Welding Joint Types Explained: The 5 Basic AWS Joints, Groove Designs & How to Choose the Right One
Master all welding joint types: butt, tee, corner, lap and edge joints, groove preparations, CJP vs PJP, common…
READ ARTICLE →A fabrication shop once quoted a routine 50 mm pressure vessel head seam as a double-V joint, and the welder burned through three days of filler metal before anyone noticed the U-groove on the drawing would have cut the job nearly in half. That single misread groove symbol is the kind of quiet, expensive mistake this article will teach you to never make again.
A fabrication shop in the Midwest once quoted a routine 50 mm pressure vessel head seam as a double-V joint. It was not until the welder was three days into the job that they realized the symbol called for a double-U groove. The U prep would have roughly halved the deposited metal on that seam. Nobody got fired, but everybody got a lesson: groove geometry is not a minor detail. On heavy plate, it is one of the biggest levers for cost and quality.
That mistake is far more common than it needs to be. Before defaulting to a straight chamfer, it pays to know exactly what a U groove weld is, when its dimensions fall under code such as AWS D1.1, and how it stacks up against other groove preparations.
A groove weld is any weld made in a prepared opening between two members. A U groove weld takes that opening a step further by machining a curved, U-shaped profile into both edges. Instead of the straight chamfers found on a V or bevel groove, the walls sweep down to a small, radiused root.
This radius is the reason it earns its keep. It allows the arc to reach the bottom of the joint without wedging into a tight vertex, gives the molten pool room to wet out, and helps smooth the transfer of stress through the root. On thick sections, those rounded corners mean far less weld metal is needed to fill the joint, which is exactly what would have saved those three lost days.
All welds fall into two families. Fillet welds join overlapping or crossing members in a triangular cross-section. Groove welds join members edge-to-edge in a prepared channel, and they are the choice when the joint must carry full load through the thickness of the plate.
The AWS groove family includes several standard preparations: square, V, bevel, U, J, and flare grooves. Square grooves rely on penetration alone and are limited to thinner plate. V and bevel grooves use straight chamfers, while U and J grooves use curved profiles. Flare grooves take advantage of the natural curve of a rolled or bent surface, making them common in HSS and tube work.
One detail sets the U groove apart: a U groove prepares both members of the joint, symmetrically. Bevel and J grooves only prepare one member. That symmetry is why a U groove symbol never needs a broken arrow, as outlined in AWS A2.4 Section 6.4.1. It also carries over to shop drawings, where AWS joint designation number 6 is a single-U-groove weld and number 7 is a double-U-groove weld. For bridge and highway fabrication, those numbers come up often.
Quick Note: Not every groove weld penetrates the full thickness. AWS D1.1 separates them into complete joint penetration (CJP), which fuses the entire joint, and partial joint penetration (PJP), which only fuses a specified depth. A CJP’s size is taken as the thinner part joined, with no credit for reinforcement, while a PJP must have its depth clearly stated on the drawing.
Groove selection is rarely about metallurgy alone. It is about time. According to industry cost breakdowns, labor and overhead account for roughly 85% of total weld cost, while filler metal is only about 10 to 15%.
That flips the usual assumption: cutting machine time for a U prep is often far cheaper than burning that metal with an arc. Fewer passes, fewer starts and stops, less distortion, and less consumable changeout all add up. If you are still weighing your filler options, our guide on what electrodes are in welding can help round out the equation.
For thin plate, a straight chamfer may be more than enough. For heavy, load-bearing seams, the curved geometry of a U groove weld is designed to do the same job with less arc time and less weld metal. It will not save money if the prep is unnecessary, but when the code calls for it, skipping it can cost far more than machining it ever will.
The rest of this guide will dig into the specific dimensions, symbols, and decision points that put that lesson into practice without repeating the one that cost that shop three days.
Before anyone strikes an arc, a U groove is just a set of numbers on a drawing. If you can read those numbers, you can check a fit-up in thirty seconds and catch a bad edge before it becomes a repair. If you can’t, you’ll be guessing. So let’s take the joint apart and name every piece of it.
Most groove geometry comes down to a handful of terms, and they show up in every standard from AWS to ISO. Here’s the plain-language version:
Notice that a U groove has more knobs to turn than a V, and that flexibility is exactly why it exists. A plain V gives you one choice: open the angle wider for better access, and pay for it in filler metal. A U lets you hold a narrow included angle while still leaving a wide, curved pocket the electrode can actually reach.
For anyone working to AWS D1.1, the prequalified single-U complete-joint-penetration butt joint detail is designated B-U6 for shielded metal arc welding. The numbers are worth memorizing:
| Detail | Process | Groove angle | Root face | Root radius | Root opening |
|---|---|---|---|---|---|
| B-U6 (SMAW) | Stick | 45° flat; 20° flat and overhead | 1/8 in (3 mm) | 1/4 in (6 mm) | 0 to 1/8 in (0 to 3 mm) |
| B-U6-GF | GMAW / FCAW | 20° flat | 1/8 in (3 mm) | 1/4 in (6 mm) | 0 to 1/8 in (0 to 3 mm) |
Two things jump out. First, the root face and root radius stay fixed even when the angle changes, because those two features guarantee the arc can reach the bottom. Second, the angle narrows dramatically for out-of-position and semiautomatic work: 45° drops to 20° when the welder isn’t fighting gravity. Under ISO 9692-style joint preparation tables, heavy-plate U preparations follow the same logic: a narrow included angle, a small root face, and a generous root radius, all held to tight tolerance.
Note: The four numbers to write on every U-groove fit-up checklist: groove angle (α), root face (f), root radius (r), root opening (R). If a fit-up doesn’t match all four, don’t tack it.
A sharp V bottom forces the arc into a wedge. The sidewalls converge to a point, the arc has to climb those converging faces, and heat concentrates in a tiny volume. That makes fusion inconsistent and leaves a high-stress notch right where cyclic loads concentrate. A radiused root spreads the same arc energy over a broader, smoother surface. The puddle wets out along a curve instead of a corner, sidewall fusion becomes repeatable, and the finished root profile has no sharp re-entrant angle to act as a stress raiser.
The root face matters just as much. Too thin, and the arc blows through before the joint is filled. Too thick, and you get lack of fusion at the bottom because the process simply can’t melt that much unassisted metal. Typical maximum penetration tells the story: roughly 3 mm for SMAW and GTAW, about 6 mm for GMAW, and up to 15 mm for submerged arc welding (TWI design guidance). Hold a 3 mm root face with a stick electrode and you’re at the edge of what the process can do, which is precisely why D1.1 caps it there. That’s also why groove prep exists at all: no common process penetrates thick plate on its own.
U and J preparations cost more to machine than a flame-cut bevel, and they give the arc a slightly tighter approach at the bottom. In exchange, they consume noticeably less weld metal and produce less angular distortion and residual stress than an equivalent V (Hobart Brothers groove weld guidance). We’ll put real numbers on that trade in the next section, but the geometry is the whole argument: a narrower angle filled over a radiused root is less metal, less heat, and less pull. If you’re setting up equipment for this kind of work, our MIG welder setup guide covers the parameter discipline that tight angles demand.
Master these terms and the rest of the joint, the symbol, the pass sequence, and the inspection criteria all start to line up.
A groove is only half the story. The other half lives on the drawing, in the little arrow-and-flag language that tells the welder what shape to cut, how deep to go, and who is responsible for the root. Get the U groove symbol wrong and you get one of two expensive outcomes: a joint welded to the wrong geometry, or a joint that looks right but was never specified for full penetration.
The governing reference is AWS D1.1/D1.1M, which leans on the symbol vocabulary defined in AWS A2.4. The U-groove symbol itself appears in AWS A2.4 Figure 4.1, sitting alongside the square, V, bevel, J, and flare variants. If you can read a V, you can read a U, because the mechanics of the reference line do not change. What changes is the extra geometry a curved root demands.
Everything starts with one rule. Weld symbols placed below the reference line apply to the arrow side of the joint, the side the arrow points at. Weld symbols placed above the reference line apply to the other side. That convention is the same for a U groove as it is for a butt weld on a sheet metal joint, and it is the first thing every junior detailer should tattoo on their forearm.
Two numbers ride to the left of the symbol. Groove depth is written as D, and for partial-joint-penetration (PJP) work, the weld size S appears in parentheses next to it. A PJP call for a 20 mm groove machined to 14 mm of required fusion reads, in effect, “U symbol, D14 (S14)” — the detailer is telling you exactly how much of the groove must actually be filled with weld metal.
Now the counterintuitive part: on a complete-joint-penetration (CJP) call, both D and S are deliberately omitted. That omission is not laziness. When D and S are absent, the symbol specifies full penetration through the entire joint thickness, regardless of how the edges were machined. Adding a size to a CJP call would quietly downgrade it to a PJP in the eyes of anyone reading the print, and you would lose the throat thickness the design depends on. Read our related explainer on welding procedures, positions, passes, and inspection to see how that distinction plays out in a live WPS.
This is the rule that separates people who actually read standards from people who guess. AWS A2.4 Section 7.3.3 requires that groove radii and root faces of U- and J-groove welds be specified by cross-section, detail, or other data, with the reference carried in the tail of the welding symbol. The little envelope-shaped tail after the reference line is not decoration. It is where you point to a specific joint detail, a detail number from AWS D1.1 Figure 5.1, or a shop standard sheet that defines the 6 mm radius and the 2 mm root face you actually want cut.
Quick Note: The three most common U-groove symbol errors fabricators make: (1) omitting the tail reference, so the machinist guesses the radius and root face; (2) writing D and S on a CJP call, which changes the penetration requirement; and (3) flipping the symbol to the wrong side of the reference line, putting the machined U on the face that has no torch access.
Picture a 40 mm butt joint, double-U, CJP. The arrow points to the left member. Above the reference line sits a U symbol; below it sits an inverted U symbol. No D, no S, because full penetration is required from both faces. In the tail: “DETAIL A / D1.1 FIG 5.1-5.4” — and Detail A carries the radius, root face, and the mandatory note that appears in the AWS joint detail legend: “Gouge root to sound metal before welding second side.” That single line of tail text is what obligates backgouging on every prequalified U-groove CJP joint under D1.1.
Double-check the side rule, confirm whether the tail reference actually exists in the package, and resist the urge to add a size to a CJP call. Do those three things and your U groove symbol will mean the same thing to the machinist, the welder, and the inspector, which is the entire point of having a symbol in the first place.
Every groove style is a trade between what the edge prep costs and what the weld costs. The U groove is not automatically the smarter choice, it just wins in a specific band of thickness and volume. Here is the honest scorecard.
Weld metal volume is where the argument gets real. On heavy plate, a single-V groove with a 60-degree included angle has a cross-sectional area of roughly 1.49 in². Narrow that same joint to 45 degrees and the area drops to about 1.12 in², a 24% reduction in filler for the same plate thickness. Flip it to a double-V at 60 degrees and you are down to about 0.84 in², a 43% cut against the original single V.
That is not a rounding error. Weld metal volume in a V prep scales roughly with the square of thickness, so the penalty compounds as plate gets heavier. On 20 mm plate a double V costs around half of a single V, and a U preparation costs less again, because the radiused bottom keeps the upper region narrow while still leaving enough room for the arc to reach the root. This is exactly why the shop story in our U groove weld guide played out the way it did: same seam, different prep, wildly different consumable burn.
Note: Rule of thumb thickness bands — under 10 mm: square or single V, never a U. 10 to 20 mm: single V, bevel, or double V. 20 to 50 mm: double V, narrow-angle V, or start weighing a U. Above 50 mm: U or double-U usually pays for itself. Above 100 mm: double-U is the default heavy-section answer.
Here is the trade nobody sugarcoats. V and bevel grooves can be produced economically by machining or flame cutting, which is why they dominate general fabrication. A torch and a track burner can put a bevel on 40 mm plate in minutes for pennies of gas. U and J grooves must be machined, and a milled or planed radius profile costs real setup time on a CNC profile mill or a planer. For a one-off bracket or a short run of frames, that machining quote alone can outweigh every dollar of filler you would save.
Preparation cost also depends on batch size. A single set-up that runs twenty identical heavy girders amortizes the cutter programming across all of them. A single pressure vessel head seam that will never be repeated does not.
| Criterion | Single V (60°) | Bevel / J | Single U | Double-U |
|---|---|---|---|---|
| Prep method | Flame or machine | Flame or machine | Machine only | Machine both edges |
| Filler volume | Highest | High | Low | Lowest |
| Distortion | High, one-sided | High, asymmetric | Moderate | Low, balanced |
| Arc access at root | Excellent | Excellent | Good with root face | Good both sides |
| Fit-up precision | Forgiving | Forgiving | Tight radius tolerance | Tightest |
| Best thickness band | 3–20 mm | 3–40 mm | 40–100 mm | 50 mm and up |
A single-U keeps preparation to one edge per member, so fit-up is simpler and only one side needs gouging access later. A double-U halves the deposited volume again and lets you alternate sides to cancel angular distortion, but it doubles the machining, requires backgouging on both faces, and demands access to both sides of the joint. On a heavy butt joint where both edges sit on the table and are accessible, the double-U usually wins on distortion. Where the assembly can only be reached from one side, the single-U is the practical call.
The distinction is symmetry. The J groove prepares only one member, leaving the other square, which makes it ideal when one edge is inaccessible or when you are joining a machined part to plate. The U groove prepares both members, which makes it the symmetric choice when both edges are accessible on a heavy butt joint and you want an even wall profile for consistent fusion on each pass. A double-J pair approximates a U, but it leaves two sharp internal corners where stress concentrates; the single continuous radius of the U flows stress more smoothly, which matters under cyclic load.
Quick Note: Where the U groove loses — thin plate, tight budgets for one-off parts, jobs without a mill, and out-of-position welding where the narrower angle leaves little room to manipulate the electrode. In those cases a bevel or narrow-angle V is simply the better engineering decision, consistent with groove selection guidance from TWI and the AWS AWS D1.1 prequalified joint details.
The short version: prep cheap and fill heavy with a V on light work; machine once and fill sparingly with a U on heavy work. The crossover sits where filler volume starts to outweigh the cost of a cutter, which is why estimating a job by groove cross-section times joint length is the habit that keeps U groove joints profitable.
Knowing how to cut a U groove is only half the job. The other half is knowing when one is worth cutting at all. Pick a U groove on a 12 mm bracket and you have paid for machining you will never recover in filler metal. Skip it on a 60 mm pressure vessel seam and you will burn roughly three times the wire needed to fill a narrower profile. Here is a decision framework you can run in about two minutes at the drawing board.
Start with thickness, because it drives everything else. Practical fabrication guidance places U and J preparations on heavy walls, commonly above about 25 mm (1 in), and especially where wall thickness exceeds 1 in. Below that band, the cross-sectional area difference between a 60-degree V and a U profile is small, so machining, edge handling, and extra fit-up time outweigh the saved filler. Above it, the math flips hard: a U groove’s narrow parallel-walled channel grows far more slowly in area than a V’s widening wedge, and that gap compounds with every additional millimetre of thickness.
Feel the difference this way: a 60-degree single V on 40 mm plate carries roughly three to four times the fill area of a U profile at the same depth, depending on root opening and root face. Multiply that by a four-metre seam and by wire density, and you are comparing tens of kilograms of consumable per joint. Machining a radius on two edges costs a fixed amount; filler savings scale with joint length. That is why long seams are where U grooves earn their keep.
Look at the industries running heavy, long, or highly loaded joints and U preparations appear everywhere:
| Application | Why the U groove fits |
|---|---|
| Pressure vessel shells and heads | Thick walls, full penetration required, high fatigue and pressure cycling |
| Heavy plate girders and bridges | Long CJP flange and web seams, distortion control matters |
| Offshore tubulars and jacket members | Large diameter, cyclic sea loading, automated SAW work |
| Mill rolls and forging builds | Very heavy sections where fill volume becomes the dominant cost |
| Heavy wall pipe and piping | Narrower channel means fewer passes and better control out of position |
On tubular work there is a useful boundary to remember: submerged arc welding is applicable to diameters greater than 24 in (610 mm), and tubular groove joints intended for welding are generally detailed with a groove angle of not less than 60 degrees. Below that diameter, or where access forces manual welding, a bevel or V preparation is often the more realistic call. For pipe-specific procedures and positions, our guide to pipe welding procedures walks through the WPS, pass, and inspection side in detail.
Next question on the drawing: does this joint actually need complete joint penetration? A CJP groove weld is a far more demanding call than a PJP, and it drags preheat, backgouging, NDT, and documentation along with it. Groove design is a balance of penetration, cost, and distortion, and the U groove simply makes CJP cheaper on heavy plate.
Fatigue changes the answer completely. AWS D1.1 prohibits partial joint penetration groove welds subject to tension normal to their longitudinal axis where design criteria indicate cyclic loading could produce fatigue failure. In plain language: if that seam sees repeated tension, a PJP is off the table no matter how cheap it looks, and you are moving to a CJP detail where a U preparation starts to make sense.
Access is the other filter. A U groove’s narrow angle is unforgiving in overhead or vertical positions because torch angle and travel speed have less room for error. If the joint can only be welded overhead, review the technique limits in our overhead stick welding guide before committing to a tight U profile on the drawing.
Quick Note: Run this five-point checklist before specifying a U groove: (1) Thickness — above ~25 mm / 1 in? (2) Access — can the welder reach and hold torch angle in the narrow channel? (3) Batch size — one joint or a series? (4) Distortion tolerance — is angular distortion a measurable risk? (5) Machining availability — can you hold the root face and radius tolerance on site, or must you sub it out? If three or more answers favour U, price it out properly.
Estimators often fixate on thickness and forget volume of work. For one-off fabrication, the single V usually wins economically: no machining setup, no radius gauges, no special tooling, just a torch and a bevel cut. For series work and heavy sections, the double V or U preparation pays back quickly, because the machining cost is amortised across every unit and the filler savings multiply with each joint.
There is also a codes-and-paperwork reality. Prequalified joint details in AWS D1.1 and the groove definitions in ASME Section IX only help you if your WPS points at the right detail and your procedure qualification record backs it up. Our guide to weld prep and acceptance issues and the ISO 9692 joint preparation summaries are useful cross-checks when a client asks why the drawing says U instead of V.
Note: Rule-of-thumb thickness bands: under 25 mm — single V or bevel; 25 to 50 mm — double V often beats both, but U is competitive on long seams; above 50 mm — U or double-U, where filler volume and distortion dominate the cost. Confirm against the prequalified details your WPS actually cites.
The short version: choose the U groove when the wall is thick, the seam is long, the joint is critical, and machining is available. Everything else is a V in disguise.
Edge preparation is what separates a shop that wins bids from a shop that eats rework. A welder can run a beautiful cap, but if the groove was rough-cut, out of tolerance, or welded over a contaminated root, the joint is compromised before the first bead goes in. Here is the workflow, from the cutting table to the backgouge.
A square, laser, or plasma cut can produce a clean straight edge, but a U groove has a radiused bottom and a curved wall profile that no thermal cutting torch reproduces reliably. Oxy-fuel and plasma cutting also leave a heat-affected surface, dross on the lower edge, and an oxide layer that must be removed before welding. On thicker plate, that affected surface should be ground back to sound metal. As a practical rule, flame cutting roughs out the bulk material; the finished profile comes from a machine tool.
That is why milling or planing is unavoidable for U and J preparations. The machining of the weld preparation dictates that dimensions, particularly the root face thickness, are controlled far more closely than flame cutting allows, and that tighter control is what gives you accurate fit-up at the fit-up table. Common approaches include:
The economics matter too. Groove preparation is a machining cost, not a welding cost, and shops that quote the job honestly compare that machining bill against the filler metal savings discussed in our earlier cost breakdown.
Even a freshly milled edge can sit in the shop for a week and collect oil, mill scale, and rust. Cut edges must be cleaned of slag and oxide before the root pass. On plate that has been flame roughed, any residual heat-affected surface should be ground back, because that layer produces porosity and lack of fusion right where it hurts most: along the curved groove wall and at the root face. Wipe the joint out with a clean brush and solvent, keep it dry, and do the final cleaning the day you weld.
Note: Pre-weld fit-up inspection checklist: (1) root opening held to the WPS tolerance along the full length; (2) root face thickness within spec on both members; (3) radius profile checked against a template, no flat spots or gouges; (4) joint cleaned to bright metal, free of oxide, dross, oil, and moisture; (5) alignment or hi-lo between members within the code limit.
Tack the joint with the same process, filler, and preheat as the production weld. Tack spacing and settings follow the same rules as any structural tack-up; our MIG tack weld guide covers spacing and sizing in detail. Tacks land where they will not interfere with the root pass, and tack removal by arc-air gouging leaves a notch that must itself be ground and inspected before the root covers it.
Backing strips are permitted by many prequalified details but change the joint economics and add a strip that must be fused. Where a backing strip is used, machine its edges to match the groove radius and confirm fusion at both ends of the strip. Where it is not used, you are committing to the backgouge sequence.
Troubleshooting: Gap grew after tacking? Do not close it by heating and hammering. Check for tack shrinkage pulling the plates, remove the offending tacks, re-shim to the specified root opening, and re-tack from the opposite side in shorter, staggered increments. If the gap exceeds the WPS allowance, the joint needs an engineered repair, not a fitter’s improvisation.
The AWS joint detail note for double-groove U joints requires gouging the root to sound metal before welding the second side. In practice you weld the first side, flip the member (or reach through), and arc-air or grind out the root along its full length until you expose clean, unfused, shiny metal. Only then do you weld from the second side.
Why so strict? That backgouge removes any root lack-of-fusion or trapped porosity before it is sealed inside the weld where no inspection can reach it. Skipping the backgouge is the single most common reason a CJP U groove fails ultrasonic testing, and as our pipe welding procedure guide explains, evidence of complete joint penetration is exactly what the inspector is verifying. The gouge itself must be sound, dry, and ground clean, and the second side is welded to the same WPS, preheat, and interpass requirements as the first.
The message is simple: a U groove rewards shops that treat preparation as engineering, not as a preliminary chore. Machine it, clean it, fit it to tolerance, and gouge the root before you close it, and the joint will reward you for decades of service.
Picture two welders running the same 38 mm double-U box girder seam, same WPS, same machine settings. One lays tight stringer beads and finishes in eleven passes. The other drags a wide weave across the whole groove and needs nineteen. The second job comes out over-deposited, visibly bowed, and back from the inspector with a distortion note. U groove geometry is forgiving, but the technique still decides the result.
Process selection follows access, thickness, and productivity rather than the groove shape itself. In a shop running structural steel, the usual pattern is a GTAW or SMAW root pass when there is no backing strip, a hot pass to knock the chill off, then SMAW, GMAW, or FCAW to fill. Long flat seams on heavy plate are candidates for SAW, where the buried arc eats up weld metal volume far faster than a stick electrode can. As Hobart Brothers’ groove welding guidance puts it, the process has to deliver full fusion in the root and then deposit the remaining volume economically.
Root passes deserve special respect in a U groove. The radiused bottom holds a wider puddle than a sharp V toe, so the arc wants to wander sideways instead of biting straight down. Run the root slightly cooler, keep the electrode or torch angle pointed at the groove walls, and watch the leading edge of the puddle rather than the crater. On a 3/32 root with E7018, the settings in our 3/32 E7018 amperage guide are a solid starting point. Once the root is sealed and visually clean, the hot pass fills and re-melts any trapped slag before the fill layers go in.
Quick Note: Typical pass sequences by thickness (single-U, SMAW, flat): 12–20 mm = root + 1–2 fill + cap; 25–40 mm = root, hot pass, 3–6 fill passes, cap; 50–80 mm = root, hot pass, 8–15 fill passes, cap, usually as a double groove. Over 80 mm, switch to a double-U or double-V and add SAW on the flat side.
Fill passes in a U groove should be stringer beads with a slight overlap, not rainbow weaves. The reason is geometric: larger included angles increase weld cross-sectional area, weight, cost, distortion, and residual stress while adding virtually no strength, so the angle is kept as small as practical while still achieving fusion. That narrow profile means the electrode has less room to maneuver. TWI’s design notes make the same point about groove geometry, and The Fabricator’s weld prep advice echoes it on the economics.
In practice, a narrow U demands better travel speed and angle control than a wide V. Too much drag and the arc climbs one wall, leaving the opposite side cold; too slow and you overfill, then grind the cap back. Weave only when the groove is wide enough to keep the puddle under about three electrode diameters, and limit weave width on out-of-position work. Position matters: AWS D1.1 prequalified single-U details give a 45° included angle for flat-position work, and the angle drops to 20° in the details qualified for flat and overhead SMAW. Always match the drawing detail to the actual position, because electrode access in the narrow bottom is the real limiting factor. Our overhead stick welding technique guide covers the body mechanics that out-of-position groove work requires.
Using a double-sided joint configuration such as a double groove significantly reduces the time and filler metal required on thicker material, and alternating passes on each side further reduces distortion. Each bead shrinks as it cools, so laying three beads on the front and then three on the back lets the second side pull the first side back toward straight. Double-groove welds may have grooves of unequal depth, but the depth of the shallower groove must be no less than one-fourth the thickness of the thinner part joined, a rule stated in AWS D1.1/D1.1M.
Hold interpass temperature within the WPS range, usually a 150–250°C band for carbon steel, and resist the urge to bump amperage to “make up time”. Extra heat input buys penetration you already have and distortion you do not want. Cap height should land flush or slightly proud; grinding a humped cap is wasted labor and a hidden source of residual stress.
On a 60 mm roll housing split in half for repair, the crew cut the U angle, tacked both sides, and welded front-back-front-back in three-beat cycles. Total passes: fourteen. A neighboring job done one side at a time on the same thickness took twenty-one passes and came off the table 3 mm out of square. Same groove, same filler, different sequence, completely different outcome.
A U groove weld is a precision joint. That curved root radius and machined root face give you arc control and clean stress flow, but they leave almost no room for sloppy habits. When a U groove fails inspection it is rarely a mystery: geometry drifted out of tolerance, a step in the sequence got skipped, or the welder was handed a groove that was never prepared correctly. Here is what goes wrong, why it happens, and how to keep it off your test plate.
Skipping the backgouge. On a CJP joint welded from both sides, the second side cannot be laid over an unverified root. AWS D1.1 joint detail notes require evidence of complete joint penetration, obtained through visual examination and confirmed by NDT such as radiographic or ultrasonic testing. Backgouging to sound metal is how you earn that evidence, and it is why first-side root problems are cheaper to fix than most people assume.
Over-narrow included angle. Making the U too tight to save on filler backfires. An excessively small included angle makes consistent fusion difficult and produces excessive dilution, which leads to crack sensitivity with certain alloys. The repair costs far more than the filler saved.
Wrong root face, prequalified limits ignored, and overwelding. A root face outside the prequalified detail means the joint is no longer covered by the WPS you are working to. Overwelding deserves special mention: because the relationship between weld size and cross-sectional area is parabolic, the cost penalty accelerates sharply as size grows, so a cap run 2 mm oversize across a long seam is not a small overage. For the documentation and pass-sequence side, our pipe welding procedures guide covers WPS structure in depth.
Removing weld tabs incorrectly. Cutting a tab off with a torch and grinding flush leaves a notch that reads as an arc strike or crack on radiography. Tabs should be removed by grinding or air-carbon-arc cutting followed by grind-out, then blended and re-inspected.
Note: 5 mistakes that get a U groove rejected: (1) no backgouge before the second side, (2) root face held outside tolerance, (3) included angle narrower than the WPS detail, (4) hi-lo beyond the code limit, and (5) weld tabs ground into a visible notch instead of removed cleanly.
Visual inspection comes first and catches most geometry problems: profile, undercut depth, overlap, misalignment, and cleanliness. After that, CJP groove welds are verified by radiography or ultrasonic testing. You cannot declare complete joint penetration by intent; you demonstrate it. Qualification testing under AWS D1.1 backs this up with root, face, and side bend specimens, reduced-section tension tests, and macroetch examination, each with defined acceptance criteria. A shop that skips procedure qualification and relies on welder reputation is betting on nothing measurable.
One closing thought: most U groove rejects are prevented by the same habits that stop defects in simpler joints. Correct amperage, correct filler, correct preheat and interpass control. If any of those fundamentals feel shaky, revisit them before you commit to a 60 mm seam, not after the radiographs come back.
A U groove is chosen because it saves filler metal on heavy plate. What it also quietly demands is arc time: a 40 mm single-U joint can mean hours of continuous deposition inside a deep, narrow channel where fume has a harder time escaping than it does off a shallow V. That is the trade-off nobody puts on the drawing. The health and quality obligations that come with it are real, and they are the part of the job that gets signed off by someone other than you.
Welding fume is not one thing. It is a shifting mix of metal oxides from the base metal, the consumable coating or core, and the shielding gas, plus gases like ozone and nitrogen oxides. On mild steel the dominant concerns are iron oxide and manganese; on stainless, nickel, chromium, and in some processes hexavalent chromium, Cr(VI). The deep, narrow U profile and long fill hours are exactly the conditions that push personal exposure upward.
The numbers worth knowing:
For PPE selection, filtration, and respirator fit, our welding PPE guide covers the full kit in detail.
Quick Note: Ventilation minimums and extractor positioning: treat 2,000 CFM per welder as the floor, not the target. Position the extraction inlet as close to the arc as the heat allows, ideally within 1 to 1.5 times the duct diameter, on the fume side of the welder’s breathing zone, never directly behind the arc where it pulls fume across the operator’s face. If the room cannot supply make-up air, add it before adding hoods.
Heavy U groove work often lands inside tanks, vessels, or large-diameter pipe, which turns a ventilation question into a confined-space question: permit-required entry, atmospheric monitoring, forced ventilation with the discharge routed away from intakes, and a standby person outside. Deep groove positions also force awkward posture for hours, which is where burn and strain prevention matters as much as fume control.
Every hour of arc time sits on top of paperwork that must already be in place. The inspector will ask for, in roughly this order:
Note: The documents an inspector will ask for: approved WPS, PQR, welder qualification with continuity, preheat/interpass log, consumable control record, fit-up inspection sheet, backgouge sign-off, and the NDT report proving CJP. Missing one stalls sign-off regardless of how good the weld looks. Our pipe welding procedures guide walks through this document chain end to end.
Attributing requirements correctly matters: AWS D1.1 defines the structural requirements (D1.1/D1.1M), OSHA defines the exposure limits, and NIOSH defines the health recommendations. None of them substitute for each other, and the shop that treats all three as one checklist is the shop that passes inspection and keeps its people well.
The biggest cost of a groove weld isn’t the wire or shielding gas. It’s time. Roughly 85 percent of what a groove weld costs is labor and overhead, not consumables. This means every cubic millimeter of groove you cut is mostly bought in welder-hours, not kilograms of filler. For heavy plate, that single fact determines whether a job turns a profit or quietly loses money.
Note: A U groove weld is a full- or partial-penetration groove joint where both members are edge-prepared into a curved, radiused U profile instead of a straight chamfer. This reduces filler volume and improves stress flow when compared to a V groove on heavy sections.
You don’t need expensive estimating software to see if a U prep pays off. All it takes is a few calculations:
These numbers speak for themselves. Moving from a single-V to a double-V on the same thickness typically cuts filler by about 40 to 43 percent, since the depth is split between two shallower grooves. A U prep can reduce it even further, as its curved walls stay nearly parallel for most of the depth instead of flaring outward. As Hobart Brothers’ guidance on groove weld preparation and TWI’s joint design notes on welding joint design point out, the cheapest joint is the one that deposits the least metal while still meeting its penetration requirement.
Even the best prep will fall apart without discipline on the floor. A few small habits will protect both your margin and your weld quality:
Nearly every cost driver for a U groove weld is locked in before metal is cut. Thickness, groove type, root opening, single or double-sided, and backgouging accessibility will all outweigh small tweaks at the station. Review these details before sending an RFQ, and if the drawing is ambiguous about the root face, radius, root opening, or groove angle, clarify it against AWS D1.1 in writing.
The right answer isn’t always a U, either. Batch size, available machinery, access, and distortion tolerance matter just as much. A one-off usually favors a simple single-V, while series production on heavy sections can justify a machined U. When the joint can be turned, a double-V often beats both on cost per kilogram of deposited metal, a point echoed in The Fabricator’s coverage of economical bevel cutting.
Quick Note: Copy this 10-point checklist onto your job traveller to lock in your margin before fit-up.
The U groove weld isn’t something to specify for looks. It’s a tool to be deployed when the arithmetic proves it will deliver a sound, fully penetrable, and inspectable joint for the least total cost. By deciding the joint at the drawing stage, holding your tolerances, and doing the math upfront, the U groove stops being an expensive surprise and starts doing exactly what it was designed to do: putting less metal, less heat, and less stress into a joint that has to last.
A U groove weld is a groove weld where both members are prepared with a curved, U-shaped cross-section featuring a machined root radius and a small included angle instead of a straight V chamfer. It is used mainly on thick plate where the narrow profile deposits far less weld metal than a V groove.
Use a U groove on heavy sections, typically above about 25 mm and especially beyond 1 inch of thickness, where the filler metal and distortion savings exceed the cost of machining the edge. For thinner plate or one-off work, a single or double V groove is usually cheaper because it can be flame cut.
For the B-U6 SMAW detail, AWS D1.1 specifies a 45 degree groove angle in flat position (20 degrees for flat and overhead), 1/8 inch (3 mm) root face, 1/4 inch (6 mm) root radius, and a root opening of 0 to 1/8 inch (0 to 3 mm). The GMAW/FCAW version B-U6-GF uses a 20 degree angle in the flat position.
The U-groove symbol (AWS A2.4 Figure 4.1) is a U shape placed on the reference line, and because both members are prepared, the arrow does not need a break. Per A2.4 Section 7.3.3, the root radius and root face must be called out by cross-section, detail, or other data referenced in the tail of the symbol.
A U groove generally must be machined by milling, planing, or a CNC profile cutter because the curved radius and controlled root face cannot be produced reliably by oxy-fuel or plasma cutting. V and bevel preps, by contrast, can be flame cut economically.
A U groove prepares both members of the joint symmetrically, while a J groove prepares only one member, which is why J-groove symbols use a broken arrow. Both require machining and both offer lower weld volume than V grooves, but the J groove is chosen when only one side can be machined or accessed.
Yes. All prequalified U-groove CJP joints in AWS D1.1 require the root to be gouged back to sound metal from the second side before the back weld is deposited. This removes any root lack-of-fusion or slag trapped on the first side before it becomes an internal defect.
The most common are lack of fusion along the curved sidewalls, incomplete root penetration, porosity from oxide or slag left on the machined edge, undercut at the weld toes, and misalignment or hi-lo at fit-up. Tight root face control, clean edges, correct torch angle, and backgouging prevent nearly all of them.
Heavy groove welding means long arc hours, so fume exposure is a real concern: OSHA’s welding fume PEL is 5 mg/m3 and hexavalent chromium is limited to 5 micrograms per cubic meter over 8 hours. Local exhaust ventilation kept within about 18 inches of the arc and NIOSH-approved respirators when engineering controls fall short are the standard controls.