Lack of Fusion in Welding: Causes, Detection, Prevention, and Repair (Complete Guide)

The Hidden Defect: Why Lack of Fusion Is Welding’s Silent Weak Point

The weld looked perfect. The bead was smooth, the toes tied in cleanly, and the inspector’s flashlight found nothing to flag. No spatter, no undercut, no obvious porosity. The part was signed off, shipped, and installed. Months later, on a vibrating pump base, a hairline crack wandered out of the weld and split the bracket in two. The culprit had been hiding in plain sight the whole time — a thin dark line running along the fusion boundary, a lack of fusion that no amount of squinting would ever reveal.

That is what makes lack of fusion (LOF) welding’s silent weak point. Welders and inspectors are trained to spot surface flaws. Spatter can be chipped off, undercut can be measured, and a wandering bead is obvious to anyone who has run one. But LOF leaves almost no external signature. The weld can look textbook-perfect on the outside while the inside is barely holding on.

At its core, lack of fusion is simple to define: deposited weld metal that fails to bond metallurgically with the base metal or with a previous weld pass. The arc melted the filler, the puddle froze, and everything looked fine — but the two surfaces never truly joined. There is contact, not cohesion. And bonded or not, the joint is carrying load. As the ESAB welding defects guide makes clear, an incomplete bond is a defect even when the surface looks flawless.

Those unfused interfaces behave like built-in stress concentrators. Instead of flowing smoothly through the weld, stress piles up at the sharp edges of the unwelded gap. That makes LOF a prime crack-initiation site, and the danger climbs sharply under cyclic loading — the push-pull of vibration, traffic, pressure cycles, or repeated thermal expansion. A static joint might tolerate a small flaw for years. A fatigue-loaded joint may not tolerate it for weeks.

Compare that with a cosmetic defect. Spatter, arc strikes, and a slightly convex bead are easy to see and easy to judge. LOF is the opposite: it hides at the root, along the sidewall, or between passes, where only a destructive test or volumetric inspection can reach it. That is exactly why the defect survives visual acceptance and then shows up as a fracture in service.

LOF is also more common than most people assume. The American Welding Society lists incomplete fusion alongside porosity, cracks, and undercut as one of the most frequently encountered weld discontinuities in inspection work (see the AWS Inspection Trends discussion). One fabricator running flux-cored arc welding reported reject rates of roughly 20% for fillet welds and about 10% for groove welds, with incomplete fusion at the root and slag inclusions topping the defect list. Those numbers are not a small quality hiccup — they are a rework bill and a safety concern at the same time.

This guide untangles the whole picture. We will define LOF precisely and separate it from its close cousin, lack of penetration. We will trace the root causes, from low current and wrong welding polarity to a worn contact tip, dirty plate, and sloppy fit-up. We will walk through detection methods — visual, radiographic, ultrasonic, and phased array — and see what AWS D1.1, ASME, and ISO 5817 actually accept. Then we cover why it matters in fatigue and fracture terms, how to prevent it with disciplined technique and WPS control, how each process behaves, and how to repair a defect properly the first time.

You do not need to be a certified inspector to take this seriously. Whether you are a hobbyist laying your first fillet — building real welding skills one bead at a time — or a fabricator pushing thousands of parts, understanding LOF is the difference between a weld that looks done and a weld that is done.

What Exactly Is Lack of Fusion? Types, Terminology, and How It Differs From Lack of Penetration

Lack of fusion (LOF) is a discontinuity where the weld metal fails to bond metallurgically with the base metal or with a previously deposited weld pass. The two surfaces meet, but their atoms never joined — no shared grain structure, no genuine connection. That missing bond is what makes LOF so deceptive: the bead can look smooth and the profile perfect while the joint is only partially fused.

Clearing Up the Terminology

Inspection codes treat incomplete fusion as a broad family defined by both cause and appearance. Inside that family sit three distinct defects: lack of fusion, incomplete joint penetration (also called lack of penetration, or LOP), and overlap. Knowing which label applies changes everything downstream — the acceptance call, the root cause, and the repair strategy.

The words themselves cause friction. Some shops use “incomplete fusion” as the umbrella term; others say “lack of fusion” for the same idea. AWS D1.1 treats lack of fusion and incomplete joint penetration as separate items, while ISO 5817 folds fusion defects into its own quality levels — as the ESAB welding defects guide stresses, matching the term to the defect comes first.

Overlap is a bonding failure too, but it leaves a surface defect rather than a buried one: the weld metal spills over the joint toe without fusing to the base metal. See our guide to the overlap welding defect.

The Three Types of Lack of Fusion

Lack of fusion shows up in three predictable places, and each points to a different root cause:

• Sidewall lack of fusion — the weld fails to bond to the groove face. It usually traces back to low heat input, a poor travel angle, or heavy mill scale on the plate edge. Correct polarity and amperage help here.
• Inter-run (interpass) lack of fusion — a new pass fails to fuse to the one beneath it. Fast travel speed and insufficient overlap between beads are common culprits.
• Root lack of fusion — the root pass rides over the joint root without melting into it, often because the root gap is too tight or current is too low.

All three are hidden-fusion failures — the metal is there, the bond is not. That is why a bead can sail through a casual visual check and still be rejected by a bend test or ultrasonic inspection. Building the feel to avoid them takes deliberate practice, as we cover in our welding practice guide, and the same fusion gaps show up clearly in our guide to good and bad MIG welds.

How LOF Differs From Lack of Penetration

The confusion between LOF and LOP makes sense, because both are fusion-related and both read as sharp, linear signals on a radiograph. The difference is location and mechanism. Lack of penetration occurs when the weld bead does not extend fully through the joint thickness. It is a depth problem — the root is never reached, so the effective throat shrinks, which directly lowers static strength and fatigue life.

Lack of fusion is an interface problem. The bead may reach full depth, yet still fail to bond to a sidewall, a prior pass, or the root face. So LOP lives along the root line for the joint’s full length, while LOF hides at whatever interface went cold.

Why Terminology Matters

Radiographers group these together as linear indications — defined as any indication longer than three times its width, a category that includes cracks, lack of penetration, lack of fusion, and elongated slag inclusions, per this NRC reference document. That shared category is exactly why the names get muddled. But the codes treat them separately, with different acceptance limits and different repair expectations — as recent AWS inspection guidance highlights.

Root Causes: Why Lack of Fusion Happens in the First Place

Lack of fusion rarely announces itself with a bang. It sneaks in quietly, the result of a handful of small conditions lining up along the same weld pass. Understanding why it forms is the difference between chasing defects forever and controlling them at the source.

At its core, fusion happens when molten filler and molten base metal merge into one continuous pool of liquid metal. Three things must be true at once: enough heat to melt the base metal, surfaces clean enough for the liquid metal to wet them, and an arc that physically reaches the joint long enough to do its job. Break any one, and you open the door to lack of fusion (LOF).

That’s why LOF is so often a process-control problem, not just welder error. A welder can have flawless hand skills and still produce LOF if the machine is set wrong, the fit-up is off, or the procedure wasn’t respected. ESAB’s welding defects guide identifies low current and heat input, excessive travel speed, incorrect torch or electrode angle, long arc length, and surface contamination like rust, scale, paint, and oil as the primary causes of lack of fusion. Notice how many of those sit outside the welder’s wrists.

Parameter Causes: When the Machine Undercuts the Weld

Amperage is the master variable. Too little current means the arc can’t melt the base metal, so the filler lays down on a cold surface and never bonds. Excessive travel speed does the same — running the arc away before the metal melts and merges.

Long arc length scatters heat over a wider area, dropping energy density. The deposit looks wide and flat, but the sides may never have melted. Wrong polarity throws off where heat lands, which is why polarity deserves real attention — our breakdown of TIG welding polarity (DCEN, DCEP, and AC) explains how the wrong setting starves the workpiece of heat. The same applies to stick and flux-core, where our guide to DCEN vs DCEP polarity for flux core covers how an incorrect reversed setup changes penetration and fusion behavior entirely.

Technique Causes: Where the Arc Actually Points

Even with perfect parameters, technique can ruin fusion. An incorrect torch or electrode angle drives the arc into one plate and away from the other, so the far side never melts. Welding straight down or dragging at the wrong lead angle pushes heat ahead of the puddle and starves the root. Some welders run a lazy stringer with no dwell at the sidewalls, and the toes never tie into the base metal.

A proper weave or slight pause at each sidewall lets heat soak into both edges. Skip that pause and you get the thin, dark interface line that hides until a bend test or radiograph exposes it.

Preparation Causes: Contamination Blocks Wetting

Melted metal won’t wet a dirty surface. Rust, mill scale, oil, paint, and moisture act as barriers between the puddle and base metal. They burn off, spit, and leave a film that stops the two metals from merging. Cleanliness isn’t cosmetic; it’s a fusion requirement. Tight groove faces and narrow root openings make it worse, trapping slag and preventing the arc from reaching the root. Undersized roots contribute to trapped slag, poor penetration, and LOF, so they must be opened up to the WPS-required dimension. When fit-up is off, no amount of skill downstream can recover the bond.

Machine and Fixture Causes: The Overlooked Contributors

Hardware fails quietly too. A worn contact tip lets the wire wander, throwing off arc direction and current transfer — see our guide to contact tip selection and wear. Faulty wire feed causes stubbing or erratic deposition. The wrong electrode for the base metal undermines the arc’s ability to melt. Poor fixturing lets parts shift, opening gaps mid-pass.

Then there’s habit. Welders often set parameters based on what “felt right” last time rather than what the welding procedure specification (WPS) calls for. Inspectors should verify welders work within the acceptable ranges the WPS lists — not just close to them. Building that discipline is a skill in itself, which is why deliberate welding practice pays off. Reading the arc, checking the tip, and honoring the WPS keep fusion where it belongs.

How to Detect Lack of Fusion: Visual Inspection, Radiography, Ultrasonic, and Phased Array

Lack of fusion (LOF) hides where inspectors rarely look: at the interface between weld metal and base metal or a previous pass — a flat, tight gap that can be microns wide yet sealed on the outside. The cap looks smooth, the toes look fine, and under radiation it may barely cast a shadow.

Why Visual Inspection Alone Can’t Confirm It

Visual inspection is fast and cheap, but it only sees surfaces. On its own it can never confirm lack of fusion — it can only flag warning signs. When checking a bead, watch for:

• Rope-like, humped, or convex beads that suggest the arc rode high without melting the sidewalls
• High crowns and narrow stringers that point to low heat input or fast travel
• Visible straight lines or unfused edges along the weld toe
• Uneven toes and irregular fusion boundaries

These clues should trigger a full volumetric inspection. Our guide to good and bad MIG welds covers the same clues, as does our flux core vs. MIG comparison.

Radiography (RT): Good for Volumetric, Weak for Planar

Radiographic testing passes X-rays or gamma rays through the weld onto film or a digital detector. It excels at volumetric defects — porosity, slag, voids. But LOF is planar and oriented through the thickness, so the beam can run nearly parallel to the crack face. A tight, aligned lack of fusion may produce only a faint linear indication or vanish. RT catches LOF only when the beam angle aligns; it misses many tight planar defects.

Ultrasonic Testing (UT): The Preferred Tool for LOF

Ultrasonic testing is generally better than radiography for planar flaws such as lack of fusion, because the defect is oriented through the thickness. A shear-wave beam traveling at an angle strikes that flat interface broadside and reflects a strong echo straight back to the probe. In many cases, RT misses what UT catches.

Phased Array Ultrasonic Testing (PAUT)

Phased array takes UT further by electronically steering and focusing multiple beam angles from one probe. That sweeping maps the flaw like a small scan, giving length, depth, and through-wall position rather than a blip on a screen. Here the research gets convincing.

NASA built a flux cored arc welded pressure-vessel sample and implanted a side-drilled hole about 40 mm long and 1.61 mm in diameter to simulate lack of fusion, then evaluated it with phased array. Their published work documents the technique’s performance against that reference flaw (NASA technical report). Separately, PNNL phased-array studies reliably detected flaws about 10% to 50% through-wall in thick-section components using 0.8 to 2.0 MHz probes. A NIST-backed reliability study also examined detection reliability by flaw position.

Magnetic Particle and Dye Penetrant: Surface-Connected Only

Magnetic particle testing (MT) and liquid penetrant testing (PT) are quick and cheap, but they only find LOF that breaks the surface — a surface-connected flaw. If the unfused interface is buried mid-wall or between passes, both come up empty. Treat them as screening tools, never as proof a weld is sound.

Destructive Testing: The Ground Truth

Sometimes the fastest way to confirm LOF is to cut the weld apart:

• Bend tests open the fusion line and expose the dark, unfused plane
• Macro-etch sections and acid-etches the weld so interface flaws show as distinct lines
• Nick-break fractures the weld to reveal internal fusion on the break face

These destructive methods are the reference standard — and why qualification practice built through structured welding practice matters so much.

Probability of Detection: No Method Is Perfect

Every NDT method has a probability of detection (POD), which depends on flaw size, orientation, and technique. That’s the core lesson from the NASA and PNNL work: detection isn’t binary. Serious fabrication specs therefore combine visual inspection, a volumetric method suited to the flaw type, and periodic destructive verification.

For heavy, repetitive work, automation raises consistency — see robotic submerged arc welding. But technology only goes so far: the inspector, technique, and procedure must match the defect you’re hunting.

Code Acceptance Criteria: What AWS D1.1, ASME, and ISO 5817 Say About Lack of Fusion

Here’s the uncomfortable truth about lack of fusion: most codes don’t treat it as a “maybe” defect you can measure and argue about. They treat it as a rejection. The logic is straightforward — LOF is a planar, crack-like discontinuity that concentrates stress, and unlike porosity, which codes often tolerate in small scattered amounts, it gets almost no grace at all.

That’s why a weld that looks flawless can still fail an inspection. LOF often hides at the sidewall, between passes, or at the root, staying invisible until a bend test, an ultrasonic scan, or a service failure exposes it. Before you can judge a defect, you have to know which rulebook applies.

The Linear vs. Rounded Indication Framework

Codes sort discontinuities by shape because shape predicts behavior. Rounded indications — gas porosity, most slag — are volumetric, with measurable length and width. Linear indications are sharp and planar, and they behave like built-in cracks.

Under ASME Section VIII Div. 1 radiography rules, a linear indication is any indication longer than three times its width. That definition sweeps in cracks, lack of penetration, lack of fusion, and elongated slag inclusions. Shape matters because a planar defect concentrates stress at its tip and can propagate under cyclic loading — exactly why inspectors treat it so harshly.

Why so strict? A planar flaw has almost no cross-sectional area to blunt stress. Porosity is a rounded void that redistributes load around it; lack of fusion is a sharp-edged separation that opens like a zipper. That difference is the entire reason codes separate the two categories. It’s also why visual “looks good” is never enough, as our guide to good and bad MIG welds shows.

The rejection rule in ASME is blunt: cracks, lack of penetration, and lack of fusion are unacceptable regardless of length. No length allowance, no arithmetic — they’re rejectable on sight.

ISO 5817 and Quality Levels

ISO 5817 takes another route, defining quality levels B, C, and D, with B the most stringent and D the least. It sets limits by quality level for steel arc-welded joints, but lack of fusion remains among the most tightly controlled defects. Buyers and specifiers pick a level based on service — the harsher the fatigue or pressure demand, the higher the level.

AWS D1.1, Tension, and Compression

AWS D1.1 handles acceptance differently as a structural steel code, tying criteria to weld type and loading condition. Complete joint penetration welds carrying tension are held to the tightest requirements, while certain discontinuities in compression zones or statically loaded members may be permitted within specific limits. The tension-versus-compression idea runs through every code: a discontinuity sitting across a tensile stress field is far more dangerous than one in a compression region. AWS inspection guidance makes the same point: the joint and its loading drive acceptance.

ASME Section IX and Qualification

ASME Section IX governs procedure and welder qualification, and it’s stricter still. For ultrasonic examination of qualification welds, cracks, lack of fusion, and incomplete penetration are unacceptable. Visual qualification coupons must show complete joint penetration with complete fusion of weld metal and base metal. In plain terms, a procedure that produces LOF on a test coupon simply isn’t qualified — which is why disciplined, repeatable technique, the kind covered in our welding practice guide, matters as much as the machine settings.

Why Lack of Fusion Matters: Fatigue, Fracture, Rework, and Real-World Consequences

A weld that looks flawless on the outside can still be a failure waiting for the right moment. That is the cruel truth of lack of fusion: the defect hides at the interface, passes visual inspection, and ships to the customer — until a load cycle, a pressure surge, or a cold winter morning finally opens the seam. By then the repair isn’t a grinder and a re-weld; it’s an investigation, a recall, or worse.

From a Thin Line to a Catastrophic Crack

An unfused interface is never just a cosmetic gap. It reduces the effective load-bearing area of the joint and acts as a sharp stress concentrator, because the geometry of a planar flaw forces load to funnel through the bonded metal around it. In a normally loaded weld that stress spreads evenly; around a lack-of-fusion line it can spike several times higher, and a flaw oriented perpendicular to the principal stress is the worst case of all.

Under cyclic or fatigue loading, that concentrated stress becomes a crack initiation site. Fatigue cracks start microscopically, grow a little with every cycle, and can reach critical size with no visible warning — which is exactly how weld cracks turn into sudden, catastrophic fractures rather than slow, inspectable leaks. When the defect is lack of penetration, the effective throat is reduced, lowering both static strength and fatigue life. That is why codes treat planar defects more severely than rounded porosity of similar size, as outlined in the ESAB welding defects guide.

Where the Consequences Hit Hardest

Industry consequences vary by sector, but the mechanism is identical:

• Pressure equipment and pipelines: an unfused seam is a leak path for gas or liquid; the NRC and other regulators treat these as safety-significant flaws requiring repair.
• Structural steel: bridges, buildings, and cranes rely on weld continuity; a fatigue crack can grow to a fracture-critical condition.
• Automotive: spot and seam welds carry crash loads; a weak interface compromises safety and durability.
• Aerospace: fracture-critical hardware demands near-zero planar defects, as NASA research on weld flaw reliability has long emphasised.

In ductile pipelines a through-wall crack often gives leak-before-break warning, but a planar defect at a girth weld can instead drive a fast-running fracture along the seam.

The Business Cost: Scrap, Rework, and Recalls

The engineering stakes translate directly into money. A defect found in the shop means scrap and rework; a defect found downstream means inspection, logistics, and reputation damage. Unidentified weld defects increase warranty risks and recalls, and when they’re caused by bad welds the automotive industry alone absorbs an estimated 9.9 billion USD a year. That figure is why major manufacturers and bodies like AWS — alongside the joint IBM–AWS drive to bring AI into inspection — are pushing smart, data-driven weld inspection to catch flaws early, a shift covered in AWS Inspection Trends.

The financial case for prevention is brutal once you tally everything. Scrap loses the material and the labour already invested. Rework consumes a second pass, often in a confined or already-inspected area. Third-party NDT, documentation, and schedule slippage add indirect costs that dwarf the weld itself. Recalls multiply all of it across every unit in the field, plus legal exposure and brand damage that no inspection budget can undo. The earlier the flaw is caught, the cheaper it is to fix — a rule that holds from a $50 bracket to a $50,000 weldment.

Automation helps, which is why shops invest in robotic submerged arc welding to hold parameters repeatably, and why disciplined welding practice and clean contact tip maintenance matter far more than a final visual pass — whether you run a hand torch or gasless MIG. Catching the defect at the arc, not at the bench, is what keeps a fabrication shop profitable.

Prevention First: Technique, WPS Discipline, Fit-Up, and Preparation

Lack of fusion is rarely a mystery—it’s a process failure, and process failures have known causes. That’s good news, because prevention comes down to discipline: following the qualified Welding Procedure Specification (WPS) instead of guessing at settings that merely feel right. Prevention is cheaper than repair, and it begins long before the arc even strikes. Get this one habit right and most other fixes fall into line.

WPS Discipline: The Single Most Powerful Prevention Tool

The WPS ties amperage, voltage, travel speed, arc length, preheat, and technique into a proven recipe. AWS points out that many welders still set parameters based on what they think works rather than the procedure, which is why inspectors should confirm the operator is running within acceptable WPS ranges. Follow the procedure, and lack of fusion loses its favorite hiding place. It is also a living document: if a parameter proves wrong on the floor, the fix is to revise it through engineering, not to quietly dial in a new number.

Control Heat Input and Technique

Poor fusion usually starts with a cold, fast, or wandering arc. ESAB recommends increasing current or reducing travel speed, shortening the arc, and dwelling at the sidewalls to drive the weld pool into the joint faces. Keep your torch or electrode angle consistent, push rather than drag where the process allows, and avoid excessive weave—wide, fast manipulation overheats the center and starves the sidewalls. Set amperage for the real joint, not a guess; our 3/16 E7018 amperage guide shows how thickness and position shift the number. Polarity matters too, whether you’re running flux-core DCEN vs DCEP or sorting out TIG polarity. On thick sections, preheat keeps the base metal warm enough to fuse cleanly.

Joint Prep, Fit-Up, and Fixturing

You cannot fuse what you cannot reach. Clean every joint face to bright metal—mill scale, rust, oil, and paint all block wetting and fusion. Cut the correct bevel angle, set a proper root gap, and confirm the joint offers adequate access so the arc can actually touch both sidewalls; a deep, narrow groove with a poor aspect ratio is a textbook setup for lack of fusion. Fixture parts to hold that fit-up against distortion, and verify it again after tacking.

Consumables, Equipment, and Training

Consumable and machine care do the quiet work. Keep low-hydrogen electrodes dry in a rod oven, match polarity to the process, and maintain a clean contact tip with steady wire feed—worn tips and erratic feeding create the arc wander that causes misses. Then train deliberately. When defects appear, AWS’s advice is to perform a root cause analysis and deliver corrective training, not simply demonstrate the technique once and hope it sticks; our guide to building real welding skills explains how structured practice beats blind repetition.

Prevention Essentials at a Glance

• Weld to the qualified WPS and verify you’re inside its parameter ranges.
• Clean joints to bright metal and cut the correct bevel, gap, and access.
• Use enough current, a short arc, and a steady sidewall dwell.
• Dry low-hydrogen electrodes; clean contact tips; consistent wire feed.
• Verify preheat and interpass temperatures on thick or crack-sensitive steel.
• Log parameters and inspect the first production welds of every shift.

Finally, stack the odds in your favor with a complete system, not a single fix: qualified welders monitored by an AWS Certified Welding Inspector, proven procedures, high-quality consumables, and a clean, contaminant-free environment where the arc, the metal, and the operator all stay within the lines of the WPS. None of these steps is glamorous. But together—technique, discipline, preparation, and training—they close every door lack of fusion uses to sneak into a finished weld.

Process-by-Process Fixes: MIG, TIG, Stick, Flux-Core, and Submerged Arc

Every welding process has a different personality, and that personality shapes exactly how lack of fusion sneaks in. Heat source, shielding, and slag system all change the risk profile — so the corrective levers change too. The good news is that each process usually has one or two dominant culprits. Fix those first, then chase the smaller variables. Here’s how to fight LOF process by process.

GMAW/MIG: Feed Consistency and Tip Condition

MIG fusion problems usually trace back to the wire feed and the contact tip. A worn, egg-shaped, or spattered tip makes the wire wander, throwing arc length and direction off so the arc never bites the sidewall. Check tip bore, replace it on schedule, and verify drive rolls aren’t crushing or slipping the wire. Keep stickout short and steady, because long stickout drops voltage and cools the arc right where you need heat. Polarity matters as well: DCEP drives deeper penetration, so a polarity mix-up quietly removes the fusion you planned. Read our contact tip welding guide for wear patterns, then confirm your bead shape against good and bad MIG welds.

Flux-Cored: Slag Control and Interpass Cleaning

Self-shielded and gas-shielded FCAW both run on slag. Incomplete fusion at the root and slag inclusions are common causes of high fillet-weld reject rates, especially between passes where slag carry-over starves the next layer of fusion. Slag left in a corner acts like a wall: the puddle rolls over it instead of fusing to it, and the defect hides until the weld is cut or fails. Wire-brush and chip every interpass, watch your drag angle (which controls slag coverage versus penetration), and select polarity deliberately. For the trade-offs, see our breakdowns of DCEN vs DCEP for flux core and gasless flux-cored welding.

GTAW/TIG: Heat Input and Filler Timing

TIG fusion depends on patient heat input and clean metal. Too little amperage or too fast a travel speed leaves the sidewalls cold. Dabbing filler before the puddle is wet traps cold laps. Clean to bare metal — oxides and oils are fusion killers — and use a dedicated stainless brush, not the one you use on carbon. Polarity is decisive: the wrong setting shifts heat away from the work. Learn the details in our TIG polarity guide (DCEN, DCEP, and AC).

SMAW/Stick: Amperage, Angle, and Manipulation

With stick, electrode amperage, angle, and manipulation determine whether the arc reaches the sidewalls and root sufficiently to fuse them. Run too cold, whip too wide, or hold a lazy angle, and the arc flicks along the surface without fusing. Keep the rod tight to the puddle, use arc force to push into the joint, and match amperage to diameter. As a rule, don’t weave more than about two to three times the electrode diameter; wide, slow weaves let the arc sit on already-molten metal instead of the cold parent. Our 3/16 E7018 amperage guide shows how.

SAW: Flux Handling and Travel Speed

Submerged arc fuses deeply when flux is dry, coverage is adequate, and travel speed is matched to current. Contaminated or recycled-dirty flux causes porosity and fusion loss, while running too fast outruns the puddle and leaves the toe unfused. Store flux sealed and dry, and don’t let a dust layer or fines build up in the recovery system. For high-duty automation, see robotic submerged arc welding and confirm technique fundamentals in our welding practice guide.

Bottom line: fix the variable that controls fusion for that process — feed and polarity for wire, slag discipline for flux and stick, heat and cleanliness for TIG. For a deeper look at defect classification, the AWS CWI weld defect reference and ESAB’s welding defects guide are both worth bookmarking.

Repair and Rework: How to Properly Fix a Lack-of-Fusion Defect

Here’s the hard truth that trips up a lot of welders: lack of fusion is not a weld you can fix by laying more metal on top. The defect is a cold, unbonded boundary buried inside the joint. Adding a cap over it doesn’t fuse that boundary — it just hides it under a fresh bead and gives it more stress to work with. The only reliable repair is complete removal followed by a qualified re-weld.

Regulatory guidance backs this up. NRC documents are blunt that for cracks, lack of fusion, and excessive porosity, the unacceptable portion must be completely removed and replaced using approved repair procedures. Welding over a crack without removal will not fix the issue and can lead to further failure, because the original discontinuity remains a stress riser.

The Standard Repair Workflow

1. Locate and size the defect. Use the NDT method that originally flagged it — UT, phased array, or RT — to map the full extent before you touch the metal. Guessing is how repairs get redone.
2. Excavate to sound metal. Grind or air-arc gouge until every indication is gone and you reach clean, fused base metal. Confirm by re-checking.
3. Re-prepare and clean. Restore the joint geometry, then remove slag, oxides, oil, and moisture. This is the same prep discipline that prevents the defect in the first place — see our guide on welding practice for the fundamentals that keep repairs from looping.
4. Re-weld with a qualified WPS. Match the original qualified welding procedure or use a documented repair procedure approved by the engineer.
5. Re-inspect. Test the repaired area with the same method and acceptance criteria used originally.
6. Document everything. Record location, method, filler, and inspector sign-off.

Excavation depth matters more than most people expect. A shallow grind that leaves even a sliver of the unfused boundary behind will survive the repair, the re-inspection, and then fail in service. The AWS Inspection Trends discussion of weld cracks reinforces the same principle: the discontinuity has to go entirely, not be bridged.

When Repair Is Impractical or Prohibited

Not every defect is fixable. In tube and pipe work, cracks may be cause for outright rejection of the component rather than repair, especially in thin-wall or high-pressure service. Repairing weldments of hardened or thick-section steel introduces its own risk — preheat and post-weld heat treatment decisions, as outlined in resources on welding hardened steel, can make a repair more dangerous than the original defect because they create new heat-affected zones.

For critical-service tube fittings, the accepted answer is often replacement rather than excavation. If a repair is permitted, preheat the joint to the WPS-specified minimum, control interpass temperature, and apply post-weld heat treatment where required.

Technique Tips That Make Repairs Hold

• Re-establish the correct polarity and settings — see our breakdown of TIG welding polarity for precision repairs.
• Weld hotter and slightly slower than the original to guarantee fusion on the re-weld.
• Keep contact tips and consumables fresh; a worn tip on a repair can destabilize the arc. Our contact tip welding guide covers wear diagnosis.
• Never chase a defect with repeated passes — each attempt adds restraint and stress.

A repair weld should be treated as seriously as the original fabrication: qualified procedure, clean metal, verified inspection, and a signed record. Cut corners, and the second failure arrives faster than the first.

Common Mistakes, Training Tips, and Final Takeaways

The bead looked fine — and that is exactly the trap. Across this guide we have seen lack of fusion hide at sidewalls, root faces, and between passes, surviving visual inspection only to surface on a bend test or under load. The encouraging truth is that nearly every LOF failure traces back to a handful of repeatable human errors. Fix the habits, and you prevent the defect.

The Mistakes That Keep Causing Lack of Fusion

When shops audit their weld failures, the same culprits appear again and again:

• Chasing speed over penetration. Traveling too fast and dragging the arc leaves the toes cold — the base metal never reaches fusion temperature and the puddle simply rides over the joint.
• Ignoring the WPS. The procedure controls current, voltage, travel speed, and technique. Freelance it and you have no baseline when a defect appears.
• Skipping interpass cleaning. Slag, oxide, and spatter between passes create a physical barrier that traps a fusion line which never heals.
• Welding dirty or scaled metal. Mill scale, rust, oil, and paint disturb wetting and arc stability. Clean to bright metal along the joint.
• Bad fit-up and root gaps. Excessive gap, mismatched edges, or a tight root you cannot reach all block proper root fusion.
• Worn consumables. A tired contact tip or damaged liner causes erratic feeding and arc wander — cheap insurance to replace, as we cover in contact tip maintenance.
• Wrong polarity. Running DCEP where DCEN is called for, or guessing on TIG, relocates the heat — heat balance shifts more than most welders expect (polarity explained). Confirm settings every time instead of trusting memory.
• Assuming visual inspection proves soundness. A smooth cap says almost nothing about the interface beneath it — learning what a truly sound weld looks like is only the starting point.

Training That Actually Sticks

Most LOF problems are skill and discipline problems, so training is the highest-leverage fix. Weld defect classification helps here: defects fall into appearance-based groups (incomplete fusion, porosity, cracks, undercut) and cause-based groups (incorrect parameters, improper joint preparation, inadequate technique, contaminated materials, metallurgical defects). Sorting a defect by cause keeps you from fixing the symptom instead of the source.

Root cause analysis follows the same discipline: collect the data, identify potential causes, test each hypothesis, then implement corrective actions that stop recurrence. Deliberate repetition — not just arc time — is what builds the habit; our guide to building real welding skills covers structured drills that target penetration and fusion. And a single demonstration is not training. Corrective coaching should continue until the welder can accurately measure the joint and repeatedly produce conforming welds on several samples — not just after one good pass.

Your Immediate Action Checklist

1. Read the WPS before striking an arc and post it at the station.
2. Verify polarity and machine settings against the procedure.
3. Clean joint faces to bright metal and remove all interpass slag.
4. Check fit-up, root gap, and edge alignment; measure with gauges, not guesses.
5. Inspect the contact tip, liner, and electrode; replace worn parts.
6. Confirm travel angle and speed produce visible toe wetting.
7. Re-clean between every pass.
8. Document defects, run root cause analysis, and verify the fix with samples.

Final Takeaways

Lack of fusion is not mysterious. It is predictable, detectable, and — above all — preventable. The welders and shops that beat it share the same habits: they respect the WPS, they prepare and clean religiously, they verify settings instead of guessing, and they treat every pass as one that must fuse, not just fill. Industry guidance on weld defects agrees that the majority of fusion problems are process and preparation issues, not bad luck.

So the next time a bead looks perfect, ask the harder question: is it fused? Verify it, measure it, and document it. Do that consistently and lack of fusion stops being a silent weakness — it becomes a problem you rarely create and always catch.

Frequently Asked Questions

What is lack of fusion in welding?

Lack of fusion is a weld defect in which the deposited weld metal fails to metallurgically bond with the base metal or with a previously deposited weld pass. The unfused interface acts as a sharp discontinuity that concentrates stress and can initiate cracks, particularly under cyclic loading.

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

Lack of fusion is a failure to bond between weld metal and base metal or between passes, and it can occur at the sidewalls, between passes, or at the root. Lack of penetration (incomplete joint penetration) specifically means the weld bead did not extend fully through the joint thickness, which reduces the effective throat.

Can lack of fusion be detected by visual inspection?

Usually not conclusively. Surface-connected lack of fusion may show as a visible line or toe, but most lack of fusion is internal and hidden. Visual clues such as high crowns, rope-like beads, and uneven toe lines may suggest a problem, but radiography or ultrasonic testing is needed to confirm it.

Why is ultrasonic testing better than radiography for lack of fusion?

Lack of fusion is a planar defect oriented primarily in the through-thickness direction, which radiographic beams often pass through without showing a clear indication. Ultrasonic testing is much more sensitive to planar, crack-like discontinuities, and phased array ultrasonic testing can size and characterize them accurately in thick sections.

Is lack of fusion always a rejectable defect?

Under most codes it is treated very strictly. ASME pressure-vessel and Section IX criteria deem cracks, lack of penetration, and lack of fusion unacceptable regardless of length, and AWS D1.1 generally rejects incomplete fusion in primary structural welds. The applicable code, service conditions, and loading always govern.

What are the main causes of lack of fusion?

The main causes are insufficient heat input (too little current or too fast travel), a long arc length, incorrect electrode or torch angle, failure to dwell at the sidewalls, and contamination such as rust, mill scale, oil, or paint. Poor fit-up, narrow root gaps, and worn consumables also contribute.

How do you fix lack of fusion in a weld?

You must remove it completely, usually by grinding or air-arc gouging to sound metal, then clean and re-weld using a qualified procedure. Welding over a lack-of-fusion defect without removing it does not restore fusion and can lead to further failure. The repair is then re-inspected and documented.

How can lack of fusion be prevented?

Prevent it by following the qualified WPS, using adequate current and the correct travel speed, keeping the arc short, dwelling at the sidewalls, and preparing clean, correctly beveled joints with proper root gaps. Training welders to work the sidewalls and verifying parameters with gauges or meters round out the prevention plan.

Authoritative Sources & Further Reading

• https://www.aws.org/magazines-and-media/inspection-trends/2025/august/it-aug-25-the-answer-is
• https://www.aws.org/magazines-and-media/inspection-trends/2024/may/it-may-24-feature-02-weld-cracks
• https://esab.com/lt/eur_en/esab-university/articles/welding-defects-guide
• https://www.nrc.gov/docs/ML1322/ML13225A409.pdf
• https://www.nrc.gov/docs/ML2407/ML24079A326.pdf
• https://www.nrc.gov/docs/ML1215/ML12157A671.pdf
• https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.252
• https://www.osha.gov/sites/default/files/publications/OSHA_FS-3647_WELDING.pdf
• https://www.iso.org/standard/72747.html
• https://www.nist.gov/publications/web-based-data-analysis-methodology-estimating-reliability-weld-flaw-detection-location
• https://ntrs.nasa.gov/api/citations/20120012877/downloads/20120012877.pdf
• https://awscwi.com/weld-defect-classification