Incomplete Fusion in Welding: Causes, Detection, Repair & Prevention

What Is Incomplete Fusion? A Plain-English Introduction

You finish a clean-looking bead on a critical structural joint. The inspector runs a transducer along the toe, and the screen lights up with an echo you did not expect. The weld looks perfect on the surface — smooth, even, neatly stacked ripples. But underneath, the metal never actually bonded to the sidewall. That hidden gap is incomplete fusion, and it is one of the most underestimated causes of weld failure in the field.

Incomplete fusion — often shortened to lack of fusion (LOF) — is a weld discontinuity where the deposited weld metal fails to bond, or fuse, completely with the base metal or with a previously deposited weld pass. The metal is there, and it may look sound, but at the interface it simply did not melt into what it was supposed to join. The result is a cold, unbonded boundary hiding inside the joint. Because metal can be deposited without bonding, incomplete fusion is a bonding failure first and a strength problem second.

Why a Planar Discontinuity Behaves Like a Crack

Here is the detail that makes incomplete fusion so serious: it is a planar, two-dimensional discontinuity. Porosity is rounded and volumetric, so it scatters stress around its edges. Incomplete fusion is flat and sharp-tipped, so it concentrates stress the way a crack does. Under cyclic loading, the unfused interface becomes a stress concentration and a natural crack-initiation site, which is exactly the risk the ESAB welding defects guide warns about. A defect that starts sharp tends to grow sharp.

Discontinuity vs. Defect: An Important Distinction

Many welders use “discontinuity” and “defect” as if they were synonyms. They are not, and the difference matters. The AWS B1.10M/B1.10:2016 Guide for the Nondestructive Examination of Welds lists incomplete fusion (Section 4.5) and incomplete joint penetration (Section 4.6) as distinct discontinuities — interruptions in the normal structure of a weld. AWS then defines a defect as a discontinuity that, by nature or accumulated effect, renders a part unable to meet minimum acceptance standards. In plain terms: every defect is a discontinuity, but a discontinuity is not necessarily a defect. A small, isolated flaw may be perfectly acceptable; that same flaw in a fatigue-critical joint may be a rejection.

Quick Note: This is why the phrase “a discontinuity is not necessarily a defect” is written into the standards language. Acceptance depends on size, location, orientation, and the code governing the job.

Why It Matters: The Invisible Bond Failure

Incomplete fusion can be entirely subsurface. Because the discontinuity often sits at a sidewall, a root, or between passes, it can pass a visual exam behind a clean-looking bead and still be there. What it quietly steals is effective throat thickness — the actual bonded cross-section carrying load. A weld can measure full size on the outside and behave like a much thinner joint on the inside. That is the trap: good appearance is not the same as good fusion. If you want to sharpen your eye for surface cues, our guide to good and bad MIG welds walks through what a sound bead should actually look like, and our breakdown of the overlap welding defect covers another bonding failure that shares similar root causes.

What This Article Covers

Over the next sections we will move from theory to practice. We will separate incomplete fusion from lack of penetration, map where it hides in a joint, trace the root causes, and compare detection methods from visual and PT through RT and UT/PAUT. Then we will look at real consequences, prevention by process, repair by excavation and re-welding, acceptance standards, and the tips that keep it out of your work in the first place.

Whether you are chasing certification through focused welding practice or running production on the floor, understanding incomplete fusion is understanding why a weld fails when it “looked fine.”

Incomplete Fusion vs. Incomplete (Lack of) Penetration: Clearing Up the Confusion

Spend a single week on a fabrication floor and you will hear these two terms swapped as if they mean the same thing. They do not. Incomplete fusion and incomplete penetration are distinct defects with different geometry, different causes, and — critically — different reliability when you go hunting for them with nondestructive testing.

Here is the clean split. Incomplete fusion is a failure to bond. The weld metal never coalesced with the base-metal sidewall, or it failed to fuse into the previously deposited bead. Incomplete joint penetration (also called lack of penetration, or LOP) is a failure to reach. The weld simply did not extend through the full joint thickness at the root. Your earlier guide to welding practice treats these fundamentals as the foundation of good technique, and this distinction is a big reason why.

TWI draws the line precisely. In its guidance on lack of sidewall and inter-run fusion, TWI defines these imperfections as cases where weld metal fails to fuse with the joint sidewall or fails to penetrate the previous weld bead. A separate page on incomplete root fusion or penetration explains that incomplete root fusion means the weld failed to fuse one side of the joint at the root, while incomplete root penetration means both sides of the root region remain unfused.

Geometry is where the two defects live. Fusion defects hug the fusion boundary — vertical sidewalls, bevel faces, and the valleys between stacked beads. Penetration defects live only at the root, on the joint centerline. In a single-V groove, that boundary is a pair of sloped planes climbing both bevel faces, while the root sits flat along the bottom. That single difference explains why they respond so differently to inspection.

A Side-by-Side Comparison

Both defects rob the joint of load-carrying capacity, but through different mechanisms. ESAB’s welding defects guide notes that lack of penetration reduces the effective throat thickness, lowering both static strength and fatigue life. Fusion defects, by contrast, attack the integrity of the bonded interface itself — the joint may be full-thickness yet still tear apart along an unbonded sidewall.

Detection is where the confusion gets expensive. Radiography is generally less sensitive than ultrasonic testing for detecting lack of fusion, because these tight, planar, often oriented flaws may not present a useful density change on film. That is exactly why a UT or phased-array scan — like the toe transducer in the inspector’s hand — routinely finds what an X-ray passed.

Standards treat them differently too. Incomplete root penetration is not permitted for ISO 5817 Quality Level B (stringent) or Level C (intermediate), with only a limited exception for partial-penetration butt welds welded from both sides. Because lack of penetration cuts into the throat, many codes rank it as the more serious of the two in thin sections, where there is little material to spare. The acceptance route follows the geometry, not the label.

Finally, do not confuse an unintended fusion defect with a deliberate technique. Overlapping welding: technique vs. defect explained covers that grey area well, and the companion article on the overlap welding defect shows when an intentional overlap crosses into rejection territory. Getting fusion clear in your head is the first step to fixing it.

The Types and Locations of Incomplete Fusion in a Weld

Not all incomplete fusion hides in the same place. Each type has its own address inside the joint, its own signature in cross-section, and its own favourite set of processes. The AWS D1.1 structural welding code treats these as distinct defects, and the companion NDT guide, AWS B1.11M/B1.11:2015, illustrates incomplete fusion at the groove face, between weld beads, and between the weld and base metal. Knowing which one you are hunting changes how you inspect, and where.

The Four Recognized Categories

1. Lack of sidewall fusion. The bead fails to bond to the beveled groove face. In cross-section it looks like a sharp, straight interface with a thin gap between deposited metal and parent plate; on film it appears as a dark, straight-line indication hugging the fusion line. It dominates single-pass and narrow-groove work, and it is a hallmark of GMAW and FCAW run cold or fast. TWI explains that too narrow a joint preparation causes the arc to be attracted to one sidewall, starving the opposite wall of heat.
2. Lack of inter-run (interpass) fusion. A cold line left between successive beads. The next pass is laid over a bead that never got hot enough to melt, usually because of heavy slag, low current, or a fast travel speed. On a radiograph it reads as a straight indication running parallel to the layer boundaries.
3. Lack of fusion between the weld and base metal. The deposit sits on the plate surface instead of melting into it. You see it along the joint faces and at the toes of fillets, especially with cold starts and sluggish travel.
4. Root fusion defects. Incomplete fusion at the root, frequently confused with lack of penetration. TWI separates the two in its guidance on incomplete root fusion or penetration, because the cause and the fix differ.

Fillet and groove geometry decides which category dominates. A tight bevel angle, a thick root face, or a short root gap narrows the arc’s access, while restricted joint access compounds the problem — TWI notes that incomplete fusion is more likely when access to the joint is restricted. Open the bevel too little and the arc simply cannot reach the bottom corners; open it too much and burn-through, distortion, and frantic weaving take over. That is why weld tube fittings, with close tolerances and limited torch room, are so fit-up sensitive.

Location tells you what to correct. Sidewall problems point to amperage, drag angle, and bevel width; inter-run problems point to interpass cleaning and preheat; root fusion points to fit-up, root gap, and root-face thickness. Because FCAW carries slag, poor cleaning between passes is a frequent enabler — the same discipline behind clean gasless MIG welding applies.

Arc Blow and the FCAW Field Case

In ferromagnetic steels, magnetic arc blow can deflect the arc away from the joint face, producing a lack of fusion that mimics a geometry problem — and the surface can still look perfectly fine. In AWS Inspection Trends (August 2025), a heavy structural fabricator reported roughly a 20% reject rate on FCAW fillet welds and about 10% on groove welds. Digging in, they found incomplete fusion at the root of large single-pass fillet welds and incomplete fusion plus rope-like beads in groove welds — the classic fingerprint of a process fighting fit-up and travel-speed trouble, not one bad machine setting. The lesson: incomplete fusion is rarely random — it follows geometry, access, and technique.

Root Causes: What Actually Produces Incomplete Fusion

Incomplete fusion rarely has a single mysterious cause. In the field it usually traces back to four buckets: what you set on the machine, how you move the torch, how the joint was prepared, and what sits on the metal surface. Understand the physics and you stop chasing symptoms instead of fixing them.

Why the Bond Fails: A Quick Physics Check

Fusion is not about how much metal you deposit—it is about energy reaching the joint interface. The arc must raise the base metal to its melting point, and the pool must then wet the sidewall through capillary action. Weld metal poured into a gap that never melted just sits there as a mechanical ledge, not a metallurgical bond. That is why a bead can look full and sound and still tear out under load.

1. Welding Parameter Problems

This is the most common bucket. Too little current or heat input fails to melt the sidewall, so the filler simply sits on top. ESAB lists low current/heat input, excessive travel speed, incorrect torch angle, long arc length, and contamination as principal causes. The flip side is just as damaging: TWI notes that a travel speed too high for the current raises the risk of lack of fusion, while too high a current or too low a travel speed causes the weld pool to flood ahead of the arc. That flooding is the killer—molten metal runs out in front, insulates the joint, and the arc never digs into virgin base metal. Long arcs do similar damage, encouraging preferential melting along one side and shallow penetration. A voltage mismatch compounds it: too much voltage for the wire feed speed lengthens the arc, too little buries it and cuts penetration. Polarity belongs here too; it decides where the heat concentrates. Our guides to TIG polarity and polarity for flux core welding break down how that heat balance shifts fusion. Repeatable settings come from deliberate welding practice.

2. Technique Problems

Good settings mean nothing if the torch wanders. Wrong work and travel angles push the arc toward one plate and starve the other. Weaving too wide—or too fast—moves the pool before it can wet each sidewall. Lincoln Electric puts it plainly: if the electrode is not aimed properly at the root, or if the contact tip-to-work distance (CTWD) is inconsistent, root lack of fusion is far more likely. Inconsistent CTWD changes current and arc stability shot to shot, so a weld that looks fine in one inch can be cold the next. Worn or undersized contact tips make that worse—see our contact tip welding guide. The fix is deliberate manipulation with a short, consistent arc and a brief dwell at each sidewall—just long enough for the pool to bite.

3. Joint and Fit-Up Problems

You cannot fuse metal the arc cannot reach. A bevel that is too narrow, a root face that is too thick, or a root gap that is too tight leaves the arc blind to the sidewall. Misalignment stacks heat on the high side and hides the low side. Insufficient back gouging on double-sided joints leaves un-fused metal buried in the root that no later pass can fix.

4. Contamination and Material Problems

Rust, mill scale, paint, oil, and moisture all block wetting. The pool cannot bond to a dirty surface—it beads up instead of spreading, producing lack of fusion even with perfect travel and heat. Magnetic arc blow, common with DC on thick or residual-magnetized steel, deflects the arc off the joint line. Wrong polarity compounds the trouble, sending heat away from where it is needed.

The through-line is control: heat, aim, preparation, and cleanliness. Nail all four and incomplete fusion becomes the exception, not the default.

How Incomplete Fusion Is Detected: Visual, PT, MT, RT, and UT/PAUT

Finding incomplete fusion is a game of matching the right tool to the right geometry. No single method catches everything, and the inspector who understands each technique’s blind spots is the one who keeps a weak joint out of service. On a critical structural joint, though, the eye is only the first filter; the real verdict comes from volumetric and planar NDT. A well-organized quality-control environment — the kind described in our guide to welding facilities — gives you the lighting, access, and documentation that inspection depends on.

Visual Examination: Surface Only

Visual examination (VT) is fast, cheap, and brutally limited: it only sees surface-breaking defects. A poorly tied-in bead or an obvious toe line may show up, but subsurface lack of fusion (LOF) stays invisible. Still, VT catches the setup errors — wrong travel angle, low current, dirty plate — that create LOF in the first place. Reading bead clues is a core skill, and our guide to good and bad MIG welds shows how much the surface tells you. Building that eye takes deliberate welding practice.

Penetrant and Magnetic Particle Testing

Liquid penetrant testing (PT) pulls dye into any surface-connected opening, revealing LOF that breaks the surface but nothing buried inside. Magnetic particle testing (MT) works only on ferromagnetic steels, detecting surface and near-surface sidewall defects along the fusion boundary. Neither method reaches deep internal planes. Clean, paint-free surfaces are mandatory, and PT fails on porous or coated areas.

Radiography: Volumetric Strength, Planar Weakness

Radiographic testing (RT) uses X-rays or gamma rays to reveal volumetric defects — porosity, slag, and inclusions. Its weakness is the tight, planar, unfused interface. According to research catalogued by OSTI, radiography is not as sensitive as ultrasonics for detecting lack of fusion and can barely detect it unless it is associated with sizable inclusions or voids. Detection improves when the radiographic angle of incidence matches the bevel angle of the weld.

AWS B1.10M/B1.10 makes the geometry explicit: RT can detect discontinuities whose major planes are aligned parallel to the radiation beam, but it usually cannot detect laminations or cracks lying parallel to the plate surface. That blind spot is exactly where many bevel-side lack-of-fusion defects hide. You can review the standard through the AWS guide to nondestructive examination of welds.

Ultrasonic and Phased Array: The Planar Specialist

Ultrasonic testing (UT) is the preferred method for planar discontinuities. AWS B1.10M notes that UT can detect cracks oriented in any direction, provided the sound beam is essentially perpendicular to the crack’s major axis. Phased array ultrasonic testing (PAUT) sharpens this advantage by steering a focused beam through many angles in a sectorial scan, so a tight, unfused bevel face lights up where RT stayed silent.

The FHWA’s Phase II study, FHWA-HRT-14-075, developed phased-array acceptability criteria for steel bridge welds and noted that the defects found mainly consisted of lack of fusion, porosity, and cracks. AWS D1.1 and the performance-demonstration requirements linked to AWS D1.8 steer inspectors toward qualified UT for demand-critical joints, where a missed planar flaw is a fracture origin waiting to happen.

Beam orientation decides everything. RT is strongest when the beam runs parallel to the plane of the discontinuity, so a bevel-aligned LOF hides from a perpendicular shot. PAUT escapes that trap by scanning multiple angles, sweeping until a planar reflector responds. But PAUT is not magic. It depends on operator skill, calibration, and a written procedure; a poorly set up scan misses as much as RT does.

Match the method to the defect. For surface and near-surface checks, start with VT, then PT or MT. For volumetric flaws, RT is efficient. For planar LOF and cracks, go straight to UT or PAUT. Planar flaws are easy to confuse — an overlap defect can mimic LOF on a scan, so cross-check against our overlap welding defect guidance and the welding procedure.

Why Incomplete Fusion Is Dangerous: Consequences and Real-World Failures

Incomplete fusion rarely announces itself. The bead looks full, the cap is uniform, and the weld survives a casual visual once-over. Yet the bond that should tie the filler metal to the sidewall or the previous pass simply never formed. What remains is a planar discontinuity — a knife-edge gap trapped inside the joint.

Reduced Effective Throat and Lost Strength

Any unfused area removes load-bearing cross-section. When fusion is missing at the root or along a sidewall, the effective throat shrinks, and the joint no longer carries the strength its design assumed. Lack of penetration compounds the problem by reducing the effective throat further, which lowers both static strength and fatigue life. A joint that calculates as adequate on paper can be quietly undersized in reality.

For a comparison of how a different bonding failure plays out, see our guide to the overlap welding defect, where metal sits on top of the surface instead of fusing into it.

Stress Concentration: Why Planar Beats Volumetric

This is the heart of the danger. Porosity is volumetric — a rounded void that blunts stress and spreads it over a larger area. Incomplete fusion is planar — a sharp, crack-like slit with a tiny tip radius. That geometry concentrates stress enormously, and as ESAB notes in its welding defects guide, unfused interfaces act as stress concentrators and potential crack initiation sites, especially under cyclic loading.

Under repeated stress, a planar defect becomes a crack starter. AWS Inspection Trends notes that incomplete fusion in the root promotes cracking — and that cracks are generally the primary cause of catastrophic failure in structures and components.

Service Loading Decides the Timeline

Static service is forgiving; cyclic service is not. A pipeline under steady internal pressure may tolerate a small unfused area for years. Add vibration, thermal cycling, or pressure surges and the stress concentration begins to work the tip of the defect, extending it a fraction at a time until it reaches critical size. In corrosive environments, the open gap becomes a crevice that traps moisture and accelerates corrosion.

Hardened or high-carbon steels raise the stakes. Their heat-affected zones are crack-prone, so an unfused interface is more likely to propagate — see Can You Weld Hardened Steel? for how material behavior changes the risk.

Leak Paths, Standards, and the 20% Rule

In pressure containment, a continuous unfused path through the wall is a leak waiting to open, even if the hydro test held. Standards try to bound how much is tolerable: Los Alamos NL ES&H criteria specify that the length of unfused bead or layer areas shall not exceed 20% of the pipe circumference or total weld length. That is an acceptance limit, not a goal — it exists because inspectors know fusion defects behave like pre-cracks. A slow weep in a chemical line can do more damage than a full rupture.

Liability, Rework, and the Defect That Passes Inspection

The financial sting is real. When an inspector flags incomplete fusion, the repair means excavation, re-welding, and re-inspection — plus downtime and a paper trail. The engineering consequences explain why: a defect that passes inspection is not a defect that passes service. Many failures trace back to fusion flaws that sat below the detection threshold, then grew under load. From a liability standpoint, that is the nightmare scenario.

Surface appearance offers false comfort here — a lesson that shows up again in our breakdown of good and bad MIG welds, where the most dangerous flaws are often the ones you cannot see. Solid technique and disciplined welding practice keep the bond intact in the first place, which is far cheaper than chasing a crack later.

Prevention by Process: Stick, MIG, Flux-Cored, TIG, and SAW

Every welding process produces incomplete fusion (LOF) for its own reasons, so prevention must be process-specific. The universal rule, as TWI’s guidance on lack of sidewall and inter-run fusion explains, is that sufficient current is needed for the arc to penetrate the joint sidewall and the previously deposited runs. How you reach that point differs with the torch, stinger, or gun in your hand.

SMAW (Stick)

With stick, electrode choice and technique decide sidewall access. Match the electrode diameter to the root opening so the arc can reach both sidewalls and the root; an oversized rod physically blocks the angle you need. Run enough current for good penetration—too little amperage, and the toes simply sit on top of the base metal. A slight work angle that drives the arc into the sidewall, plus a comfortable drag angle, keeps fusion tight. On E7018 work, our guide to 3/16 E7018 amperage shows how the right setting and polarity change the bead entirely.

GMAW/MAG (MIG)

MIG and MAG reward consistency. Control contact-tip-to-work distance (CTWD)—long stick-out drops voltage and amps, so the arc cools and stops biting the sidewall. Aim at the root, not at the joint face, and avoid rushing travel speed; a fast pass lays metal against cold walls. Keep the gun angle steady and the shielding gas flowing. On thin or dirty steel, cleaning to bright metal before you strike matters as much as the settings. If you’re new to the process, How to Use a Wire Welder covers the fundamentals, while our contact tip guide explains why a worn tip wrecks arc stability.

FCAW (Flux-Cored)

Flux-cored welding brings its own pitfalls. Self-shielded wire can leave a rope-like, humped bead that hides root lack of fusion; the fix is to dwell at the sidewalls and confirm polarity is set correctly for the wire you’re running. Gas-shielded FCAW behaves more like MIG but still demands clean surfaces and a stable travel speed. Our overview of gasless MIG welding and the DCEN vs DCEP polarity guide both walk through setup choices that directly affect fusion.

GTAW (TIG)

TIG is precise but unforgiving of sloppy technique. Hold a consistent, short arc length—a long arc lets the heat spread without penetrating. Feed filler steadily into the leading edge rather than dropping it into the puddle, and remember polarity changes where the heat lands. Our guide to AC or DC for TIG welding steel explains the difference; for DCEN precision work, TIG polarity settings decide whether you get fusion or a cold lap.

SAW (Submerged Arc)

Submerged arc is a high-deposition process where inter-run fusion depends on bead placement and travel speed. Keep flux depth and coverage consistent, control travel speed so each pass overlaps the last, and place the bead so it washes into the previous pass’s toe. In automated setups, small tracking errors can push a bead off the joint centerline—an argument for monitoring systems in robotic submerged arc welding.

Universal Best Practices

Across every process, the fundamentals repeat. ESAB’s prevention list—increase current or reduce travel speed, shorten the arc, dwell at the sidewalls, prepare bright metal, and ensure a suitable bevel and access—applies whether you’re running stick or SAW. Lincoln Electric notes that complete fusion depends on proper joint dimensions and the correct depth-to-width ratio of the root pass. Widening or normalizing joint prep gives the arc room to work.

Before blaming technique, confirm the machine is actually delivering the amperage the WPS calls for—duty cycle, cable length, and work-lead connections all steal current. Following a structured approach, like the habits in our welding practice guide, builds the muscle memory that keeps fusion consistent job after job.

Repairing Incomplete Fusion: Excavation, Re-Welding, and Verification

A repair for incomplete fusion is never just “run another pass over it.” Because the flaw sits beneath a bonded surface, welding straight over the top traps a hidden discontinuity inside the joint. AWS Inspection Trends is blunt about this: welding over a crack or unfused region without removing it will not fix the issue and can lead to further failure. The correct workflow is remove, verify, clean, re-weld, and re-inspect.

Step 1: Confirm and Map the Defect

Before touching a grinder, nail down exactly where the flaw starts and stops. Use the NDE method that found it — RT, UT, PAUT, MT, or PT — and mark both ends on the part. A map prevents the classic mistake of excavating too little and leaving the tail of the defect behind.

Step 2: Excavate to Sound Metal

Remove the defect using localized gouging or grinding, following the agreed welding procedure. TWI notes that remedial action normally requires removal by localized gouging or grinding followed by re-welding as specified in the agreed welding procedure. Excavate with a slightly wider, smoother profile than the defect itself so the re-weld has clean sidewalls to fuse to. Gouge with care — arc gouging can smother the area you actually need to inspect.

Remove It or Accept It? Reading the Code

Not every indication forces a repair. Codes such as AWS D1.1 set acceptance criteria based on flaw type, size, and location. But acceptance is not the same as safe-for-everything. In pressure equipment, cyclic-loaded structures, and anything with an engineering critical assessment, incomplete fusion is treated as a planar, sharp-tipped flaw that can grow — so it comes out entirely, even when it measures inside the allowable limit. When in doubt, get the engineer of record to sign off before you weld past it.

Step 3: Verify Removal

Confirm the defect is gone with MT or PT, or a documented visual check where the code allows. A magnetic particle test is especially useful on ferromagnetic groove faces. Document the result with a photo or map before you weld, because once metal goes back in, that record is your only proof of removal. Don’t skip this — a missed fragment becomes a new flaw with a fresh weld around it.

Step 4: Clean and Re-Weld

Grind, brush, and degrease the excavation until you see bright base metal. Then re-weld to a qualified WPS with the correct technique and travel angle. Building sound sidewall fusion is a skill, not a guess — sharpening those habits on the bench pays off in the field, which is exactly what our Welding Practice guide covers. For double-groove welds, put the D1.1 back-gouge rule to work: the depth of the back gouge on the second side equals base metal thickness minus the first-side preparation depth, plus 1/8 in. Skimping here is a leading cause of incomplete joint penetration.

Step 5: Re-Inspect and Fix the Root Cause

Re-inspect the finished repair with the same NDE method used originally. But inspection alone is not a cure. If the root cause is not corrected, the flaw returns on the next part. AWS’s “The Answer Is” case study walks a corrective-action loop: identify the specific issue, review the quality system, check WPS parameters, and provide corrective training. The same loop explains why some shops keep repairing the same joint on every unit — nobody ever fixed the fit-up, the amperage, or the operator technique that produced the flaw. A repair that ignores this loop is just deferred rework — the kind of chronic defect we also see around overlap.

Treat repair as a process, not a pass. Excavate fully, verify honestly, re-weld correctly, and close the loop on the cause.

Incomplete Fusion Compared With Other Defects and Acceptance Standards

Judging incomplete fusion fairly means placing it beside the defects it is routinely confused with. They all steal capacity, yet each forms differently, sits in a different location, and demands a different response. What matters most is how a discontinuity behaves under load and whether the governing code allows it.

Overlap and undercut are surface-visible flaws you can spot on a finished bead — our guides to the overlap welding defect and to overlapping welding dig into telling a real defect from a deliberate technique. Comparing surface cues on a good and bad MIG weld trains your eye for these toe-line issues fast — a skill that sharpens with deliberate welding practice.

Overlap differs from incomplete fusion in a subtle but important way: with overlap the weld metal is present but draped over the surface, whereas with incomplete fusion the metal is present but never bonded at all. Undercut is not extra metal at all — it is missing base metal, a notch that focuses stress at the toe. TWI places sidewall and inter-run fusion loss in the same family as root fusion loss: all are planar, unbonded interfaces.

Why Planar Defects Are Treated Harsher Than Rounded Ones

A rounded pore leaves the surrounding metal intact; a planar flaw peels open like a page. Incomplete fusion, lack of penetration, cracks, and slag lines all create sharp edges that concentrate stress. That is why codes tolerate a surprising amount of porosity yet treat lack of fusion and cracks far more strictly, and why fatigue-rated joints get the tightest limits.

Porosity is the classic rounded gas pocket, and radiography catches it superbly because round voids show clearly as darker spots. Here is the trap: tight, planar lack of fusion is often oriented parallel to the radiation beam, so RT can sail right past it. Ultrasonics and phased array see those planar voids far better. That detection gap shapes acceptance criteria, which is why many codes pair film with ultrasonic testing on critical joints.

Acceptance is never universal — it is service-dependent. ISO 5817 defines three quality levels — B (stringent), C (intermediate), and D (moderate) — that set limits for lack of penetration. Lack of or incomplete root penetration is not permitted for Quality Level B or C, with an exception for partial-penetration butt welds welded from both sides.

AWS D1.1 sets UT acceptance-rejection criteria separately for statically loaded and cyclically loaded nontubular connections, because a flaw that is tolerable under static load may be unacceptable under fatigue. In pressure vessels, TWI notes incomplete root penetration is not permitted during manufacture, yet it may be allowable in pipework depending on material and wall thickness.

This is why AWS B1.10M/B1.10 reminds us that a discontinuity is not necessarily a defect — acceptance depends on the specified standard and the service conditions. Critical industries frequently demand complete elimination. A pipeline girth weld, a pressure vessel, and a decorative railing can carry the same discontinuity under three different verdicts. Service load, material, consequence of failure, and the specified code all move the line. Never assume a repair is optional or required without checking the contract standard. When in doubt, treat a planar fusion flaw as a crack until the code says otherwise.

Practical Tips, Common Mistakes, and Key Takeaways

Incomplete fusion rarely announces itself. It hides at sidewalls, between passes, and at the root, waiting for load or fatigue to expose it. The good news: with disciplined procedure and technique, it is almost entirely preventable. Ten sections in, the pattern is clear. Incomplete fusion traces back to a handful of controllable variables — heat, aim, cleanliness, and discipline. Here is the field-ready version you can tape inside your hood.

Pre-Arc and In-Progress Checklist

• Read the WPS before striking an arc. Verify amperage, voltage, travel speed, electrode diameter, and polarity against the qualified range — not against what felt right last week. An AWS Inspection Trends discussion found many welders set parameters by habit rather than the WPS acceptable ranges.
• Check fit-up and bevel dimensions. Measure root gap, included angle, and land. A too-narrow bevel crowds the arc and starves the sidewalls of heat.
• Keep the joint clean. Mill scale, oil, rust, and moisture all interfere with wetting.
• Match electrode diameter to the joint. Oversized electrodes in tight prep cannot reach the sidewalls; undersized ones waste heat.
• Control CTWD (contact tip-to-work distance). Long stickout drops current and pushes you toward cold lap; keep it within the WPS window. See our guide to contact tip welding for wear and setup tips.
• Aim at the root and dwell at the sidewalls. Pause long enough to let the puddle bite the walls, then move. Technique — not luck — is what fuses metal. Lincoln Electric notes less experienced welders are more prone to lack of fusion precisely because technique controls electrode aim and arc length.
• Avoid excessive travel speed. Rushing leaves the arc ahead of the molten puddle and produces cold lap.
• Watch for arc blow. Wander, spatter, and stubbing are warning signs — adjust ground position or switch to AC.
• Inspect with the right method. Visual examination is the least expensive technique but is limited to surface discontinuities (AWS B1.10M/B1.10). For planar flaws like LOF, use UT or PAUT.

The Mistakes That Keep Causing Incomplete Fusion

• Assuming a good-looking bead means good fusion. A smooth cap proves nothing about what happened underneath. Learn to read bead shape in our good and bad MIG welds guide.
• Welding over a defect. Trapping slag or an existing crack under a new pass multiplies risk.
• Ignoring the WPS. It exists to keep you inside qualified ranges.
• Using too-narrow prep. The arc simply cannot reach the sidewall.
• Trusting RT alone to find LOF. Radiography struggles with planar defects oriented against the beam; UT is the better tool.

Troubleshooting at a Glance

The Takeaway

No single tip replaces practice, but together they close the gap between a weld that looks done and one that actually is. Incomplete fusion is hidden, planar, and dangerous — but it is not mysterious. It comes from heat that never reached the sidewall, an arc aimed just off the joint, or a procedure quietly ignored. Across TWI, ESAB, and AWS guidance, one recommendation recurs: qualified procedures and continuous welder training. Build those habits deliberately through our welding practice guide, and stay safe with proper protective eyewear. Review the guides linked above, apply the checklist on your next job, and treat every arc as a chance to get a little better.

Frequently Asked Questions

What is incomplete fusion in welding?

Incomplete fusion (also called lack of fusion) is a discontinuity where the deposited weld metal fails to bond completely with the base metal or with a previously deposited weld pass. It is a planar defect, so it concentrates stress like a crack and can seriously reduce weld strength and fatigue life even when the surface looks fine.

How is incomplete fusion different from lack of penetration?

Incomplete fusion is a failure to bond along sidewalls or between passes, while lack of penetration (incomplete joint penetration) means the weld did not extend through the full joint thickness at the root. Both reduce load capacity, but they occur at different locations and are detected with different reliability.

What causes incomplete fusion?

The main causes are too-low current or heat input, travel speed too high for the current, long arc length, incorrect torch or electrode angle, narrow or poorly prepared joints, contamination like rust and oil, and magnetic arc blow. Poor welder technique and working outside WPS parameters are frequent contributing factors.

Can incomplete fusion be seen visually?

Often it cannot. Many incomplete fusion defects are subsurface, so a weld can look clean and still hide an unfused interface. Visual examination only finds surface-breaking indications, so ultrasonic testing is usually required to reliably detect and size planar fusion defects.

Is radiography good at detecting incomplete fusion?

Not especially. Research and industry guidance note that radiography is less sensitive than ultrasonics for lack of fusion and may only detect it when associated with sizable inclusions or voids. Detection improves when the radiographic angle of incidence matches the weld bevel angle.

How do you fix incomplete fusion?

You confirm and map the defect with NDE, excavate it by grinding or gouging to sound metal, verify removal, clean the area, re-weld using a qualified procedure, and re-inspect. Welding over the defect without removing it does not fix the problem and can leave a larger hidden flaw.

Will incomplete fusion cause a weld to fail?

It can. Unfused interfaces act as stress concentrators and crack initiation sites, especially under cyclic loading, and they reduce the effective throat thickness. In critical, pressure, or fatigue-loaded service, even small planar defects can lead to premature failure.

How can incomplete fusion be prevented?

Prevent it by following a qualified WPS, using adequate current and heat input, controlling travel speed and CTWD, aiming the arc at the root, dwelling at sidewalls, preparing joints correctly, and cleaning all surfaces before welding. If problems persist without arc blow, amend the procedure or retrain the welder.

Authoritative Sources & Further Reading

• https://www.aws.org/standards-and-publications/codes-and-standards/d1-1
• https://pubs.aws.org/p/853/b110mb1102009-guide-for-the-nondestructive-examination-of-welds
• https://www.aws.org/magazines-and-media/inspection-trends/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://www.twi-global.com/technical-knowledge/job-knowledge/weld-defects-imperfections-in-welds-lack-of-sidewall-and-inter-run-fusion-041
• https://www.twi-global.com/technical-knowledge/job-knowledge/weld-defects-imperfections-incomplete-root-fusion-or-penetration-040
• https://www.lincolnelectric.com/en/welding-and-cutting-resource-center/process-and-theory/weld-fusion-vs-weld-penetration
• https://www.osti.gov/biblio/21260276
• https://www.fhwa.dot.gov/publications/research/infrastructure/structures/bridge/14075/index.cfm
• https://engstandards.lanl.gov/esm/welding/vol2/WFP%202-01-Att-4-R1.pdf
• https://esab.com/gb/eur_en/esab-university/articles/welding-defects-guide
• https://www.nrc.gov/docs/ML0916/ML091600498.pdf