Weld Joints: Types, Symbols & Drawing Prep

A weld joint is the junction where two members meet, and a weld is the metal that joins them. AWS recognizes five basic joint geometries: butt, lap, tee, corner, and edge. Each geometry can carry several weld types, most often fillet or groove welds. On a drawing, the weld is specified with a welding symbol from AWS A2.4, or from ISO 2553 where the ISO system governs. The arrow marks the joint, the reference line places the weld on the arrow side or the other side, and the numbers around the symbol state the size, the length, and the spacing. A complete callout names the weld type, the side, the size, and the extent, because a symbol that leaves any of these out hands the decision to the shop floor.

Joint geometry versus weld type

The two words are often swapped, and the swap causes real drawing errors. A joint is the junction of the members, the place where their faying surfaces meet. A weld type is the shape of the metal that joins them. AWS defines the joint as the junction of the workpieces, before and after joining, and each welding symbol applies to one joint: the joint its arrow points at. The joint geometry comes from the part views; the symbol adds the weld.

One joint geometry can host several weld types. A tee joint is usually joined by fillet welds, the triangular welds laid into the corner between the two members, but in thick or heavily loaded material the same tee joint can carry groove welds that penetrate the member. A butt joint on thin sheet takes a square-groove weld with no edge preparation, while the same joint on thick plate needs a V, bevel, U, or J preparation to let the weld reach through the thickness. The choice among weld types turns on the thickness, the load path, the access from each side, and the code that governs the work. The welding hub covers how those forces and positions shape the decision.

Most work splits into two weld families. Fillet welds fill an angle between surfaces and are sized by their leg length. Groove welds fill a prepared channel between edges and are sized by their depth or effective throat. Beside them sit the specialty types: plug and slot welds join overlapping members through holes, spot and seam welds come from resistance processes, and surfacing welds build layers onto a surface. The definitions for each term sit in the manufacturing glossary. The material being joined shapes both weldability and filler choice, which the welding materials page covers.

The five weld joint types

Butt joints

A butt joint places two members in the same plane, edge to edge. It is the most common joint in structural and pressure work because it keeps the load path in one plane and wastes no overlap material. On thin material, up to roughly an eighth of an inch in steel, a square butt joint needs no edge preparation at all: a tight fit or a small root opening is enough. The exact limit depends on the process and the position. As thickness grows, the arc cannot reach the root of a square edge, so the edges are prepared into a groove.

Groove preparation is where butt joints branch. A V preparation chamfers both members, simple to cut with a torch or plasma and the usual choice for moderate thickness. A bevel preparation chamfers only one member, which halves the preparation and the filler but leaves an asymmetric weld. The arrow of the symbol points at the chamfered member to say which side gets the treatment. U and J preparations curve the edge instead of chamfering it, which costs more to machine but needs less filler in thick plate. A double preparation, double V, double bevel, or double U, balances shrinkage from both sides on the heaviest sections. For example, a one-inch plate joint moves from a single V to a double V or a U when distortion control and filler volume start to matter, and that decision belongs on the drawing, not in the shop.

Lap joints

A lap joint overlaps one member on another, and it is the easiest joint to fit up: no edge preparation is needed, and the fit-up tolerance is forgiving. It is welded with a fillet weld along the edge of the overlapping member, on one side for light duty or both sides when the joint must seal or carry load. Lap joints suit members of different thicknesses, common in sheet metal assemblies, and the sheet metal fabrication page covers the work that uses them most.

The trade-off is the eccentric load path. Because the members are offset, the load passes through the weld in shear rather than straight tension, so a lap joint needs more weld metal than a butt joint of equal strength. The overlap gap is also a crevice that traps moisture and corrosive media, a real issue in anything washed or exposed, and both faces of the overlap are common sites for the porosity and lack-of-fusion defects the welding defects page catalogs.

Tee joints

A tee joint brings one member to the middle of another at a right angle, forming a T. It is the standard geometry for brackets, stiffeners, gussets, and tube-to-plate connections. The usual weld is a fillet on each side of the standing member, and AWS D1.1 ties the minimum size of those fillets to the material thickness, so an underspecified tee weld can fall below code even when it looks generous.

Thick or highly loaded tee joints graduate to groove welds, often a double bevel with fillet welds superimposed, when both sides are accessible and the load justifies the preparation. The risk specific to tees is lamellar tearing, a crack that runs through the thickness of the through member when weld shrinkage strains it across its rolled layers. Heavier joint restraint and thicker plate raise the risk, which is why critical tee connections get attention to weld sequence and to the properties of the through member. The material side of the problem sits on the materials page.

Corner joints

A corner joint meets two members at their ends to form an L, as distinct from the tee, where one member lands at the middle of the other. Corner joints dominate frames, boxes, and enclosures, and they come in two forms. An open corner leaves a visible V at the outside, usually filled with a fillet weld, and a closed corner butts the edge of one member to the face of the other, welded from the outside and sometimes from the inside where access allows.

Corner joints in light sheet are often welded as edge or corner-flange welds. Where the corner must present a flush face, for appearance or for a sealing surface, the drawing adds a contour symbol to the weld symbol to call for a finished face. Corner joints carry load best when the weld closes the corner fully. An intermittent or single-sided corner weld leaves a flexible, crevice-prone edge, which matters for anything galvanized, washed, or sealed.

Edge joints

An edge joint joins two members at their edges with their surfaces parallel or nearly so, most often the bent flanges of folded sheet. Because the surfaces lie in parallel planes, the joint naturally takes an edge weld, a light weld along the contacting edges, or a groove weld where the bent flanges form a V between them. Edge joints are common in light-gauge work and appear throughout ducting, enclosures, and panel assemblies.

They are the lightest of the five joints and rarely carry a primary load path, since the parallel overlap gives little weld depth and the joint is prone to the crevice and slag-inclusion problems of any lapped geometry. Where an edge joint must seal, the weld runs continuous and the callout says so, either with a length covering the full joint or with an all-around symbol where the joint really wraps the part.

How the AWS welding symbol works

The welding symbol is a small assembly with fixed rules, and AWS A2.4 standardizes every part of it. A horizontal reference line anchors the symbol. An arrow runs from the reference line to the joint, and where the arrow meets the line, a tail can carry process notes, a welding procedure reference, or any other data with no assigned place on the symbol. The weld symbol itself, the pictograph that names the weld type, sits against the reference line. Its position is the load-bearing rule of the whole system: information for the arrow side of the joint goes below the reference line, and information for the other side goes above it. A2.4 states this directly, and it holds no matter which way the arrow points and no matter which side of the assembly the reader sits on.

A symbol below the line welds the side the arrow touches. The same symbol above the line welds the far side. Symbols on both lines weld both sides.

Terminology distinguishes the welding symbol from the weld symbol, and the distinction saves confusion when a drawing is disputed. The welding symbol is the whole assembly, reference line, arrow, tail, dimensions, and supplementary marks. The weld symbol is only the pictograph of the weld type, the triangle for a fillet or the V-shaped line pair for a V groove. A drawing note that says a weld symbol sits on the wrong side means the pictograph, while a dispute about the welding symbol can involve any element of it.

Two supplementary marks appear constantly. A flag at the arrow-and-line junction means a field weld, made during erection or installation rather than in the shop. Older drawings used a filled dot for the same meaning. An open circle at the same junction means the weld runs all around the joint. Both marks change what the fabricator is asked to do, so they belong only where they are meant. One further rule prevents a quiet error: each welding symbol applies to the joint its arrow points at, so a connection made of several distinct joints, such as a plate passing through a slotted tube, needs a symbol for each joint rather than one symbol read broadly.

A last element handles sequencing. Where passes on both sides must follow an order, or where a back weld follows a root, multiple reference lines stacked away from the arrow are read in order, the line nearest the arrow first. The welding process itself, MIG, TIG, or Stick, is not named by the weld symbol. It belongs in the tail or in the procedure document, and the welding types comparison covers how that choice is made.

Reading dimensions on weld symbols

Numbers ride at fixed positions around the symbol, and the positions are the same on every drawing. For a fillet weld, the weld symbol is a right triangle with its perpendicular leg always on the left. The weld size, which for a fillet means the leg length, sits to the left of the triangle. The weld length, when it is shorter than the joint, sits to the right. When the legs must be unequal, both dimensions are given, and the drawing indicates which leg is longer.

Intermittent fillets carry two numbers to the right of the symbol, segment length first, pitch second. The pitch is the trap: in the AWS system pitch is measured center to center between segments, not as the clear gap between them. A callout of 2-4 means two-inch segments on four-inch centers, which leaves a two-inch gap between welds. A shop that reads the four as the gap delivers half the intended weld count. The number of segments, where it matters, sits in parentheses above or below the symbol.

Groove welds carry more data because the preparation itself is dimensioned. The depth of the groove preparation sits to the left of the symbol. The effective throat, the weld size that counts for strength, sits in parentheses after that depth when it differs. The root opening, the intentional gap between members at the root, is shown inside the symbol, and the groove angle sits below or inside the symbol as the joint geometry requires. Where penetration must be proven through the joint, a melt-through symbol across the reference line calls for root reinforcement on the back face. A backing bar symbol, marked with an R when the bar must be removed after welding, calls for backing that supports a full-penetration root.

The ISO system, specified in ISO 2553, shares the pictographs but reads differently, and the differences are exactly where drawings go wrong. The usual ISO layout uses a dual reference line: a solid line with a dashed companion line. The weld symbol sits on the solid line for the near side and on the dashed line for the other side, rather than above and below a single line. ISO 2553 also documents a single-line system, so the standard itself carries two coexisting conventions in separate clause sets, which is one more reason the drawing set must name its governing standard. Fillet sizing differs as well: ISO prefixes the size with a z when it is a leg length or an a when it is a throat thickness, where AWS states the leg length with no prefix. Intermittent pitch differs too: ISO measures the clear distance between segments, where AWS measures center to center.

Contour and finish marks complete the symbol. A contour mark, flush, convex, or concave, states the required face shape. A finish letter after it states the method: G for ground, M for machined, C for chipped, H for hammered. These marks earn their place only where the face must be finished, since every mark added is a requirement the shop must meet and inspect.

Calling welds out on drawings

Drawing preparation is where real disputes concentrate, and the practice rules are consistent across sources. Name the governing standard on the drawing set, in the notes or the title block, and use it alone; mixing AWS and ISO conventions on one sheet is a documented source of fabrication disputes. The general habits that make drawings unambiguous are covered in technical drawing requirements. The same discipline that names a thread standard, size, pitch, and class, as the thread standards guide shows, applies to welds: state the weld type, the side, the size, and the extent, every time.

Size deserves the first check. A fillet weld called out with no size has no default to fall back on, because AWS practice gives fillets no unspecified default: an unsized fillet leaves the size to the fabricator. A groove weld specified without dimensions reads toward complete joint penetration. Neither reading is usually what the designer intended. State fillet sizes explicitly, reference a noted schedule where the drawing set has one, and where partial joint penetration is acceptable on a groove, state the depth or effective throat so the symbol cannot be read as CJP. Complete joint penetration is specified in one of three ways: groove symbols without dimensions, the letters CJP in the tail, or a backing or back weld symbol placed opposite the groove symbol.

Two trap cases come from the structural literature. The weld-all-around symbol suits a joint that really runs around the whole connection with constant geometry. At three-dimensional corner configurations the weld type can change along the path, fillet along a face and flare-bevel around a curved edge, so each distinct joint deserves its own symbol rather than one all-around mark. And at a slotted tube accepting a plate gusset, a double fillet symbol specifies welds on the outside and the inside walls of the tube, not the two exterior welds beside the plate that most designers intend. AWS guidance for these connections calls for separate single fillet symbols for each exterior joint, or a note in the tail. For example, a slotted HSS brace-to-gusset connection is drawn with four single fillet symbols when all four exterior welds are wanted, or the tail note replaces them where the pattern is unmistakable.

Structural work adds a code layer. AWS D1.1 requires the welding symbols on contract documents to be those of A2.4, and it sets minimum fillet weld sizes by material thickness, so the structural drawing set inherits both the symbol rules and the size floors. Inspection closes the loop: acceptance of a coded weld runs against the code criteria, not against appearance, and the quality inspection practices document how that acceptance is verified. The design review pass, the habit of checking every joint for access, fit-up, and both-sides welding before release, is part of the design for manufacturing discipline. Welding itself is one step in the larger metal fabrication chain, which is the right place to see how joint choices interact with cutting and forming upstream.

Joint types and their symbols at a glance

The table below lines up the five joint geometries, the groove variants of the butt joint, the welds and symbols that typically specify them, and the edge preparation each implies. Read it as the map from geometry to callout: the joint comes from the part views, the symbol comes from AWS A2.4, and the edge preparation column is what the shop must do to the members before the weld goes in. Where a row lists a groove symbol, the drawing must also carry the groove dimensions, the depth, the angle, and the root opening, or point to a prequalified joint detail that supplies them.

Weld callout checklist

Run this list against any drawing that carries welds before it releases to a fabricator.

• The governing symbol standard is named on the drawing set, and only that standard appears on it.
• Every fillet weld carries a size, stated at the symbol or by reference to a noted schedule.
• Every weld symbol sits on the correct side of the reference line for the side intended, checked against the arrow, not against the near side of the assembly.
• Groove welds state the preparation depth, the angle, and the root opening, or reference a joint detail that states them.
• Partial versus complete joint penetration is explicit, by dimensions, by the CJP tail note, or by a backing or back weld symbol.
• Intermittent welds give length and pitch, and the pitch is read center to center under AWS rules.
• Weld-all-around and field-weld marks appear only where the joint and the location truly warrant them.
• Complex connections, such as slotted tubes, carry a symbol per joint rather than one symbol read broadly.
• Contour and finish marks appear only where the finished face must be controlled.
• The tail carries the process or procedure reference where the weld is procedure-controlled.

A callout that passes this list tells the shop exactly what to build and tells the inspector exactly what to accept, which is the whole point of the symbol system: the joint is designed once, on the drawing, instead of negotiated at the bench.