Convex Weld: Causes, Code Limits, Fatigue Risks, and How to Fix Excessive Convexity

Table of Contents

What Is a Convex Weld? Defining Weld Face Geometry

You finish a fillet weld, step back, and the bead looks strong — rounded, proud, sitting tall over the joint. Your gut says more metal equals more strength. Then the inspector walks over with a convexity gauge and frowns. That rounded crown you were proud of may be the single most common reason a weld gets rejected, and it has far less to do with strength than most welders assume.

A convex weld is one whose face rises above the straight line joining the two weld toes. Draw an imaginary line from toe to toe across the face. If the surface bulges up above that line, the weld is convex. If it dips below, it is concave. If it sits level with the line, it is flat. That toe-to-toe line is your reference for everything that follows.

Fillet Convexity vs. Groove Reinforcement

On a fillet weld, this rounded crown is called face convexity. On a groove weld, the same geometry goes by a different name: reinforcement height, or simply the crown. Same bulge, two terms. The codes give each its own dimensional limits, which is why welders and inspectors sometimes talk past each other.

Convex Is Geometry, Not Automatically a Defect

Here is the point that trips up new welders: “convex” is a purely geometric description, not a verdict. A slightly convex face is common, often unavoidable, and frequently acceptable. Plenty of welds that look “nice and full” are perfectly within spec.

What gets welds rejected is excessive convexity — a crown so tall and narrow that it breaches the limit set by the governing code. Once it crosses that line, convexity becomes a rejectable condition, and the fix is counterintuitive: you remove material rather than add it. The LANL Engineering Standards Manual lists “Overlap or Excessive Convexity” as a condition corrected by removing excess weld material. Too much of a good-looking thing is a defect.

Every major code inspects profile. AWS D1.1, ISO 5817, and ASME Section IX and B31 piping codes all set dimensional limits on weld face geometry. The exact accept and reject numbers differ, but the message is consistent: profile is a quality characteristic, not an afterthought.

Quick Note: Convexity and reinforcement describe the same bulge on different joint types. A fillet weld has face convexity; a groove weld has reinforcement. Both are measured against the toe-to-toe line, and both carry upper limits.

What This Guide Covers

In the sections ahead, we break convexity down from every practical angle:

What causes a crowned face across MIG, TIG, flux-cored, and stick welding
The code limits governing convexity in AWS D1.1, ISO 5817, and ASME
How to measure and inspect convexity in the field with a gauge
Why excessive convexity concentrates stress and shortens fatigue life
When a convex weld is acceptable, or even preferred
How to fix it through grinding, parameter changes, and technique

If you want a visual starting point, our guide to good and bad MIG welds shows what a healthy bead profile looks like next to common defects. Pair that with deliberate time on welding practice fundamentals, and judging profile by eye becomes second nature. Get the toe-to-toe line clear in your head, and every convexity call gets easier.

Convex, Concave, and Flat: How Weld Face Profiles Compare

A fillet weld’s face is the visible surface that runs from one toe, across the crown, to the other toe. That face can bulge outward, sit almost dead flat, or dip slightly inward. Each shape tells you something about heat input, travel speed, and how the load will flow through the joint. Comparing the three profiles side by side is the fastest way to understand why inspectors care so much about a crown. Once you can name the shape, you can predict how it behaves under load.

The Three Basic Face Profiles

Convex (crowned). The face bulges outward above the straight line joining the two toes. Extra filler sits on top like a rounded hill. It looks “strong,” but the metal above the design throat does nothing for strength.

Flat (mitre). The face is essentially a straight line between the toes, forming a triangle that resembles a mitre joint. This is the reference geometry the codes use to define a proper fillet.

Concave. The face curves inward, dipping below the toe-to-toe line. In controlled amounts it is acceptable and even preferred for fatigue, but too much and the throat shrinks below the required size.

Leg, Throat, and Why Geometry Matters

A fillet weld is sized by its legs — the two fused dimensions along each plate. The theoretical throat is measured from the root to the hypotenuse of the largest right triangle that fits within the weld face. For an equal-leg fillet, that works out to about 0.707 times the leg length. Unequal legs shift the numbers, but the principle stays the same: only the inscribed triangle carries the design load.

This is the key point: excessive convexity raises the actual face height but does not increase the effective throat beyond the design leg size. The load still travels through the same small triangle near the root. All that extra crown adds weight, cost, and a steeper toe angle — nothing structural.

Profile	Face shape	Effect on throat	Stress at the toe
Convex	Bulges above toe line	No gain past design leg	High — sharp transition
Flat	Straight toe-to-toe	Matches nominal throat	Moderate
Concave	Dips below toe line	Can fall below required throat	Low if throat is kept

How Each Profile Guides the Load

A flat or slightly concave face provides a smoother transition at the toe, lowering the stress concentration factor. The load path blends gradually into the base metal instead of hitting a notch. That is why fatigue-sensitive joints often favour a slightly concave cap. Think of it this way: the smoother the face, the more gradually the stress spreads into the parent metal, and the longer the weld survives repeated loading.

A highly convex bead does the opposite. The steep angle where the crown meets the plate acts like a built-in stress riser, concentrating stress exactly where cracks like to start. Under cyclic loading, that sharp toe is a fatigue crack waiting for a reason.

Quick Note: Neither extreme is safe. A slightly concave weld is fine, but one that cuts the throat below the required size is just as rejectable as a grossly convex bead — both fail the same acceptance criteria.

Don’t Confuse Convexity With Overlap

New welders often point at a rounded bead and call it convex when it is actually overlap. Overlap (cold lap) is a separate discontinuity — molten metal that rolls over the toe without fusing to the base plate. It can look crowned, but under the surface it is a lack-of-fusion defect, not a geometry issue. The two demand different fixes.

Getting the profile right comes down to reading the puddle and managing heat and travel speed. Practice across positions builds that eye; a solid routine like the one in our welding practice guide helps you hold consistent face geometry instead of chasing it after the fact. Codes such as AWS D1.1 and ISO 5817 set the numeric limits that turn these shapes into pass-or-fail decisions.

Why Convexity Happens: Root Causes Across Welding Processes

Every convex bead tells the same story: the molten pool froze before it could flatten and wet into the base metal. Instead of spreading, the metal piled up in the center and solidified into a rounded crown. The machine changes, the process changes, but the physics behind that crown do not. Convexity is almost always the fingerprint of a cold, slow, or heat-starved welding condition.

Understanding that single idea makes diagnosis far easier. When you see a high, ropey bead, you are not looking at “extra strength.” You are looking at heat that arrived too slowly, spread too little, or both. The tricky part is that each process starves the pool in its own way.

GMAW / MIG: Cold Arc, Slow Hands

In MIG, convexity usually comes from a mismatch between how fast metal is fed and how much heat is available to spread it.

Travel speed too slow. When you creep along, the pool grows faster than it can wet out. Metal stacks up and freezes into a tall crown. Speeding up spreads the same deposit over more length and flattens the profile.
Voltage too low for the wire feed speed. Voltage sets arc length and how widely the arc wets the joint. Crank wire feed speed up while leaving voltage low and you get a stiff, short arc that deposits a narrow, ropey bead that never flattens.
Excessive contact-tip-to-work distance. A long stick-out raises electrical resistance, which drops voltage at the arc. The result is a noticeably colder, more convex bead. Keeping stick-out consistent is one of the simplest fixes, and it’s a core reason contact tips and CTWD deserve regular attention.
Poor torch angle. Angle the gun so it pushes heat away from the base metal and you starve the toes of the weld. A steep drag angle or a lazy, angled torch both reduce wetting and raise the crown.
Wrong polarity. MIG runs on DCEP, which puts more heat into the workpiece. Reverse it and the arc behaves differently and wets poorly.

Listen to any MIG setup: a crisp, consistent crackle means a stable arc, while a harsh sputter paired with a high bead usually means you are running too cold.

Flux-Cored: The Same Cold Arc, Plus Slag

Self-shielded and gas-shielded flux-cored wires inherit every cold-arc problem MIG has, then add slag. Low voltage, slow travel, and long stick-out still produce a raised, ropey bead. Molten slag then blankets the bead and freezes over it, locking in the shape before the metal settles. A heavy, domed slag cap often hides a convex weld underneath. Get polarity wrong for your wire and the arc runs colder still, so confirm DCEN versus DCEP for your specific flux-cored wire before blaming technique.

Stick (SMAW): Low Heat and a Short Arc

Stick welding punishes slow, timid technique with thick, rounded beads. The usual suspects:

Amperage too low. Undersized current can’t melt enough base metal to let the puddle wet in, so it sits on top like a caterpillar.
Arc too short. Holding a tight arc concentrates heat into a narrow spot and reduces the spread you need for a flat face.
Wrong electrode angle. Point the rod so it directs heat away from the joint and the toes stay cold and unwetted.
Weaving too slowly. A slow weave pauses the arc long enough to dump extra metal in one spot, building that stacked, rounded look.

The cure is a hotter, more confident rod: sufficient amps, a slightly longer arc, and steady motion through the weave.

TIG: Starved Pool, Overfed Filler

TIG gives the clearest view of convexity forming because you watch every drop go in. Insufficient current plus slow travel produces a stacked, rounded bead instead of the smooth, wet-out, mirror-like puddle good TIG is known for. Filler is the other lever: add it too fast, or dip too much on each dab, and the metal has no time to flow to the toes. Set the right TIG polarity and heat balance first, then let travel speed and dab timing follow the puddle rather than fighting it.

Quick Note: Across every process, convexity points to one of three shortages: not enough amperage, too little voltage or arc length, or travel too slow for the heat available. Fix the heat balance before you reach for the grinder.

Read the bead and you can trace it back to a setting or a motion. Most convexity fixes live in the same handful of dials and habits, and the next section shows how code limits judge the result. Building that diagnosis skill early is exactly what deliberate welding practice is for.

Code Limits on Convexity: AWS D1.1, ISO 5817, and ASME

A rounded bead might look strong, but what decides whether it passes is the acceptance criteria your contract calls out. Codes don’t ban convexity outright — they put a hard number on how much is allowed, and that number changes with the joint, the code, and the loading. Three families of rules cover most work: AWS for structural steel, ISO for international fabrication, and ASME for pressure and piping.

What AWS D1.1 Allows

For structural steel, the governing document is AWS D1.1. It limits fillet weld face convexity with a simple formula:

C ≤ 1/16 in (2 mm) + 0.07 × w

Here, C is the maximum convexity above the theoretical face, and w is the specified leg size — the leg you were asked to produce, not the one you happened to lay down. That distinction matters, because you don’t get extra allowance for piling on metal.

Run the numbers for a common 5/16 in (8 mm) fillet:

0.0625 in + (0.07 × 0.3125 in) = 0.0625 + 0.022 = 0.084 in (about 2.1 mm)

So on that joint, anything more than roughly 0.084 in of crown above the theoretical face is excessive. The limit scales gently with leg size:

Specified leg, w	Max convexity, C
1/4 in (6 mm)	0.080 in (2.0 mm)
5/16 in (8 mm)	0.084 in (2.1 mm)
3/8 in (10 mm)	0.089 in (2.3 mm)
1/2 in (13 mm)	0.098 in (2.5 mm)

Notice how small those allowances are. On a 3/8 in fillet you’re only permitted about 3/32 in of crown — less than most welders would guess.

Quick Note: The formula uses the specified leg size. A bead stacked on top of a bead doesn’t buy a larger allowance — the limit stays tied to the drawing, not your travel speed.

Groove Weld Reinforcement Limits

Convexity shows up on groove welds too, where it’s called reinforcement — the crown of metal sitting above the plate surface. D1.1 caps that height as well, and the allowance generally grows with the width of the groove face. The key principle is that these are profile limits: a weld can be perfectly fused and still be rejected because the face geometry sits outside the table. Profile and fusion acceptance are separate checks, and both have to pass. Because reinforcement adds nothing to the throat, an oversized cap is pure cost — extra metal, extra heat, and more residual stress for no gain.

ISO 5817 and Its Quality Levels

Outside the US — and increasingly on projects that specify international standards — ISO 5817 is the reference for weld imperfections in steel. It sorts allowable convexity into three quality levels:

B — the highest requirement, for the most demanding, often dynamically loaded work.
C — the intermediate level, common on general fabrication.
D — the most permissive level, suited to non-critical, statically loaded members.

The standard tabulates maximum convexity against leg size for each level, so B allows the least crown and D the most. If you’re used to US practice, note that ISO separates quality by level rather than by a single formula. The current version is ISO 5817:2023, the 4th edition published in February 2023, replacing ISO 5817:2014.

ASME and Pressure/Piping Codes

For pressure equipment and piping, ASME takes over. The ASME Boiler and Pressure Vessel Code (BPVC) governs vessels, while B31.1 covers power piping and B31.3 process piping. These documents carry their own weld profile requirements, and their limits don’t always match D1.1 or ISO. B31.3, for instance, requires reinforcement to blend smoothly into the base metal. A pressure boundary is judged by its own code, so a bead that’s fine on a structural bracket may not clear on a process line.

Which Code Wins?

Here’s the rule: the contract specification wins. When a project invokes a code, that code and the owner’s supplementary requirements govern the work. If several documents touch the same joint, the more restrictive applicable limit wins — and if the engineer specifies tighter convexity for fatigue or appearance, that’s the number you build to. Reading a profile critically is part of telling good and bad MIG welds apart on your own bench.

It’s also worth remembering the lineage behind these rules. AWS D1.1 was first published in 1928 as the Code for Fusion Welding and Gas Cutting in Building Construction. A century of field failures and fatigue testing shaped the current AWS D1.1/D1.1M:2025-AMD1 edition — which is why a limit that looks arbitrary on paper is really a hard-won safeguard. When in doubt, read the specification, then build the bead to fit it. Fundamentals like these are exactly what a structured approach to welding practice builds over time.

How to Measure and Inspect Convexity in the Field

Measuring convexity is less about fancy equipment and more about good light, a clean surface, and knowing exactly where the theoretical face should be. Before any gauge touches the weld, the surface has to be free of slag, spatter, and mill scale—crown height readings on a dirty bead are worthless.

Field Tools That Get the Job Done

The workhorse is the fillet weld gauge set. Most sets include several leaf-style gauges sized to common leg lengths, and crucially, they include a convexity step—a small notch machined specifically to measure the crown height above the theoretical face. You press the gauge against the joint, and if the crown rides above the step, the weld is over the limit.

Related tools cover the rest of the profile:

Bridge cam gauge — measures leg length, throat, and convexity in one pass; the “cam” rides over the crown while the shoulders sit on the plate.
Hi-Lo gauge — checks high/low misalignment and internal fit-up, which affects what you see at the toes.
Dedicated convexity gauges and profile templates — transparent templates let you compare the real face against an accept/reject outline at a glance.

The Math Behind the AWS Formula

AWS D1.1 ties convexity to the actual weld size: the maximum allowed crown is 0.07 times the fillet weld size plus 0.03 in. (0.8 mm), so bigger welds are allowed taller crowns. For a 1/4 in. fillet, that works out to about 0.0475 in. of allowed crown—roughly 3/64 in. The formula exists because crown height scales with bead size; a flat limit would unfairly reject large welds while missing tiny ones. Measure the true leg first, then cap the crown to the matching limit. The AWS D1.1 structural welding code tabulates these values so you never have to guess.

Visual Inspection Sequence

Good practice is methodical:

Clean the weld — remove slag, spatter, and smoke; a wire brush and grinder get you to bare metal.
Set the lighting — bright, angled, and shadow-free. Raking light reveals the toes and any undercut or overlap a head-on view hides.
Read the theoretical face and toes — find where the weld should sit geometrically, then judge crown, toe angle, and blending.
Measure — apply the gauge, confirm leg size, then check convexity against the code limit.

Set the inspection up like a controlled task: stable bench, raking task light, and clean PPE. A well-planned bay makes readings repeatable and keeps sparks and grinding dust away from your gauges, as covered in our guide to setting up welding facilities. Never handle freshly ground or cut metal without protective eyewear for welding, and keep your lenses clean so you can actually see the toes.

Quick Note: A high, narrow bead can hide a cold lap at the toes, so a clean gauge reading doesn’t guarantee fusion.

When to Move Beyond Visual

A weld profile can pass dimensional measurement yet still hide lack of fusion, so visual limits are only a proxy for process control. Gross convexity often signals trapped slag, poor fusion, or a process running cold—exactly the conditions NDT is built to confirm. When you suspect trouble, escalate:

PT (penetrant testing) — surface-breaking cracks and porosity.
MT (magnetic particle) — surface and near-surface defects in ferromagnetic steel.
UT (ultrasonic) — internal lack of fusion, slag, and porosity that a proud crown conceals.
RT (radiography) — volumetric defects, with the caveat that it orients poorly to some planar lack-of-fusion flaws.

Document everything: weld ID, gauge readings, leg size, convexity, code clause, and inspector sign-off. The correlation between quality level and acceptance level for different NDT methods is spelled out in ISO 17635, and the acceptance limits themselves live in ISO 5817. LANL guidance is blunt about the fix: excessive convexity requires mechanical removal of excess weld material, not simply adding more weld.

The Real Cost of Excessive Convexity: Fatigue and Stress Concentration

A convex weld rarely fails the day it’s made. It fails months or years later, usually at 2 a.m., on a bridge deck, a trailer frame, or a machine that never stops cycling. That’s the real cost of a crowned bead: not the metal you added, but the notch geometry you created at the toe.

Why the Weld Toe Is the Weakest Link

When a fillet or butt weld crowns upward, the transition from weld face to base metal becomes abrupt. Instead of a gentle, blended slope, you get a sharp crease, a geometric notch. Under load, stress lines can’t flow smoothly through that region; they bunch up at the toe. That’s a stress concentration, and in welded joints it’s where fatigue cracks overwhelmingly begin, along the toe or at a stop-start region where the profile jumps abruptly.

Ferreira (1989) showed the mechanism clearly: increasing the weld toe radius and reducing the toe angle lowers the stress concentration factor and raises fatigue strength. In everyday terms, picture a highway curve. A generous radius spreads the load like a long sweeping bend; a sharp toe forces the same traffic through a hairpin, so everything accelerates locally.

The stress concentration factor is a multiplier, not a minor correction. A sharp toe can push local stress several times above the nominal value, which is exactly why fatigue strength, not yield strength, governs the design of a cyclically loaded joint.

Hot-Spot Stress and Fatigue Design Categories

Modern fatigue design moves past nominal stress and uses structural hot-spot stress, evaluated from principal stresses perpendicular to the weld toe within roughly a ±60 degree range. IIW-based hot-spot fatigue curves for gusset joints typically fall in the FAT71 to FAT80 range at 2 million cycles. Translation: at two million cycles, the joint survives a stress range of only about 71 to 80 MPa.

Compare that with the parent metal and the penalty becomes obvious. In a study of 6005 aluminum alloy MIG lap joints, the welded joint’s fatigue strength was 32.75 MPa versus 119.78 MPa for the base metal, a mere 27.34% of the base metal. Welded joints routinely sit far below base-metal fatigue strength, and added convexity widens that gap by sharpening the toe geometry.

Quick Note: Convexity itself rarely governs static strength. What governs fatigue is the toe notch that excessive crown creates.

Why Cyclic Loading Punishes Convexity

Static loading and cyclic loading treat a convex weld very differently. A single overload cares mostly about net section and material strength, so a fat bead often passes the day it’s built. Fatigue cares about geometry and repetition. Every stress cycle pries at that notch, and millions of cycles turn a tiny concentration into an initiating crack.

Extra crown metal also makes the weld cross-section locally stiffer than the adjoining plate, creating an abrupt change in stiffness and a sharper toe angle. The load path has to negotiate that change, and the toe becomes the point of highest local stress.

That’s why fatigue-sensitive structures treat toe profile as a design variable, not a cosmetic detail:

Bridge girders and decks under relentless traffic cycles
Trailer and chassis frames flexing over millions of road miles
Vibrating machinery, pumps, and rotating equipment
Offshore platforms and pipelines loaded by waves and pressure

The fabricated members behind these joints begin as rolled plate and shapes, so upstream quality feeds directly into toe profile. If you want that full picture, our guide to metal rolling and welding shows how the processes work together, while consistent welding practice keeps toes smooth and blended rather than sharp.

Why Static Joints Still Reject Gross Convexity

Even when fatigue isn’t the design driver, an inspector who flags gross convexity usually isn’t worried the bead will break. They’re reading it as a process-control red flag: heat input too low, travel speed too slow, wrong polarity, or poor manipulation. Any of those can bury lack-of-fusion, cold lap, or overlap under that rounded crown. A convex weld is often a symptom, not the disease.

A crowned bead hides small sins, and it hides them literally: the rounded profile can mask an overlap defect or cold lap that a flatter face would expose immediately. Those defects, not the convexity measurement itself, are what actually fail under load. For a visual breakdown of healthy versus unhealthy beads, see our guide to good and bad MIG welds.

When a Convex Weld Is Acceptable — or Even Preferred

Not every rounded bead in your shop is destined for the grinder, and a convex profile is not automatically a defect. The reality is more nuanced: modest reinforcement within code limits is normal, frequently unavoidable, and in certain applications genuinely desirable. What matters is the joint type, the leg size, the loading, and above all the governing specification.

A Modest Crown Is Normal and Often Unavoidable

When you deposit a fillet, gravity and surface tension pull the molten pool into a slightly rounded shape. A small crown is simply what a healthy bead looks like on many joints. Within code limits, that crown does not increase the effective throat beyond the design leg size, so you gain no structural credit for it. In other words, extra metal above the theoretical throat line is “free” material the code ignores when calculating strength. It adds cross-section and a bit of stiffness, but it does not make the joint stronger than its designed leg dimension.

This is where good technique pays off. Controlling travel speed, work angle, and heat input keeps convexity inside allowable bounds instead of spilling over. If you are refining those fundamentals, the complete guide to building real welding skills walks through the deliberate practice that tightens up bead profile. Sharpening your eye helps too: comparing your profiles against reference images, like those in our guide to telling good and bad MIG welds apart, makes it easier to catch a crown drifting out of range.

Code Notes That Exempt Convexity

Structural codes are not monolithic. AWS D1.1 includes notes clarifying when welds are exempt from reinforcement and convexity limitations, covering specific joint configurations, certain weld sizes, or particular loading conditions. These exemptions matter because they tell you where a rounded crown is not a rejection risk at all.

Rules also evolve. A minimum allowable convexity requirement was eliminated from a schedule for outside corner joints in a later D1.1 edition, a clear sign that what counts as “correct” gets revisited as experience accumulates. That is a good reason to work from the current edition of your governing code rather than relying on shop-floor folklore.

Quick Note: A convex profile that sits within the reinforcement limit is compliant. The defect only appears when the crown exceeds the code’s convexity allowance for that joint and size.

Where Convexity Earns Its Keep

Some applications want a thick, crowned deposit. Wear-facing and hardfacing are the classic examples: the goal is abrasion resistance, not a structural fillet, so operators intentionally build up a heavy, rounded layer of hard metal. Here convexity is the point, and judging it by fillet-weld rules would be a category error.

Repair welds can also tolerate a fuller profile. When you are restoring a worn or damaged area, a slightly proud bead may be acceptable under the governing specification, especially where matching the original surface matters less than soundness. The deciding factor remains the spec, not the shape alone.

Don’t Fall for “More Metal Is Stronger”

The tempting logic — a bigger, rounder bead must be tougher — falls apart in practice. A weld’s capacity is governed by its effective throat and leg size, and code allowances already bake in the reinforcement that does not count toward strength. Piling on extra metal beyond the limit just adds weight, heat, distortion, and stress-concentration risk without adding load capacity.

So treat convexity as a design and inspection question, not a gut feeling. Read the governing code, check the joint and thickness, and let the specification decide. Where a slight crown is permitted, leave it and move on. Where it is not, that is your cue to fix it, which we cover next.

Fixing Excessive Convexity: Grinding, Parameters, and Technique

You’ve measured the bead, the numbers don’t clear the acceptance limits, and the inspector is waiting. Fixing it is a two-part job: repair the weld you already have, then change the settings so the next one comes out right. Do the second part well and you’ll spend far less time doing the first.

Repair: Grind It Down, Don’t Pile It On

Correcting excessive convexity requires mechanical grinding to restore the profile. That’s the whole game. Re-welding on top of a convex bead typically makes it worse: you add heat, add more metal to an already swollen crown, and finish with a taller bead plus fresh distortion.

Start by confirming the weld is actually rejectable. Acceptance limits vary by code and application, so check the relevant clause in AWS D1.1 or ISO 5817 before you touch it. Weld profile gauges include a convexity step, which makes the check quick and repeatable right at the bench instead of a guess from a photo.

Then work the profile back with a grinder:

Bring the crown down to flat or slightly convex. A near-flat face with a smooth toe transition is the target, not a deeply dished-out valley.
Feather the toes. Blend the edges into the base metal so there’s no visible line. Sharp grinding marks or notches defeat the purpose by creating stress risers.
Respect the leg size. Don’t remove so much that the weld drops below the required throat or leg dimension. Measure the legs after grinding.
Keep it cool. Light passes with a wheel matched to the steel avoid tempering the surface or gouging a channel you then have to fill.

Quick Note: Grinding throws just as many sparks as welding. Wear tinted protective eyewear and a face shield during repair work, then swap back to your welding lens when you strike an arc.

Prevention: Fix the Variables, Not the Bead

Most convexity traces back to a puddle that’s too cold, too slow, or carrying too much metal. Correct the input and the profile largely corrects itself.

Raise voltage relative to wire feed speed. In GMAW, this softens the arc and flattens the bead, spreading the puddle instead of stacking it into a crown.
Raise travel speed, moderately. Faster travel reduces puddle pile-up, but push too far and you trade convexity for undercut and lack of fusion.
Reduce wire feed or current. Less metal deposited per unit length means less height in the bead.
Set the contact-tip-to-work distance. Too long drops voltage and stalls wetting; too short crowds the arc. This is a low-cost fix that addresses many convexity complaints, so check your contact tip setup and wear first.
Fix torch and work angles. The wrong angle steers molten metal onto one plate and can’t wet the other. A consistent angle lets the bead flatten on its own.
Check polarity and fit-up. The right polarity for the process and a tight, properly prepped joint remove the gaps that force you to fill and crown.

Process-Specific Tuning

Every process has its own lever for flattening the bead. Small adjustments go a long way:

Process	Adjust this
GMAW (MIG)	Raise voltage around 0.5–1 V, shorten CTWD, add a slight push angle, trim wire feed speed.
FCAW (flux-cored)	Nudge voltage up, confirm DCEN vs DCEP, and keep the contact tip clean. Our guide to gasless MIG welding covers this setup in detail.
SMAW (stick)	Run slightly more amperage, hold a marginally longer arc, and move faster. See typical 3/16 E7018 amperage settings.
GTAW (TIG)	More current, less filler wire, a slightly longer arc, and steady travel. Polarity matters here — review TIG welding polarity.

Aluminium behaves differently again. Its cleaning action and puddle wetting shift with polarity, so if you’re working on it, review how DC welding aluminium changes bead shape before you start tuning.

Test on Scrap Before the Real Weld

Don’t dial in new settings on the production part. Weld a short coupon, change one variable at a time, and measure each bead with the gauge. Write down what worked, because the same numbers usually transfer.

Deliberate welding practice on scrap is how you stop fighting convexity and start anticipating it. Once your ear and eye know what a soft arc looks and sounds like, the gauge becomes a formality.

Common Mistakes, Practical Tips, and a Convexity Checklist

Most convexity problems aren’t mysteries. They come from a short list of habits repeated across the shop, and once you can name them, they’re easy to catch. Below are the recurring errors to eliminate, the checks that flag them before they reach inspection, and the routines that keep profiles in spec shift after shift.

The Mistakes That Keep Showing Up

Confusing convexity with overlap. A rounded crown that sits high but still wets into the base metal is convexity. Metal that rolls past the toe without fusing is overlap, a different defect with different consequences. Using the two terms interchangeably sends the wrong repair instructions to the floor, and welders end up grinding when they should be re-welding.

Judging welds by appearance alone. This is the dangerous one. A convex bead can visually mask lack of fusion at the toes, so it can sail through an appearance check while hiding a serious defect. Visual inspection is a screening tool, not a verdict.

Adding metal to fix a cold bead. Piling a second pass over a cold, overly convex bead doesn’t add strength. It adds mass and buries the real problem: insufficient heat or an excessively slow travel speed.

Ignoring WPS voltage and travel ranges. Slow travel speed and low voltage are the two most common parameter causes of excessive fillet convexity. Operators who treat the procedure as a suggestion drift outside both without noticing until a gauge or an inspector catches it.

Welding over mill scale or moisture. Contaminants change wetting and arc behavior, worsening profile and inviting porosity.

Measuring without cleaning the weld. Slag, spatter, and silicon islands throw off a convexity gauge. Clean the crown before you measure it.

Quick Note: Weld profile is treated as a proxy for process control, so a profile failure often triggers a broader WPS review rather than a quick touch-up. That’s why AWS D1.1 and ISO 5817 both put hard numbers on face profile.

Pre-Weld Checklist

Read the WPS; confirm voltage, amperage, and travel-speed ranges.
Dial in parameters on scrap before touching the part.
Inspect the contact tip and liner; replace worn consumables.
Clean the joint to bare metal — no mill scale, oil, or moisture.
Set a steady work angle and a comfortable gun position.
Verify fit-up and gap against the drawing.
Stage the correct gauge and tools at the workstation.

Post-Weld Checklist

Clean slag and spatter from the weld face.
Measure convexity with the correct gauge for the joint size.
Compare the reading to the applicable code limit.
Inspect both toes for overlap or lack of fusion.
Confirm the repair plan before grinding anything.
Record the measurement and disposition.

Shop Habits That Build Consistency

Consistency is boring by design. Maintain contact tips and liners on a schedule, control travel speed with a steady rhythm instead of racing, and hold a consistent work angle across the whole bead. Keep a calibrated gauge within arm’s reach so measuring is never a chore. The goal is to make the correct profile the default, not something you chase weld by weld.

A few skill guides are worth bookmarking: the how to use a wire welder guide covers travel and technique fundamentals, and the spot weld cutter bit reference is handy when you need to remove a bad weld cleanly before redoing it.

For Supervisors and Inspectors

Document profile nonconformance with the measurement, the code clause, and a photo. Track convexity rejections by welder and process to spot trends early, and review those trends weekly rather than waiting for a quality meeting. Then coach rather than just reject: most excessive convexity traces back to travel speed or voltage, so point the operator at the parameter, not the person. A five-minute conversation at the machine prevents the next ten rejections — and it’s far cheaper than grinding.

Conclusion: Getting Convexity Right on Every Weld

Here’s the single takeaway from everything above: match your weld face geometry to the governing code limit, not to how the bead looks. A convex weld isn’t inherently strong or weak because it’s crowned. It’s acceptable or rejected based on measurable profile, and that profile sits inside the spec you’re welding to. Once you internalize that, the convexity gauge stops being an inspector’s tool and becomes yours.

Think back through the causes. Too much wire feed, travel speed that’s too slow, a long arc, low current, and a gun angle that dribbles metal onto the joint all pile metal into the center of the face. The result looks robust but hides a real problem: those tall, pinched weld toes are the dominant driver of fatigue life in cyclically loaded joints. Geometry, not volume, decides how a weld carries stress. That’s why fatigue research keeps circling back to toe shape as the controlling factor. A fat, rounded crown also traps slag and can mask incomplete fusion underneath, so what looks like extra insurance is often the opposite.

Quick Note: A crowned bead is a signal, not a sentence. When you see one, check your parameters and technique before you reach for the grinder. Fixing the process is faster and cheaper than rework.

The Workflow That Keeps You Out of Trouble

Recap the practical loop. First, know your governing spec — whether that’s the AWS D1.1 convexity formula, an ISO 5817 quality level, or an ASME acceptance criterion. Read it before you assume a crown is fine, because a profile that’s perfectly acceptable in one application can fail another. Second, dial in amperage, WFS, and travel speed so the face flattens naturally as you weld. Third, check profile on every bead with a gauge — regular checks catch the problem before an inspector ever walks up. Treat these three steps as one repeating cycle, not a one-time setup, since heat, fit-up, and position all shift as a job progresses.

Remember the exceptions, too. There are legitimately cases where a slight crown is fine or even preferred, so don’t grind a sound weld just to satisfy your eye. And when convexity genuinely exceeds limits, correcting technique and parameters beats grinding or full rework on both time and cost.

Build the Habit, Own the Bead

Good geometry is a skill, not a lucky pass. It comes from repetition, from watching the puddle instead of the arc, and from adjusting on the fly. If you want to sharpen those instincts, work through our welding practice guide and study how polarity and settings shape bead profile on processes like flux core and wire welding.

So tune the parameters, check the profile, and treat every crowned bead as feedback rather than failure. That habit separates welders who pass inspection from welders whose work never gets questioned. Keep learning with our full skill and technique resources, and stay sharp on welding safety while you’re at it. Get convexity right, and you’ll get a lot more than geometry right — you’ll build the kind of consistency that holds up under load, under inspection, and over the life of the structure.

Frequently Asked Questions

What is a convex weld?

A convex weld is one whose face rises above the straight line joining the weld toes, giving a crowned or rounded profile. On fillet welds this is called face convexity; on groove welds the equivalent feature is reinforcement height. It becomes a defect only when it exceeds the limits set by the governing code.

What is the maximum convexity allowed on a fillet weld under AWS D1.1?

AWS D1.1 limits fillet weld face convexity to C <= 1/16 in (2 mm) + 0.07 x w, where w is the specified leg size. For a 5/16 in fillet this is roughly 0.084 in (about 2.1 mm) above the theoretical face. Values above the limit are rejectable regardless of fusion quality.

Is a convex weld stronger than a flat weld?

No. The effective throat of a fillet weld is based on the design leg size, so extra crown height does not add usable strength. Excessive convexity actually raises the stress concentration at the toes and can reduce fatigue life. A flat or slightly convex face is generally preferable for structural joints.

Why does my MIG weld look like a rope with a high crown?

A high, ropey crown usually means the arc is running cold or too slow. Common causes are low voltage relative to wire feed speed, travel speed that is too slow, excessive contact-tip-to-work distance, or a torch angle that pushes heat away from the base metal. Tuning voltage up slightly and increasing travel speed usually flattens the bead.

Can I just weld over a convex bead to fix it?

No. Adding more weld on top of a convex bead typically makes the profile worse and can trap slag or lack of fusion. The correct repair is to remove excess material by grinding, restore a flat or slightly convex profile, and blend the toes smoothly into the base metal.

Does ISO 5817 cover convexity?

Yes. ISO 5817 sets limits for weld imperfections, including profile features, across three quality levels designated B, C, and D, with B being the highest requirement. ISO 5817:2023 is the current edition and applies to fusion-welded joints in steel, nickel, titanium, and their alloys.

Why is excessive convexity a problem in fatigue-loaded structures?

A crowned bead creates an abrupt section change and a sharp toe angle, which concentrates stress exactly where fatigue cracks initiate. Research shows that increasing the toe radius and reducing the toe angle improves fatigue strength. This makes profile control critical for bridges, machinery, trailers, and other cyclically loaded joints.

How do I measure convexity on a weld?

Use a fillet weld gauge set with a convexity step, a bridge cam gauge, or a profile template, and compare the measured crown against the code formula for the specified leg size. Clean the weld first, use good lighting, and verify the theoretical face and toe locations before measuring.