What Causes Undercut in Welding? Causes, Fixes & Limits

A fabricator once watched a Certified Welding Inspector reject a clean-looking handrail hours before delivery — the only issue was a hairline groove running along the weld toe, barely visible under the shop light. That groove was undercut, and it is one of the most common, most underestimated reasons welds fail inspection.

What Is Undercut in Welding, and Why Does It Happen?

If you have ever had a weld rejected for a defect you could barely see, undercut is usually the culprit. The American Welding Society defines it plainly in AWS A3.0, Standard Welding Terms and Definitions: undercut is “a groove melted into the base metal adjacent to the weld toe or weld root and left unfilled by weld metal.” In everyday language, the arc carved a small notch into the plate next to the weld, and the filler metal never flowed back in to level it off.

That single sentence covers almost all of the mechanics. The arc did exactly what an arc does: it melted steel. The problem is that it melted more base metal than the puddle could refill before it solidified. What remains is a sharp, crease-like valley running along the edge of the weld bead.

Toe Undercut, Root Undercut, Fillet and Groove Welds

Undercut shows up in two distinct places, and inspectors treat them differently:

• Toe undercut — a groove melted into the base metal along the edge of the weld face, at the point where the weld meets the plate. This is the far more common type and the one most people mean when they say “undercut.”
• Root undercut — a groove melted into the base metal at the root of a joint, often on the back side of a groove weld. It is harder to see and, because it sits inside the joint, more dangerous per unit of depth.

It also behaves differently by joint type. On groove welds, undercut typically runs along the top edge of the weld, where the arc bites into the bevel face. On fillet welds, it most often appears on the vertical leg of a horizontal fillet weld, because gravity pulls the molten puddle down and away from the upper toe. Along a horizontal groove weld, you will usually find it on the top edge of the weld, where heat concentrates and the puddle is thinnest. Undercut can appear in any process, from stick and MIG to TIG and submerged arc, and on any alloy, including the aluminum joints discussed in guides like Aluminum MIG Welding: The Complete Practical Guide, where the narrow freezing range of aluminum makes an unfilled notch particularly easy to freeze in place.

Why the Groove Stays Unfilled: The Mechanical Picture

Think of what happens in the half-second the arc passes a joint edge. The arc column, several thousand degrees at its core, melts a small crater into the plate. Surface tension holds that crater as a liquid notch. Filler metal — wire, rod, or electrode — is supposed to flow in and replace the volume that was melted. If the puddle is moving too fast, the arc is aimed at the plate instead of the puddle, or the current is high enough to gouge deeper than the filler can fill, the notch cools and freezes exactly where it was. Nothing closes it. The weld looks solid from a distance, but a cross-section reveals a V-shaped crease at the toe.

As SSAB’s welding defect reference notes, the usual suspects are excessive current, too fast a travel speed, an incorrect electrode angle, and an arc that is too long. Later sections break each of those down, but the underlying principle never changes: the arc melted base metal faster than filler metal replaced it.

Why Undercut Is Taken Seriously

The reason this small groove carries oversized consequences is geometry. Undercut is a notch, and a notch concentrates stress. Where a smooth weld toe spreads load evenly into the plate, an undercut creates a stress riser at its root, a point where local stress runs several times higher than the nominal stress in the member. That is why undercut is a fatigue-driven problem more than a static-strength problem: cyclic loading initiates cracks at the notch root long before the section would fail in a single overload. Research compiled in the International Journal of Fatigue repeatedly shows weld toe geometry dominating fatigue life in welded joints.

What Causes It: A Preview

Undercut almost never comes from one single mistake. It comes from a combination, and those combinations cluster into four groups covered in the rest of this article:

• Parameters — excessive amperage or voltage, travel speed that outruns filler deposition, and heat input stacked up from preheat or repeated passes.
• Technique — wrong work and travel angles, over-wide weaving without dwell at the toes, long arc length, and filler aimed at the puddle center instead of the toe.
• Process specifics — how SMAW, GMAW/FCAW, GTAW and SAW each develop their own undercut signatures.
• Situation and joint design — out-of-position welding, poor fit-up, magnetic arc blow, thin material, and contaminated surfaces.

Learn to recognize that unfilled groove for what it is, and you are already halfway to preventing it. The next section looks at the physics of the arc that cuts it.

The Physics: How the Arc Actually Cuts That Groove

Undercut is born in the first fraction of a second of arc contact, long before anyone looks at the finished bead. Picture what an arc actually is: a column of ionized plasma that packs thousands of degrees of heat into a spot only a few millimeters across. The AWS handbook Physics of the Welding Arc makes an important point about that column — it does not just sit on the surface. It actively melts a crater into the base metal directly beneath and slightly ahead of it. When that crater sits at the edge of a joint, you are looking at the birthplace of undercut.

From there the sequence is predictable. The arc melts base metal at the toe of the joint faster than filler metal arrives to replace it. Surface tension and gravity then do what they always do with a molten puddle: they pull the liquid inward, toward the center of the weld, rounding the puddle’s edges away from the joint face. That leaves behind a small notch of melted, unfilled base metal along the toe. If the puddle moves on, or the arc swings away, before filler flows into that notch and the metal solidifies, the notch freezes open. What remains is a permanent groove — base metal melted away and insufficient filler deposited to fill the resulting depression, which is exactly how SSAB’s welding defect guidance defines the condition.

Notice what that means. Undercut is not really a “cut” in the sense of material being gouged out and hauled away. Nothing leaves the joint. It is a filling failure. Metal was melted and pulled out of position, and the space it vacated was never refilled.

Voltage, Amperage, Arc Length and Heat Input

Four variables decide whether that notch freezes open or gets filled, and they rarely act alone.

Arc voltage is the usual suspect. Voltage controls the width and shape of the arc cone: raise it, and the arc gets wider, flatter and more aggressive across the joint. A wider arc melts a wider band of base metal at the toe than the filler can possibly cover, which is why high-voltage beads look smooth and pretty in the center yet carry notched edges. Amperage works in the perpendicular direction — it sets penetration depth. Too much current digs the crater deeper than the puddle can refill, so even a correctly aimed arc can undercut if the machine is turned up too far for the material thickness.

Arc length behaves much like voltage, because a long arc adds resistance and spreads the arc column out. Stretching the arc to gain visibility or control dumps heat sideways instead of into the puddle. Travel speed completes the picture: move too fast and the puddle simply outruns the deposition rate, with the arc already melting the next inch of toe before the wire or electrode has caught up with the last one. Heat input ties everything together. More amps, more volts and slower travel all feed extra energy into the joint, and that energy keeps the notch molten longer. While the notch stays liquid, gravity and puddle surface tension keep pulling metal away from the toe. The Michigan Technological University welding discontinuities training materials describe this same melt-then-drain sequence as the core mechanism behind undercut on both fillet and groove welds.

Parameter	What changes in the arc	How the notch forms
Arc voltage	Wider, flatter arc cone	Melts a wider groove at the toe than filler can cover
Amperage	Deeper penetration	Digs the crater deeper than the puddle refills
Arc length	Diffuse, unfocused arc column	Spreads heat sideways and cools puddle control
Travel speed	Less dwell time per point	Puddle outruns filler deposition
Heat input	Slower cooling, larger puddle	Notch stays molten longer and drains open

Why Weaving Makes a Small Problem Worse

Weaving adds dwell time at the toes, which is good, but it also widens the puddle, which is risky. The technique works only if the welder pauses for a brief moment at each side of the weave, letting filler metal flow into the corner before reversing direction. Abrupt oscillation without that momentary pause leaves the melted edge unfilled, and the notch freezes on the return swing. A wide, fast weave is essentially a machine for producing undercut: it spreads heat over more surface area while depositing metal at the same rate.

Thin material makes all of this less forgiving. The margin between “penetrated” and “melted through” is tiny, and the puddle can collapse faster than filler can chase it, as anyone who has welded thin aluminum sheet with MIG quickly learns. Processes that carry higher deposition rates refill a widened puddle more reliably, which is part of the appeal of flux core welding when coverage speed matters.

SSAB’s practical guidance on preventing undercut boils the physics down to five recurring culprits: too-high travel speed, incorrect oscillation, incorrect torch or work angle, too-high voltage and too-high heat input. Notice what those five have in common — every one of them is something the welder controls. That is encouraging, because it means undercut is not bad luck. It is a parameter or technique decision, and the next sections break down exactly which one to change, starting with machine settings.

Welding Parameter Causes: Current, Voltage, Speed and Heat Input

If there is one bucket that catches most undercut complaints, it is this one. Michigan Technological University’s welding discontinuities training lists too-high welding parameters, inadequate filler addition, incorrect work angle, and too-fast travel speed as the four primary undercut causes, and three of those four are dial settings rather than hand skills. The Iowa Department of Transportation is even more direct: in IM 558, undercut is attributed primarily to welding technique, excessive welding currents, or both. In other words, before you blame the welder’s hands, read the machine.

Excessive Amperage: Deeper Than the Joint Needs

Amperage controls how deeply the arc penetrates. Push the current above what the joint requires and the arc bites a deeper crater into the base metal at the fusion line than the filler wire or electrode can fill on the next pass. The bead still looks acceptable from above, but the toes are undercut because the melted area is simply larger than the deposited volume. This is most common when a welder bumps the dial to compensate for a cold lap or lack of fusion problem elsewhere, solving one discontinuity by creating another.

Excessive Voltage: Wide Bead, Notched Toes

Voltage shapes the arc cone. Higher voltage gives a longer, wider, flatter arc that spreads heat sideways, so the puddle washes out toward the toes instead of staying stacked in the middle. The result is a broad, thin bead with a sharp notch where it meets the plate. SSAB’s defect guidance associates high arc voltage and high current with undercut for exactly this reason: the arc is doing more melting than the filler is replacing.

Voltage and wire feed speed travel together in GMAW. If the machine is set hot and the wire feed is slow, you get the worst of both worlds: a wide arc cone and a starved puddle. Balancing the two is the single most effective parameter fix, and it is covered in depth in our Aluminum MIG Welding: The Complete Practical Guide, where the margin for error is unusually small.

Travel Speed: When the Puddle Outruns the Filler

Travel speed is the most overlooked variable because it feels like a free fix. Moving faster seems neater, and it thins the heat-affected zone. But travel speed also sets how much filler metal lands per inch of bead. Move too quickly and the puddle is dragged along faster than the wire or rod can fill it, leaving a shallow unfilled trough along the trailing edge of the toes. The official correction is straightforward: undercut caused by parameters is fixed by slowing travel speed and making sure enough filler metal replaces the melted base material.

Excessive Heat Input and Preheat

Heat input rises with current, voltage, and preheat, and it also climbs with every additional pass over the same area. Excess heat keeps the puddle fluid longer after the arc has moved on, which gives the unfilled groove more time to open up and less time to be capped by the next pass. Multi-pass fillet welds on thick, preheated plate are a classic setting for this: each pass remelts and widens the previous toe notch. For heat-sensitive materials such as aluminum, where the window between “fluid” and “collapsed” is narrow, this effect arrives much sooner.

The Parameter Troubleshooting Order

When undercut shows up mid-job, there is a sensible sequence to follow. Slow down first. Travel speed is the easiest change to make and the one that immediately restores filler deposition per inch. If undercut persists, then reduce current and voltage back toward the qualified range for the joint thickness. Only then question the setup: stick-out, preheat, electrode diameter, or a machine that has drifted out of calibration. Chasing voltage first usually just swaps one problem for another.

Observed symptom	Likely parameter cause	Correction
Sharp notches at both toes of a flat, wide bead	Voltage too high for the wire speed	Lower voltage or raise wire feed until the bead tightens up
Deep penetration with heavy spatter and notched toes	Excessive amperage	Reduce current to the WPS range for the plate thickness
Thin, washed-out bead with visible troughs	Travel speed too fast	Slow down; increase dwell at each toe
Undercut appearing on later passes only	Accumulated heat input, high preheat	Let the joint cool, reduce interpass temperature, verify preheat
Undercut on one side, cold lap on the other	Parameter imbalance with arc direction	Rebalance current and correct work angle together

One last practical point: qualified parameters are not suggestions. Electrode selection matters too, since a rod sized or classified incorrectly forces you to run the machine hotter than the joint wants, as discussed in our Alloy Steel Welding Electrodes: Selection, Use & Safety guide. Set the machine to the procedure, hold the speed, and undercut from parameters very nearly disappears.

Technique and Operator Causes: Angle, Weave and Arc Length

Set the machine dial correctly and you can still ruin a weld with your hands. Undercut is at heart a deposition problem: the arc melts a notch into the joint edge, and the welder must refill that notch with filler before it freezes. Aim the arc at the plate instead of the puddle, or move the puddle faster than filler can keep up, and the notch survives. That is why technique tops nearly every troubleshooting list, including Portland Community College’s weld quality standards, which name improper technique, arc length too long, abrupt oscillation, amperage too high, base material too hot and travel speed too fast as the recurring causes of undercut.

Work Angle and Travel Angle: Where the Arc Is Actually Pointing

Two angles govern almost everything. The work angle is the side-to-side orientation of the electrode or gun relative to the joint. The travel angle is the fore-and-aft lean relative to the direction of travel. Push the arc into the joint edge rather than into the developing puddle and you are cutting the toe directly, with no filler there to heal it.

Stick and MIG angles differ because the processes behave differently. Most SMAW work runs with a trailing or drag angle of roughly 10–15 degrees, letting molten slag and weld metal flow back over the crater instead of the arc carving forward into cold base metal. With GMAW, a slight push gives a flatter, wider bead in flat position while a drag deepens penetration; for out-of-position and vertical-up work, drag is almost always correct. Whatever process you run, the arc should point at the puddle you are building, not the plate you are about to melt. If the arc and the filler wire are not converging on the same spot at the toe, one of them is wrong. The same discipline matters even more on aluminum, where the heat sink moves so fast, as covered in Aluminum MIG Welding: The Complete Practical Guide.

Weave Width, Speed and Side Dwell

A weave bead puts the arc on the toes twice per cycle, which is why it is also the fastest way to undercut them. Two mistakes compound:

• Weaving too wide. Exceed the reasonable weave width for your electrode and the puddle at the edge never reaches full fusion depth before the arc swings back across. The center builds up while the edges stay notched.
• Weaving too fast. Abrupt oscillation, in the PCC standards’ words, gives no time for filler to transfer. The arc flicks to the toe, the metal there goes liquid, and the arc is already elsewhere before filler arrives.

The correction is a brief pause at each toe of the weave. That dwell lets filler flow into the groove edge and lets surface tension pull the puddle flat against the plate. Narrow stringer beads avoid the problem entirely, which is why so many procedures prefer stringers over weaves on critical joints.

Arc Length and Filler Placement

A long arc spreads heat over a wider area, so the arc cone melts a broader band around the joint edge while delivering less concentrated energy into the puddle. The result is a wide, shallow, cratered bead with notched toes. Keep the arc tight: roughly one electrode diameter for stick, and the shortest practical stick-out for MIG, where excessive contact-to-work distance raises voltage and flattens the bead. On the filler side, direct the rod or wire at the toe and the leading edge of the puddle, not the center of the crater. Filler aimed mid-weld simply piles up where the weld was already sound.

Stance, Craters and Tie-Ins

Body position matters more than it sounds. If your travel hand reaches past its comfortable range, your angle drifts and your speed turns jerky midway through the pass, so plan your body so the whole bead runs from a stable, square stance. Out-of-position welding adds gravity: overhead and vertical welds drain the puddle away from the joint edge, and the instinct is to speed up, which makes undercut worse. Welders who hold All Position Welding Rod Guide angles with a smaller electrode and reduced current get better results than those who simply hurry. Craters and tie-ins are the other quiet source: stopping abruptly or restarting on top of the previous weld leaves a cold notch, so overlap the last inch rather than striking an arc right at the crater edge.

Symptom on the finished bead	Likely technique cause	Correction
Undercut on one side only, consistently	Work angle off center or filler aimed at the wrong toe	Re-square the work angle; point the filler at the deficient toe
Undercut on both toes of a weave	Weave too wide or too fast, no side dwell	Narrow the weave; add a pause at each toe
Shallow, wide bead with ragged edges	Arc length too long	Tighten arc length; shorten MIG stick-out
Notches at every stop-start	Abrupt crater and tie-in technique	Back-track on the crater; overlap restarts
Undercut only in difficult positions	Stance drift, gravity drainage, hurried travel	Slow down, reposition the body, reduce current slightly

One important caveat: if the undercut is uneven in a way that follows the joint rather than your hand, suspect magnetic arc blow instead of technique. Undercutting is usually due to the technique employed by the welder, though magnetic arc blow can also be a factor, and that distinction decides whether you correct the operator or the ground placement.

Process-by-Process Causes: SMAW, GMAW/FCAW, GTAW and SAW

Undercut is undercut no matter which torch is in your hand, but how it forms shifts with the process. The heat source, the way filler metal arrives at the puddle, and the shielding arrangement all change which variable goes wrong first. That is why a welder who runs clean beads with a stick can suddenly be chasing toe undercut after switching to flux core. Here is what specifically triggers the groove in each of the four common arc processes.

SMAW (Stick): Current, Arc Length and Arc Blow

Stick welding is mechanically simple, which makes its undercut causes easy to name. Excessive amperage for the electrode diameter drives the arc deeper than the joint needs, eating a notch at the toes that deposited metal never catches up with. An arc length that is too long behaves like added voltage: the cone widens, the bead flattens, and the edges undercut while the center looks fine. Fast travel compounds both, because the puddle moves ahead of the filler deposition.

Electrode choice matters more here than in wire processes. A fast-freezing, deep-penetrating rod run at the high end of its current range on a thin joint edge carves base metal faster than the operator can fill it, and an electrode sized too large for the plate has the same effect. Rod selection and current settings go together, which is why guides like the Alloy Steel Welding Electrodes: Selection, Use & Safety resource emphasize matching classification to material thickness before striking the arc.

Then there is arc blow. On DC work, magnetized parts or a poorly placed ground can push the arc to one side, dragging the puddle off course and leaving undercut along a single edge of the bead, a telltale one-sided pattern inspectors learn to spot quickly. AWS CWI field guidance on undercut inspection also lists arc blow among the practical causes seen on the job.

GMAW and FCAW: Voltage, Wire Feed Speed and Gun Angle

Wire processes have more interdependent dials, so their undercut causes are usually a mismatch rather than a single wrong setting. AWS CWI field guidance identifies incorrect voltage and amperage, with voltage too high as the typical offender, as a primary cause of undercut across SAW, GMAW, MCAW and FCAW. High voltage widens the arc cone, produces a wide, flat bead and leaves notched toes on both sides.

Wire feed speed deserves special attention because it is the amperage control on a constant-voltage machine, and it fails in two opposite directions:

• Too fast: the wire feeds in faster than the arc can melt it, the pool is chilled and freezes prematurely, and the bead edges never tie in smoothly, leaving undercut along the edges.
• Too slow: the joint is starved of filler. The arc burns directly into base metal with too little deposited metal to bridge the notch it creates, especially at the toes.

Gun angle and contact tip-to-work distance finish the picture. A work angle aimed at the joint edge instead of the puddle, or an excessive stick-out that quietly raises voltage beyond the dial setting, produces undercut even when the machine readout looks correct. The same balance principles apply in aluminum work, as covered in Aluminum MIG Welding: The Complete Practical Guide.

GTAW: Torch Angle and Filler Discipline

TIG gives the operator the greatest manual control of any arc process, which makes it highly effective for preventing undercut, but only when filler addition is consistent. The concentrated arc can melt a clean notch at the joint edge in a fraction of a second, and nothing replaces it if the rod is not already aimed at that toe. Typical failure points are a torch work angle pointing the arc into the plate rather than the joint, pausing too long at the edge without adding filler, and dipping the rod behind the leading edge of the puddle. On thin material the margin is small: heat input that would be harmless under MIG flux can blow an edge open under TIG. Training standards for precision work, as seen in Aerospace TIG Welding: Standards, Techniques, and Training, build filler timing into qualification from day one.

SAW: High Heat, Flux and Travel Conditions

Submerged arc welding hides the puddle under slag, so undercut tends to be discovered at inspection rather than in progress. High voltage combined with high current produces a wide, deeply penetrating bead whose toes cut into the parent metal. Incorrect flux, meaning the wrong size or type for the current and travel speed, or coverage that breaks down and lets the arc expose itself, changes bead shape and can leave the toes unfilled. Travel speed too fast, or an unstable arc caused by wrong contact tip-to-work distance, has the same effect. Because SAW runs at high deposition, the temptation is to raise current and voltage together; that pairing is one of the most reliable ways to create undercut in a single pass.

Process	Most likely undercut driver	Check first
SMAW	Excessive current, long arc, fast travel, arc blow	Electrode size vs. current range; arc length
GMAW/FCAW	Voltage too high; wire feed imbalance	Wire feed speed and stick-out
GTAW	Torch angle; filler not reaching the toe	Work angle and filler timing
SAW	High voltage/current; flux or travel mismatch	Flux condition and travel speed

The pattern across all four processes is the same: more arc energy than the filler metal can replace, delivered at a spot the operator did not guard. Match the machine to the joint, keep the arc aimed at the puddle, and the groove never gets the chance to freeze open.

Situational and Joint-Design Causes: Position, Magnetism and Material

Two welders can run identical settings on the same machine and still get different results, because parameters are only half the story. The situation around the joint — the position you are welding in, how the parts fit together, whether the steel is magnetized, whether the wind is blowing — decides how the puddle behaves. When welders tell us undercut “just showed up” even though the dial was right, the cause is almost always one of these situational factors.

Gravity: The Out-of-Position Problem

Undercut is most frequently found on the vertical leg of a horizontal fillet or on the top side of a horizontal groove weld, and both cases share the same physics. In a horizontal fillet, the arc naturally cuts upward into the vertical plate while gravity simultaneously pulls the molten pool down and away from that same toe. If the pool does not refill the notch before it freezes, the groove stays open. SSAB’s welding defect guidance lists excessive current and incorrect travel speed as primary causes, but position amplifies both.

Overhead is worse still. Gravity drains the pool away from the toe before solidification, so overhead work demands lower amperage, a shorter arc, and slower travel than the same joint in flat position. Vertical-up requires enough travel speed to keep the pool from sagging, but not so much that the filler never catches the arc. Selecting an electrode actually rated for all-position work makes a real difference here, as covered in our all position welding rod guide.

Joint Geometry, Fit-Up and Restraint

Poor fit-up quietly engineers undercut before the arc is even struck:

• Tight or absent root opening: the arc digs into plate edges instead of reaching the joint bottom, over-penetrating the faces and notching the toes.
• Edge mismatch (high-low): the arc favors the proud edge, melting it back while leaving the recessed side unfilled.
• Heavy, fully restrained joints: thick sections act as heat sinks, and the stored shrinkage stress can literally peel the solidifying toe away from the plate.
• Excessive preheat: preheat is necessary on many steels, but too much of it keeps the base metal near melting point longer, widening the arc crater and giving the puddle more time to run away from the toe.

Contaminated Surfaces

Paint, mill scale, oil, rust and oxides destabilize the arc. The arc wanders toward the cleaner metal, wanders back, and cuts irregular notches along the edge it keeps revisiting. This is especially punishing on aluminum, where the oxide layer melts around 3,700°F while the base metal melts near 1,220°F — a mismatch that makes the arc jump unpredictably. Thorough mechanical and chemical prep is non-negotiable; our aluminum cleaner for welding walkthrough explains the prep sequence. Coated or painted steel should always be stripped back at least an inch beyond the weld zone, and burning coatings also raise fume hazards that OSHA flags under its welding, cutting and brazing standard.

Magnetic Arc Blow

On DC work, arc blow happens when current cannot find a straight path to the ground clamp and must pass through the weld puddle instead, deflecting the arc sideways. The deflection concentrates heat on one toe and starves the other — producing lopsided undercut that repeats run after run. Magnetized parts, unbalanced ground placement and the end of a weld path are the usual triggers.

Wind, Shielding Gas and Thin Material

Outdoors, a modest breeze can strip shielding gas from the arc, ionizing the path and making the arc wander and spit. Use wind screens, increase gas flow modestly, or move to a flux-cored or self-shielded wire. Thin material has its own trap: it melts faster than filler can replace it, so the edge of the sheet simply disappears into the puddle. Drop amperage, slow down, and favor stringer beads. The aluminum MIG welding guide covers these settings in depth for the thinnest, most unforgiving alloys.

Why Undercut Matters: Fatigue, Stress Concentration and Code Limits

Undercut is easy to dismiss because it looks like a cosmetic blemish: a thin line at the weld toe, no wider than a scratch. It is not cosmetic. That groove is a pre-made notch cut into the part exactly where the load path already concentrates, which is why acceptance limits exist at all and why an inspector will stop a job over something the fabricator can barely see.

Why a Shallow Groove Becomes a Fatigue Crack

Metal does not fail because a surface looks rough. It fails because a sharp geometric change forces the local stress far above the average stress across the section. A weld toe is already a stress raiser thanks to the abrupt transition from bead to plate; undercut sharpens that transition into a notch with a small root radius, and stress rises sharply at the notch root. The result is that a crack can start at a much lower applied load than the same joint would tolerate with a smooth, ground toe. Research published in the International Journal of Fatigue found that the reduction in fatigue strength caused by weld toe undercut is at least twice that of comparable joints without undercut when measured against a flush-ground plate reference.

The orientation matters too. Undercut running transverse to the direction of principal stress behaves like a crack starter pointed across the load path, and the crack usually initiates at that notch and grows through the thickness. Undercut running parallel to the load is less damaging but still knocks down the fatigue rating of the joint. This is why cyclically loaded structures get far tighter limits than beams that only ever sit still, and why a groove tolerated on a handrail would never pass on a crane chord or a bridge member. It is also the reasoning behind the visual acceptance rules in CWI undercut inspection practice.

AWS D1.1 Visual Acceptance Criteria

In the United States, most structural steel work is judged against AWS D1.1, Structural Welding Code – Steel, which sets undercut limits in its visual inspection table. The limits split by loading condition, and the cyclic numbers are dramatically tighter than the static ones:

Connection	Material thickness	Maximum undercut depth
Static, non-tubular	Under 1 in (25 mm)	1/32 in (0.8 mm)
Static, non-tubular	1 in (25 mm) and thicker	1/6 in (1.6 mm)
Cyclic, primary tension member	Any	0.01 in (0.25 mm)
Cyclic, all other members	Any	1/32 in (0.8 mm)
Tubular connections	Governed by D1.1’s separate tubular table	Check the table for your joint and thickness

Two details in that table cause arguments in the shop. First, the static limits come with accumulated-length provisions: the code does not judge each inch of undercut in isolation, so a long run of near-limit undercut can still fail even though every spot gauge reading sits inside the number. Second, 0.01 in is a genuinely strict tolerance. It is thinner than a sheet of paper, and a recent D1.1 amendment added a short-weld factor to that accumulated-length logic so short welds no longer inherit the same total allowance as a long seam. Confirm the exact wording in the edition your project specifies rather than trusting a remembered rule.

Tubular work is scored against its own acceptance table rather than the plate rules above, which means the same groove can pass on a beam and fail on a tubular joint. Fabricators working in regulated sectors see the same pattern: aerospace welding programs, tank construction, and boiler shops each run to a written standard that supersedes general practice.

Stricter Standards: API 650 and ISO 5817

D1.1 is not the floor for everything. Tank fabrication to API 650 is stricter still, capping undercut at 0.4 mm (about 1/64 in) on vertical joints — half the depth D1.1 accepts on a static structural connection. Storage tanks are cyclic structures in practice: they fill and empty, pressure breathes, and the weld toes see repeated stress ranges, so the tighter number is deliberate.

Internationally, ISO 5817 grades weld quality into three levels — B, C and D — with B being the most demanding. Undercut allowances shrink as you move up the levels, so a weld accepted at level D would be rejected at level B without changing a thing about the part itself. Which level applies is a contractual decision, not a welder’s preference.

That is the practical takeaway: undercut matters because it converts a smooth joint into a notch that shortens fatigue life, and because the number that decides your pass or fail is written down somewhere specific. Know that number, measure against it, and treat anything deeper as a repair rather than a debate.

How Undercut Is Detected and Measured in Inspection

Undercut is one of the few weld discontinuities that can look clean and still fail. The groove is shallow, often hidden under spatter or a shadow line, and it sits exactly where the eye tends to slide past. That is why inspection practice treats undercut as a measurement problem, not an opinion problem. The inspector is not asking “does this look okay?” but “how deep is this groove, and what does the governing code allow?”

Visual Testing Is the First and Primary Method

Virtually every weld inspection begins with visual testing (VT), and for undercut it is usually all that is required. Undercut is a surface-breaking discontinuity, so it is directly observable once the weld is clean and properly lit. As CWI inspection guidance on undercut emphasizes, visual examination is the primary acceptance method, with other techniques brought in only when the groove cannot be seen or measured reliably.

Before any inspection starts, the surface has to be prepared. Wire brushing, grinding off spatter, and removing paint or coatings expose the weld toe clearly. On aluminum work, pre-cleaning matters even more, since oxide film and residue can disguise the toe line; the preparation steps in Aluminum Cleaner for Welding: How to Prep Aluminum Right follow the same principle of giving you a clean, readable surface before you judge a weld.

Lighting is the second half of the setup. The inspector needs strong, directional light aimed roughly along the weld rather than straight down, because a shallow notch casts a visible shadow only at a low angle. A dim shop light directly overhead will hide a 0.01 in groove that a raking flashlight reveals instantly.

The Tools: Gauges, Not Guesses

Three small tools do most of the work:

• V-WAC fillet gauge — a stamped steel or stainless gauge with pre-cut notches in standard depths. The inspector lays it across the weld toe and checks whether the notch in the gauge seats over the undercut. It is fast, cheap, and standard on fillet welds.
• Bridge cam gauge — a pocket-sized multi-tool that measures fillet leg size, weld reinforcement, and undercut depth with a single device, using its sliding probe and graduated scale. Because it handles several dimensions at once, it is the tool most inspectors carry by default.
• Undercut depth gauge / wire feeler check — for very shallow grooves, a set of thin wires or a dedicated depth probe can confirm a depth the gauge alone cannot resolve confidently.

The measurement technique is simple but specific: bridge the gauge across the groove so its reference surfaces rest on sound base metal on both sides of the toe, then read the depth of the notch below that reference line. If the gauge tees into the groove instead of bridging it, the reading is wrong and usually optimistic.

Why a 0.01 In Limit Needs a Gauge

AWS D1.1 allows as little as 0.01 in (0.25 mm) of undercut on primary tension members of cyclically loaded connections. Picture that dimension: it is roughly the thickness of a sheet of paper. No inspector can reliably distinguish 0.008 in from 0.012 in by eye, and lighting conditions change the apparent depth of the same groove from hour to hour. The limit is not guessable, so the gauge is not optional. The acceptance criteria are applied through the standard’s visual inspection clause and documented under the project’s quality records; the AWS D1.1 structural welding code sets out both the limits and the examination requirements.

When NDT Is Brought In

Magnetic particle (MT) and liquid penetrant (PT) testing are occasionally used when the weld toe is hard to access or when the inspector needs to confirm crack-like indications alongside undercut. Both reveal undercut as an elongated indication tracing the weld toe rather than giving a depth measurement, so they confirm extent, not severity. Ultrasonic and radiographic testing are rarely useful for undercut itself because the groove is surface-breaking and open; they are used instead when undercut is suspected as an initiator for a deeper defect. As SSAB’s welding defect reference notes, undercut is fundamentally a surface condition tied to the weld toe, which is why its evaluation stays anchored in visual methods.

Documentation: What Gets Recorded

A measurement without a record proves nothing later. Every finding should capture the joint or weld ID, the specific location along the weld, the measured depth in inches or millimeters, the instrument used, and the pass or fail determination against the governing criterion. Inspectors keep these records with the project quality documentation so a repair can be verified and an audit can trace the decision. Welders doing their own self-check benefit from the same habit: All Position Welding Rod Guide: Types, Uses & Tips reinforces that out-of-position passes are exactly where undercut hides, and a quick gauge check before the pass cools to room temperature catches problems while they are still cheap to fix.

How to Fix Undercut and Prevent It for Good

Undercut rarely means scrapping the part. In most shops it means one careful repair and a change in habit before the next bead goes down. The good news: the corrective path is well established, and the prevention side is mostly about a handful of repeatable behaviors.

The Repair Path, Step by Step

Start with AWS D1.1’s basic repair practice: clean the area and deposit additional weld metal, grinding first where the groove needs to be smoothed out. The practical sequence looks like this:

1. Clean and examine. Wire-brush or grind away slag, spatter, oil and paint so the full length of the groove is visible. You cannot repair what you cannot see.
2. Grind the notch into a uniform groove. Michigan Tech’s welding defect guidance recommends light grinding of the affected area to a consistent, smooth profile before any filler is added. The goal is to remove the sharp re-entrant corner and give the new bead somewhere to tie in, not to gouge out base metal.
3. Weld with a qualified procedure. Repair welds must be made under a qualified welding procedure, using a smaller electrode or lower wire settings and reduced amperage than the original pass. A repair bead that runs as hot as the pass that caused the undercut will simply cut a new notch.
4. Blend and re-inspect. Dress the repair flush, then re-inspect using the same method originally used on the joint.

Two details get skipped more than they should. First, preheat: if the WPS calls for preheat on the original joint, the repair needs it too, especially on restrained or heavy sections where a cold repair bead risks hydrogen cracking. Second, documentation: record the joint ID, the location, the measured depth before repair, the procedure used and the re-inspection result. An undocumented repair is, for practical purposes, no repair at all.

The single most reliable field fix is also the simplest: slow the travel speed down and add enough filler to replace the base metal the arc melted. Almost every undercut event is, at heart, a shortfall of deposited metal.

The Prevention Playbook

Prevention is cheaper than repair, and it comes down to a short list you can run before every pass:

• Dial parameters to the WPS. Match amperage and voltage to the material thickness rather than to what “feels right.”
• Slow the travel speed. Let the puddle wet the toes before moving on.
• Use stringer beads with side dwell. A narrow bead with a brief pause at each toe fills the notch; a wide, fast weave creates one.
• Hold correct work and travel angles. Keep the arc pointed at the puddle, not the joint edge, and aim the filler rod or wire at the toe.
• Match wire feed speed to voltage. A balanced arc gives a calm puddle that can reach the edges. Check dial-in order in our Aluminum Flux Core Welding: What Actually Works in 2026 guide for how out-of-balance settings behave on thin aluminum.
• Choose the right electrode size. Smaller electrodes on thinner material cut down penetration depth. Our Alloy Steel Welding Electrodes: Selection, Use & Safety guide covers how consumable choice changes arc characteristics.
• Prep and clean the joint. Contaminants destabilize the arc and encourage erratic fusion at the toes. Proper cleaning is covered in Aluminum Cleaner for Welding: How to Prep Aluminum Right.
• Practice deliberately. Stringer beads with a conscious two-count dwell at each side build the muscle memory that stops undercut at the source. Structured instruction, like what’s covered in an Aluminium Welding Course: What You Learn, Costs & Careers, shortens that learning curve considerably.

For broader discontinuity definitions and repair philosophy, the MTU welding discontinuities webinar is a useful reference. Fix the groove once, then change the behavior that made it.

Conclusion: A Practical Undercut Checklist for the Shop and the Field

Undercut looks like a small problem until a welder has to explain a rejected handrail, a repaired pressure vessel nozzle, or a cracked crane attachment. The encouraging news from everything covered in this guide is simple: most undercut is preventable, and the causes distill down to just two dominant factors. As SSAB’s welding defect guidance puts it, the arc is melting base metal that the weld puddle never gets around to refilling. In practice, that comes down to incorrect voltage and amperage settings (usually voltage too high or current too aggressive) and incorrect electrode or gun angle directing heat into the joint edge instead of the puddle.

Everything else, from travel speed to arc blow to poor fit-up, amplifies those two. Fix them and you eliminate the large majority of the undercut a shop sees.

The Seven-Point Undercut Checklist

1. Know the limit before you strike the arc. Ask which specification governs the job and what depth it allows. AWS D1.1 visual acceptance criteria differ sharply between static and cyclically loaded connections, so the number you are aiming for changes with the drawing. Check AWS D1.1 and the project specification before the first pass, not during inspection.
2. Dial parameters to the WPS. Set amperage, voltage, polarity, travel speed and stick-out as the qualified procedure requires, and verify rather than trust the machine’s display. If you need a refresher on balancing wire feed speed against voltage, revisit the parameter sections above.
3. Hold angle and speed steady. Work angle and travel angle decide where the arc energy lands. Push, drag or vertical-up correctly for the process, keep the arc length short, and slow down enough for the puddle to fill the notch the arc just cut.
4. Watch the toes while weaving. During any weave, dwell long enough at each side for filler metal to wash into the toe. When weaving gets wide and fast, undercut follows the wand. Stringer beads with a brief side dwell are the reliable habit here.
5. Check with a gauge, not your eyes. Clean the weld, light it well, and bridge a V-WAC or bridge cam gauge across the toe. A 0.01 in cyclic limit is invisible to naked-eye judgment, and a feeler wire rarely lies about depth.
6. Repair early, with qualified metal. Grind the groove into a smooth, uniform taper and re-weld it under a qualified WPS using a smaller electrode or reduced current so the repair bead does not cut new undercut. Shallow exceedances sometimes disappear under grinding alone. For process and filler choices, resources like the Alloy Steel Welding Electrodes guide help match consumable to repair conditions.
7. Document everything. Record joint ID, location, measured depth, method and pass/fail result. Visual acceptance alone is not always the finish line: undercut that passes visual criteria still has to be evaluated together with other discontinuities in demand-critical zones before the joint is released.

Undercut also rarely travels alone as a skill. Welders who develop undercut systematically usually need repetition rather than a lecture, and structured practice on all-position technique or a formal aluminium welding course tends to correct the angle habits faster than any written procedure can.

The path forward from here is the same one inspection and fabrication run on: qualify the procedure, confirm the welder can hold it, and measure instead of assume. Learn the acceptance logic, practice the puddle control, gauge your work honestly, and undercut becomes what it should be, a discontinuity you rarely produce rather than a reason your work gets rejected.

Frequently Asked Questions

What causes undercut in welding?

Undercut happens when the arc melts base metal at the weld toe faster than filler metal fills it back in. The main causes are excessive current or voltage, travel speed that is too fast, incorrect torch or electrode angle, too-long arc length, and abrupt weaving without pausing at the toes.

What is the difference between undercut and underfill?

Undercut is a groove cut into the base metal beside the weld toe or root. Underfill is a groove weld where the weld face or root surface sits below the adjacent base metal surface — the joint itself has not been filled to the required profile.

How much undercut is acceptable per AWS D1.1?

For statically loaded non-tubular connections, D1.1 allows up to 1/32 in (1 mm) deep undercut in material under 1 in thick, with accumulated-length exceptions, and up to 1/6 in (1.6 mm) in material 1 in thick or heavier. Cyclically loaded primary tension members are limited to 0.01 in (0.25 mm).

Is undercut always a rejectable defect?

No. AWS A3.0 distinguishes a discontinuity from a defect — undercut only becomes a defect when it exceeds the acceptance criteria of the applicable code or project specification. Shallow, isolated undercut within limits can be accepted.

Why is undercut dangerous for fatigue life?

Undercut acts as a sharp notch at the weld toe, concentrating stress where fatigue cracks typically start. Research shows weld toe undercut can cut fatigue strength by at least twice as much as an equivalent joint without undercut, which is why cyclic-loading codes set much tighter limits.

How do I stop undercut when MIG welding?

Reduce voltage if the bead is wide and flat, slow your travel speed, hold a slight drag angle, and balance wire feed speed with travel speed so the pool always has enough filler. Pause briefly at each side if you are weaving.

How is undercut repaired?

Grind the groove to a uniform contour, clean it, apply preheat if the WPS requires it, then deposit a stringer bead using a smaller electrode at reduced amperage so the repair does not create new undercut. Re-inspect using the same method as the original inspection.

How do inspectors measure undercut depth?

Most use a V-WAC fillet gauge, bridge cam gauge, or a dedicated undercut depth gauge bridged across the groove. Visual testing is the primary method, with magnetic particle or penetrant testing used when a groove must be distinguished from an elongated indication.