Welding Oxygen Cylinder: Sizes, Safety Rules, Storage & Handling Guide
Complete guide to the welding oxygen cylinder: sizes, pressures, CGA fittings, OSHA storage rules, hydrostatic testing, safety mistakes…
READ ARTICLE →A fabrication shop spent three days welding a 10-gauge hopper, only to watch the seam peel open at the edge a week after delivery — the weld looked fine in the middle, but the edge had been sheared dirty, mis-fit, and burned back by too much heat. That is the silent problem with welded edges: 90% of the joint’s trouble starts at the very edge most welders never think twice about.
A small fabrication shop spent three days welding a 10-gauge hopper for a feed mill. The long seams looked textbook: consistent ripples, good color, no visible porosity. Then, about a week after delivery, the seam peeled open — not in the middle, but right at the edge. The plate had been sheared dirty, the fit-up gap varied from nothing to a sixteenth, and the welder had burned the edge back with too much heat and too slow a travel speed. The middle of the weld was fine. The edge was never given a chance.
That failure is why welded edges deserve far more attention than they get. The edge is the one place on a joint where fusion, fit-up, heat control, and inspection all meet at the same millimeter. It is where the arc starts, where it terminates, where the crater forms, where the plate is thinnest, and where a crack will find its way out if you leave it a reason to. Get the edge right and the rest of the weld tends to follow. Get it wrong and no amount of pretty beads in the middle will save the part.
“Welded edges” is one of those phrases that means several different things depending on who is asking. One welder is thinking about edge preparation — squaring, beveling, and cleaning the plate before the arc strikes. Another means an edge joint, the AWS joint type where two parallel members are joined along their exposed edges. A third is picturing edge welds on bellows, diaphragms, or thin enclosures. A fourth is buying welded-edge conveyor belting or folded, seam-welded architectural panels. Search results mash all four together, which is exactly how people end up reading the wrong answer with confidence.
This guide untangles all of it. Codes treat edge quality as a first-class acceptance issue: AWS D1.1 dedicates an entire figure to “Edge Discontinuities in Cut Material,” spelling out notches, fins, tears, and cracks that must be removed before welding even begins. If the code takes edges that seriously, so should your shop.
Quick Note: This article covers welded edges in fabrication — prep, joint types, sizing rules, defects, applications, settings, and safety. It also briefly covers welded-edge products such as conveyor belting (per ISO 14890) and folded-and-seam-welded architectural panels, so you know the full scope of the term before you hit a page that only answers half your question.
Section by section, we move from definition to execution: what welded edges actually mean on a shop floor, how to prepare them, which edge joint detail to choose, the edge distance and weld-size rules codes really enforce, the defects that show up at edges and how to stop each one, where edge welding shows up in real jobs, the machines and settings that produce clean edges, the safety picture, and a final troubleshooting checklist.
Along the way we lean on established references and practical data. The Bureau of Labor Statistics found that of 370 lost-time welding- and heating-tool injuries in 2018, more than one-third — 140 cases — were thermal or heat burns, and the median injured worker lost 10 workdays beyond the day of injury. Edges concentrate that heat, spatter, and UV, which is why safety is not a throwaway paragraph here.
Expect short paragraphs, bold subheadings, comparison tables, and quick-note boxes throughout, plus practical references to our related guides: Welding Defects: Undercut for acceptance limits, TIG Welding Sheet Metal for thin-edge settings, and Welding PPE: The Complete Guide for the gear that keeps burns off your arms and arc flash out of your eyes.
Short Note: Read the headings in order if you are learning, or jump straight to the section matching your problem — a burn-through on thin edge? Go to defects and settings. An inspector’s question about undersize welds? Go to the weld-size rules.
One promise before we start: no filler, no fluff, no recycled definitions. By the end you should be able to look at a sheared edge, know exactly what it needs, and weld it with the confidence that the seam will still be closed a week after delivery.
Ask five welders what “welded edges” means and you will get five different answers. One talks about beveling plate, another about joining sheet metal along its edge, a third about conveyor belting. None is wrong. The phrase is genuinely overloaded, and that ambiguity is why it turns up in search results, blueprints, and shop arguments carrying completely different meanings.
Pin the four distinct meanings down now, and every drawing, WPS, and trade article afterward makes far more sense.
The most common shop-floor usage is simply getting the edge ready to weld. When a fabricator says “the edges are prepped,” they mean the plate has been cut square or beveled, the shear-hardened zone has been ground off, and mill scale, oil, and rust are gone. Edge preparation is not the weld itself; it is everything that happens before the arc is struck.
A weld is only as good as the metal the arc has to fuse, and a sheared, oxidized, or mis-fit edge guarantees a mediocre joint regardless of welder skill. Section 3 covers preparation in depth, but hold this thought: preparation is a meaning of “welded edges,” not an afterthought.
AWS D1.1 recognizes five basic joint types — butt, tee, corner, lap, and edge. The edge joint is the least understood of the five. Two (sometimes more) members are clamped so their faces are parallel and flush, and the weld is deposited across their exposed edges, fusing them together. Think two sheets of 16-gauge steel butted face to face and welded down the exposed end, or a tube capped and welded along its rim.
Edge joints are everywhere in sheet-metal work, light fabrication, and enclosure building, yet rarely discussed in beginner material. A common classroom exercise — a single-V edge joint on 3/8 in practice plate, prepared with a 22.5° bevel per side and a 1/4 in root gap — is how welding curricula aligned to AWS D1.1 introduce the geometry. Procedure-wise, thin material gets a square-groove edge joint; as thickness climbs, the groove gets beveled. New to joint set-up? Our 3/32 E7018 amperage guide walks through how electrode choice and joint geometry interact.
Subtly different from the edge joint: an edge weld is a fusion weld made along the edges of overlapping members, and it is often autogenous — no filler metal, just the arc melting the two edges together. This is the process behind edge-welded bellows, diaphragms, filter cans, hermetic enclosures, and thin-walled sensor housings.
The stakes here are unusually precise. ASME Boiler and Pressure Vessel Code Section IX carries dedicated edge-weld geometry rules precisely because of this category, and in thin-foil bellows work individual layers can be as thin as 0.005 in. At that thickness, “weld size” becomes a question of how far the melt zone wicks into the foil, not how big a fillet you can stack. Edge welding is also the process family behind laser edge sealing of battery modules, which production lines run at speeds hand welding cannot approach.
Quick Note: “Welded edge” is also a product term, not just a process term. Welded-edge conveyor belting — covered by ISO 14890, the current international specification for rubber- or plastics-covered textile conveyor belts — has its edges sealed to resist moisture uptake and spill. Folded-and-seam-welded architectural panels and welded edge banding on stainless sheet use the same idea: the exposed edge is closed by a weld instead of left raw. None of this relates to woodworking PVC edge banding, despite the confusingly similar name. If someone asks you to “weld the edges” on a cabinet job, ask which meaning they mean before you pick up a torch.
Strip away the jargon and all four meanings share one physical truth: the edge is where fusion, fit-up, and heat control all concentrate. Edges are thinner than the plate they belong to, so they heat faster, melt sooner, and cool differently. A rough or hardened edge resists fusion; a sloppy gap sends the arc elsewhere; a bevel too steep starves the root, and one too shallow buries it in lack of fusion. That hopper seam from our opening story failed at the edge for exactly this reason.
It is also why edge work is where machine setup matters most — our MIG welder setup guide covers the basics that carry directly into edge-joint work.
Keep these six words bookmarked. Every section that follows — prep, joint selection, weld-size rules, defects — leans on them constantly, and once they are second nature the rest of the guide reads much faster.
Every failed welded edge tells the same story in reverse: someone skipped ten minutes of prep, and the joint paid for it later. Fit-up, heat control, and travel speed all sit downstream of the condition of the metal you are about to fuse. A rough, hardened, out-of-square edge gives inconsistent root openings, unpredictable penetration, and a weld that fuses shallow while looking fine on the surface.
No two cutting processes leave the same edge, so the method that shaped your plate tells you how much grinding comes next.
| Method | Edge condition | Prep before welding |
|---|---|---|
| Shearing / stamping | Deformed, work-hardened zone, burr or fin on the punch side, smeared grain | Grind or machine off the fin and hardened layer |
| Band or cold sawing | Clean and square, fine tooth marks | Deburr both faces; usually ready to fit |
| Plasma cutting | Slight bevel drag, dross on the bottom edge, narrow thermal zone | Grind dross, dress the face to bright metal |
| Oxy-fuel cutting | Oxidized, decarburized layer, wide heat-affected zone | Grind or machine to sound metal; consider preheat on heavy plate |
| Laser cutting | Very clean, narrow HAZ, minimal dross | Light deburr, then fit |
| Waterjet | No thermal damage, sharp burr on exit | Deburr only; no hardened layer |
Shearing is fast and cheap, and that is why it gets overused. The blade smears and work-hardens a thin band along the cut, and the burr on the exit side acts as a stress raiser. Weld straight over that layer and you invite lack of fusion, toe cracks, and brittle starts. Grinding the fin off is not cosmetic — it removes crack-initiating material. The same goes for any notch left by cutting: a gouge at the root is a crack waiting for an arc.
The root face, or land, is the small flat edge at the bottom of the bevel — too thick and the root never fuses, too thin and it burns away before you get there. The root opening is the gap between members, sized so the arc can reach the joint root. Thin sheet runs best with no bevel and a tight fit; heavier plate wants a land around 1/16 to 3/32 in, with the opening matched to the process. For tubular work, AWS D1.1 qualification becomes critical once dihedral angles drop below 30°, because arc access and fusion reliability fall off sharply there (AWS D1.1/D1.1M).
The AWS D1.1 fabrication clause is blunt: surfaces and edges to be welded must be smooth, uniform, and free from fins, tears, and cracks (AWS D1.1:2020). When cutting or handling damages an end preparation, the code accepts two remedies — build it back up by welding, or re-cut it by machining or grinding. Los Alamos National Laboratory’s welding procedure WFP 2-07 states it directly: if root-opening and fit-up tolerances cannot be achieved, end preparations may be built up by welding or re-prepared by machining or grinding (LANL WFP 2-07). A gouged edge is repairable; a cracked or laminated one is not.
AISC’s weld-clearance studies recommend cutting doubler plates square below roughly 3/8 in thick, then a 30° bevel on heavier plate to cut deposited weld volume. At 3/4 in thick, the same research recommends 15° so a standard GMAW nozzle can still reach the joint root (AISC weld-clearance research). The pattern holds: steep bevels when the arc must get deep, shallow bevels when access or weld volume is the constraint.
Quick Note: Grinding belongs where edges are thin, cosmetic, or low-duty. A milling machine earns its keep on thick plate, repeat bevels, and tight squareness tolerances. Hand-grinding forgives geometry but is hard to hold consistent — a bevel face that wanders 5° along its length changes penetration from one end of the seam to the other.
If you are new to dialing in the machine itself, our How to Set Up a MIG Welder walkthrough covers the settings that make a well-prepped edge actually cooperate.
Short Note: Laser-cut edges cost more up front but routinely eliminate an hour of hand grinding per part — on a 200-piece run, machined or laser-cut bevels usually pay for themselves before the first weld is laid. On stainless, chrome-moly, or any pressure-boundary joint, pay for the machine edge.
Prep is the one phase where time spent always returns as better fusion, less distortion, and fewer grinding repairs at the end. For thin sheet, it is even more unforgiving, which is why TIG welding sheet metal starts with edge condition long before torch settings.
Two sheets of 16-gauge stainless clamped edge to edge, a MIG torch, and a welder who had never been taught the difference between an edge joint and a butt joint. It happens more often than shops like to admit. The joint detail you pick at the fixture decides your heat input, your distortion, your inspection method, and whether the part passes. So before you strike an arc anywhere near an edge, let’s walk through the edge-related joints you’ll meet in fabrication and the code language behind each one.
AWS D1.1 defines an edge joint as two or more members placed side by side (parallel or nearly so) and joined at their exposed edges. Think of two plates clamped face to face like a book closing, with the weld laid across the top where both edges line up. You’ll see it in brackets, hinges, lap plates, and light sheet assemblies where loading is modest and the geometry invites it.
Edge joints come in two flavors:
The subtlety is where the arc is aimed. In a thin square-groove edge joint, the arc naturally melts both members equally. In a heavy edge joint with a single bevel, the welder must direct the arc against the groove face, which is exactly why tee joints with square grooves don’t get the same treatment.
Quick Note: D1.1 permits groove preparation on either or both members of a corner joint, as long as the basic groove configuration is unchanged and enough edge distance is maintained to prevent excessive edge melting. Bevel one side or both, but leave enough metal at the edge or the arc will eat the corner before the groove fills. The full clause sits in the AWS D1.1:2025 Structural Welding Code.
A corner joint puts two members at an angle, edges meeting at the outside corner, welded on the outside, inside, or both. It’s the joint behind nearly every box, cabinet, hopper shell, and duct transition. Fillet-welded outside corners are the most common; full-penetration outside corner groove welds appear where the corner carries real load or acts as a pressure boundary.
The edge-distance rule matters most here. A corner welded too close to a trimmed edge will blow out, warp, or leave undercut right where stress concentrates. That’s why fitting the corner square and grinding back a ragged sheared edge before tacking isn’t optional fussiness.
Hollow structural sections are formed, not machined, so the rounded corners of a rectangular tube give you a natural groove called a flare. Welding a plate to that rounded corner produces a flare-bevel groove weld. Because the radius is consistent, D1.1’s prequalified details account for it — the commonly referenced design size for flare-bevel groove welds at rectangular HSS corners is 5/8 of the wall thickness. Round HSS corners produce flare-V groove welds with their own sizing. The practical lesson: on thin-wall HSS, that 5/8 rule often leaves you with a very small effective throat, so size your weld and loads accordingly. For technique in heavier structural work, our 3/32 E7018 amperage guide covers settings that keep thin throats from overheating.
Not every edge detail is prequalified. A square-groove tee joint is explicitly not prequalified under D1.1 where the arc cannot be directed against the base metal — fusion simply can’t be relied on. AISC research notes this limitation in its discussion of weld clearances. If your edge geometry puts the arc in that situation, you owe your shop a qualified WPS, not a guess.
Tubular work has its own trap. Any tubular joint with a dihedral angle less than 30° falls under AWS D1.1 Subclause 9.15.4.2 and requires WPS qualification with the specific process and technique you’ll actually use. Sharp acute corners on elliptical or formed HSS joints quietly put you there. Pipe and tube fabricators should read our pipe welding procedures guide before assuming a structural WPS covers the joint.
For very thin sheet, the cleanest edge weld uses no filler at all. Two edges butted together, the arc fusing both faces into a single bead. This is autogenous edge welding, the process behind edge-welded bellows, diaphragms, filter cans, and laser-welded battery enclosures. It demands tight fit-up, zero gap variation, and stable control — which is why GTAW and laser or electron-beam systems dominate this niche. If you want the TIG side of that skill set, start with the complete guide to TIG welding sheet metal. Where two sheets overlap instead of butting, resistance spot welding is a faster alternative; our spot welder guide explains the physics.
| Joint | Best For | Watch Out For |
|---|---|---|
| Edge joint | Parallel thin members, brackets, hinges, enclosures | Limited throat; needs edge distance and fixture control |
| Square butt joint | Flat panels, duct runs, full-thickness seam welding | Requires backing, purge, or tight gap control to avoid burn-through |
| Lap joint (edge fillets) | Overlapping sheet, patch repairs, wear plates | Over-size edge fillets tear out — size to thickness rules |
| Autogenous edge weld | Bellows, battery cans, hermetic seams | Zero tolerance for fit-up gaps; needs precision control |
No single joint wins everywhere. Edge welding beats everything when the part is a sealed thin-walled enclosure that must stay lightweight and leak-tight. A lap joint with edge fillets wins on repairability. A butt joint remains king when you need the full strength of the base metal across the seam. Choose the detail first, then set the machine — not the other way around.
Here is what separates a fabricator from someone who just melts metal: knowing where a weld is allowed to sit. Every rule in this section traces back to one piece of physics. An edge is the only place in a plate where heat has nowhere left to travel, so push the arc close and the corner melts away with the puddle falling off the side. The codes turn that into numbers.
AISC 360 Table J3.4 sets minimum edge distances: the shortest gap from the centroid of the weld group to the edge of the connected part. Its purpose: stop edge melting and blowout during welding, and leave sound metal behind the weld to carry load. The values depend on how the edge was made — about 3/4 in for sheared material up to 1/2 in thick, 5/8 in for rolled or planed edges up to 3/4 in thick, and 2 in or more on heavy plate.
Sheared edges get the larger number because shearing tears metal rather than cutting it, leaving a hardened zone and burr that detonates under the arc and drags impurities into the puddle.
Corner joints get the same warning in different words. AWS D1.1 corner-joint provisions allow groove preparation in either or both members when the configuration is unchanged, but only when edge distance is adequate to prevent excessive edge melting. No edge distance, no allowance.
This is the rule most shops break without realizing it. AWS D1.1 and AISC J2.2 limit the maximum fillet-weld size along the edges of lap joints like this:
So a 3/16 in plate takes a 1/8 in fillet maximum, not the fat 3/16 you might naturally deposit. Why do oversized edge fillets tear out? Fillet size is measured along the leg, but strength comes from the throat. On thin plate, extra filler simply melts the corner. Extra filler falls off as an icicle, the edge rounds into a teardrop, and a nominal 3/16 fillet delivers a 1/16 throat plus a cracked nib. The weld toe ends up off the plate, where cracking starts.
Quick Note: Undersize is more forgiving than you think. D1.1 visual acceptance permits fillet legs measuring 1/16 in under the specified size over up to 10% of the weld’s total length, and AISC research on weld clearances found measured undersize fillets that still passed visual acceptance outright. Chase perfect legs elsewhere; know where the tolerance saves you.
A weld that stops is a weld that can crack. D1.1 flags terminations near edges in tension because the crater is the weakest, most segregation-prone part of the bead, and a free edge gives any crack a surface to run to. Fix it with back-stepped craters, crater fill at every stop, and end returns that wrap 2× the weld size around the corner so the stop lands in compression-bearing material. Such a stop pairs with undercut at the weld toe as the most common edge defect in structural work.
Thin edges conduct heat poorly — there is no mass to sink it — so they reach melting point long before the plate around them. Your compensation toolkit:
Weld termination near edges in tension, distorted edges, and mismatch after blowout all trace back to one mistake: treating an edge like the middle of a plate. For amp and rod specifics, see the stick welding thin metal guide.
| Material thickness | Edge treatment | Typical weld size |
|---|---|---|
| 22–26 ga (0.03–0.05 in) | Square, deburred; chill block | Autogenous or 1/16 in |
| 16–10 ga (0.06–0.13 in) | Square groove; grind shear zone | Fillet ≤ t − 1/16 in |
| 1/8–1/4 in | Square or shallow single-V | Fillet ≤ t − 1/16 in under 1/4 in |
| 3/8–3/4 in | Single-V 30°–37.5° per side, root face 1/16–3/32 in | Full plate thickness |
| 1 in and heavier | Double-V, U or J; multi-pass with interpass cleaning | Per WPS; preheat required |
Quick Note: Welded wire mesh and perforated sheet are the extreme case: there is barely any edge. Do not weld to the outermost wire or perforation. Clip the mesh back, clamp it to a solid frame or keeper strip, and weld to that instead; fusing the last wire is the fastest route to burn-through and a warped panel.
Learn these rules once and they show up everywhere: the limits that keep a duct seam from burning back also protect a pressure-vessel nozzle and a stainless panel. Codes read as restrictive, but they are mostly experienced welders writing down what happens when you ignore them.
Every welder has stared at a finished edge seam and felt that sinking moment. Almost every edge defect traces back to three things: heat, fit-up, or preparation. Once you can name a defect and its root cause, preventing it becomes routine. Here are the offenders one by one.
Symptoms: a groove or notch melted into the base metal right at the weld toe, along the plate edge.
Root cause: too much amperage, too long an arc, or a travel angle that directs the arc away from the edge. Thin edges give the arc nowhere to go but sideways.
Prevention: drop amperage 5–10%, shorten the arc, and slow travel just enough to let filler metal fill the crater left by the arc. On thin sheet, our TIG sheet-metal settings guide covers the finesse techniques that matter here.
Repair: grind the notch smooth and re-weld with corrected settings. Deep undercut in a tension member may require grinding out the whole segment. We have a complete deep dive in our undercut defects guide.
Symptoms: holes, excessive thinning, or the edge literally melting away (melt-back) before filler can catch up.
Root cause: heat input exceeds what a thin, poorly-supported edge can absorb. No backing, no chill block, and a gap in fit-up compound the problem fast.
Prevention: back the edge with a copper chill block, use stitch or back-stepping techniques, reduce wire feed and voltage, and tighten fit-up gaps. If you’re still chasing blow-through, our stick welding thin metal guide and sheet-metal rod selection guide explain the tradeoffs.
Repair: fill the hole with a series of overlapping tacks or short passes, never one hot pass.
Symptoms: toe cracks at the weld boundary, crater cracks at the stop-start point, or lamellar tearing beneath the weld on heavily restrained sheared plate.
Root cause: high restraint and shrinkage stress, sulfur-rich steel with poor through-thickness ductility, or simply a crater left unfilled at the edge.
Prevention: fill craters every time, use lower-hydrogen consumables like E7018 (see our 3/32 E7018 settings guide), apply appropriate preheat on thick sections, and grind off hard sheared zones instead of welding directly over them.
Repair: cracks are never “weld over.” Cut or grind completely past both ends of the crack, then re-weld.
Short Note: Never weld over an edge crack. Cracks propagate under the new weld’s shrinkage stress. Grind out to sound metal, verify with dye penetrant, then re-weld. If a crack keeps reappearing in the same spot, you have a restraint or metallurgy problem, not a technique problem — stop and escalate to engineering.
Symptoms: a cold, unfused line between weld and base metal, often invisible from the surface but obvious in bend or section tests.
Root cause: arc aimed too far into the joint, travel speed too fast, insufficient current, or an oxide-covered edge (aluminum especially).
Prevention: aim the arc at the edge, clean down to bright metal, and bump amperage slightly.
Repair: gouge or grind out the unfused length and re-weld.
Symptoms: icicles hanging off the back side, root mismatch (one plate sitting higher than the other), or a blown-out root.
Root cause: too much heat, gaps in fit-up, or clamping that forced plates into misalignment mid-weld.
Prevention: control heat input, fixture properly, and check alignment after tacking, not just before. If you’re unsure how much heat your process really puts in, our MIG setup guide walks through the numbers.
Symptoms: a raised fin along the shear line, a hard, brittle band just below the shear edge, or notches left by cutting torches or saws.
Root cause: shearing cold-workens and partially hardens the edge; cutting leaves stress-concentrating notches.
Prevention: AWS D1.1 addresses this directly. AWS D1.1/D1.1M includes Figure 5.1, “Edge Discontinuities in Cut Material,” which sets allowable notch depths in sheared or cut plate before welding. Grind off fins and the hardened zone, or machine the edge entirely. D1.1 also permits building up deficient end preparations by welding and re-preparing them by machining or grinding — a legitimate code-approved fix.
Repair: grind notches smooth and blend beyond the hard zone.
Symptoms: the edge waves, cups, or pulls out of plane after welding.
Root cause: unbalanced heat, wrong weld sequence, or no fixturing.
Prevention: back-step sequences, balanced weld placement, proper tack spacing, and chill blocks. Understanding how tack welding interacts with pull forces helps enormously — our MIG tack weld guide covers spacing and sizing.
Visual inspection is your first and cheapest line of defense. AWS D1.1 Table 6.1 sets the practical limits: undercut is generally limited to 1/32 in (0.8 mm) depth, and fillet welds may run up to 1/16 in under required leg size over no more than 10% of the weld length. Anything beyond that is rejectable.
Magnetic particle (MT) and dye penetrant (PT) testing find surface-breaking cracks the eye misses, especially hairline toe cracks on ground edges. If a structural edge crack is suspected, test before you grind.
Radiography and ultrasonic testing get trickier near edges. D1.1 requires special image quality indicator (IQI) and edge-block placement when radiographing near plate edges, because the geometry generates false indications that look like defects but aren’t. Skip this and you’ll chase phantom cracks all day.
Quick Note: Grinding out and re-welding is legitimate for isolated, shallow defects — undercut, a short lack-of-fusion line, or a small crater crack. Scrap the part (or send it to engineering) when the same defect repeats after two repair attempts, the crack touches the weld root or base metal, the defect sits in a pressure boundary or dynamic-load member, or grinding would thin the section below minimum thickness.
One habit worth building: log edge defects by type and location. Patterns emerge fast — usually pointing back to one cutting machine, one operator, or one WPS. That data is how shops move from firefighting to prevention.
Edge prep theory only matters until you see where the work lives. Once you look for welded edges, they are everywhere: duct running across a warehouse ceiling, the chute that eats rock all day, the brushed-steel counter of a coffee shop, the battery pack of an electric car. Each of those jobs fails or succeeds at the same place — the last eighth of an inch of the plate.
On a rooftop HVAC job the crew joins 24-gauge galvanized transitions to a 16-gauge curb adapter. The gauge change is where trouble starts: the thin side burns back if you linger, the thick side will not fuse if you rush it. The fix is fit-up discipline more than amps — tack evenly from the center outward, the pattern in our MIG tack weld guide, then run short stitches instead of one continuous bead.
Duct edges also show the classic cosmetic defect: a ripple of undercut along the seam that reads as a dark line under paint. It is a textbook edge-discontinuity complaint, and almost always a travel-speed and gun-angle problem rather than a metal problem.
Take a mining chute lined with AR plate. Its lining is welded at the edges so material slides across a continuous surface instead of catching on a step. Weld it proud and you have built a wear ridge that sheds into the seam; weld it undercut and you have started a crack that propagates under impact loading. Wear-facing edge welds are ground flush, then checked visually and by magnetic particle testing before service.
The same logic covers aggregate hoppers, grain bins and dust-collector funnels — anywhere a flowing product rides across a welded boundary. The weld is not just structural; it is the wear surface.
Shell courses, nozzles and corner details put welded edges under pressure, literally. Here edge prep is governed by code: AWS D1.1 requires that surfaces and edges to be welded be smooth and uniform, free of fins, tears and cracks, and it permits building up a defective end preparation by welding or re-preparing it by machining or grinding. Nozzle and run-on details follow clearance guidance from AISC, the reference shops use when a weld lands too close to a connection heel. For the workflow from procedure to inspection, see our pipe welding procedures guide.
A café job shows it in miniature: a 1,200 mm stainless back-bar cladding panel, folded at the corners and seam-welded along the fold. The point of an architectural welded-edge panel is that the exposed boundary is smooth, rigid and weather-tight rather than a sharp open cut. The weld is ground through a grit progression and brushed to match the surrounding grain, so the seam disappears.
Elevator cab panels, column covers and rainscreen cladding work the same way: fold-and-weld gives thin sheet a stiff edge no bare shear cut can match, and it removes the raw edge that rusts first outside.
In food and pharma plants the specification is blunt: welded, ground flush, then passivated. The reason is crevice corrosion. An open cut edge or unblended weld toe creates a pocket where residue sits, bacteria colonize and clean-in-place solutions cannot reach. Grinding flush and passivating removes those sites and restores the passive chromium-oxide layer that makes stainless corrosion-resistant.
At the other end of the size scale are parts where the edge weld is the product. MIL and aerospace-style edge-welded bellows are built from stacked foil diaphragms fusion welded at their edges, producing a sealed part that flexes thousands of cycles. Filter cans, sensor housings and hermetic enclosures rely on the same continuous edge seam.
The newest example is electric-vehicle battery packs, where fiber-laser edge welding has become standard for pack lids — a continuous leak-tight seam replaces the gasket entirely. Module covers use the same approach, which is why spot and seam welding physics is now relevant reading for automotive fabricators. Thin foil edge work fits GTAW naturally; our TIG sheet metal guide covers the settings.
Security fencing, machine guards and concrete reinforcement panels are mesh assemblies finished with a welded edge bar or bound frame. Here edge distance essentially disappears: the weld lands on a wire, not a plate, so weld size follows wire diameter and settings stay conservative to avoid burning the intersection off.
Conveyor belting is the industrial cousin of all the above. ISO 14890:2026 sets cover and construction requirements for textile conveyor belting used on flat or troughed idlers, and edge treatment is part of that picture: a sealed or welded edge controls spillage and stops moisture working into the carcass. Same principle as the cladding panel — the edge wears first, so it is reinforced first.
Quick Note: When should you fold instead of weld? A folded edge wins on thin sheet where stiffness, hand-safety and speed matter — duct flanges, cabinet returns, cladding returns. A welded edge wins where the joint must be sealed, ground flush, load-bearing or machined later. If the edge is customer-visible and needs a continuous sealed seam, weld it; if it only needs to be safe and rigid, fold it.
Watch these applications long enough and the pattern is identical: the seam may run three metres, but the failure, the leak or the complaint always starts at one end of it. That is why edge prep pays off on every job here.
Welding an edge leaves very little room for error. With minimal material to absorb heat, a thin edge can burn back in seconds, while a restrained, thick edge can crack if it’s rushed. To avoid either outcome, the process, settings and heat control must all work in unison.
Not every arc process is suited for edge work. Each one has a specific range where it shines:
For steel edge and corner joints, ER70S-6 wire with GMAW in DCEP and spray transfer is a common baseline. The following range is widely used on the shop floor:
| Parameter | Typical Setting |
|---|---|
| Process / polarity | GMAW, DCEP (electrode positive) |
| Filler | ER70S-6, 0.030–0.045 in |
| Amperage | 190–280 A |
| Voltage | 22–30 V |
| Wire feed speed | 350–600 ipm |
| Contact-tip-to-work distance | 1/2 to 3/4 in |
| Transfer mode | Spray (short-circuit for thin edges) |
Keep your stickout consistent. Pull it too far and you’ll lose penetration, push it too close and you’ll burn the tip and drop out of spray. If you’re starting from scratch, our MIG setup walkthrough lays out the correct dial-in order.
Amperage tells you half the story. Heat input is what determines if an edge warps, burns back or fails to fuse. Calculate it using the standard formula:
Heat input (kJ/in) = (60 × Amps × Volts) ÷ (1000 × travel speed in ipm)
For example, 220 A at 25 V at 12 ipm yields roughly 27.5 kJ/in, while the same settings at 20 ipm drop to about 16.5 kJ/in. Travel speed and weave width are your cheapest levers to adjust it. Preheat, however, is code-driven. The AWS D1.1 structural welding code ties minimum preheat to base-metal thickness and ambient temperature, requiring material below freezing to be warmed before starting.
Quick Note: On stainless edges, back purging with argon isn’t a luxury. Without it, chrome-moly and stainless roots will develop sugaring (oxidized scale) that can’t be fully removed by grinding.
Copper backing bars pull heat away while shaping the root, making them standard for thin stainless and aluminum edges. Chill blocks clamped along the face work in tandem to prevent waviness. To seal a purge, a well-fitted dam is required. For more on this, see our gas line welding purging guide.
When assembling an edge, tack in a back-step sequence with small, evenly spaced tacks to counteract shrinkage. Our MIG tack weld guide breaks down proper spacing and sizes for clean joints.
Between passes, remove slag, silicates and oxides by wire-brushing or grinding. Trapped contaminants at an edge are a direct path to lack of fusion. Light peening of a restrained edge can relieve stress, but only when used cautiously: keep it to the first pass with a round-nosed hammer, and never peen the root or cap indiscriminately. Always terminate with a crater fill to prevent crater cracks.
For the finish, grind along the edge, not across it, using a grit progression until the bead is blended. Stainless edges should then be passivated or electropolished to restore their corrosion resistance. Lastly, never begin without proper protection. Edge welding can throw spatter back at you, so check your full kit using our welding PPE guide before striking an arc.
Edges concentrate heat, spatter and light in a narrow band, which is exactly why they hurt people more than the middle of a weld does. A bead running along a plate edge throws spatter sideways, exposes your glove cuff to radiant heat, and tempts you to peek at the joint without your hood fully down. None of that is theoretical. The injury data is blunt about how badly welders get hurt.
In the Bureau of Labor Statistics heat-burn study, more than two-fifths of injured workers suffered third-degree burns and about one-sixth required hospitalization averaging six nights. Read that again: the “minor burn” category is smaller than people assume. Welding burns are rarely flat, clean, first-degree injuries — they are often deep, from molten spatter that sticks to skin through a hole in a glove or a gap at a cuff.
Edge work raises the odds because spatter has a clear escape path off the workpiece instead of bouncing back into the joint. The fix is boring and effective: dry, hole-free gloves, leathers with no open cuffs, and a pair of welding PPE chosen for the position you are actually welding in. Wet gloves conduct heat; hole-free gloves stop the molten droplet before it finds skin. If you want the full gear breakdown, see our what does a welder wear guide.
The UV from an arc is intense enough to burn skin the way a sunburn does — arc burn, or “welder’s flash” on the eyes. OSHA 1910.133(a)(5) requires proper filter lenses, and shade selection guidance is published in AWS F2.2. At an edge, your hood is more likely to be flipped up mid-pass for a fit-up check, so get in the habit of a hand-shield or auto-darkening lens rather than a quick unprotected glance.
Fumes are the slow hazard. Manganese and hexavalent chromium — the latter from welding stainless and chromoly — are respiratory and neurological concerns that accumulate over years. OSHA 1910.252 sets the ventilation floor, and general ventilation plus local exhaust at the joint is the practical answer. For process-specific fume profiles, our MIG welding safety precautions guide covers what each transfer mode produces.
Edge welding often happens on large, grounded, already-installed fabrications — tanks, duct runs, equipment frames. That is a shock hazard when the workpiece is energized indirectly, when leads are draped across sharp edges that can nick insulation, or when you weld in damp conditions. Keep connectors tight and insulated; our welding connectors guide explains why a loose lug heats up and fails.
This is where edge work turns deadly. Striking an arc near a container with residual product is a classic fatality pattern. OSHA’s public accident files include a case where an employee suffered third-degree burns about 6 inches in diameter when a flange exploded as the arc was struck during pipe repair. Tank-explosion fatalities during welding appear repeatedly in the same archive.
The compliance map is straightforward:
Supporting practice notes live in university safety material on welding health and safety and AWS procedure documents such as AWS WFP-2.07 on erection welding safety. Pipe and vessel edge joints carry extra purging and hot-work obligations — see gas line welding safety and purging rules and our pipe welding procedures guide.
Short Note: Before striking an arc near an edge: confirm the container is cleaned, purged and gas-free; check for flammable residue within 35 feet; stage a charged extinguisher; verify ground clamp placement away from the joint; confirm dry gloves and leathers; check ventilation or a respirator; and brief a fire watch for 30 minutes after the last arc.
One last practical point: burns still happen despite good gear, so know the response. Our weld burns on arms guide covers first aid and when to escalate, and overhead edge work is where spatter risk peaks — spats and a neck guard earn their keep there.
Every step of a welding job leads back to the edge. It’s the most common point of failure, but also the easiest to protect. Most issues with welded edges come from a small group of avoidable mistakes. Correcting them will improve the quality of hoppers, ducts, chutes, tank shells, and any other joint you fabricate.
Quick Note: The t minus 1/16 in rule exists to prevent the melted edge from pulling away on sub-1/4 in material. Because of this, undersize edge fillets are still considered a code concern. For this reason, undercut along an edge is one of the most frequent causes for rejection of an otherwise clean looking weld.
| Symptom | Probable cause | Fix |
|---|---|---|
| Burn-through at edge start | Too much heat, no backing, arc lingering | Add chill block, raise travel speed, back off amperage 10–15% |
| Toe cracking on sheared edge | Hardened shear zone left in place | Grind or machine the hardened layer off, then re-prep |
| Undercut along edge toe | Travel too fast, angle too steep, amps too high | Slow down, ease the electrode angle, add a small weave |
| Lack of fusion at root | Gap too tight, bevel too narrow, wrong polarity | Open root slightly, widen bevel, verify DCEP and wire type |
| Edge waviness after welding | Uneven heat, poor fixturing, no sequence | Tack in a balanced sequence, clamp to a strongback, skip-weld |
| Sugared stainless edge | No purge or root shield | Purge with argon, verify oxygen before striking arc |
| Crater crack at termination | Arc pulled off abruptly | Use crater-fill technique, overlap the last 1/2 in |
Short Note: D1.1 permits limited undersize fillet welds — up to 1/16 in under the required size over 10% of a weld’s length — but never treat that allowance as a substitute for correct edge sizing. Check the current clauses in AWS D1.1/D1.1M before you rely on it.
Hand-grinding is acceptable for short runs, mild steel, non-pressure work, and gentle bevels when an operator can hold geometry. However, pay for machined or laser-cut edges when the bevel angle must be held to a tolerance, when thick plate makes grinding slow and inconsistent, when the alloy is corrosion-sensitive, or when the joint is a pressure boundary. A laser edge may cost more per part, but often saves hours of fit-up and rework.
Some decisions cannot be left to field judgment. Escalate when the joint carries dynamic or cyclic loads, forms a pressure boundary, joins stainless in food or pharmaceutical contact, involves chrome-moly or preheat-dependent alloys, or falls under structural or seismic criteria where AISC clearances and edge-distance rules apply. Bringing engineering in early prevents a WPS from becoming retroactive paperwork.
The success of any seam starts with the edge. Focus your efforts on proper prep, correct sizing, controlled heat, and a thorough inspection of the toe. These fundamentals will outlast any quick fix. If you’re still building your skillset, pair this checklist with The Complete Guide to TIG Welding Sheet Metal, and ensure your protection is up to par by reading Welding PPE: The Complete Guide along with MIG Welding Safety Precautions. For fit-up details that directly affect edge alignment, see our MIG tack weld guide. Next in this series, we’ll expand these practices to heat input and distortion control across full assemblies.
It generally means an edge that has been prepared and fused by welding — either an edge joint (parallel members joined at their edges), an edge weld along overlapping edges, or a finished part whose exposed edge is seam-welded (as in welded-edge panels or belting). Context decides which meaning applies.
An edge joint is a joint type where two parallel members are welded together at their aligned edges. An edge weld is a fusion weld placed along the edges of overlapping members — common in bellows, diaphragms and thin enclosures. The AWS D1.1 code treats the edge joint as one of the five basic joint types.
AISC 360 Table J3.4 gives minimum edge distances for welds to prevent burn-through and edge melting, and AWS D1.1 requires enough edge distance to avoid excessive edge melting during welding. As a working rule, the fillet leg plus a small margin of base plate beyond the weld toe is the bare minimum — more is always better on thin material.
On material less than 1/4 in thick, AWS D1.1 limits the weld to the plate thickness minus 1/16 in. On 1/4 in and thicker material, the full thickness may be used. Oversize welds on thin edges tear the melted edge out instead of strengthening it.
Use short-circuit GMAW or TIG with autogenous fusion, keep the bevel small or square, support the edge with a copper or aluminum chill block, shorten arc time with faster travel, and stagger tacks to control heat buildup. Lower amperage helps, but travel speed and heat sinking matter more.
Edge cracks usually come from undercut at the weld toe, crater termination, dirty or hardened sheared edges, or too much restraint. Grind the crack out completely, dress the edge back to clean metal, verify the edge distance and bevel, then re-weld with proper crater fill — and use MT or PT to confirm the crack is gone.
For visible or sanitary work, a welded-and-ground edge eliminates the sharp, porous cut face and any crevice where corrosion can start — which is why food and pharma equipment specs call for ground, passivated welded edges. For simple non-visible brackets, a clean cut or fold is cheaper and perfectly acceptable.
Yes — heavy textile conveyor belting is specified for cut, moulded or welded edge finishes depending on the belt type, and the current international specification is ISO 14890:2026 (3rd edition, February 2026). Lighter thermoplastic belts commonly use hot-welded edges to seal the carcass against moisture.
A welding helmet with the correct filter shade (shade selection guidance is in AWS F2.2 and required by OSHA 1910.133), safety glasses underneath, dry hole-free leathers, cuff-less pants, and spats or covered shoes to keep hot slag out. Edges throw spatter in unusual directions, so full skin coverage is non-negotiable.