects than to correct them, and TWI’s work on repair statistics shows how repair volume adds up in fabrication. Because welding is a special process under ISO 9001:2015 — its quality cannot be readily verified after the fact — quality depends on process controls, which is why ISO 3834 specifies quality requirements for fusion welding of metallic materials and complements ASME and AWS codes. HSE cites British Standards and ASME IX as among the most commonly used QA/QC documents for welding.

Build prevention into the job:

• Qualified WPS and PQR for every joint.
• Welder qualification matched to process and position.
• First-article and in-process inspection, not just final.
• Cleaning and joint-prep discipline: no oil, rust, or mill scale.
• Parameter discipline — voltage, amperage, travel speed, gas flow.
• Equipment maintenance: contact tips, liners, drive rolls, gas flow.
• A working QMS per ISO 3834 and ISO 9001.
• Reading the WPS as a contract, not a suggestion.

Note: Every re-weld adds thermal cycles. If a joint needs more than one repair, investigate why — repeated repairs degrade toughness and can void assumptions in the original design.

Common Mistakes That Cause Repeat Defects

• Grinding a defect deeper without re-verifying removal.
• Repairing with an unqualified procedure or unqualified welder.
• Chasing a pinhole with extra passes instead of excavating.
• Running parameters from memory instead of the WPS.
• Skipping post-repair inspection because “it looked good.”

Handled properly, a repair returns a joint to code. Handled casually, it becomes the next bad weld example — and the one that fails in service. Keep your welding practice sharp, maintain your shop per good welding facilities planning, and protect yourself with proper protective eyewear while you grind and gouge. Prevention is cheaper than every repair you will ever make.

Bad Weld Examples: Your Final Checklist and Key Takeaways

Every bad weld example we’ve covered, from a pinhead of porosity to a crack hiding beneath a cap pass, shares one thing: it failed its acceptance criteria. Ugly-but-sound is a cosmetic problem. Sound-but-ugly is no problem at all. Unsound-but-pretty is a structural one, and it’s the one that hurts.

Here’s the irony worth remembering. Because weld quality cannot be readily verified after the fact, prevention through process control is the most effective route to weld and weldment quality. You don’t inspect your way into good welds—you build systems that make them, then confirm the system worked.

Your One-Page Bad Weld Checklist

Run this before, during, and after every joint. It’s short by design, so there’s no excuse to skip it. Print it, tape it to your helmet, and treat it as a ritual rather than a formality.

• Know the standard. Read the drawing, code, and acceptance criteria first. If you don’t know what “acceptable” means, you cannot reliably hit it.
• Clean the joint. Oil, rust, mill scale, and moisture are porosity engines. Clean metal is the cheapest quality control there is.
• Verify the WPS. Confirm process, filler, polarity, and settings against the procedure. Polarity errors quietly destroy fusion—see how TIG polarity choices change everything.
• Check fit-up. Gaps, misalignment, and tight root spacing cause incomplete penetration before technique even matters.
• Maintain equipment. Worn contact tips, dirty liners, and drifting gas flow sabotage bead after bead.
• Watch the arc and bead shape. Travel speed, work angle, and profile tell you in real time what no after-the-fact test can.
• Inspect your own work first. Find the pinhole before your inspector does.
• Document defects. Track what fails and where, so patterns—not blame—drive the fix.

Notice what’s missing: none of these steps demands exotic equipment. They demand attention and consistency, which is exactly the point. Quality is a habit, not a lucky bead.

Five Highest-Impact Lessons

1. Hidden defects beat ugly ones. Porosity you can see is often cosmetic; lack of fusion and cracking you can’t see are the ones that fail in service.
2. Root causes beat blame. HSE research found the top quality-related root causes were crack-like or volumetric flaws (49%) and poor geometry or undersized welds (45%)—process problems, not careless welders.
3. Repair costs dwarf prevention. Repair welding is most economical when defects are caught early, especially by inspecting root runs in thick joints before completion. Prevention is cheaper than rework every single time.
4. Inspection is a system, not a single test. Visual checks, gauges, and NDT stack together; no one method catches everything.
5. Training and supervision are the biggest levers. With over 700,000 U.S. workers in welding and allied processes, skill and oversight move quality more than any gadget.

Keep Building the Skill

Every welder who has ever chipped slag off a beautiful bead—or compared good and bad MIG welds side by side—knows the feeling. Competence isn’t about never making a bad weld—it’s about recognizing one fast, understanding why it happened, and correcting it before it ships. That’s a skill, and skills are trained, not wished for.

The welders who improve fastest aren’t the ones with the steadiest hands; they’re the ones who review their own failures honestly and go back to the table with a specific fix in mind.

So practice deliberately. Burn rod on scrap, cut coupons, watch your deliberate practice routines improve, and treat every defect as information. Do that, and bad weld examples stop being embarrassments and start becoming your best teachers. Keep building. The bead you’re proud of next year starts with the one you fix today.

Frequently Asked Questions

What is the most common bad weld defect?

In practice, porosity, incomplete fusion, and slag inclusions top most fabrication lists, with profile defects like undercut and overlap close behind. HSE’s review of real weld failures found crack-like or volumetric flaws in 49% of cases and poor weld geometry or undersized welds in 45%. The answer depends on process, joint type, and position.

Is a pretty weld always a good weld?

No. A weld can have an excellent appearance and still contain a lack of fusion, incomplete penetration, or a tight crack that is invisible on the surface. Appearance is a useful screening signal, but only inspection methods such as ultrasonic or radiographic testing can confirm internal soundness.

What is the difference between a discontinuity and a defect?

AWS defines a discontinuity as any interruption in the typical structure of a material, such as a lack of homogeneity. A defect is a discontinuity, or group of discontinuities, that makes the part unable to meet the applicable acceptance criteria, so every defect is a discontinuity but not every discontinuity is a defect.

Why do so many welds fail with lack of fusion?

Lack of fusion usually comes from insufficient heat input or incorrect technique, such as travel or work angles that let the arc ride the puddle instead of the sidewall. Wrong amperage or wire feed speed, poor fit-up, and excessive travel speed all reduce the energy available to melt the base metal and previous passes.

How much does it cost to repair a bad weld?

Research summarized by AISC found that repairing a 100–200 mm defective region costs about 1.5–3.5 times the labor of making a meter of original weld, and 5–10 times for difficult welds. There is also a 20–30% chance that a repair itself will need to be re-repaired, so prevention is far cheaper than rework.

Can you just weld over a bad weld?

Not safely. Welding over a defect often traps slag, porosity, or lack of fusion beneath the new pass and can make the joint worse. The correct approach is to excavate the defect completely, verify removal with a suitable inspection method, then re-weld to a qualified procedure and re-inspect.

Which inspection method detects internal weld defects?

Ultrasonic testing and radiographic testing are the two main methods for internal defects. UT is better for tight planar defects such as cracks and lack of fusion, while RT is better for volumetric defects such as porosity and slag inclusions, which is why codes sometimes require both.

Are undersized welds dangerous?

Yes. An undersized weld has less cross-sectional area to carry load, so it can fail by overload or fatigue even if the weld metal itself is sound. HSE found poor or undersized weld geometry in 45% of reviewed weld failure cases, making it one of the leading quality deficiencies.

Authoritative Sources & Further Reading

• https://www.hse.gov.uk/research/rrhtm/rr1215.htm
• https://www.hse.gov.uk/Research/assets/docs/rr1215.pdf
• https://www.twi-global.com/media-and-events/insights/repair-statistics-of-manufacturing-and-fabrication-welding
• https://www.aisc.org/media/aq0jqzzn/the-effect-of-repair-welds-on-service-performance.pdf
• https://www.cdc.gov/niosh/welding/about/index.html
• https://www.osha.gov/sites/default/files/publications/OSHA_FS-3647_WELDING.pdf
• https://www.aws.org/magazines-and-media/inspection-trends/it-aug-25-the-answer-is
• https://www.fhwa.dot.gov/bridge/t514011.cfm
• https://link.springer.com/article/10.1007/s40194-019-00822-0
• https://esab.com/al/eur_en/esab-university/articles/welding-defects-guide
• https://www.cdc.gov/niosh/engcontrols/ecd/detail43.html
• https://pubs.aws.org/Download_PDFS/D1_1_2020SecondP_PV.pdf