Overlap Welding Defect: Causes, Detection, and Prevention

The weld passed the eye test — a clean, even bead with a nice crown — until the inspector ran a light along the toe and a thin shadow appeared where metal should have fused. That shadow was an overlap, and it turned a finished joint into a grind-out and rework job. Overlap is one of the quietest weld defects in fabrication: cheap to prevent, expensive to miss, and genuinely dangerous under cyclic load.

The Weld That Looked Perfect Until Someone Shone a Light on It

The bracket came off the table looking like a training poster. A 6 mm fillet with an even ripple pattern, consistent legs on both sides, almost no spatter, and a toe line clean enough that the welder photographed it before setting the part on the rack. Same joint, same machine, same wire as a hundred times that month. Nothing about that bead suggested trouble.

Then the inspector walked past holding a flashlight almost flat against the plate and ran the beam along the toe. A thin dark shadow appeared where the weld metal met the base plate, no wider than a fingernail clipping, running maybe 40 mm before it vanished. She marked it, measured it, and wrote it up. Overlap.

The welder’s first reaction is the one nearly everyone has: that bead looked fine. And it did. That is exactly what makes overlap so uncomfortable. It hides in plain sight, sitting on top of the joint instead of gouging into it, so the weld appears fuller and stronger than it really is. A cold lap or roll-over at the toe reads as a slightly generous bead to an untrained eye and as a rejected part to someone who knows where to point a light.

If you have been handed a nonconformance report for a weld that looked good, this guide is for you. If you write those reports, it is for you too.

The Four Questions This Article Answers

That bracket inspection raises the questions readers bring to this topic again and again:

• What actually is overlap? Not “weld metal in the wrong place,” but a specific geometry with a specific failure mechanism underneath it.
• Why does it happen when the bead looks fine? Because the visible surface says almost nothing about whether the metal underneath bonded to the base plate.
• Is it genuinely rejectable? Sometimes yes, sometimes no. It depends on the code, the quality level, and the service the part will see.
• How do I stop it? With process settings, technique, and cleaning habits far cheaper than rework.

Overlap is formally described as the protrusion of weld metal beyond the weld toe or weld root, a definition laid out in welding reference material and repeated across inspection standards. On the floor you will also hear cold lap, roll-over, or cold roll. Those terms are largely nonstandard and get used loosely for several different things, which is part of why arguments break out. One more complication worth flagging up front: when an overlap is very small, it can be genuinely hard to see without good light and a trained eye.

Who This Is Written For

Welders who want to stop producing the defect instead of arguing about it. Engineers and supervisors writing procedures and training new hands. QC inspectors who need language for a nonconformance that survives pushback. Fabricators pricing work and discovering rework eats the margin. And hobbyists making the jump from “it holds” to code work, where a joint is judged against a document rather than a feeling. Where a standard says something specific, you get the clause and a link. Where shop practice diverges from the paperwork, you will hear about that too.

What’s Ahead

We will start with the geometry of an overlap and why the unfused boundary under that lip matters more than the shape you can see, then work through why heavy, cold weld pools produce it, including the confusing contradiction in the literature about whether too much or too little current is to blame. From there we compare overlap against undercut, lack of fusion, and excessive convexity. The middle of the guide covers real inspection workflow, from the flashlight test to profile gauges to magnetic particle testing, then the acceptance criteria in ISO 5817 and AWS D1.1 that decide pass or fail. After that: why overlap shortens fatigue life and invites corrosion, a process-by-process prevention playbook, how to repair one properly, and the quality system habits that keep it from coming back.

If your shop runs aluminum, some of the same fusion and cleaning principles apply, and the technique side is covered well in this guide to aluminum MIG welding, where travel angle and cleaning discipline make an outsized difference.

First, though, we need to be precise about what that shadow along the toe actually is, because the difference between a cosmetic blemish and a crack starter comes down to a few tenths of a millimeter of unfused interface.

What Is an Overlap Welding Defect? Definition, Geometry, and the Cold Lap Confusion

Strip away the shop-floor language and overlap has a plain definition: it is an imperfection at the toe or root of a weld where molten metal flows onto the surface of the parent metal and solidifies without ever fusing to it. The same thing can happen over a previously deposited bead. You end up with a lip of weld metal resting on top of cold, unmelted steel — material that looks joined but is only touching. Overlap shows up in both fillet and butt welds, which is one reason it appears in so many different fabrication contexts rather than being a quirk of one joint type, as described in this welding technology reference on overlap.

The Geometry: A Lip, an Interface, and a Sharp Edge

Cut an overlapped weld through the cross-section and the picture is consistent. The weld metal folds over the plate edge like a wave curling on a beach. Directly beneath that overhang is a boundary that never reached melting temperature — the two surfaces are in contact but metallurgically separate.

This is why the unfused boundary matters far more than the shape. A porosity void is a rounded cavity, and stress flows around it reasonably well. An overlap interface is planar, extends into the joint at a shallow angle, and terminates in a sharp edge. That combination is textbook stress concentration territory, which is exactly the point made in this review of geometric shape imperfections. The lip is the visible problem; the sharp, unfused plane underneath is the real one.

Cold Lap, Roll-Over, Cold Roll: Sorting Out the Names

Different regions and codes use different words for the same physical condition, and the overlap is genuine — pardon the pun.

• Cold lap is the most common shop term, especially in pipe and pressure work. It emphasises that the metal was cold or unfused when the next layer arrived.
• Roll-over and cold roll tend to describe the profile aspect — the rounded, overhanging edge that rolls onto the base metal without wetting it.
• AWS treatment is the awkward part. Overlap is not always given its own numeric acceptance row. Instead, it gets caught either as incomplete fusion, because fusion genuinely is missing at the interface, or as a weld profile violation, because profile rules require a smooth transition into the parent metal and an overhanging lip is not smooth.

So the answer to “what do I call this?” depends on the code book open on the desk. The physical feature does not change; the clause it violates does.

Where Overlap Actually Appears

Overlap loves gravity and loves a heavy pool. The usual suspects:

• Fillet weld toes, particularly the lower leg in the horizontal-vertical position, where the puddle runs downhill onto cold plate.
• Groove weld caps, right at the parent metal line where the last bead met an unmelted surface.
• The root side of pipe welds, where a starved keyhole or a rushed root leaves metal lapping over the bore instead of fusing through.
• Restart points, where the arc re-establishes on a cold crater edge.
• Wide weave beads, where the arc lingers mid-weave and the puddle reaches the toe with almost no heat left to give.

Aluminium deserves its own mention here, because the tenacious oxide layer raises the fusion threshold and makes cold lap unusually easy to produce. If you work with it, the prep advice in how to prep aluminium right is directly relevant to preventing it, and the transfer-mode discussion in this aluminium MIG welding guide explains why the pool behaves the way it does.

Why Overlap Does Not Make the Weld Bigger

Here is the part that catches people out. Fillet weld size is not the widest measurement you can find on the surface. It is defined by the largest isosceles right triangle that can be inscribed inside the actual weld cross-section — the convention laid out in AWS reference material such as the AWS welding handbook. Metal sitting outside that triangle adds weight, consumables, and distortion risk. It adds nothing to strength.

A joint that measures 10 mm across the face may only contain a 7 mm effective triangle once the rolled-over lip is discounted. That is how an overlap can pass a casual “looks big enough” check while the throat is quietly undersized, a point defect guides commonly illustrate with cross-section diagrams, including this Kobelco weld defect reference.

Why Overlap Happens: The Root Causes Behind a Cold, Heavy Weld Pool

Almost every overlap comes down to one physical situation: a pool of molten metal sitting on parent metal that is too cold to fuse, with nothing driving heat into the toe ahead of it. Amperage, angle, mill scale, and position are all just levers that create or worsen that condition. Group the levers into three layers and diagnosis gets fast.

Layer 1: Process Parameters That Build Too Much Pool

On a fillet weld, too much current does one thing very well: it deposits metal faster than the arc can wet the toe ahead of it. The pool balloons, gravity pulls the lower leg down and out, and the leading edge rolls onto base metal the arc never had time to heat. Excess wire feed speed in GMAW and FCAW produces the same result, since the wire melts off faster than travel can distribute it.

Travel speed is the parameter welders overlook most often. Move too slowly and metal piles up in front of the arc, so the toe is buried under liquid instead of being washed by it. The pool grows, the arc’s penetrating force is spent under the deposit rather than at the fusion line, and the bead edge freezes as a lap instead of a wet-in. Kobelco’s defect guide lists precisely this cluster: too low amperage, too slow electrode manipulation, too short an arc or too low arc voltage, and inappropriate travel and work angles.

Arc length and voltage matter because they set the shape and force of the arc cone. A long arc spreads heat over a wide area, flattens the pool, and loses the directed plasma jet that pushes into the corner. A short arc concentrates energy and wets better, provided the voltage is matched to the wire and the transfer mode.

The Too-Hot vs. Too-Cold Contradiction

Read enough literature and you hit a wall: some sources blame excessive current, others blame low current. Both are right, because both end in the same place. High current builds an oversized pool that dulls the arc’s ability to reach the toe. Low current produces a sluggish, cold pool that never wets the sidewall at all. Pipe welding training material, for instance, lists low current as a cause and recommends higher travel speed, higher current, a smaller electrode diameter, or a change in electrode angle.

Weld Australia’s review of geometric imperfections frames it as manipulation failure: a large, cold pool that gravity reshapes before solidification when the electrode or gun is handled poorly. That is the unifying mechanism — pool volume outrunning arc force and heat at the fusion boundary.

Layer 2: Technique, Position, and Gravity

Vertical-up and horizontal-vertical work the defect hardest. In the 2F and 4F positions the lower toe is under the full weight of the pool, and any pause or excess weave lets metal sag over unfused plate. Wide weaves compound it: the arc lingers in the middle, the edges cool, and the puddle rolls out at the shelf line.

Angle is the other quiet culprit. Point the electrode or gun so the stream pushes molten metal ahead of the arc and you have painted liquid onto cold steel. Semi-automatic work adds push-versus-drag errors — an aggressive push angle flattens the bead and skates it over the toe rather than digging in.

Layer 3: Material Condition, Wire, and Setup

Contamination is where many “welder error” verdicts turn out to be wrong. Tightly adherent oxides and mill scale prevent wetting, so the bead sits on top instead of fusing — the classic on aluminum, where the oxide film melts far hotter than the base metal, and why proper aluminum cleaner and surface prep matters so much. Rust, oil, paint, and zinc coatings do the same job.

Add wire and gas problems: wet or rusty wire, a kinked liner causing erratic feed, excessive stick-out, or shielding loss from drafts. Magnetic arc blow can shove the pool sideways onto a cold toe. Poor fit-up creates a gap-dependent joint with an awkward, narrow shelf that no reasonable angle can wet.

Cause	Mechanism	First Adjustment
Current / WFS too high	Pool volume exceeds arc reach at toe	Reduce amps or WFS, re-tune voltage
Current too low	Cold, sluggish pool never wets sidewall	Raise current, smaller electrode
Travel too slow	Metal accumulates ahead of arc	Increase travel speed
Long arc / low voltage	Diffuse arc, no directed force	Tighten arc length, match voltage
Bad work/travel angle	Pool pushed onto unfused plate	Square up, correct push/drag
Mill scale, oxide, oil	No wetting or fusion at interface	Grind to bare metal before welding
Poor fit-up	Awkward toe geometry	Fix gap and alignment

Overlap vs. Undercut, Lack of Fusion, and Excessive Convexity: Telling Them Apart

Almost every overlap argument on a shop floor starts the same way: “That’s not overlap, that’s undercut.” Or convexity. Or just a heavy cap. These features crowd the same real estate, the weld toe and the fusion boundary, and they are not cousins. Different mechanisms, different inspection routes, different code posture. Mislabel the feature and you either reject a sound weld or ship a crack starter.

Undercut: A Groove Melted and Left Unfilled

Undercut is a groove melted into the parent metal along the toe or root that the weld metal never filled back in. It is a loss of base metal, the exact opposite of overlap’s excess. Usual drivers: an arc that is too long, current set too high, travel speed outrunning the puddle, or poor dwell at the edges of a weave, where the sides get melted and then abandoned.

The bigger split is code posture. Most standards tolerate some undercut. ISO 5817 caps it at roughly 0.5 mm at the most stringent quality level and up to 1 mm at the moderate level for material thicker than 3 mm. Overlap rarely gets that courtesy. It is usually prohibited or tightly capped rather than measured against a generous allowance.

Lack of Fusion: Same Failure, Different Address

Incomplete fusion and lack of fusion live inside the joint or at a boundary: sidewall, interpass line, root. The cause is insufficient heat input to melt the mating surface, or a barrier that stops wetting, such as mill scale, rust, oil, paint, or a tenacious oxide film. Aluminum is the classic offender, and the prep advice in this guide to aluminum cleaner for welding applies directly, because an uncleaned oxide layer will refuse to fuse no matter how pretty the bead looks.

Most codes attack this head-on. AWS D1.1 Table 8.1 requires complete fusion between adjacent layers of weld metal and between weld metal and base metal, and treats incomplete fusion as unacceptable. Overlap is the surface-breaking cousin of that same failure: metal sat on a cold surface instead of fusing to it. That is why inspectors working under fusion-based criteria can reject overlap without reaching for a dedicated overlap clause. The practical difference is visibility. Lack of fusion hides; overlap announces itself.

Excessive Convexity: A Profile Problem With a Familiar Notch

Excessive convexity is too much metal piled on a fillet or cap, crowned instead of flat or slightly concave. Like overlap, it builds a sharp notch at the toe, and in a multi-pass groove it interferes with the next pass: slag traps against the shoulder and the following bead bridges instead of fusing. The mechanism is geometric, not a fusion failure.

That distinction drives the language. Convexity limits are usually numerical, a maximum height expressed as a function of weld size, while overlap is often prohibited outright or limited to a fraction of the weld width. One is a measurable profile tolerance. The other is a fusion defect wearing a profile defect’s clothes.

Defect	What it looks like	Primary mechanism	Typical detection	General code posture
Overlap	Overhanging lip at the toe, sharp shadow line beneath	Molten metal flows onto cold, unfused base metal	Visual testing, PT/MT for confirmation	Prohibited or tightly limited (ISO: level D only)
Undercut	Groove or channel melted into parent metal	Arc too long, current too high, travel too fast	Visual, profile gauge	Limited allowance at most quality levels
Lack of fusion	Often invisible; line or plane at sidewall or interpass	Insufficient heat input or surface contamination	UT, RT, macro section; MT if surface-breaking	Not permitted, complete fusion required
Excessive convexity	Rounded, humped bead profile	Excess deposition, slow travel, poor manipulation	Visual, fillet weld gauge	Numeric maximum, varies by code

Why Inspectors Argue at the Fillet Toe

A proud reinforcement pass and a shallow overlap can look identical in a photograph. The tiebreaker is contact, not shape. Run a low-angle light along the toe: overlap throws a thin, continuous dark shadow line where the lip stands off the base metal. Lay a profile gauge or fillet weld gauge across the toe and it will rock or show a gap under the overhang instead of sitting flush. Inspection guidance agrees on where to look, namely the lower leg of fillet welds, especially in the horizontal-vertical position, and the final or reinforcement pass of groove welds. That is exactly where these arguments start.

If the feature is small and genuinely ambiguous, a light touch with a grinder settles it. Convexity blends into a smooth, fused transition. Overlap opens a line, peels a sliver of metal, or reveals a bright unfused face underneath. Document the as-welded shape first, though, because the profile is the evidence.

How Overlap Is Detected and Measured in the Real World

Overlap rarely announces itself. It sits at the toe as a thin shadow or a slight overhanging lip, sometimes as a bead that looks almost proud of the joint. The good news is that overlap is usually a surface-breaking imperfection, which makes it one of the more findable defects in fabrication, provided the inspection is done properly with the right light and the right number.

Start With Visual Testing: It Carries the Load

Visual testing (VT) catches overlap in most real cases because the unfused lip breaks the surface at the weld toe or root. That shapes the inspection strategy: before anyone rolls out a magnetic particle yoke or books a radiographer, the weld has to be cleaned and examined carefully.

Preparation matters more than most people expect. Wire brush or grind away spatter, silica islands, flux residue, and loose scale. You cannot evaluate a toe profile you cannot see. Once the weld is clean, set up low-angle lighting, holding a torch or inspection light almost parallel to the surface rather than overhead. Raking light exaggerates height differences, which is exactly what an overhanging lip creates. A 2x to 5x magnifier helps on small fillets and thin-wall work.

The practical check is the shadow test. Run the light along the toe. A properly fused toe transitions smoothly into the parent metal. An overlap throws a thin, dark line or shadow under the lip because the metal has rolled over and left a gap beneath it. That shadow is the tell. It needs no gauge and is the single most useful habit a welder can build before calling for inspection.

Measuring What You Found

Finding overlap is the easy half. Expressing its size in a way that matches the acceptance criteria is where things get fiddly. For fillet welds, a fillet weld gauge or profile gauge is the standard toolkit. The fillet gauge checks leg length and throat; the profile gauge, with its wire fingers or contoured comb, replicates the actual cross-section so you can see how much metal sits outside the theoretical triangle. A pit or undercut gauge with a pointed blade can quantify a lip, though it is awkward on a curved toe.

The real difficulty is resolution. A marginal overlap may be only a fraction of a millimeter proud of the surface, sitting next to mill scale of similar tone and spatter of similar height. The boundary between a fused, slightly convex toe and an unfused lip is not always crisp to the eye or the gauge. That is why the shadow test and a magnifier often outperform a quick gauge reading, and why experienced inspectors photograph the condition before deciding.

Measurement is usually expressed relative to the weld, not as a flat millimeter number. Under ISO 5817, the overlap limit h is compared against 0.2 times the weld width b, so the same absolute overhang can pass on a wide cap and fail on a narrow one. Record the weld width, record the overhang, and do the arithmetic rather than eyeballing it. The ISO 5817 standard is the reference to work from, and a good ISO 5817 imperfection guide helps translate the table into shop-floor language.

When Eyes Are Not Enough: Surface and Volumetric NDT

Visual testing gets you most of the way, but some overlaps hide in geometry that is hard to see, such as a tight root, a restart point, or a toe tucked into a corner. Magnetic particle testing (MT) is the natural next step on ferromagnetic steel. It is excellent at surface-breaking linear indications, and an overlap lip behaves like a linear indication because the unfused interface interrupts the magnetic field. Penetrant testing (PT) serves the same role on austenitic stainless and aluminum, where MT does not apply. Both are fast and sensitive to tight, toe-hugging features that a light might miss under scale.

Volumetric methods are a different story. Radiography is poor at a tight, planar overlap oriented roughly parallel to the beam. The radiation passes along the unfused plane rather than through a thickness change, so there is little contrast to record. This limitation is covered in Kobelco’s weld defect reference and in general reviews of geometric shape imperfections. Ultrasonic testing does better with planar flaws when the probe angle is right, but toe geometry can make coupling and interpretation awkward. On overlap, VT and MT carry most of the load, and an RT film that looks clean is not proof of fusion.

Documenting the Finding

A nonconformance that says “overlap at toe” is nearly useless three weeks later when the joint is painted and the welder has moved on. Good documentation is specific and repeatable.

• Photograph with scale. Put a steel rule, a gauge, or a calibrated reference in frame. A close-up without scale proves nothing about size.
• Record location along the weld. Use a tape measure from a defined datum, or a clock position on pipe work, so the repair crew can find the spot.
• Note the weld and joint ID. Include weld number, drawing reference, spool or member mark, and the WPS in use.
• State the measurement and the criterion. “Overhang approximately 1.2 mm on a 6 mm wide cap, exceeds 0.2b at quality level C” tells the reader exactly why it was rejected.
• Record the disposition. Repair, grind and blend, or accept by concession, and note who approved it.

When the base metal is aluminum or a specialty alloy, surface prep and cleaning before inspection follow their own rules, because the same contamination that hides a defect also causes it. The prep principles in this guide to aluminum cleaner for welding apply to inspection as much as to welding.

Finally, remember what the inspection is deciding. Overlap is a discontinuity until the applicable acceptance criteria say otherwise. Detecting it well, measuring it consistently, and documenting it clearly turns a shadow at the toe into a defensible engineering decision rather than an argument.

Acceptance Criteria: What AWS, ISO, and Other Codes Actually Allow

Most welders learn acceptance criteria backwards. They spot a defect first, then reach for a code to decide whether it passes. Experienced inspectors work the other way around: they establish the applicable standard, quality level, and project requirements before they look closely at the weld. Only then does a discontinuity become a defect.

Quality levels exist for a practical reason. A warehouse rack and a crane boom do not carry the same risk, so they should not carry the same acceptance threshold. Requiring every joint to meet aerospace-grade soundness would drive cost and rework without improving safety. Accepting every joint to the loosest level would put people and structures at risk. The code sets the menu; the designer or project specification picks the level.

ISO 5817: A Narrow Allowance at Level D Only

ISO 5817 classifies overlap as imperfection reference 506 in ISO 6520-1. The rule is blunt: overlap is not permitted at quality levels B or C. At quality level D it is allowed only when the height of the overlap, h, does not exceed 0.2b, where b is the weld width. A wide cap can tolerate a slightly taller lip than a narrow bead, but the allowance remains small.

ISO 5817 quality level	Overlap permitted?	Limit
B	No	Not permitted
C	No	Not permitted
D	Yes, within limits	h ≤ 0.2b

The standard also separates short imperfections from systematic ones. A short imperfection is a total length not exceeding 25 mm in any 100 mm of weld, or 25 percent of the length for welds shorter than 100 mm. Systematic imperfections are repeated along the weld and are permitted only at quality level D. This ISO 5817 imperfection guide is useful for seeing how those categories are applied, and Annex B adds further requirements for welds subject to fatigue.

That last point is where many quality plans go wrong. Level D is not a blanket pass. If the joint is cyclically loaded, the fatigue annex can tighten the rules considerably. Quality level selection should happen before production starts, based on loading, service conditions, and the consequences of failure, not after a nonconformance appears.

AWS D1.1: Caught by Fusion or Profile, Not a Numeric Row

AWS D1.1 does not include a neat numeric overlap row in its visual acceptance table. Instead, overlap is caught by two broader requirements. The first is complete fusion: the weld must fuse with the base metal and with previous passes. An unfused lip is a failure of that requirement. The second is weld profile. Table 8.1 points to the profile rules, which require smooth transitions and reject unacceptable overlap at the weld toe.

On the shop floor, this means a D1.1 inspector often writes “incomplete fusion” or “unacceptable weld profile” rather than “overlap.” The distinction matters during repair. Grinding the profile smooth may satisfy the profile clause, but it does not restore fusion if the lip was never bonded to the base metal underneath.

Other Codes and the Contract Trap

Some standards are more permissive than D1.1. AWS D9.1 for sheet metal, for example, accepts conditions that would be rejected on heavier structural work. That reflects thinner material, lower loads, and different service demands. It is a fit-for-purpose argument, not permission to ship a poor profile.

The final authority is usually the project specification. Contract documents, drawing notes, and client quality plans can be stricter than the base code. They may prohibit overlap outright, require a higher quality level, or mandate additional inspection. Always confirm the acceptance criteria in writing before welding, and make sure the welder, inspector, and project engineer are reading the same document.

The practical takeaway is simple. ISO 5817 allows a small overlap only at level D and prohibits it at B and C. AWS D1.1 catches it through fusion and profile requirements. Other codes vary, but none of them turn overlap into a feature. Confirm the quality level, confirm the contract, and judge the weld against both.

Why Overlap Matters: Fatigue, Corrosion, Coating Failure, and Fitness for Service

An overlap can pass a drawing check and still be the most dangerous feature on a joint. Acceptance criteria decide whether to repair it. Service conditions decide why that matters. The same small lip can be a nuisance on a static bracket and a crack starter on a vibrating frame.

The notch at the toe: a crack starter in plain sight

Look at a fillet weld toe under a magnifier. An overlap is weld metal rolled over the surface without fusing, leaving an overhanging lip and an unfused interface beneath it (OpenWA). That interface is planar, sharp, and located where stress flow turns from base metal into weld. Geometric imperfections like this concentrate stress far more aggressively than a rounded pore.

A pore has a smooth boundary. An overlap is closer to a pre-existing crack. One heavy load may only cause local yielding, but thousands of cycles can open a fatigue crack from the sharp tip with little visible warning.

Static strength versus cyclic life

That is why a static proof test can mislead. A bracket loaded a few times may show no problem with a rejectable overlap. A machine frame under constant vibration is different: damage accumulates at the stress raiser with every cycle, and the toe is already the highest-stress location.

ISO 5817 notes that quality level selection should account for mode of stressing, service conditions, temperature, environment, and consequences of failure. Annex B adds requirements for welds subject to fatigue. In plain language, a table value is not a risk assessment. A level that suits a guarded bracket may be wrong for a cyclically loaded crane boom.

The hidden size problem: effective throat and misleading profiles

Fillet size is not the outermost metal. It is the largest right triangle that fits inside the actual cross-section. Overlap adds a lip outside that triangle: extra weight, no extra throat. A visual check can be fooled because the weld looks bigger than it is. The inspector sees a wide leg; the load path sees a shorter throat. Calculated stress rises and the real safety margin becomes unknown.

Corrosion and coating: the crevice you cannot clean

The lip creates a crevice. Moisture, chlorides, and oxygen-concentration cells collect there; stainless steel suffers because the tight gap starves the passive film of oxygen. Carbon steel gets crevice attack and early coating breakdown. Blasting and painting cannot reliably reach under the lip. Paint bridges it, leaving a void and a sharp edge that chips first. Once coated, the overlap is hidden from the visual inspections that might have caught the corrosion.

On aluminum, where cleaning and oxide removal decide whether the weld wets, a folded toe traps moisture where even good aluminum cleaning and preparation cannot reach.

Pressure, leaks, and hidden paths

At the root of a tank or pipe weld, a cold lap can be a leak path. It may pass an initial test, then weep after thermal cycling opens the interface. In food or high-purity service, even a non-leaking crevice is a contamination trap. This is a cleanability and containment issue as much as a strength issue.

Three scenarios, three risk profiles

Trailer frame. Road vibration loads the bracket toe again and again. An overlap is a fatigue crack starter, and a static overload check proves nothing. Remove and reweld; do not blend and paint.

Stainless food-contact tank. The risk is crevice corrosion and trapped product. Passivation cannot reach the gap, and cleaning chemicals may not either. A root leak adds containment risk. The repair must restore a smooth, cleanable surface.

Cyclically loaded crane component. Highest consequence. The joint already demands strict quality and fatigue-aware detailing. A toe overlap is a crack initiation site on a component whose failure can injure people. Inspection, repair, and documentation all need to be tighter.

Fitness for service: when the weld is already in service

If replacement is impractical, an engineer can assess whether a known overlap is tolerable for the remaining life. That may require advanced NDT, fatigue or fracture analysis, and operating limits. It documents acceptable risk; it does not make the flaw disappear.

The consequence-based view

Treat overlap as a fusion and profile problem whose consequences change with service. On cyclic joints, call it a crack starter. In corrosive service, call it a crevice. In tanks and piping, call it a potential leak path. Inspect before coating, set the quality level by consequence, and never let a wide-looking bead convince you the throat is adequate.

Prevention Playbook: Fixing Overlap Process by Process

Overlap is rarely caused by one bad setting. It is usually a pool-control problem sitting on top of a fusion problem: too much molten metal in the wrong place, moving too slowly, over base metal the arc never properly wet. That is why the fix has to match the process. A change that helps a short-circuit GMAW bead can make a SAW pass worse. Use the guidance below as a starting point, then confirm every change against your qualified procedure.

SMAW: Keep the Arc Close and the Bead Lean

For stick welding, prevention centers on using amperage within the electrode manufacturer’s range, controlling arc length, and holding work and travel angles that keep the arc force pointed at the joint. Too long an arc spreads heat, loses directional force, and lets the pool roll over the toe. Keep the arc tight enough to hear a crisp crackle, not a soft hiss.

For a fillet, a work angle near 45 degrees splits heat between both plates; a slight drag angle helps direct the pool. Keep weaves narrow and pause briefly at each edge so the toe wets without piling. If the pool is large and cold, lower the current or increase travel speed. If you are moving into stick work, this practical look at why SMAW still wins covers the fundamentals that make arc control easier to learn.

GMAW and FCAW: Match Wire Feed to Voltage

Wire feed speed and voltage must be balanced. Too much wire for the voltage produces a stubbing, cold, heavy pool. Too little wire with excessive voltage gives a long arc that sprays metal onto cold base metal without fusing it. Set the pair from the procedure, then judge the bead: a convex, ropey deposit with a shadow at the toe means the pool is outrunning the arc.

Travel angle matters. A severe push angle can push molten metal ahead of the arc; a drag angle generally gives better directional control, though the right choice depends on transfer mode and position. Inductance or arc-control settings on short-circuit transfer affect puddle fluidity and penetration, so use them to stabilize the pool rather than to compensate for poor fit-up. Short-circuit transfer on thick plate in the vertical-up position is an overlap factory: low heat input, heavy deposition, and gravity all work against fusion. Switch to a spray or pulsed mode where the procedure allows, or reduce deposition and use stringers. The Kobelco defect guide is a useful visual reference for distinguishing overlap from other profile faults.

SAW: Control Deposition Before It Controls You

Submerged arc welding deposits metal fast, which is exactly why overlap appears at high travel speeds and heavy passes. Match current, voltage, and travel speed so the bead profile stays flat to slightly convex without a rolled edge. Excessive voltage widens the arc and can produce a shallow, wide cap with poor toe fusion; too slow a travel speed simply piles metal. Keep flux clean and dry, and remember that a deep flux burden can hide the toe until the slag is removed.

GTAW: Do Not Overfill the Cap

GTAW has limited arc force compared with other processes, so heavy single-pass caps are risky. Keep current and travel speed matched to plate thickness, add filler at the leading edge of the pool rather than dropping it into the arc, and avoid feeding too much rod at once. A large, slow-moving pool can roll over the toe before the base metal has a chance to fuse. When in doubt, use a smaller bead and more passes.

Pipe Welding: Watch the 5G Lower Toe

In the horizontal-vertical position, gravity pulls the pool toward the lower toe. Control heat input, keep the root and cap passes lean, and adjust electrode angle so the arc force drives metal into the joint rather than over the shelf. Pipe training material recommends reducing electrode diameter or changing the electrode angle when arc force is pushing molten metal onto unfused base metal; the OpenWA pipe welding chapter explains that mechanism clearly.

Universal Levers That Work in Every Process

• Clean to bare metal. Remove mill scale, rust, oil, paint, and galvanizing. Oxides and scale prevent wetting and are a direct cause of overlap.
• Fix fit-up first. A gap that is too tight or an uneven root creates an awkward toe that no travel angle can rescue.
• Use preheat where required. It improves fusion and lets the pool wet out instead of freezing on cold plate.
• Sequence the beads. Balance heat input and avoid burying a cold toe under the next pass.
• Choose stringers over wide weaves on critical joints; narrow beads are easier to keep fused.

Geometric imperfections such as overlap often trace back to a combination of travel speed, deposition rate, and joint geometry, as this review of geometric shape imperfections explains.

Quick Troubleshooting Reference

Symptom	Likely change
Heavy, convex bead with a lazy pool	Lower current or wire feed speed; increase travel speed
Toe shadow on one side of a fillet	Adjust work angle toward the cold toe; clean the surface; reduce pool size
Overlap in vertical-up short-circuit GMAW	Reduce deposition; use stringers or change transfer mode
Rolled edge on a SAW cap	Increase travel speed; review voltage and current balance
Overlap at a pipe lower toe	Change electrode angle or diameter; reduce heat input

Repairing Overlap: Grinding, Rewelding, and Getting the Sign-Off

Once an overlap has been written up, the conversation shifts from “what is it” to “what do we do with it.” This is where well-meaning welders get into trouble, because the shop-floor instinct is to run a grinder over the lip, smooth it out, and call the job done. On a cosmetic part that might pass. On code work it usually does not.

The reason is simple. An overlap is not a shape problem with a fusion problem hiding underneath. It is a fusion problem that happens to be visible as a shape. Grinding away the overhanging lip does nothing to the cold, unfused interface beneath it unless you grind past that interface entirely.

What the Codes Actually Expect

The standard repair position is removal and replacement of the defective portion, not reshaping. Iowa DOT’s guidance on weld discontinuities states this directly: overlap and incomplete fusion are corrected by removing and replacing the defective portion of weld. The same logic appears in documents covering excessive convexity, with one crucial difference. Convexity may be corrected by grinding to a smooth profile because the metal is fused and only the contour is wrong. Overlap requires removal of the unfused condition itself. You are cutting out a plane of non-fusion, not sanding a bump.

That distinction drives the whole repair sequence. AWS D1.1 requires repairs to be made with a qualified welding procedure and, where the code demands it, by a qualified welder working within the limits of their qualification. A repair is a weld like any other: it gets a WPS, a qualified hand, and inspection against the same acceptance criteria as the original joint.

The Practical Repair Sequence

1. Map the extent. Clean the area and use low-angle light to find where the lip starts and stops. Overlap often runs further than the first shadow suggests, especially along a horizontal-vertical toe. Mark the boundaries with soapstone.
2. Remove to sound metal. Grinding works for shallow, accessible overlap. For deeper sections, air-carbon-arc gouging is faster and gets under the lip without smearing it. Either way, the excavated surface must be smooth and fully fused, with no remnant of the original interface left behind.
3. Verify the excavation. Where the procedure or inspector requires it, confirm the cavity is clean with penetrant or magnetic particle testing. A linear indication at the bottom of a ground-out groove means you stopped too soon.
4. Reweld. Deposit with the qualified procedure, matching the original joint configuration. Base metal removed during repair must be compensated with additional weld metal to restore full size. You cannot grind a groove into the throat and leave it shallow.
5. Re-inspect. Same method, same criteria, same documentation. A repair that has not been re-inspected has not been completed.

When Blending Alone Is Defensible

There is a narrow window where grinding without rewelding is acceptable. It requires three things at once: the overlap is shallow, the remaining profile still meets the drawing size and the applicable quality level, and the code in use explicitly allows profile correction by grinding. The exposed surface must then be confirmed fused, not assumed. If the excavation leaves the throat or leg below the required dimension, you are now welding whether you planned to be or not. On vertical-up repair work, check that your electrode and technique suit the position; this all position welding rod guide is a useful refresher before you start.

The Two Ways Repairs Fail

Failure one: grinding the throat away. The welder chases the lip until it disappears and takes the joint’s strength with it. The profile looks excellent. The fillet is undersized. That is a dimensional nonconformance created by the fix for another nonconformance.

Failure two: welding over the lip. A pass is laid on top of the unfused overhang, the defect gets buried, and the surface now looks perfect. Visual inspection will never find it again. That buried planar flaw is exactly the fatigue initiation site you were trying to eliminate, now screened from every future visual check.

Documentation and Repair Rates

Record each repair with the weld ID, location along the joint, length, removal method, whether PT or MT was used, the WPS reference, the welder ID, and the re-inspection result. Photograph it with a scale in frame. Then watch the repair rate itself. Repeated thermal cycles from gouging and rewelding on the same joint can degrade toughness in the heat-affected zone, so a joint repaired several times should trigger a procedure review rather than another attempt. A climbing repair rate on one joint type, or one welder, is your earliest warning that a process problem exists upstream. The quality standard you are working to, whether ISO 5817 or a project specification, is the yardstick for sign-off, not the inspector’s eye.

Common Mistakes, Quality System Fixes, and Key Takeaways

Most overlap problems are not welder problems. They are system problems that surface at the arc, and the same handful of mistakes show up in shops of every size.

The Mistakes That Keep Overlap Alive

Judging a weld by appearance alone. Overlap hides at the toe, in a shadow line a casual glance reads as a crisp edge. Signing off on looks instead of a proper low-angle visual examination approves exactly what an inspector will find later. Visual inspection is the baseline method and is required on production welds under AWS D1.1, with RT and UT specified only when contract documents call for them — but “visual” means a real examination with good light, not a walk-by.

Blaming the welder when the cause is upstream. Mill scale, oil, galvanising, paint, a joint gap that left a shelf at the toe, arc blow from a bad ground: no technique fixes any of that. If the plate arrived contaminated, the welder inherited the defect.

Adjusting amperage without touching travel speed. Turn the heat up and the welder slows, so the pool gets heavier. Turn it down and the arc loses the force to wet the toe. Both roads end at the same lip; travel speed, angle, and arc force have to move together.

Chasing a number from a forum. “Set it at 180 and drag” from a stranger with a different power source is not a procedure. The only number that counts is the one your procedure qualification supports and the specification accepts.

Grinding overlap and calling it repair. Removing the visible lip does not remove the unfused interface beneath it unless you grind past it to sound metal with a smooth transition. A blended-over cold lap is a buried planar flaw with a thinner throat — worse than what you started with.

Treating a permissive code as permission. Sheet-metal standards such as AWS D9.1 are more tolerant of profile imperfections. That tolerance is a fit-for-purpose judgement for light-gauge, low-stress work, not licence to ship a poor profile on a cyclically loaded structure.

Quality System Countermeasures That Actually Work

Individual vigilance fades by Friday afternoon. Systems do not.

• Written WPS and qualified welders. A procedure stating current range, voltage, travel speed, position, and cleaning requirements gives you something to audit. Without it, every fix is folklore.
• First-piece inspection every shift. Ten minutes catches a changed liner, a rolled drive roll, or a swapped gas cylinder before it becomes a shift of rework.
• A checklist that names the code and quality level. “Weld looks good” is not a criterion; ISO 5817 level C or AWS D1.1 Table 8.1 is. That table permits limited undersize on fillet welds — up to 1/16 in for welds from 1/8 in to 3/16 in, provided the undersize portion does not exceed 10 percent of weld length — while overlap is generally rejected as incomplete fusion or a profile violation.
• Calibrated gauges. A worn fillet gauge reading a false leg size is how undersized throats get approved.
• A photo-based nonconformance log. Every rejection gets a photo with a scale reference, a location along the weld, and the parameters at the time. Patterns appear within weeks — usually one position, one machine, or one fit-up station.
• Periodic macro cross-sections. Cut, polish, and etch a production sample now and then. A macro of overlap beside a good toe teaches more than any lecture, and confirms that what you believe is fusion actually is.

Cleaning deserves its own line item. Aluminium and stainless punish poor preparation faster than carbon steel, so the same discipline applies to filler and surface prep — the principles in this aluminum cleaner for welding guide carry straight over to any shop running both materials. For heavy removal during repair, understanding air arc welding and carbon arc gouging matters, because a gouged cavity left undressed is just a new place for lack of fusion to start.

Train the Eye, Not Just the Hand

Teach the shadow test — a low-angle light along the toe until the lip casts a distinct line — and make it part of every self-check. Teach toe wetting explicitly, so the welder understands the goal is a concave, feather-edged transition, not a rounded bead sitting on the plate. And let welders see a macro cut from their own work. Once someone has looked at a crack-like planar gap under a lip at 10x, they never unsee it.

Keep the reference material honest, too. The same condition can be an acceptable discontinuity on one connection and a rejectable defect on another, depending on thickness, loading, and acceptance table limits. That is why the checklist names the code and quality level instead of relying on memory.

Key Takeaways

Overlap is two problems wearing one coat: a fusion failure at the toe and a profile violation on top of it.

1. Check cleaning. Bare metal at the toe — no scale, oil, or coating.
2. Check pool size. Too much metal and the arc cannot reach the fusion boundary.
3. Check arc force and angle. A pushing angle floods the toe; the right work angle wets it.
4. Check travel speed. Slow travel piles metal, fast travel starves the toe.
5. Confirm against the specification. Code, quality level, and contract documents decide acceptance — not the drawing alone, and never a forum post.

Clean it, control the pool, aim the arc, move at the right speed, and measure the result against the standard you agreed to before the first arc was struck.

Frequently Asked Questions

Is overlap in welding always a defect?

No. Overlap is a discontinuity, and it only becomes a defect when it exceeds the acceptance criteria of the applicable code or specification. ISO 5817, for example, permits limited overlap at quality level D but not at levels C or B, while AWS D1.1 work typically rejects it through the complete-fusion or weld-profile requirements. Always check the project specification before deciding.

What is the difference between overlap and cold lap?

They usually describe the same visible condition, but the terminology matters. Overlap is the standard term for weld metal protruding beyond the weld toe or root without fusing to the base metal, while cold lap is a nonstandard or shop term that AWS-linked guidance often folds into incomplete fusion. Because codes treat incomplete fusion differently from a profile irregularity, the label can change the acceptance decision.

Does overlap weaken a weld if the bead looks big?

Yes, in the ways that matter most. Fillet weld size is measured by the largest right triangle that fits inside the actual cross-section, so metal overhanging the toe adds weight but not throat. That unfused lip is also a sharp notch at the point of maximum stress, which is why overlap shortens fatigue life even when static strength looks adequate.

Can I just grind overlap off and pass inspection?

Sometimes, but only if grinding removes the unfused interface and leaves a profile that still meets the required size and a smooth transition. If you grind away throat thickness or leave a remnant of the unfused lip, the joint is worse than before. Many codes and inspection guides call for removal and replacement of the defective portion, followed by re-inspection.

What causes overlap on vertical-up fillet welds?

A heavy, slow-freezing pool combined with an electrode or gun angle that pushes molten metal onto cold base metal. Gravity does the rest before the arc can fuse the toe. Reducing pool volume through current and travel speed, tightening your weave, and adjusting the work angle so arc force drives metal into the joint rather than over it are the usual fixes.

Is overlap detectable by radiography?

Poorly, in many cases. Overlap at a toe is a tight, planar, surface-breaking feature, and radiography is weakest on planar defects aligned with the radiation beam. Visual testing, sometimes with magnetic particle or penetrant testing, does most of the work, while ultrasonic methods are better suited to buried planar defects than to a thin toe lip.

Does ISO 5817 allow any overlap?

Yes, at quality level D only, where the overlap height h must not exceed 0.2 times the weld width b. It is not permitted at quality levels C or B. Remember that quality level selection belongs to the designer and should be fixed before production, based on loading, service conditions, and the consequences of failure.

How do I stop overlap from coming back after a repair?

Treat the repair as a symptom and fix the process. Re-check joint cleaning, confirm the WPS parameters against what the welder is actually running, verify travel angle and speed, and inspect the first few inches of the reworked joint before letting production continue. If overlap recurs across multiple welders on the same job, the problem is usually the procedure or the fit-up, not the individual.