Undercut Welding Defect: Causes, Effects, and How to Fix It

A fabricator signs off on a beam column late Friday, convinced the welds look clean. On Monday, an inspector’s undercut gauge finds a 1.5 mm groove running along the weld toe — a defect invisible from three feet away that can now scrap the assembly or force a costly repair. Undercut is one of the most common, most underestimated, and most frequently rejected weld imperfections in structural steelwork, and understanding it separates welders who pass inspection from those who keep reworking joints.

What Is an Undercut Welding Defect? Definition and Types

At its most basic, an undercut welding defect is base metal that has been melted away at either the weld toe or the weld root and was never refilled by weld metal. This leaves behind a sharp, continuous groove or notch that cuts into the original base metal, rather than building up over it. The key point here is that this is lost base metal, not excess weld metal. Undercut appears at the weld toe (the edge of the weld bead where it meets the base metal) or at the weld root, forming that distinct notch in the surrounding material.

Before classifying undercut as a problem, it’s important to understand the difference between a discontinuity and a defect. Per AWS terminology, a discontinuity is any interruption in the uniformity of a weld, while a defect is a discontinuity that fails to meet the applicable acceptance standard. This means undercut is only a rejectable defect when it exceeds code limits, not just for being present. For structural work governed by codes such as AWS D1.1, Table 8.1 sets specific undercut limits at 1/32 in (1 mm) depth for material under 1 in (25 mm) thick and 1/16 in (2 mm) for material 1 in thick or greater. These measurable values are what separate an acceptable profile from one that is out of compliance.

Toe Undercut vs Root Undercut

The two most common locations for this defect are the weld toe and the weld root. Toe undercut is found along the outer edge of a weld bead, running parallel to it where the weld meets the face of the base metal. This is by far the most seen in fillet and groove welds. Root undercut, on the other hand, occurs at the root of a joint, most often in open root groove welds where penetration is attempted from one side. With root undercut, the melted groove is formed in the base metal at the root opening and goes unfilled, creating a notch that is often more difficult to visually detect without proper inspection.

Longitudinal, Transverse, and Joint Type Variations

Undercut isn’t just defined by where it’s at, but also by its orientation. Longitudinal undercut runs parallel to the direction of the weld, following the length of the bead. Transverse undercut, while less common, runs across the weld direction and can be found at tie-ins, corners, or around abrupt changes in the weld path. These can be especially easy to introduce when weaving through tight geometry.

This defect also behaves differently depending on the joint. In fillet welds, toe undercut is the primary concern, as it directly cuts into the leg side of the base metal. In groove welds, you can see toe undercut along the face bevel edges, or root undercut within the root pass of the joint. When working with more crack-sensitive materials such as those paired with alloy steel welding electrodes, keeping these profiles tight is especially important due to the reduced margin for error. Even for lighter gauge work such as the applications covered in an aluminum MIG welding setup, the same notch principle still applies.

Ultimately, undercut is a loss of base metal, not a lack of weld metal. Its sharp profile is what gives it its biggest impact, which will be explored in the next section. The type, location, and depth all determine whether it’s simply a profile imperfection, or a code-defining defect that cannot be ignored.

Why Undercut Matters: Structural Strength and Fatigue Risk

At first glance, undercut looks like a cosmetic blemish, a thin shadow along the weld toe that disappears under a coat of paint. That impression is what gets fabricators into trouble. An undercut is not just missing metal, it is a sharp notch cut into the base metal at the exact point where the load path transitions from one plate into the weld. Every structural consequence of undercut flows from that geometry.

There are three ways it hurts a joint: it removes load-bearing cross-section, it concentrates stress at its root, and it can cut fatigue life by more than half.

Reduced Effective Throat and Cross-Section

Fillet and groove welds are sized by their effective throat, the shortest distance from the weld face to the root that actually carries load. Undercut at the toe eats into the base metal just outside the weld face, so the joint is lighter than the weld profile suggests. An inspector measuring only the reinforcement sees a full weld; the load-carrying section tells a different story.

On a lightly loaded static joint, a shallow undercut may never matter. On a heavily loaded tension member, that removed metal directly lowers the net section available to carry force, and the fusion that should have tied the weld to the plate is gone along the fusion line. In the worst case, the weld toe stops functioning as a bonded edge and starts functioning as a pre-cracked one.

A Sharp Notch Sitting in the Load Path

Stress does not flow evenly through a joint. It crowds around geometric discontinuities, and the sharper the discontinuity, the harder it crowds. A properly profiled weld toe has a gentle radius that lets stress spread out gradually. An undercut replaces that radius with a narrow, often sharp-rooted groove, and the local stress at the root can climb to several times the nominal stress in the plate.

This is why undercut is judged far more harshly than a slightly convex or slightly concave profile. A profile issue changes stress mildly. A notch changes it dramatically, and it does so continuously along the full length of the weld rather than at one spot.

Fatigue: Undercut Behaves Like an Initial Crack

Most welded structures do not fail from a single overload. They fail from repeated loading, where a tiny flaw grows a little with each cycle until the remaining section can no longer carry the load. Fatigue cracks almost always start at the weld toe, and anything that mimics a crack there steals life from the structure.

Research from the University of Illinois quantified just how closely undercut tracks with an actual flaw. Modeling a 0.1 in (2.5 mm) undercut at a stress of 20 ksi, cycles-to-failure dropped from roughly 442,000 to about 171,600, a loss of more than 60% of fatigue life. The same study treated undercut as essentially equivalent to a 0.1 in initial crack, which is a striking result: the arc, not the service load, produced the starter flaw (University of Illinois fatigue report).

That is why fatigue-critical industries, from aerospace welding to offshore structures to bridge fabrication, hold weld toes to a much higher standard than general structural work. The flaw that ruins fatigue life is invisible from three feet away.

Why Codes Are Strictest in Tension and Cyclic Members

Codes do not treat every location equally, because the stress state decides how much damage a notch does. The AASHTO/AWS D1.5 Bridge Welding Code limits undercut to 0.01 in (0.25 mm) where it runs transverse to tensile stress, one of the tightest allowances in welding inspection, precisely because a notch oriented across the stress flow is the most damaging configuration.

Direction matters as much as depth. Research at the University of Texas at Austin found that welds wrapped around stiffener corners carrying up to 1/8 in undercut did not show compromised fatigue performance, a finding that informed the relaxed tolerances in the 2025 edition of D1.5 (UT Austin fatigue study). Those corners sit where undercut runs largely along, not across, the principal stress lines.

Note: A 0.03 in undercut running perpendicular to a tension member’s stress flow can be more dangerous than a deeper one running parallel to it. Always check the direction, the load type, and the governing code before deciding a weld is acceptable.

The practical rule is simple: a groove on a bridge girder in tension, a crane hook, a pressure vessel seam, or any member seeing reversal is a reject until measured and dispositioned. The same groove on a lightly loaded, non-cyclic bracket may fall inside tolerance. Understanding that distinction is what separates a considered repair decision from a reflexive grind-and-re-weld, which we will return to later in undercut repair options.

Main Causes of Undercut in Welding

Undercut rarely comes from one dramatic mistake. It builds up from small deviations that each leave a little more base metal melted than refilled. Whether the groove at the weld toe gets filled back in depends on heat balance, travel speed, arc direction, and how cleanly the molten metal wets the sidewall. Shift any one of those and the notch stays open. Here are the root causes inspectors see most often.

Excessive Welding Current

Amperage is the most common single driver of undercut. Too much current widens the arc cone and melts away more base metal than the filler volume can replace, so the deposited metal never catches up with the crater the arc leaves behind. Excess current also increases arc blow and spatter, which makes the puddle harder to hold against the toes. The typical pattern is a sharp, narrow notch that deepens as the pass continues, worst at the end of a long run when the plate is already hot.

Often the current was raised to compensate for something else entirely: a sluggish puddle caused by too-short arc length, dirty metal, or travel speed that was too slow. That is why AWS D1.1 and qualified welding procedure specifications lock amperage, voltage, and travel speed into defined ranges instead of leaving them to operator judgment.

Travel Speed That Is Too Fast

Travel speed that is too fast concentrates arc energy at the leading edge of the weld and gives the molten metal insufficient time to flow into and fill the groove melted at the weld toe. The pool ends up trailing the arc, and its trailing edge freezes before it can wet the sidewall and level out. The telltale sign is undercut that appears when a welder runs at production pace and vanishes at a slower speed on the very same joint with identical settings.

Wrong Electrode Angle, Arc Length, and Electrode Size

• Electrode angle: a work angle tilted too far toward the plate pushes the arc into the side wall instead of the joint root, while a travel angle dragged too far back lets the pool run ahead of the arc and freeze against an unfilled edge.
• Arc length: a long arc is effectively higher voltage. It widens and softens the arc, spreading heat over a broader area than the filler can cover.
• Oversized electrode: a rod or wire larger than the joint demands produces an arc wider than the groove needs. The operator then has to spread the bead, and the arc washes past the fusion line at the toes. Electrode selection matters as much as settings, which is why guides such as this alloy steel welding electrodes selection guide put diameter and type first.

Poor Technique at Stops and Starts

Craters and tie-ins are classic undercut zones. When a welder strikes an arc in the middle of a previous bead without backing up slightly, or breaks the arc abruptly without pausing to fill the crater, the restart leaves a notch exactly where the two passes meet. Wrapping a weld quickly around a stiffener corner or running too fast into a clip end produces the same result: the arc leaves the plate before the filler metal arrives.

High Heat Input on Thin Material

Thin sections cannot sink or spread heat, so a setting that is merely aggressive on 10 mm plate will blow a clean groove into 2 mm stock. This is especially unforgiving on aluminum, where high thermal conductivity pulls heat away fast but a low melting point lets the arc erode the toe before the filler catches up. Parameter control on thin aluminum is covered in detail in this aluminum MIG welding guide.

Gravity in Out-of-Position Welding

Vertical and overhead positions increase undercut risk because gravity pulls the weld pool away from the edges. In overhead work the pool sags toward its center, leaving the toes starved; in vertical-up runs, an arc held too long in one spot lets the molten metal slump downward instead of wetting sideways. Correct rod selection and manipulation for these positions is the subject of this all position welding rod guide.

Contaminants and Poor Surface Prep

Mill scale, rust, oil, paint, and primer all destabilize the arc and interfere with wetting. On aluminum the problem is more stubborn: the surface oxide film melts at a far higher temperature than the base metal, so an arc traveling over an uncleaned surface wanders and refuses to tie into the toe. Thorough cleaning before the arc ever strikes is why proper aluminum surface preparation is treated as a welding step, not a housekeeping one.

Cause	What happens at the arc	Quick shop check
Excessive current	Arc broadens; base metal melts faster than filler fills	Verify amperage against the WPS
Travel speed too fast	Pool freezes before it wets the toe	Slow down and re-run the joint
Wrong angle or long arc	Heat directed into the side wall	Watch the arc cone, not the puddle
Oversized electrode	Arc wider than the joint requires	Match diameter to plate thickness
Out-of-position work	Gravity pulls the pool off the edges	Reduce puddle size, tighten arc
Dirty base metal	Arc wanders, metal will not wet	Clean to bright metal before welding

The pattern across all of these is the same: more heat, less control, or less time for filler metal to catch up. Correcting the cause at the machine or in the hand technique is always cheaper than grinding and re-welding the notch afterward.

Undercut by Welding Process: SMAW, GMAW, FCAW, and GTAW

Undercut does not care which process you run. The same notch on a stick-welded column flange turns up on a MIG-sealed corner, a flux-cored fillet, and a TIG root. What changes is why it forms and which knob you reach for first, because each arc pushes the pool differently and the fix that cures one process can make another worse.

SMAW (Stick Welding): Slag Freezes Before the Metal Fills Back In

In shielded metal arc welding, the coating shields the arc and leaves a slag blanket over the bead, and that slag is viscous. Travel fast and the arc cuts a channel along the weld toe while the pool behind it has no time to flow sideways and refill before the slag freezes over the hollow. Fast travel plus high current is the classic stick-welding recipe for undercut.

Electrode type matters. Fast-freezing cellulosic electrodes such as E6010 and E6011 set quickly, which suits fast open-root passes but leaves less time for the pool to wet the toes. Fill-and-freeze low-hydrogen electrodes like E7018 let the metal settle, especially if you pause at each toe instead of dragging continuously. A steep angle in a groove weld concentrates arc force on one sidewall, and that melted corner never refills. Our advantages of stick welding guide covers where the process still shines.

GMAW (MIG): Balancing Wire Feed, Voltage, and Travel Speed

GMAW undercut usually comes from a mismatch between voltage and travel speed rather than one bad setting. In short-circuit transfer on thin material the arc is cool and the bead narrow, but push the voltage too high while keeping the gun moving and the arc carves the toe faster than the filler can cover it. Back the voltage down, slow slightly, and the notch disappears.

Spray transfer fills better with a continuous stream of droplets, but out of position that fluid pool fights gravity: metal sags away from the upper toe and leaves a groove. On vertical and overhead work, drop to short-circuit or pulsed spray, reduce plate heat, and keep the weave narrow.

Shielding gas is an underrated variable. CO₂-rich mixes such as 75% argon / 25% CO₂ raise arc force and penetration, deepening the sidewall bite on steel, while argon-heavy and helium mixes give a wider, flatter bead that wets the toes more gently. Gun angle interacts with that: a push (forehand) angle wets the toe better, while a drag angle concentrates heat and leaves a sharper profile. Thin aluminum narrows the margin between a clean toe and a burned one; our aluminum MIG welding guide covers it.

FCAW (Flux-Cored): High Deposition, High Risk

Flux-cored wire deposits metal fast, and self-shielded FCAW runs at currents and wire-feed speeds that would be unreasonable for stick or MIG. That energy has to go somewhere. Carry a wide weave on a heavy fillet and the arc keeps working the toes while the bead center piles up, leaving textbook toe undercut buried under slag. Hold the weave to roughly 2.5 to 3 times the wire diameter and split the fillet into passes that each wet the previous one. Gas-shielded FCAW is easier to tune but follows the same rule: wide weave plus high current equals notches. See aluminum flux core welding: what actually works in 2026 for more.

GTAW (TIG): Precision That Still Nicks the Toe

TIG has no slag to hide behind, so undercut shows up immediately under visual inspection. The usual culprit is travel speed without filler: the constricted tungsten arc cuts a clean, narrow groove, and if the rod is not dabbing into the trailing edge of the pool, nothing refills it. High arc voltage from holding the torch too high widens the arc and digs a broader notch.

TIG is often assumed low-risk, but research on thin-walled cruciform joints welded with GTAW found significant undercut that measurably reduced fatigue crack propagation life, so a clean TIG bead is not automatically a safe one. The fixes are simple: keep arc length tight, add filler at every stop and start, and slow down where the bead ties into the toe. Our aerospace TIG welding article covers how those tighter tolerances are controlled.

Process	Common trigger	First adjustment
SMAW	Fast travel, fast-freezing rod	Slow down, pause at toes
GMAW	Voltage too high for travel speed	Drop 1–2 V or slow the gun
FCAW	Wide weave at high current	Narrow the weave, add passes
GTAW	Traveling without filler	Dab filler, shorten arc

The common thread is time and filler: the arc melts faster than most welders refill. Your WPS ranges, including the AWS D1.1 structural welding code limits for current and travel technique, exist to keep you out of that window.

How to Measure and Inspect Undercut

Undercut can be one of the hardest weld profile imperfections to catch with the naked eye, especially when it’s shallow or running along a long, continuous weld toe. Because of its notch-like effect on fatigue life, inspection isn’t just about looks – it’s about quantifying depth and length against code limits. A disciplined approach to measurement is what keeps a weld acceptable or flags it for repair before it ever reaches service.

Visual Inspection as the First Line of Defense

Visual testing (VT) is always the starting point for undercut detection. With proper lighting at an angle across the weld toe, a trained inspector can often spot a shadowed groove where base metal has been melted away. This is consistent with the practices outlined in the AWS Welding Inspection Handbook, which emphasizes that lighting, eye position, and surface cleanliness are just as important as the tools themselves. Before any measurement begins, weld surfaces should be free of spatter, slag, or dirt that could hide or mimic a toe groove.

Measuring Tools: Gauges, Rules, and Comparators

Once suspected, undercut must be measured, not guessed. According to AWS training materials on visual weld inspection, the undercut gauge and the machinist’s rule are among the essential instruments for this task. An undercut gauge has precision steps or a sliding blade that is placed perpendicular to the weld axis at the toe to read the depth of the groove directly in inches or millimeters. A machinist’s rule can also be used to measure depth when combined with careful alignment, though a purpose-built undercut gauge is far more consistent across inspectors.

For shops that want to confirm profiles against known examples, comparator specimens (weld profile replicas with known undercut depths) can be used as a quick reference during VT. These are especially useful for less experienced inspectors, or when verifying measurements in the field. When sourcing inspection equipment, many shops rely on established suppliers such as Airgas Welding Supplies: The Complete Buyer’s Guide to keep calibrated gauges on hand.

Acceptance Criteria: AWS D1.1 and D1.5

The question of “how much is too much” is answered by the applicable structural code. For steel buildings, bridges, and many fabricated assemblies governed by the AWS Structural Welding Code, AWS D1.1 sets the limits for undercut. The 2025 amendment to D1.1/D1.1M continues to refine profile acceptance with a focus on fatigue performance, aligning with broader industry emphasis on notch control.

For bridge structures, AWS D1.5 carries its own acceptance criteria, which tend to be stricter in cyclically loaded bridge members due to fatigue demands. Because limits can differ between D1.1, D1.5, and a client’s project specification, inspectors must always confirm the governing code for that job – never assume the numbers are interchangeable.

Assessing Accumulated Undercut Length

One of the most misunderstood parts of undercut inspection is that codes often look at accumulated undercut length, not just isolated pockets. For welds governed by AWS D1.1, the rule is specific:

• Long welds: AWS D1.1 allows accumulated undercut up to 1/16 in (2 mm) deep for welds 12 in (300 mm) or longer, within any 12 in of weld length.
• Short welds: For welds shorter than 12 in, accumulated undercut of depth no greater than 1/16 in must not exceed weld length multiplied by 0.16.

In practice, an inspector will measure the depth of any undercut found along the toe with a gauge, mark each segment, and add them together over the evaluated length of weld. If those accumulated segments stay within the limit for that weld length, they may be acceptable. If they exceed it – or if any single undercut is deeper than the code allows for the application – it must be addressed. These calculations are exact enough that “close enough” has no place in a code-governed inspection.

Practical Inspection Tips for Consistency

Consistency between inspectors comes down to method. Always measure perpendicular to the weld axis at the deepest point of each undercut segment, verify with your gauge against a flat reference when possible, and document locations and lengths clearly on the inspection report. For crews looking to tighten their self-inspection habits, reviewing process fundamentals from resources like All Position Welding Rod Guide: Types, Uses & Tips can help welders understand how joint position affects the very toe profiles they’re trying to measure. Most importantly, don’t wait until final inspection to check – a quick gauge pass at reasonable intervals will catch developing undercut while it’s far easier to correct with technique, rather than repair.

Undercut vs Overlap and Other Weld Profile Defects

Undercut rarely shows up alone. On the same weld where an inspector flags a notch at the toe, you will often find the mirror-image problem on the other side of the bead: too much metal, sitting on the plate without ever bonding to it. Learning to tell these imperfections apart matters, because they look different, form for opposite reasons, and demand different fixes. Confuse them and you can grind away sound metal or add a second weld where none was needed.

Undercut vs Overlap: Two Sides of the Same Coin

Overlap is essentially the opposite of undercut. Instead of the arc melting base metal away at the toe and failing to replace it, weld metal flows out over the base metal surface and solidifies without fusing to it, leaving a rounded, rolled-over edge. Where undercut is a groove bitten into the parent metal, overlap is an overhang draped across it.

The causes invert too. Undercut comes from an arc that is too hot, too fast, or aimed too aggressively into the toes. Overlap comes from an arc that is too cold, too slow, too long, or held at too steep an angle, so the filler metal freezes before it can wet out and fuse. A welder chasing one by instinctively correcting toward the other can ping-pong between defects across a single pass.

Both are classified as weld profile discontinuities, and both are governed by visual acceptance criteria and weld profile schedules rather than volumetric NDE. AWS D1.1 treats them in the same family of profile imperfections, with explicit depth and length limits; the important distinction is direction: undercut is subtracted from the base metal, overlap sits on top of it.

Underfill: A Third Profile Defect With Its Own Limits

Underfill is a third profile problem, and it is easy to misread as undercut on a quick visual pass. Underfill means the weld simply did not deposit enough filler metal, so the finished face sits below the surface of the base metal. There is no notch melted into the parent metal, just a shallow trough of weld. AWS D1.1 carries underfill as its own line item in the visual acceptance tables, with depth limits that typically tighten further for the second side of a complete-joint-penetration groove weld, where underfill can act like a notch at the root of a loaded member.

Underfill usually signals travel speed that outran deposition rate, an electrode that is too small for the joint, or a welder stopping a pass short of the required face reinforcement. It is a deposition problem, not a heat-aiming problem, which is why the corrective action is more passes or a bigger electrode rather than a change in arc angle.

Porosity and Incomplete Fusion: Volume, Not Profile

Porosity and incomplete fusion break the pattern entirely. Porosity is volumetric: gas bubbles trapped in solidified weld metal, appearing as rounded cavities on the surface or inside the bead. Incomplete fusion (or lack of fusion) is an unbonded interface where the weld metal failed to coalesce with the base metal or an adjacent pass. Neither changes the external profile the way undercut does, which is why they are the domain of radiography and ultrasonic testing rather than visual gauges.

The engineering difference is subtle but important. Undercut is a sharp geometric notch at the surface, and surface notches are what fatigue cracks grow from. Porosity is generally tolerated at small, scattered sizes because rounded voids produce far lower stress concentration. Incomplete fusion is the most dangerous of the group when it exists, because it behaves like a planar crack, but unlike undercut, it usually cannot be confirmed or ruled out by eye alone. For how volumetric checks layer on top of visual acceptance, our guide to aerospace TIG welding standards walks through the full NDE stack.

Side-by-Side Comparison

Defect	What It Is	Main Causes	How It Is Found	Typical Repair
Undercut	Groove melted into base metal at the toe or root, not refilled	Excessive current, fast travel, poor electrode angle, wide weave	Visual inspection with an undercut gauge; VT per AWS D1.1	Grind to blend, or weld and reblend if beyond limits
Overlap	Weld metal rolled over the base metal without fusing to it	Cold arc, slow travel, low current, wrong angle, oversized bead	Visual inspection; rolled edge visible without tools	Grind off unfused metal, check fusion line, weld repair if needed
Underfill	Weld face sits below the base metal surface	Travel speed too fast, electrode too small, undersized passes	Visual inspection; machinist’s rule or filler gauge	Additional weld pass to restore required face height
Porosity	Gas cavities trapped in the weld metal	Contamination, insufficient gas, wet filler, arc wander	Visual for surface pores; RT or UT for internal	Gouge or grind out and re-weld if beyond acceptance limits
Incomplete fusion	Unbonded interface between weld metal and base metal or passes	Low heat, incorrect technique, cold lap, poor joint prep	RT or UT; rarely confirmable by eye	Remove defect completely and re-weld

Notice the repair column. Profile defects like undercut and underfill are largely correctable at the surface, while volumetric defects require excavation. That distinction drives cost: a shallow undercut blended by a grinder in ten minutes is a nuisance, whereas porosity or incomplete fusion inside a butt joint can mean gouging back to sound metal and adding hours of rework. It is one reason material preparation matters so much, since moisture, hydrogen, and surface oil feed several of these defects at once, the same prep discipline covered in our piece on aluminum cleaner for welding.

The practical takeaway for anyone reading an inspection report: match the repair to the defect family. Blend undercut, remove and replace overlap only after checking fusion, add metal to underfill, and treat any volumetric indication as a deeper investigation. Getting that mapping right is half of passing re-inspection the first time. For shops that also handle aluminum structural work, where profiles are inspected just as strictly, the same logic applies in our guide to aluminum boat welding.

How to Prevent Undercut: Technique, Parameters, and Procedure

The cheapest undercut repair is the one you never have to make. Nearly every groove at a weld toe traces back to three things: too much heat in one spot, too little filler metal delivered there, or filler placed somewhere other than where it belongs. Control those three and undercut rarely survives the first pass.

Begin With the WPS, Not a Friend’s Settings

Welding codes require welders to work within a qualified welding procedure specification (WPS). Under AWS D1.1, essential variables include current, travel speed, welding technique, and filler metal classification, each one changes how the pool flows and solidifies.

Running “the settings that always work” outside those ranges is not just a bad habit. On a code job it is a procedure violation, and any undercut produced is doubly yours. Check amperage range, recommended electrode diameter, voltage limits, and travel speed guidance before you strike an arc. Pay special attention to the maximum bead width the procedure allows, because wide beads and undercut travel together.

Match the Electrode or Wire to the Joint

An oversized electrode on thin material forces you to push enough current to blow the edge before you control the puddle. Choose the smallest diameter that gives a stable arc and the deposition the joint needs. Thick sections tolerate larger rods; sheet metal and open roots do not. Classification and sizing directly affect bead shape, and our alloy steel welding electrodes guide shows how to match consumables to the job. Arc length matters just as much with stick: a long, wandering arc raises voltage and heat without adding filler, and that excess heat eats the toe.

Slow Down, and Hold Your Angles Steady

Reducing travel speed slightly and holding correct work and travel angles lets the molten pool wet out and fill the groove melted at the toe. Move too fast and the arc cuts a notch while the filler arrives behind and freezes before it backfills.

• Travel angle: roughly 5–15 degrees drag for SMAW and FCAW; a slight push suits short-circuit GMAW on thin sheet. Keep it steady because a wandering gun changes arc force at the toe from inch to inch.
• Work angle: point the arc at the joint, not the plate. On a fillet, aim at the root so the puddle reaches both toes instead of washing up one side.
• Pool control: watch the leading edge of the puddle. If it climbs the base metal ahead of the arc, you are adding heat without filler.

Push versus pull interacts with shielding gas and the penetration profile. If you run GMAW on aluminum, our aluminum MIG welding guide explains how gun angle and travel direction shape the bead. The same physics apply to steel, only the tolerances are tighter.

Keep the Weave Narrow: The 2.5-to-3 Times Rule

One field rule prevents more undercut than any other: keep weave widths no greater than about 2.5 to 3 times the electrode or filler wire diameter. A wide weave drags the arc across the toe, overheats the edge, and lets the pool freeze before it wets back into the groove. With a 1/8 in. rod, that means a weave near 3/8 in.; with 0.035 in. wire, about 1/8 in.

Where you need width, use several stringer passes rather than one fat zigzag. When you do weave, pause briefly at each toe so filler wets into the corner. That short dwell closes the notch at the same time it shapes a clean profile.

Handle Restarts and Out-of-Position Work Deliberately

Undercut loves craters and restarts. Fill the crater, drag out, and break the arc on solid weld metal. Out-of-position welding is where gravity pulls the puddle away from the top toe: vertical-up and overhead joints generally call for lower current, a smaller-diameter electrode, a tighter arc, and slower, more patient travel. Our all-position welding rod guide matches rod types to position so you are not fighting a hot pool on an overhead bead.

Back the Joint When the Metal Is Thin

On thin plate or open roots, a backing strip, a small backing weld, or a chill bar gives the pool something to freeze against instead of burning the edge away, and it pulls heat out of the toe. Where thicknesses step down, shrink the bead sizes as you move onto the thinner side.

Parameter drift	How it creates undercut	Prevention move
Current too high	Arc force carves the toe faster than filler fills it	Return to the WPS amp range; step down rod diameter
Travel speed too fast	Filler arrives behind the arc and freezes open	Slow slightly and let the pool wet the toe
Weave too wide	Edge overheats before the pool backfills	Cap the weave at 2.5–3× rod diameter; run stringers
Arc too long	Voltage and heat rise without added filler	Tighten arc length; check tip-to-work distance on MIG

Prevention is a habit, not a trick: set your parameters from the procedure, pick the right consumable, hold your angles, keep the weave narrow, slow down, and back up thin joints. Do that consistently and the undercut gauge stops being a source of suspense, it simply confirms the weld you already knew was good. Start with SMAW fundamentals if stick is your process, because the muscle memory carries straight across to MIG and flux-cored work.

Repairing Undercut: Grinding, Welding, and Code Treatments

Finding undercut during inspection is not an automatic scrap sentence. Almost every structural code gives you a defined path back to acceptance, and that path turns on one measurement: how deep the groove is. Shallow undercut gets blended out with a grinder. Deep undercut gets welded back up. Get it wrong and the repair itself becomes a second reject.

Grinding: The Preferred Fix for Shallow Undercut

Blending means running a soft abrasive wheel, flap disc, or rotary file along the weld toe until the sharp notch disappears and the transition from weld face to base metal becomes a smooth radius. Done properly, grinding does more than remove a defect: it removes the stress raiser itself. Research on toe-ground joints has long shown that a smooth toe radius can restore and even improve fatigue performance over an as-welded toe, which is why undercut ranks so high on the inspection hit list in the fatigue-critical research literature.

The rules have tightened and then softened again in the bridge code. The 2025 AASHTO/AWS D1.5 Bridge Welding Code now prefers grinding undercut smooth instead of adding weld metal, on one condition: the depression left after grinding must not exceed 1/32 in (0.8 mm). This formalized what the 2020 D1.5 commentary already observed — that repair by light grinding is generally superior to a cosmetic weld repair.

Blend along the direction of the toe, keep the wheel flat rather than digging in with its edge, and stop as soon as the notch is gone. Over-grinding eats into the weld throat and base plate, creating a new thin section and a new acceptance problem. On stainless or aluminum, watch heat and contamination; smearing from a dirty wheel is a real issue, and prep discipline before and after any rework is covered in our aluminum cleaner for welding guide.

Welding Repair for Undercut Deeper Than 1/32 In

When the groove starts deeper than the grinding allowance, the fix is to put metal back, following the same discipline as the original weld rather than a quick pass:

• Dress the groove so the arc reaches the bottom of the notch — a partially filled undercut traps slag and comes back.
• Use a qualified procedure and the correct filler, with stringer beads or tight weaves instead of one wide cosmetic pass.
• Stay low on approved current and slow travel slightly so the new pass does not carve another groove at the toe.
• Clean between passes and hold preheat or interpass control where the WPS requires it, since the repair adds a second thermal cycle to the joint.
• Use a welder qualified on that procedure; most codes restrict repair welding accordingly.

On aluminum the margin is thinner still, from structural joints to wheel rims: contamination-free surfaces and tight parameter control decide whether the repair holds, which is why our alloy wheel welding guide puts so much weight on cleaning and heat input before the arc is struck.

Avoid Cosmetic Over-Welding

The temptation on a rejected weld is an extra pass that “makes it look right.” Resist it. Piling excess weld metal over a shallow groove widens the heat-affected zone, adds residual stress, can distort the assembly, and may itself be rejected under weld profile and excess-size limits. Both AWS D1.1 and D1.5 treat profile excess as a measurable condition, not a free pass. Grinding out a shallow notch is faster, cheaper, and gentler than covering it with a bead nobody asked for.

Situation	Preferred treatment	Why
Undercut at or under 1/32 in deep	Grind smooth	Removes the notch with no added heat
Undercut deeper than 1/32 in	Weld repair, then dress	Grinding alone would exceed the allowed depression
Long continuous toe undercut in a fatigue category	Weld repair plus review of parameters	Suggests a process problem, not a blemish
Shallow groove in a non-critical, cosmetically sensitive part	Cautious blending	Smooth profile without extra deposited metal

Re-Inspection and When the Engineer Must Approve

Repair closes the loop only after re-inspection. The area is examined visually, depth is re-measured with the same gauge or comparator used originally, and any required NDT is repeated on the repair, since a repair pass can introduce porosity or lack of fusion of its own. Record the location, method, and re-check result alongside the original finding.

Approval thresholds are worth checking against the governing code — the official AWS D1.1 page lists the current edition and amendments, and limits vary across structural, bridge, and railway work. As a working rule, the engineer of record or responsible inspector signs off on repair welding, on any condition outside stated limits, and on repairs that change the joint’s original design intent. Get that approval in writing before striking the arc. More on discontinuity evaluation in AWS Inspection Trends.

Common Mistakes That Lead to Undercut (And How to Fix Them)

Undercut almost never shows up at random. AWS inspection guidance consistently lists weld profile problems — undercut, overlap, slag inclusions, and elongated porosity — among the most common failures found in field quality checks, and nearly every one of those grooves traces back to a repeatable habit. AWS Inspection Trends guidance puts it plainly: when profile defects keep reappearing, the proper response is root-cause analysis, not another pass with a grinder. Here are the six mistakes that show up most often on the floor, and exactly what to change.

Mistake 1: Cranking the Heat to “Burn In” Faster

The logic feels sound — more voltage, deeper penetration, faster progress. What actually happens is that the arc carves a wider, deeper notch at the weld toe than the filler metal can refill before the puddle solidifies. The result is a clean, shiny groove that looks like great fusion but acts as a stress raiser. This habit runs wild in aluminum MIG work, where operators push voltage instead of fixing gun angle or skipping the aluminum cleaning routine the metal needs. Step your settings back toward the middle of the recommended range for your thickness, confirm penetration on a scrap piece, and only then return to production. The full parameter logic is covered in this aluminum MIG welding guide.

Mistake 2: Wrapping Welds Around Stiffener Corners

Welders love one continuous bead, so running a fillet around the end of a stiffener or tab in a single pass is tempting. But the arc decelerates and shifts angle as it rounds the corner, dwell time spikes, and the corner leg gets overheated — a textbook recipe for corner undercut that inspectors flag on sight. Stop the weld at the edge of the joint, tie in cleanly, and restart on the other side, or follow the drawing detail exactly as written. If a corner wrap is truly required, the WPS should explain how to do it. When it does not, that is a question for the engineer, not an invitation to improvise.

Mistake 3: Weaving Too Wide

A weave wider than roughly two and a half to three electrode diameters drags the arc across base metal faster than it deposits filler at the edges. The center of the bead looks perfect; both toes starve. Keep the weave narrow, and when a wide joint demands coverage, place multiple passes side by side instead of swinging one bead across. End each side-to-side cycle with a brief pause at each toe so filler can wet into the corner. Out-of-position welding makes this worse because gravity pulls the puddle away from the upper toe, which is why choosing the right all-position rod and holding a tight arc length matter more overhead than anywhere else.

Mistake 4: Ignoring the WPS Ranges

A qualified procedure under AWS D1.1 exists because someone already tested that parameter window. Welding outside it — higher current, faster travel, a larger electrode than qualified — means the procedure no longer covers what you are depositing, and undercut is one of the most predictable outcomes.

Mistake 5: Skipping the Self-Check Before Releasing the Weld

Undercut that is invisible from three feet away is still rejectable at two inches. AWS recommends that welders verify weld size and profile themselves, using gauges, before submitting work for inspection instead of trusting appearance. A fillet weld gauge or dedicated undercut gauge costs very little, and most suppliers stock them — this welding equipment guide covers what to look for when you buy inspection basics. Run the gauge along every toe before you chip, brush, and walk away. Catching a 1.5 mm groove while the plate is still in position takes thirty seconds; finding it at final inspection takes a repair procedure, paperwork, and possibly a scrapped part.

Mistake 6: Using Repair as a Substitute for Fixing Technique

The most expensive undercut mistake is behavioral: grind the groove, re-weld it, and repeat the identical defect on the next joint. Repair without root-cause analysis treats the symptom. Log what changed — electrode angle, travel speed, position, heat setting, stick-out — and you will usually find the cause within a few joints. Also remember that grinding is not free: on thin sections such as alloy wheel repair or thin aluminum fabrications, blending out a groove can cost more wall thickness than the part can spare.

Quick Reference: Mistake, Cause, Correction

Mistake	Why it undercuts	Correction
Cranking the heat	Melt outpaces filler deposition	Trim to the mid-range of the WPS
Wrapping corners	Excess dwell at the corner	Stop, tie in, restart per the detail
Wide weave	Toes starve for filler	Weave no wider than 2.5–3× rod diameter; pause at toes
Off-WPS settings	Unqualified conditions	Return to procedure values; ask before deviating
No self-check	Small grooves reach the inspector	Gauge every toe before release
Repair-only response	Defect repeats on every joint	Root-cause log, then fix the technique

None of these corrections require new equipment or extra talent. They require paying deliberate attention to the same handful of variables on every joint. Keep that habit long enough and undercut stops being an inspection problem — it simply stops happening.

Conclusion: Build a Zero-Undercut Habit

Undercut rarely announces itself. It hides at the weld toe, looks like a shadow from three feet away, and shows up only when an inspector drags a gauge along the joint. By then the weld is cold, the assembly may be stacked, and the cheapest fix has already been replaced by rework, grinding, remaking a coupon, or worse, a field repair on installed steel. The lesson running through this entire guide is simple: an undercut welding defect is a notch cut into the base metal, and a notch is where fatigue cracks begin. Everything else — the codes, the gauges, the parameter tables — exists to keep that notch from ever forming.

The Four Takeaways Worth Remembering

If you only carry a few ideas away from this article, make them these:

• Undercut is a stress concentrator, not a cosmetic flaw. It reduces the effective throat, sharpens the geometry at the weld toe, and measurably shortens fatigue life. Research on welded joints has treated toe notches as crack-like flaws for decades, which is exactly why cyclically loaded and tension members get the tightest limits.
• Prevention is dramatically cheaper than repair. Dropping amperage a little, slowing the travel speed, correcting electrode angle, and keeping weaves within roughly 2.5 to 3 electrode diameters costs nothing but attention. Grinding, re-welding, and re-inspection cost all three.
• Codes give you a number, not a vibe. Acceptance is measured in millimeters of depth and accumulated length, judged against a written table rather than someone’s opinion of what looks acceptable.
• Self-inspection catches almost everything. Trained visual inspection finds the vast majority of weld imperfections before they escalate into expensive NDT discoveries, as AWS guidance in the AWS Welding Handbook pocket reference emphasizes. A 30-second check with a gauge before you release the joint is the highest-return habit a welder can build.

Always Confirm Which Specification Governs

This is the step most welders and even some supervisors skip, and it is where projects go sideways. Undercut limits are not universal. A groove weld in a structural steel connection under AWS D1.1 is judged by one set of thresholds, and the 2025 edition with Amendment 1 (available as the D1.1/D1.1M-2025 AMD1 PDF) carries updated table values and a revision to how certain undercut is treated. Bridge work falls under AWS D1.5, where the 2025 Bridge Welding Code introduced new undercut tolerances and treatment rules. Beyond those, a fabricator’s own specification, a mill contract, a client standard, or a petroleum or pressure-code addendum can be tighter than any of them.

So before you qualify a procedure or sign off on a weld, ask three questions: which code applies, which table applies to this weld type and loading condition, and does the client spec override it? Write the answer on the traveler. Limits differ between base metal thickness, weld position, static versus cyclic loading, and whether the joint is in a tension zone, and the same groove can pass on one drawing and fail on the next.

When Undercut Already Exists: Fix It the Code’s Way

If you inherit a weld with undercut, resist the urge to simply bury it under a cosmetic bead. Over-welding adds heat, distortion, and residual stress, and it often creates a new profile problem while solving the old one. The cleaner path for shallow undercut is grinding the notch out to a smooth, blended profile — a repair route the latest bridge welding code now explicitly endorses when the depth is within limits, and one that removes the sharp root of the notch rather than covering it. Deeper undercut generally requires weld metal deposition by a qualified procedure, followed by grinding flush and, in every case, re-inspection to the same acceptance criteria as the original weld. When in doubt, stop and get the engineer’s written approval before touching the joint.

Make It a Habit, Not a Rule

The welders who never fight undercut are not the ones with the best machines. They are the ones who run within their qualified ranges, watch the arc rather than the clock, adjust technique when the position changes — a habit that transfers directly to all position welding rod selection and manipulation — and check their own work before anyone else does. Prevention is a daily routine: correct angle, controlled travel speed, moderate heat input, clean base metal, a pause at each toe, and a gauge pass at the end.

Build that routine and undercut stops being an inspection problem and becomes a non-event. Zero undercut is not an ambitious target. It is the normal output of disciplined technique.

Frequently Asked Questions

What is undercut in welding?

Undercut is a groove melted into the base metal at the edge of a weld (the toe) or at the weld root that the filler metal fails to refill. It creates a sharp notch that concentrates stress and can reduce both static strength and fatigue life.

Is undercut always a rejectable defect?

No. Undercut is a discontinuity, and it becomes a rejectable defect only when it exceeds the limits in the governing code or specification. AWS D1.1, for example, permits limited undercut within defined depth and accumulated-length limits.

How deep can undercut be under AWS D1.1?

For material less than 1 in (25 mm) thick, undercut must not exceed 1/32 in (1 mm) in depth, with exceptions allowing up to 1/16 in (2 mm) within strict accumulated-length limits. For material 1 in thick or more, the limit is 1/16 in (2 mm).

What is the main cause of undercut?

The most common cause is excessive heat concentrated at the weld toe combined with insufficient filler metal flow — typically from too much current, too-fast travel speed, incorrect electrode angle, or weaving too wide for the bead.

Can undercut be repaired by grinding?

Yes, if it is shallow. The 2025 AASHTO/AWS D1.5 Bridge Welding Code prefers grinding undercut smooth over adding weld metal, provided the repaired depression stays within allowable depth. Deeper undercut should be repaired by welding.

Does undercut affect fatigue life?

Yes, significantly. Undercut acts as a sharp stress-raiser at the weld toe, where most fatigue cracks in welded joints begin. Fatigue research found a 0.1 in undercut can reduce predicted life by roughly 60% compared with an unnotched weld.

How do inspectors measure undercut?

Undercut is measured during visual inspection using an undercut gauge, a machinist’s rule, or comparator specimens, checking both depth and accumulated length against the applicable code limits.

What is the difference between undercut and overlap?

Undercut is base metal melted away and not refilled, forming a groove. Overlap is the opposite: excess weld metal spills over the base metal without fusing to it, forming a rounded overhang at the weld edge.