Railroad Thermite Welding: How Aluminothermic Rail Joining Works
The day I watched my first thermite weld happen at track level, I understood something no textbook had…
READ ARTICLE →The drawing said 1/4-inch fillet, so Marcus ran a bead that looked perfect—smooth, even, and fast. The inspector laid a gauge against it, shook his head, and wrote undersized on the tag. Marcus had welded a 3/16-inch leg by eye, and that single sixteenth of an inch was the difference between a shipping part and a rework job.
Every welder meets a version of Marcus’s day. The bead photographs beautifully. The toes tie in clean. The spatter is minimal. And it still fails, because a weld is not judged by how it looks—it is judged by how big it is, where it sits, and whether those numbers match the drawing. Weld dimensions are the quiet line between a good part and a scrap pile. If your bead geometry is off, no amount of technique saves it; if you understand the numbers, you can catch the problem before the inspector does. That’s the mindset behind solid welding practice, and it starts with knowing what the drawing is actually asking for.
Weld dimensions are the measured values that describe a weld’s size, shape, and fit-up. On a fillet weld, that means leg length and throat thickness. On a groove weld, it means root opening, root land, bevel angle, groove depth, and reinforcement height. Across the joint, it also covers weld length, pitch, intermittent spacing, and how the parts fit together before the arc ever strikes.
Those values are not cosmetic preferences. They are contractual and structural requirements, called out on the drawing and enforced by codes such as AWS D1.1 Structural Welding Code—Steel. A 1/4-inch symbol is a legal instruction, not a starting suggestion.
That cost-and-distortion point deserves a scene. A fabricator once laid a heavy 5/16-inch multi-pass fillet across a thin plate flange because “bigger is stronger.” The weld measured fine. The flange curled anyway, the bolt holes drifted out of alignment, and the crew spent a shift straightening what a correctly sized bead would never have warped. For more on where appearance and real soundness diverge, see our breakdown of good and bad MIG welds.
Quick Note: This article is educational. The governing code and your project specification always win—when they conflict with anything here, follow the spec.
These rules aren’t new. AWS D1.1 was first published in 1928 as the Code for Fusion Welding and Gas Cutting in Building Construction (Code 1, Part A), and it remains one of the most-referenced welding codes in the world. The American Welding Society has authored more than 350 standards covering welding practices and procedures. Nearly a century of field failures, testing, and research sits behind the numbers on your drawing.
People land on this topic for four main reasons: they’re decoding a symbol on a drawing, trying to measure a weld correctly, looking up a minimum or maximum size, or figuring out why an inspector rejected a bead. This article walks through all four—starting with weld anatomy, then symbols, measuring tools, code tables, groove-weld fit-up, common mistakes, and inspection criteria. By the end, you’ll read a weld callout with confidence and catch dimensional problems before they turn into rework.
Before you can measure, verify, or argue about a weld, you need the vocabulary. Weld dimensions are spoken in a language of legs, throats, toes, and roots — precise on the drawing, but dangerously loose on the shop floor. This section builds that dictionary from the ground up.
Picture a T-joint: a vertical plate standing on a base plate, with a bead tucked into the corner. Start at the bottom of that corner and climb to the outside edge — naming each dimension as you go.
In a 90-degree equal-leg fillet, both legs are identical, and that leg length is what the drawing calls out. So “1/4-inch fillet” means a quarter-inch leg — not a quarter-inch throat. Size, technically, is the leg length of the largest right triangle that fits inside the weld cross-section, which for an equal-leg 90-degree fillet equals the leg.
Quick Note: Leg size (what the drawing specifies) and throat (what strength is calculated from) are two different numbers. Confusing them is one of the most expensive assumptions a welder can make.
The theoretical throat is the shortest distance from the root to the face — geometrically, the leg times the sine of 45 degrees. That gives the relationship every welder should memorize: theoretical throat = 0.707 × leg. A 1/4-inch fillet therefore has a theoretical throat of about 0.177 inch.
The effective throat is the number codes actually use in strength calculations. It excludes convexity — the bulge of extra metal piled on the face. For 90-degree equal-leg fillets, AWS D1.1-style tables round the theoretical throat to the nearest 1/16 inch, so a 1/4-inch leg is credited with roughly a 3/16-inch effective throat. This is why a lumpy, convex weld can still be undersized: the extra metal on top is decorative as far as the math is concerned.
AWS A3.0M/A3.0, Standard Welding Terms and Definitions, is the authoritative source for leg, throat, face, toe, root, and convexity. On site, though, “throat” often just means “the size of the weld,” which is exactly the kind of shorthand that tanks inspections.
Groove welds aren’t described by legs and throats. They’re defined by fit-up geometry:
Weld length is the total measured run of the weld. Effective length is the portion actually credited with carrying load, which for fillets excludes poorly fused ends. On intermittent welds you’ll see pitch — the center-to-center spacing of weld segments — and skip, the unwelded gap between them. Getting pitch wrong changes load capacity even when every individual bead is perfect.
Here’s the concept that ties the vocabulary together: a weld must satisfy both a size requirement and a shape/profile requirement. Convexity, concavity, undercut, and toe angle are all inspected dimensions. A bead can hit its leg dimension and still fail on excessive convexity or an undercut toe — the kind of defect that shows up in good and bad MIG welds comparisons and drives rework. Building sound profile habits comes down to deliberate welding practice — travel angle, heat, and consistency.
Note: Learn the terms as AWS defines them, then double-check what your shop actually means when someone shouts a number across the bay.
A print is a contract. Before you strike an arc, the drawing has already decided how big the weld must be—your job is to read it correctly. Spotting a 1/4-inch fillet callout is only the start. You still need to know which side it lands on, how far it runs, and what finish the engineer expects.
Every AWS symbol is built from a handful of parts, defined in AWS A2.4, Standard Symbols for Welding, Brazing, and Nondestructive Examination. Learn these parts and a chaotic-looking print becomes a set of clear instructions:
Quick Note: On a fillet symbol, 3/16 means a 3/16-inch leg—not a 3/16-inch throat. A 3/16 leg yields a throat of roughly 0.133 in, so never treat the two as interchangeable.
The size of a fillet weld goes to the left of the fillet symbol; length and pitch go to the right. Placement tells you the side: a symbol below the reference line means weld the arrow side, above means the other side, and symbols on both sides mean weld both. Equal-leg fillets carry a single number, while unequal legs are dimensioned as two figures. Get the side backwards and you’ll put metal on the wrong face—or miss a weld entirely.
Groove callouts work differently. Bevel, V, U, and J symbols describe the joint prep, and the size marks groove depth or effective throat rather than a leg. Root opening and included angle may appear beside the symbol. Read each number against the joint detail near the arrow—that’s where root gap and land are usually drawn. Groove depth and effective throat are not the same value, so confirm which one the callout is giving you.
ISO 2553 uses a double reference line: a solid line for arrow-side welds and a dashed line for other-side welds. A symbol drawn on the dashed line means only the far side gets welded. ISO also labels leg size as “z” and throat thickness as “a”, so a callout like z8 means an 8 mm leg. US prints rarely use this convention, which is exactly why mixing AWS and ISO systems on one job causes trouble.
US drawings lean on fractional inches (1/4, 3/8) or decimal inches (0.25, 0.375). Metric prints use millimetres. Converting matters: 1/4 in equals 6.35 mm, and a 3/16 in leg is 4.76 mm. Swap a fraction for a millimetre figure carelessly and you can double or halve the weld before the first bead is laid.
The symbol only tells size and length. Process, electrode, polarity, and technique come from the welding procedure specification. On pressure work, that WPS is qualified per ASME BPVC Section IX. If the callout says 1/4-inch fillet but the procedure covers 3/16, stop and ask before you weld.
Good print reading is a skill you build, much like torch control—see our guide to welding practice. And if you’re dialing in a specific rod, our notes on 3/16 E7018 amperage tie machine settings back to the size on the drawing.
You do not need a laboratory to prove a weld is the right size. A handful of inexpensive gauges, a rule, and a little arithmetic will tell you everything an inspector needs to know—and it takes less time than running a second pass you did not need.
The standard fillet weld gauge is the workhorse of dimensional checks. It carries stepped notches sized to common leg lengths—typically 1/8, 3/16, 1/4, 5/16, and 3/8 in. You press the gauge flush against the vertical member and slide a notch up against the toe of the weld. If the notch sits flat on the weld face with no light underneath, that leg is on size. If the toe cuts in above the notch, you are undersized.
There are also throat gauges molded into many gauge bodies, plus dedicated undercut gauges, bridge cam gauges, and V-WAC gauges that read undercut depth, reinforcement height, and pit depth. Together they round out the kit. The AWS A30M standard covers the measuring tools themselves.
Accuracy starts before the gauge touches metal. Spatter, silicon islands, and slag mask the true toe and distort the reading—chip and brush the weld first, or you are measuring the debris, not the bead. On a fillet, keep the gauge square to the joint and flush to the vertical plate; a gauge tilted even a few degrees will read long or short by more than the tolerance allows. This is one reason good welding practice always includes a wire brush and a chipping hammer at the bench.
Leg size is easy to see; throat is what actually carries load, and the two are not the same number. For a 90-degree equal-leg fillet:
throat = 0.707 × leg
So a 1/4 in leg gives about 0.177 in of throat. For unequal legs W1 and W2:
throat = (W1 × W2) / √(W1² + W2²)
And when the joint faces meet at angle θ:
throat = leg × sin(θ/2)
This last one is why skew joints drive welders crazy. As θ drops below 90 degrees, sin(θ/2) shrinks, so you need a bigger leg to reach the same throat—and codes such as AWS D1.1 require you to account for that reduced effective throat.
Quick Note: Leg size is what’s printed on the drawing, but throat is what the structure actually feels. Know which one your gauge—and your code—is talking about before you sign off.
Groove welds demand more than a leg check. Root opening is read with a taper gauge or feeler set. Bevel angle is checked with a protractor or adjustable bevel. Land, or root face, is read with a rule. Reinforcement height is measured with a gauge so it stays inside the code limit. Fit-up tolerances here are tighter than most welders expect, and they are set before the arc ever strikes.
If you are welding a lot of the same joint, set one gauge aside for that size and check every few inches as you go. Catching an undersized bead while it is still hot costs seconds; catching it after the paint booth costs a shift.
That last step is the one shops skip and regret. When a dimensional question comes back months later, the welder’s notes are the only proof of what was actually measured. Regular inspections catch profile problems too—good and bad MIG welds differ in shape as much as in size, and the eye notices it before the gauge does.
Every welder eventually asks the same question: how big should this weld actually be? The real answer lives in two places—the weld symbol on the drawing and the governing code. Rules of thumb get you close, but codes set hard floors and ceilings, and crossing either one costs you a part or a paycheck.
A too-small fillet fails for reasons beyond simple strength. Thin or heavy, heat-sinking material pulls heat out of the puddle fast, so a small bead freezes quickly, cools rapidly, and is more prone to cracking and lack of fusion at the root. A skinny leg also means a skinny throat—the dimension that actually carries the load—so the joint barely has the cross-section the design assumed. Codes set minimums to guarantee enough throat, enough heat input, and enough fusion.
AWS D1.1 ties minimum fillet size to the thickness of the thinner part joined. It starts at 1/8 in for material up to 1/4 in thick and climbs as sections get heavier. The takeaway: thicker steel demands a proportionally bigger leg, not just more passes.
| Thickness of thinner part | Minimum fillet size |
|---|---|
| Up to 1/4 in | 1/8 in |
| Over 1/4 to 1/2 in | 3/16 in |
| Over 1/2 to 3/4 in | 1/4 in |
| Over 3/4 to 1-1/2 in | 5/16 in |
| Over 1-1/2 to 2-1/4 in | 3/8 in |
| Over 2-1/4 to 6 in | 1/2 in |
| Over 6 in | 5/8 in |
Note that thin sheet is the exception: welding a 10-gauge bracket to a plate may be governed by workmanship, distortion, and burn-through risk rather than the table’s floor.
You cannot keep piling metal at a free edge. Where a fillet joins an edge, the maximum leg for material less than 1/4 in thick is the plate thickness itself. For 1/4 in and thicker, it is plate thickness minus 1/16 in, unless the drawing says otherwise. Weld bigger and you melt the edge, roll it over, and risk undercut—the tell-tale sign you’ve overrun the rule, because the arc has eaten into base metal that no longer has the mass to absorb heat. Plus you drag in distortion you’ll pay to straighten later.
Quick Note: These are default rules. An engineer can specify a larger weld with justification; the symbol on the drawing always wins.
Length matters as much as size. A fillet shorter than about four times its size can’t be counted at full value because the ends don’t develop full strength—and many codes set a minimum fillet length near 1-1/2 in. Intermittent welds add two more numbers: the weld length and the pitch (length plus the skip). A callout of 3–6 in means 3 in of weld on 6 in centers, leaving a 3 in skip. Both numbers are tolerance-controlled, because stretched spacing quietly reduces total effective length.
Fillet strength equals effective throat × effective length × allowable stress. For E70 electrodes under D1.1, allowable shear is 0.30 × 70 ksi = 21 ksi. Take a 1/4 in fillet: throat is roughly 0.18 in, so a 6 in weld carries about 0.18 × 6 × 21 ≈ 22.7 kips. Step up to 1/2 in and the throat roughly doubles to 0.35 in—but filler metal consumption roughly quadruples. That is the trap of “just weld it fatter”: you add cost, heat, and distortion for strength the design may already have.
Working the other way—sizing from a load—is simple arithmetic. If a joint must carry 30 kips in shear across 10 in of weld, required throat = 30,000 ÷ (10 × 21,000) = 0.143 in, which converts to a leg of 0.143 ÷ 0.707 ≈ 0.20 in, so a 1/4 in fillet covers it. That is how engineers land on the number you see on the print, and why a “little extra” is rarely free.
The fix is right-sizing. Set a consistent bead, match parameters to the size called out—on heavier 3/16 in E7018 work, amperage and polarity control leg size and tie-in—and check with a fillet gauge before moving on. Build that habit through deliberate welding practice, not guesswork. And remember the hierarchy: rules of thumb are a starting point, but the engineer’s symbol and the governing code outrank them every time.
Fillet welds are sized by legs you can measure after the bead is laid. Groove welds are a different animal. Their dimensions are set by how the joint is cut and fitted long before the arc strikes, which makes a groove weld only as good as the fit-up that precedes it. Get the geometry right and the weld almost finishes itself.
Every groove joint is defined by a handful of numbers, and each one has a job to do:
A common single-V prep for SMAW is a 60-degree included angle with a 1/16–1/8 in root face and a 1/16–1/8 in root opening. Treat those as starting points, not law—the welding procedure specification (WPS) and the governing code rule every job. On thicker plate that same prep often calls for a 45-degree or 75-degree included angle.
Quick Note: Land and gap work together. Shrink one and you usually need to open the other to hold the same penetration. Change either, and you have effectively written a new procedure.
Fit-up tolerances exist because nobody cuts to a perfect line. When the numbers drift, the defects are predictable:
That last point surprises people. Reinforcement is not free strength; it is a geometric notch at the toe where fatigue cracks begin. Structural codes such as AWS D1.1 cap reinforcement for exactly this reason.
A reality check: a 1/8 in gap variation on a 60-degree V swings the cross-sectional area you must fill by a noticeable margin, which changes pass schedule, heat input, and distortion. That is why fit-up discipline matters more on groove work than on most fillet jobs.
Open-root pipe welds are frequently beveled about 37.5 degrees per side—a 75-degree included angle—with roughly a 1/16 in land. The narrower included angle holds heat in a tighter cone, which helps control the root on a fixed joint. Fit-up, hi-lo mismatch, and gap variation are all locked in before the root pass, and the way fittings are machined and tacked shapes the groove you actually get. Our guide to weld tube fittings and fit-up welding shows how those joints come together. On heavy-wall pipe, procedures may widen or tighten the angle to suit the process—a GTAW root often prefers a slightly wider gap and a small land, while a fast root favors a tighter fit.
Backing bars, backing rings, ceramic tape, and consumable inserts all exist to control root dimensions. They support the pool, prevent burn-through, and let a root pass run hotter without dropping through; whether the backing stays or must be removed depends on the code. Pipe procedures under AWS A30M are strict about these details. Some codes require the backing ring to remain in place as part of the joint, while others demand full removal and a smooth root profile. Gauge the gap, land, and bevel before the root pass, not after—measuring is always cheaper than grinding.
Most rejected welds don’t fail because of exotic metallurgy or bad filler metal. They fail because a number was wrong, or because nobody measured it. Here are the nine mistakes that surface again and again on the shop floor, what causes each one, and how to stop making them.
A welder’s eye drifts. The same hand can wander 1/16 in or more across a shift—exactly the margin that flips a proper 1/4-in fillet to undersized. Marcus learned this the hard way. Fix: keep a fillet gauge in your pocket and check the leg every few inches, not once at the end of the pass.
Drawings usually call out the leg; inspectors often check the effective throat. Weld to the wrong one and you’re either dangerously thin or wasting metal. For a 90° fillet, theoretical throat equals leg multiplied by 0.707. Fix: confirm which dimension governs, then measure that one.
A gauge laid on a dirty bead lies. Spatter bulges the profile; slag hides the true toes; a wire edge masks undercut. A fitter once tagged three beads in a row as “cold”—until someone chipped the slag off and found the profiles were fine. Fix: chip and wire-brush the bead before you gauge it—the same clean surface you need when judging good and bad MIG welds.
Sketches often leave fillet size off completely, assuming the shop knows. Codes don’t: minimum leg size scales with the thickness of the thicker part being joined. Fix: look up the applicable minimum before you strike an arc, usually in the AWS D1.1 tables for structural work.
Doubling a fillet leg means roughly four times the deposited metal—more heat input, more distortion, higher filler cost, and slower travel that piles metal over the toe. That’s a direct path to overlap, a defect that looks like a fat weld but fails on profile. Fix: weld to the drawing, not to a feeling.
Arrow-side dimensions live below the reference line; other-side dimensions sit above it. On a complex joint, reading the symbol from the wrong side reverses the whole callout—a classic fabrication error. Fix: trace the arrow to the exact member and confirm which line carries the size before you commit.
On acute-angle joints, an equal leg doesn’t deliver the throat the engineer assumed; on obtuse joints you may over-weld. Fix: check the included angle and increase leg size so the effective throat—not the leg—meets the drawing.
A 6 mm leg looks close to a 1/4-in requirement, but it’s short by 0.35 mm—and inspectors measure. Mixed units across a drawing set cause silent, repeatable failure. Fix: identify the drawing’s primary units and convert every callout before welding.
Undercut limits in structural codes are typically on the order of 1/32 in deep, with stricter length limits for cyclically loaded members. Convexity and reinforcement have caps too, and while some specifications permit a fillet up to about 1/16 in undersize, that allowance must come from the governing document—not from habit. Fix: print the acceptance criteria for your job and keep them at the bench.
Quick Note: A gauge check takes seconds. Reworking a rejected part can cost days of fabrication, re-inspection, and rescheduling—far more than the minute you “saved.”
Dimension control is a skill you build, the same way you build arc control—see our welding practice guide for deliberate drills that make gauge checks second nature.
Marcus’s undersized fillet never made it to a shipping crate, but the inspector who tagged it relied on more than a sharp eye. Verifying weld dimensions follows a defined sequence: visual inspection first, internal testing only where the specification demands it. Knowing that sequence—and the numbers behind acceptance—separates a welder who guesses from one who knows the joint will pass.
Visual testing (VT) is normally the only method used to confirm weld size, length, and profile. Inspectors don’t just glance at a bead. They clean the surface, set low-angle lighting to catch shadows, and reach for magnification—often a 2x to 10x lens—to read toe lines and ripples clearly. Good lighting and magnification aren’t luxuries; they’re how an inspector tells a blended toe from a sharp one.
With fillet gauges, bridge cam gauges, undercut gauges, and calipers, an inspector measures:
Judging appearance is part art, part measurement—a tidy weld can still be wrong. Excess convexity, for instance, often masks a throat thinner than the leg suggests, so visual judgment must be backed by a gauge, not a compliment. Our guide to good and bad MIG welds helps sharpen your eye for profile.
A dimensionally perfect weld can still hide cracks, porosity, or lack of fusion. Non-destructive testing (NDT) catches what gauges cannot:
The method chosen depends on material, thickness, defect type, and the code’s inspection requirements. Gauges tell you whether the weld is big enough; NDT tells you whether it is sound enough. Passing one never guarantees the other.
Acceptance criteria live in the code or project spec, not in a welder’s memory. For statically loaded structural welds, common visual limits include reinforcement no more than about 1/8 in (3 mm) and undercut no deeper than about 1/32 in (0.8 mm). Those tighten sharply for cyclically loaded structures, where fatigue governs.
Note: Documented tolerance is code- and spec-specific. A 1/16 in undersize allowance or a 1/32 in undercut limit is not universal—limits shift with the governing code (AWS D1.1, D1.5, ASME, ISO), the loading condition, material thickness, and the engineer’s written requirements. Inspect to the drawing and WPS, not to a number you heard on another job.
The AWS D1.1 Structural Welding Code—Steel is the usual reference for U.S. structural work.
In fatigue-sensitive structures—bridges, crane rails, machine frames—profile matters as much as size. Excess convexity concentrates stress at the toe, and a sharp toe angle acts like a notch. Worse, heavy convexity can inflate apparent size while the throat stays deficient, so a weld looks full but carries less than the drawing requires. Controlled reinforcement and smooth toe blending reduce stress concentration and extend service life.
When a weld fails dimensional or NDT criteria, respond methodically:
Documenting each step against the welding procedure and drawing protects the shop, satisfies the customer, and gives the next welder real feedback. That paper trail turns an inspection result into a lesson rather than a punishment.
Everything in this guide comes down to one habit: treat every weld as a dimensioned part, not a lump of metal that looks about right. Marcus’s 3/16-inch leg on a 1/4-inch callout was never a skill problem—it was a measurement problem. Here is a workflow that catches that gap before the inspector does.
Most dimension failures aren’t caused by ignorance of the code. They’re caused by skipped steps under time pressure. A welder who knows the minimum-size table by heart can still miss a callout when the fit-up is rushed and the print is smudged. The workflow below is deliberately boring and repeatable, which is exactly why it works.
The payoff compounds. Size-and-profile discipline reduces rework, filler consumption, and inspection rejections across a fabrication shop. Every extra sixteenth of leg on a hundred feet of seam is wire you paid for and arc time you can’t bill twice. Those consistent legs come from repetition—the kind of deliberate practice covered in our guide to building real welding skills—so your hand knows the size before the gauge confirms it.
Dimensions don’t stop at the weld. The leg you leave behind sets the fit for the next assembly, the clearance for a bolt-up, and the stress the joint will carry in service. A dimension check at the bench protects the whole sequence—the moment the symbol and the table disagree, stop and get an answer before you burn wire.
Quality and safety travel together here. OSHA 29 CFR 1910.252 covers ventilation, eye protection, and fire precautions—the basics that keep you able to measure welds next year. Proper protective eyewear and a sensibly planned shop floor matter as much as a good gauge. A well-planned shop also limits fume and arc-flash exposure that cramped spaces make worse.
Short Note: Always verify the current code edition—for example AWS D1.1/D1.1M:2025-AMD1 and AWS D1.2/D1.2M:2026—because superseded editions are still cited on many drawings.
Marcus never welded another 3/16-inch leg on a 1/4-inch callout. He kept the gauge in his pocket, checked after every pass, and his rejection rate dropped toward zero. Measuring the weld is the cheapest insurance in the shop: a few seconds with a gauge against a tag, a grind, and a late shipment.
On an AWS-style drawing, the size next to a fillet weld symbol is normally the leg dimension of the weld, not the throat. For groove welds, the callout typically specifies groove depth, effective throat, or a complete-joint-penetration requirement. Always confirm the governing standard because AWS A2.4 and ISO 2553 use different symbol conventions.
For a 90-degree, equal-leg fillet, multiply the leg by 0.707 to get the theoretical throat. A 1/4 in fillet therefore has a theoretical throat of about 0.177 in, which codes typically round to the nearest 1/16 in for strength calculations. Unequal-leg and skew joints need different formulas.
Minimum fillet size depends on the thickness of the thinner joined part. AWS D1.1’s minimum-size table starts at 1/8 in for material up to 1/4 in thick and increases with section thickness to prevent cracking and ensure adequate throat. Your project specification or engineer may require more.
AWS A2.4 uses a single reference line with arrow-side and other-side conventions, while ISO 2553 uses a double reference line with a solid line for arrow-side and a dashed line for other-side welds. Dimension placement and some contour conventions also differ, so never assume the two systems are interchangeable.
Clean the weld, then use a fillet weld gauge to compare the leg against stepped notches for common sizes and a throat gauge where required. Measure at multiple points along the length and record the smallest reading, because a weld can be correct at one end and undersized at the other.
Reinforcement is the metal built up above the weld face or root surface beyond the required throat. Structural codes commonly limit groove weld reinforcement to roughly 1/8 in because excess reinforcement adds a stress riser at the toe without adding usable throat.
Yes. An oversized weld adds heat input, distortion, filler consumption, and arc time, and it can overrun an edge or create undercut. For equal-leg fillets, doubling the leg roughly quadruples the deposited metal, so over-welding is expensive as well as structurally unnecessary.
Yes. Beyond size, weld callouts specify length and, for intermittent welds, pitch (the distance from the start of one segment to the start of the next) and skip. Minimum length rules—often about four times the weld size—apply so short segments are not counted at full strength.