Pipe Fitting Welding: Socket Weld vs Butt Weld Guide

A 2-inch stainless elbow in a chemical skid suddenly starts weeping at the joint, and the plant has to shut down the whole batch line. The elbow itself was fine, the pipe was fine, and the welder was experienced — but the pipe had been bottomed hard into the socket with no gap, so when the joint heated up, the pipe expanded and cracked the root of the fillet weld. That single detail — a 1/16 inch gap that most people think is optional — is the difference between a fitting that lasts a decade and one that fails in the first thermal cycle. This guide walks through pipe fitting welding the way experienced fabricators actually think about it: which joint type to choose, how to prepare and fit it up, what the codes really require, and how to fix the defects that show up in the field.

Why Pipe Fitting Joints Fail More Often Than the Pipe Does

Ask anyone who maintains industrial piping where leaks come from, and you get the same answer: not the pipe, the joint. The elbow that cracked at its weld, the tee that wept at the branch, the flange that separated at the fillet, the valve body that failed where the neck met the weld. Pipe is dependable, arriving from the mill with uniform wall thickness, known chemistry, controlled manufacturing, and consistent heat treatment. The fitting joint is where all of that control ends.

So what is pipe fitting welding? Joining a fitting (elbow, tee, reducer, cap, flange, or valve body) to pipe or tubing with a permanent welded joint instead of a threaded or bolted one. The weld becomes part of the pressure boundary and must carry pressure, thermal movement, vibration, and service load for the life of the system.

Three joint families dominate:

• Butt weld: pipe and fitting are beveled and joined edge to edge with a full-penetration groove weld. Strongest and most inspectable, it keeps the bore clear, but demands accurate preparation and fit-up.
• Socket weld: the pipe slides into a machined recess and is fillet welded around the outside. Fast and forgiving, which is why it dominates small-bore instrument, utility, drain, and vent lines.
• Threaded and flanged: mechanical joints. The pipe threads into the fitting, or a flange gasket bolts two faces together. Even so, a weld-neck, slip-on, socket-weld, or lap-joint flange still depends on a weld to stay attached.

Standards govern dimensions: ASME B16.9 covers factory-made wrought butt-welding fittings from NPS 1/2 through NPS 48, while ASME B16.11 covers forged socket-weld and threaded fittings, with socket-weld classes 3000, 6000, and 9000. Geometry is standardized. Quality is not.

A Fitting Weld Is Not a Plate Weld

Welding a fitting is a different engineering problem from welding a plate or flat spool. A plate weld has room around it. A fitting weld does not.

• The geometry is tight. You weld into a corner, often at a change of direction, with limited access for electrode, torch, and wire feed.
• The heat is asymmetric. The arc sits outside the joint while the heavy fitting body acts as a heat sink beside it, so pipe and fitting do not heat or cool evenly.
• The fitting wall is often thicker than the pipe wall, so heat capacity and contraction differ. The joint is built to move, then asked not to.
• It must survive thermal cycling. A skid that heats, cycles, and cools loads the joint every time.

Every welded joint creates a stress concentration. A smooth pipe wall interrupted by a sharp fillet toe behaves like a notch: approximate factors differ sharply, roughly 1.2 for an as-welded girth weld butt joint versus roughly 2.1 for an unfinished fillet weld toe. That toe is where a socket-weld joint puts its most-loaded geometry, which is why socket welds are restricted in severe cyclic service. ASME B31.3 does not permit socket-welded joints larger than NPS 2 for severe cyclic service, because the socket crevice plus the fillet toe concentration becomes an unacceptable fatigue initiator.

Smaller-bore joints deserve extra suspicion: the ratio of fitting surface area to weld cross-section climbs as diameter drops, so more heat lands in less weld metal and distortion is proportionally larger. Socket welds are specified at NPS 2 and below because butt-welding small-bore pipe demands beveling and alignment precision that is impractical at that scale.

What the Elbow Does to the Flow

An elbow is a direction change, with consequences. Flow separates from the inner wall of the bend, forms a turbulent core, and exits with an altered velocity profile. That faster, swirling flow accelerates erosion of the internal pipe wall immediately downstream of the bend, especially where the service carries solids, particulates, or condensate. The fitting also acts as a lever arm for cyclic loading: vibration, pressure surges, and thermal expansion push on a geometry that cannot flex like a straight run.

Field repairs on piping often happen at elbows, and are usually made with a stick machine. SMAW remains the most dependable field process because it tolerates dirty metal, awkward access, and a moving job site better than anything else in the toolbox.

Three Layers of Competence

Every failure in service traces back to one of three things going missing:

1. Choosing the right joint type for the service, the size, and the inspection plan.
2. Welding it to a qualified procedure: preparation, gap, filler metal, heat input, and sequence.
3. Proving the joint through inspection and testing, not by appearance.

Miss any layer and the pipe will still be there long after the fitting is gone. The rest of this guide covers each layer in order.

Butt Weld, Socket Weld, Threaded, or Flanged: Choosing the Right Joint

Before anyone lights a torch, somebody has already decided how the pipe connects, usually from a line class, a spec sheet, or a habit copied off the last job. A fitting joint is where pressure, heat, vibration, corrosion, and inspection converge, so the four end-connection families in most pipe connection references are not interchangeable: each carries its own dimensional standard, rating philosophy, and inspection limits.

Butt Weld: The Default for Process Piping

A butt-weld fitting is forged or wrought from the same wall profile as the pipe, with ends bevelled to ASME B16.25 geometry. Dimensions come from ASME B16.9, which standardizes fittings from NPS 1/2 through NPS 24, including long-radius (1.5D) and short-radius (1.0D) elbows. The radius is not cosmetic: a 1.5D turn has lower pressure drop and less tendency to set up flow-induced vibration than a 1.0D elbow in a tight skid.

Note what butt-weld fittings are not rated by. No Class 150 or Class 3000 is stamped on the body. A butt-weld fitting inherits the pressure capability of the pipe schedule it is welded to, so the rating comes from the pipe wall and material, not the fitting. The joint is continuous metal, so full pipe strength carries through the elbow with no gasket, no thread, and no crevice to trap product.

Socket Weld: Fast, Compact, and Limited

A socket-weld fitting has a machined recess. The pipe slides in until it bottoms on the internal shoulder, then the pipe outside is fillet welded to the face of the fitting, as TWI describes the arrangement. Ratings come from pressure class rather than schedule, and ASME B16.11 covers Class 3000, 6000, and 9000 for socket-weld forged fittings. The type exists for speed: no beveling, no purging, forgiving fit-up, and a fillet often laid with a stick rod, as our guide to the advantages of stick welding notes.

The price is a permanent internal crevice at the bottom of the socket: a leak path, a corrosion trap, and the reason the joint cannot be radiographed.

Why a Socket Weld Cannot Be Radiographed

On a butt weld the beam crosses an open joint and the film shows root and lack of fusion as a density change. On a socket weld, the pipe wall inside the socket and the fitting wall around it overlap along the whole beam path. As the LANL welding manual explains, film density can no longer separate a sound root from lack of fusion, because the radiograph sees two thicknesses of steel at once. Ultrasonic testing fares no better against that geometry.

Inspection shrinks to visual plus surface methods: VT with magnetic particle for carbon and low-alloy steel, or VT with liquid penetrant for stainless and non-ferrous material. That is a real limitation, not paperwork. If a spec calls for radiography, the joint type has been chosen for you.

Threaded and Flanged Connections

Threaded fittings need no welding, so there is no heat-affected zone, no PWHT, and no procedure qualification. ASME B16.11 covers threaded forged fittings in Class 2000, 3000, and 6000 with NPT threads. The trade-offs are a restricted bore, a flow-disrupting thread profile, and a seal that depends on tape or compound, so threaded joints stay out of high-temperature and severe cyclic service.

Flanges fall under ASME B16.5, which sets pressure-temperature ratings, materials, dimensions, tolerances, marking, and testing for flanges and flanged fittings from NPS 1/2 through NPS 24. Four styles matter:

• Weld neck is butt welded to the pipe, and its tapered hub gives a gradual transition and a full-strength joint, preferred for severe and cyclic duty.
• Slip-on flanges slide over the pipe and are fillet welded: easier to install, weaker joint, generally limited to Class 150 and 300.
• Socket weld flanges take a pipe in a machined recess and are fillet welded, giving a compact joint cheaply.
• Lap joint flanges use a backing ring welded beneath the joint, so the ring is not left as a corrosion trap if the gasket leaks.

A bolted flange pair adds several leak paths, and it is the only one of the four families you can take apart later.

Attribute	Butt Weld	Socket Weld	Threaded	Flanged
Standard	ASME B16.9 / B16.25	ASME B16.11, 3000/6000/9000	ASME B16.11, 2000/3000/6000, NPT	ASME B16.5
Size	NPS 1/2–24	NPS 1/8–4	NPS 1/8–4	NPS 1/2–24
Strength	Inherits pipe schedule	Pressure class	Class, less thread engagement	Class per B16.5
Bore	Full, smoothest	Full plus crevice	Restricted	Style dependent
Fit-up	Highest: bevel, root, cap	Low: slide in, fillet	None	Medium: weld or bolt
NDE	VT, MT/PT, RT, UT	VT + MT/PT	Visual, seal check	Full on weld
Leak paths	One	Two	One, sealant dependent	Several
Best fit	Process piping, high temp, cyclic	Instrument, utility, drain, vent	Utility, low pressure, repair	Nozzles, tie-ins, disassembly

How to Actually Decide

Start with size. Above roughly NPS 2, butt weld is the default for any permanent process line.

Then read the service. High design temperature, severe cyclic duty, hygienic or high-purity product, and flow-sensitive lines point to butt weld, because a socket crevice cannot be polished out of the product path. Drains, vents, sampling lines, and instrument impulse connections up to a couple of inches belong to socket weld.

Material can override that: aluminum process lines generally avoid socket weld because of thermal expansion and oxide trapped in the crevice, so they use butt-welded joints made with purged MIG or TIG, as our aluminum MIG welding guide details. Check the inspection requirement before the price list: if the code, spec, or QA plan calls for radiography, butt weld is the only realistic answer.

Socket Weld Fitting Welding: Getting the Gap, the Fillet, and the Fit Right

Socket weld is the fastest joint in the shop and one of the easiest to get wrong. Its strength hangs on one dimension fabricators treat like a formality: the root gap, the space between pipe end and socket bottom. Get it wrong and the joint can look perfect and still crack open on the first thermal cycle.

The Gap Is Not Optional

Under ASME B31.3 paragraph 328.5.2(e), the pipe is pushed into the socket until it firmly bottoms against the internal shoulder, then withdrawn approximately 1/16 inch (1.5 mm) before welding. That withdrawal is a requirement, not a style preference. A second point is easy to miss: after welding, a gap is not required to be present and is not required to be verified. The check happens before the arc strikes, so a bad socket joint leaves the shop with no trace of the mistake.

Physics: a socket fitting body is machined from heavy forging or bar stock, and the pipe going into it is thinner. Under the arc, the thin pipe expands longitudinally faster and further than the fitting around it. With zero gap, that expansion has nowhere to go, so the pipe is driven outward against the socket shoulder and the fillet weld is bent into severe tension shear at its root and toe. Root or toe cracking shows up on the first thermal cycle.

Verifying the Gap Before the Torch Comes Out

Two methods confirm the withdrawal: a depth gauge or pin gauge inserted to confirm the pipe stands off the socket bottom, or a scribe mark on the pipe OD measured against the socket face after bottoming and withdrawing. Check at several positions around the circumference, not just at 12 o’clock. An off-center pipe shows a tight gap on one side and a wide gap on the other, distorting the fillet as it cools.

Confirm the socket itself first: bore diameter matching the pipe OD, socket depth adequate for the schedule, and a clean internal shoulder free of burrs and scale. A dry fit by hand catches a mismatched fitting early.

Sizing the Fillet Weld

ASME B31.3 Figure 328.5.2C sets minimum fillet dimensions. The leg dimension Cx must be at least 1.09 times the nominal pipe wall thickness tn, or the socket wall thickness, whichever is smaller. That gives a theoretical throat of Tc = 0.707 × Cx, which must be at least 0.77 × tn. The throat carries the load.

Some latitude is built in. A single continuous fillet weld greater than 5 mm (3/16 in.) may be undersized by no more than 1.5 mm (1/16 in.), provided the total undersized portion does not exceed 10 percent of total weld length or 50 mm (2 in.), whichever is less. Beyond that, it is a rejectable weld.

One relaxation is worth knowing: fillet and socket welds may vary from convex to concave contour. Groove welds are expected to show a smooth, even face profile, but a socket fillet can be slightly hollow within the required leg and throat dimensions. Do not dress a sound socket weld down to imitate a groove weld.

Closing the Joint

Once the gap is set, the sequence is straightforward:

1. Clean to bright metal, pipe OD and socket bore alike. Scale, oil and rust inside the socket become crevice corrosion sites you cannot reach later.
2. Tack lightly. Two to four small tacks around the clock, enough to hold the gap without loading the joint.
3. Weld continuously around the circumference rather than in separate passes that heat one zone while the rest stays cold. Control the heat, keep a short arc, stay on the leg of the fillet. Out-of-position work on small bore calls for filler suited to the position, so sort that out first. Our guide to all-position welding rods shows how filler selection changes the job.
4. Inspect for the signature defect pair. Root and toe cracking on a socket weld are classic symptoms of a missing gap. If you see either, the cause is almost never filler metal or arc length. It is the pipe bottomed hard into the socket.

One related joint deserves caution. A slip-on flange welded at the hub requires a square weld leg, because hub and pipe wall differ in thickness. Get that leg wrong and the joint cracks the same way, worsened by the preheat needed to fuse the heavier hub. The Los Alamos National Laboratory piping reference gives the dimensions for these flange weld details.

Get the gap, the leg and the throat right and a socket weld becomes one of the most reliable joints you can make on small bore piping. Miss the gap and you have built a time bomb.

Butt Weld Fitting Welding: Joint Preparation, Fit-Up, and the Root Pass

Socket welds get their reputation from the root gap. Butt welds earn theirs from geometry: a specific groove angle, sound alignment, and a deposition sequence that keeps the arc inside the joint. Get one wrong and the fitting fails quietly, then suddenly.

Groove Geometry: What ASME B16.25 Actually Calls For

End preparation follows ASME B16.25, and the numbers are specific enough to measure your own weld against them:

Wall thickness	Groove angle	Root face	Root opening
Up to 3/16 in. (4.8 mm)	None required	Not applicable	Full penetration as joined
3/16 to 5/16 in. (4.8 to 8 mm)	60 to 70 degree groove	3/32 to 1/8 in. (2.4 to 3.2 mm)	1/16 to 1/8 in. (1.5 to 3.2 mm)
5 to 22 mm (0.19 to 0.88 in.)	75 degree included, 37.5 ± 2.5 degree bevel	1.5 ± 0.8 mm (1/16 ± 1/32 in.)	1.5 to 3.2 mm (1/16 to 1/8 in.)
Over 22 mm (0.88 in.)	Compound bevel, 37.5 degrees tapering to about 10 degrees, or a J-groove	Per the WPS	Per the WPS

The 37.5 degree bevel is the number most fabricators remember. Swing the bevel a few degrees off and the arc cannot reach the root cleanly, or too much filler ends up in the groove, later showing up as lack of fusion or distortion. Past 22 mm a compound bevel or J-groove cuts deposited metal volume, residual stress, and distortion, and a fitting has far less stiffness than the pipe it connects, so more metal means more shrink and a fitting that pulls out of round.

Clean the Metal, Not Just the Joint

ASME B31.3 paragraph 328.4.1 requires surfaces to be welded to be free of paint, oil, grease, rust, mill scale, cutting slag, moisture, and other foreign matter for at least 25 mm (1 in.) back from the edge of the joint preparation. That band is where the arc passes on the way to the root, so anything left behind gets melted into the weld.

Thermal cutting is allowed: oxy-fuel, air carbon arc gouging, and plasma cutting will remove a bad end or trim a misfit. Welding over the raw cut surface is not allowed. Cut surfaces must be ground or machined back to remove the hardened layer, carburization from gouging, and oxidation scale. Air carbon arc gouging leaves a hard, dirty surface that has to come off on a lathe or grinder before any arc is struck, and the air arc welding and carbon arc gouging reference covers that side of the process in more detail.

Root Face, Backing, and Root Opening

There are two ways to handle the root: feather the edge to zero and weld against a backing ring, or grind a flat root face and weld open root with a backing bar or no backing at all. A ground root face is standard for open-root welding and gives the welder direct control of the root pass, while a feather edge against a ring is faster where the ring can stay in place permanently.

Backing rings earn their keep three ways. They act as chill rings, which is how you prevent burn-through on thin wall. They block spatter and slag from dropping into the bore, where that debris becomes an erosion or corrosion site later. And they remove the need for backing gas on materials prone to sensitization, such as stainless steel or titanium, because the ring physically shields the root. Our aluminum MIG welding guide covers the shielding and root oxidation problem that a backing ring solves for stainless in reverse.

Fit-Up: Where Good Butt Welds Are Actually Won

Fit-up determines weld quality more than any other single variable, and no procedure can rescue a joint that is out of round, offset, or tapering. Los Alamos National Laboratory’s welding manual makes the same point: weld metal must be deposited on properly prepared and aligned surfaces, and any discontinuity in the joint preparation has to be corrected before the arc is struck.

Line-up clamps come in two styles. Outside clamps sit over the pipe and pull the joint together from the top, which is fast and keeps the root open. Inside clamps reach through the bore and press against the inner wall, giving better root gap control but taking time to install. On large diameter line pipe, leave the clamp in place until most of the root bead is deposited, because the weight of the pipe can pull the joint open and start a root crack at the bottom of the joint.

Typical project limits cap internal misalignment at 0.25t where wall thickness t is under 3 mm, and 1.6 mm where the wall is thicker. Misalignment beyond that cannot be corrected by welding, because the offset concentrates stress at the joint for the life of the pipe. Consumable spacer inserts set the root gap in one motion and release on contact with the molten pool. Squareness matters too: a mitred or out-of-square end produces a V-shaped opening that welds asymmetrically, so one side ends up deeper than the other and the cap has to be ground down.

Tack, Root Pass, Back Gouge, Cap

Tack the joint lightly and evenly. Heavy tacks create their own stress concentration and can trap slag. Run the root pass hot enough to fuse the land and both sides of the groove but cool enough to control the puddle. A steady stringer beats weaving on a fitting, since weaving widens the heat-affected zone and distorts the geometry you are trying to preserve. Choosing the right consumable for the position and the material matters as well, because a vertical or overhead root pass has different melt characteristics than a flat one.

After the root is complete and the joint has cooled enough to handle, clean it. On multi-pass welds, back gouge the root on the second side to remove the oxidized, partially fused layer so the second-side root fuses properly to it. Leaving that layer in place is the single most common cause of incomplete penetration found in the field.

Finish with a cap pass that blends smoothly into the base metal without grinding into the root. Code limits cap height to roughly 1/8 in. (3 mm) on most process pipe, and over-reinforcement gets flagged because it concentrates stress and creates a crevice that traps corrosion. A slightly concave, smoothly transitioned profile is what an inspector wants to see, and undercut, overlap, or an abrupt ripple mean the cap needs dressing.

On thin wall, watch for burn-through. Too much heat on a 3/16 in. wall with a wide root gap will blow through before the puddle is controlled. Reduce the gap, use a backing ring, slow the travel speed, and keep the arc on the leading edge of the puddle. It is much easier to prevent burn-through than to weld out a hole in a fitting.

Choosing the Welding Process and Filler Metal for the Job

Choose the process first, the filler metal second. Every qualified procedure has a process at its core, and that decision drives groove angle, filler diameter, heat input, inspection method, and who may run the arc. Most pipe fitting work narrows to four processes, each with its own character.

SMAW: forgiving, portable, and slow

Stick welding earns its keep in the field. It runs on a simple machine, needs no shielding gas, and shrugs off dirty steel, light rust, and bad weather that stops a solid wire gun cold. It is the classic answer for structural repair and out-of-position fillets on a fitting stub. The costs are real: low deposition rate, heavy spatter, and slag between passes that must be chipped and ground. On thin-wall pipe and small-bore roots, a coating electrode arc is hard to control. Fabricators still reach for it anyway, and this piece explains why SMAW still wins.

GMAW and FCAW: speed, with a catch

Solid wire GMAW is the fastest way to fill a fillet weld on carbon steel, with high deposition and good out-of-position control. Its weakness is shielding: wind disturbs gas coverage, and a dirty, damp, or contaminated cylinder can put porosity into the joint. When the site will not cooperate, self-shielded flux-cored wire (FCAW) is the answer. It combines GMAW-like speed with SMAW-like tolerance, works outdoors on dirty steel, and is a staple for structural and fit-up welds. The trade is a heavy slag blanket between passes, the same logic behind all-position electrode selection.

GTAW: cleanest root, lowest output

TIG welding gives the most control and the cleanest root pass of any manual process, which is why it is the default for stainless, chrome-moly, and nickel alloy roots and tubing. The metal must be genuinely clean, and narrow bores need an inert backing gas on the root side or the root oxidizes and tints. The price is deposition rate: for a shop welding hundreds of feet of carbon steel, GTAW is too slow.

SAW: thick wall, large diameter

Submerged arc welding puts the highest deposition rate of any process into thick-wall, large-diameter work. The arc runs under granular flux that also shields the puddle, so no gas cylinder is needed. It wants flat position, horizontal-vertical, and room for flux handling and recovery. A socket weld on a 1/2 inch instrument line is the wrong place for it; a 24 inch transmission pipeline joint is exactly right.

For instrumentation and small-bore tubing, add orbital tube welding. A machine clamps around a fixed tube and drives the torch around it, producing a repeatable, uniformly fed root with less operator skill and heat variation than a hand pass.

Process	Best at	Watch out for
SMAW	Field work, dirty steel, structural fillets	Slow, spattery, slag between passes
GMAW	Fast fillet welds on carbon steel	Wind and gas quality hurt shielding
FCAW	Outdoors, rusty steel, fit-up welds	Heavy slag
GTAW	Stainless and alloy roots, tubing	Low deposition; needs clean metal, backing gas
SAW	Thick wall, large diameter, flat position	Flux handling, position limits

Reading the ASME IX filler metal system

ASME Section IX organizes qualification around three numbering systems, all three visible on the WPS and the filler metal certs. P-numbers group base metals by weldability, so a procedure qualified on one P-number metal covers others in the same group. F-numbers describe filler metal deposit strength for fillet welds and set the transition rules. A-numbers cover backing and fillet weld strength categories in the qualification tables. The practical consequence: changing F-number is an essential variable, so moving from F3 to F4 wire generally forces requalification before production work.

Heat input, preheat, and interpass

Heat input comes from voltage, current, arc time, and travel speed. It is the most useful lever for controlling burn-through on thin-wall fittings and distortion on large-bore joints. In simple form, HI = (V x A x 60) / travel speed in inches per minute, in kJ/in, with an efficiency factor for stick and SAW. High heat input on a 1/8 inch wall elbow cuts through it and bows it out of round. Low heat input on a schedule 80 tee leaves you chasing incomplete fusion and slag pockets.

Preheat slows cooling to limit hydrogen cracking, and interpass temperature caps the heat the HAZ sees on later passes. Welding fume and gas hazards are a separate issue, but the same discipline applies. Interpass must stay at or below the maximum in the qualified WPS. Exceeding it causes HAZ softening in carbon steel and sensitization and distortion in stainless steel.

Material Matters: Carbon Steel, Stainless Steel, Alloy, Duplex, and Aluminum

Most pipe fitting weld failures come down to material, not welder skill. Every alloy family rewrites the procedure: preheat, filler, interpass temperature, shielding, and purging. A procedure that works on A106 carbon steel will destroy corrosion resistance on a 316L elbow or embrittle a duplex joint.

Base Material	Typical Filler	Primary Risk	Key Control
Carbon steel A106 / A53	ER70S-6, E7018	HAZ hardening, hydrogen cracking	Preheat by CE, low-hydrogen filler
304 / 316L stainless	ER308L, ER316L	Sensitization, loss of passivation	Interpass control, L-grade filler, root purge
Duplex 2205	ER2209	Nitrogen loss, sigma phase	Low heat input, interpass max 150 °C
Chrome-moly (P11, P22), nickel alloys	Low-alloy electrodes, nickel alloy filler	Cold cracking, restraint stress	Mandatory preheat, controlled PWHT
Aluminum 6061 / 6063	ER4043, ER5356	Porosity, oxide contamination	AC GTAW, prep to bright metal, preheat

Carbon Steel: Forgiving Until It Isn’t

ASTM A106 Grade B and A53 pipe are the most forgiving materials in the shop: mild carbon content, wide preheat tolerance, endless consumables. The trap is that forgiving materials fail quietly when a parameter is wrong.

Watch the carbon equivalent (IIW: CE = C + Mn/6 + (Cr + Mo + V)/5 + (Ni + Cu)/15). Past roughly 0.35 to 0.40 the heat affected zone hardens, and hard HAZ plus hydrogen equals cold cracking. Control preheat by CE and wall thickness, use low-hydrogen consumables (E7018 or E71T-1, not E6013), and cool slowly.

Stainless Steel: The Heat Affected Zone Is the Weak Link

Stainless is not harder to weld; it is more sensitive to thermal treatment. Between roughly 450 and 815 °C (850 to 1,500 °F), chromium precipitates as carbides along grain boundaries in the heat affected zone, and sensitization strips corrosion resistance next to the weld.

Three defenses. Use L-grade fillers, ER308L for 304 and ER316L for 316, with higher chromium and nickel and low carbon to suppress carbides. Hold interpass at or below about 175 °C (350 °F), and preheat only for a specific reason. Use dedicated stainless brushes and grinding tools, since carbon steel contamination leaves iron particles that rust and pit.

Duplex Stainless: A Narrow Window That Punishes Heat

Duplex 2205 is the least forgiving material, balancing roughly equal austenite and ferrite. That balance is fragile: too much heat drives nitrogen out, and nitrogen loss plus sigma phase precipitation destroy the corrosion resistance the grade was bought for.

Practical rules: ER2209 filler, interpass at or below about 150 °C (300 °F), stringer beads instead of wide weave, no crater fills, and let the joint cool between passes. A material-specific welding procedure for high-reliability plants shows how tightly duplex parameters must be specified.

Alloy Steels: When Preheat and PWHT Stop Being Optional

Chrome-moly grades like P11 (1.25Cr-0.5Mo) and P22 (2.25Cr-1Mo) in power boilers and piping carry high hardenability and restraint, so preheat becomes mandatory and tightly specified and PWHT is often code-required. Consumable choice matters too, so alloy steel welding electrode selection deserves a look before the job starts. Nickel alloys add high thermal expansion, slower conduction, and filler matching or exceeding base metal nickel content.

Aluminum: Heat Runs Away and Oxide Never Stops Growing

Aluminum conducts heat several times faster than carbon steel, so heat leaves the puddle into plate and tube: the pool shrinks, the root fails to fuse, and porosity appears. Expect higher amperage, controlled travel speed, and preheat to clear moisture and hydrogen.

The oxide layer is the other fight: tenacious, insulating, and it reforms instantly in air. Clean to bright metal with a dedicated aluminum cleaner and aluminum-only abrasives (see how to prep aluminum for welding). For GTAW use AC, whose cathodic cleaning breaks up the oxide while DC does not. Filler must be aluminum specific, typically ER4043 or ER5356, and kept dry. Our aluminum MIG welding and aluminum flux core welding guides cover the process details.

Purging, Backing Rings, and Consumable Inserts

Nitrogen purge is standard for stainless and high-alloy root passes and must happen before the arc is struck, because the root side is inaccessible afterward. Purge dams hold the gas column in the bore, and oxygen analyzers confirm the purge.

Where purge is impractical, backing rings or consumable spacer inserts take over. Inserts come in five classes and join the root bead, giving consistent root opening without removal after tacking, which removes burnthrough and trapped slag risks. Backing rings left in place cost some corrosion performance on stainless.

PWHT relieves stress in heavy-wall alloy and chrome-moly piping and needs deliberate ramp rates and soak times. Held too long in the wrong range, it can precipitate chromium carbides in austenitic stainless and sigma phase in duplex, so there it is avoided or specifically qualified.

Codes, Welding Procedures, and Welder Qualification

Every pipe fitting weld sits inside a regulatory frame, whether the job is a chemical skid in a plant, a shipyard hydrotest line, or a warehouse handrail. Three documents do the work: the governing code, the Welding Procedure Specification (WPS), and the welder performance qualification that proves you may use it.

ASME Section IX: The Engine Behind Nearly Every ASME Code

When a spec says “weld to ASME B31.3,” the material, design, and fabrication follow B31.3, while the welding and the welder’s qualification follow ASME BPVC Section IX. Section IX defines what a WPS must contain, what a Procedure Qualification Record (PQR) must prove, and how a welder earns performance qualification.

• The WPS is the instruction. Base metal, filler metal, joint design, process, preheat and interpass temperatures, amperage, voltage, travel speed, shielding gas, cleaning, and inspection method.
• The PQR is the record. The qualification test behind the WPS: parameters used, coupon dimensions, test results, and the qualified ranges.

The PQR is the evidence, the WPS is the conclusion, and an inspector asking for your qualification on a 3 inch schedule 40 elbow joint wants both.

Choosing the Right Code for the Service

• ASME B31.1: power piping, boiler plant, steam, and high-temperature water systems.
• ASME B31.3: process piping for chemical, petrochemical, pharmaceutical, food, and general industrial systems, and the code most often met in the field.
• ASME B31.4: pipeline transportation systems for liquids, above and below ground.
• API 1104: welding of pipelines and related facilities, the oil and gas pipeline standard.
• AWS D1.1/D1.1M: structural welding, including tubular structures and pipe used structurally, with its own qualification rules.

On a B31.3 job the code dictates details that show up on the fitting. Joint surfaces must be cleaned of paint, oil, grease, rust, mill scale, cutting slag, and moisture for at least 25 mm (1 in.) back from the joint preparation edge, with alignment per the WPS. Undercut is limited to 1/32 in. and may not encroach on minimum required section thickness, and excess root penetration to the lesser of 1/8 in. or 25 percent of the nominal wall thickness of the thinner component, for wall down to 1/4 in.

Welder Performance Qualification: Why 6G Matters Most

Qualifying a welder on pipe is not a formality. The coupon is a pipe section welded in a specific position, and the exam is built so the hardest-to-weld regions get cut up and bend-tested.

Positions run 1G through 6G, and 6G opens every door. A 6G coupon is welded vertically with the pipe inclined so the test covers the transition from overhead through vertical to flat. Passing 6G qualifies you for all pipe positions and all positions on the same joint.

Rigour comes from the sectioning. The pipe is cut at specific clock positions so machined specimens land on the toughest zones: the overhead-to-flat transition near 12 o’clock, where the weld changes orientation under the arc, and the near-overhead region near 6 o’clock, where gravity pulls the pool from the root.

Bend Testing: The Most Sensitive Test in the Whole System

For AWS D1.1 qualification of a CJP groove weld on pipe, Table 4.2 sets the specimen requirements: two root bend and two face bend specimens for pipe between 1/8 in. and 3/8 in. wall thickness, additional side bends for thicker material, and diameter limits tied to coupon size. See the published AWS D1.1 Table 4.2.

Acceptance is tight. The specimen is bent 180 degrees. On the convex surface, no single discontinuity may exceed 1/8 in. in any direction. Discontinuities greater than 1/32 in. but not over 1/8 in. are summed, and that total may not exceed 3/8 in. A corner crack is limited to 1/4 in., unless a slag inclusion is evident in the specimen.

Why the fuss? Bending is the most sensitive indicator of ductility problems in the weld metal and heat affected zone. A bend test exposes hydrogen-assisted cracking that stayed microscopic in service, solidification cracking, slag inclusions, porosity, and wrong filler metal selection.

The Plate Qualification Trap

Here is a mistake that shows up on small jobs and costs owners real money: using a plate-based procedure qualification to support a pipe welding procedure. A 3/8 in. plate coupon welded flat proves nothing about a 6 in. schedule 40 elbow with higher restraint and slower travel.

Many owner and project specifications prohibit it outright. If the weld then fails a hydrotest or an RT, the defense of “but we had a plate PQR” disappears. Qualify on pipe.

For context on how these structures get written into real fabrication packages, the Los Alamos National Laboratory welding fabrication specification shows the detail expected in a WPS and its supporting records. For structural tubular or handrail work, take the AWS D1.1 route.

One last practical point for anyone choosing filler metal for all-position work: the consumable must be qualified inside the same procedure, and a weld that passes with E7018 in a shop is not automatically qualified with a different classification. Keep the PQR, WPS, and consumable aligned.

The Most Common Pipe Fitting Welding Mistakes and How to Fix Them

Most pipe fitting welding defects never announce themselves in service. They get flagged months later, during a turn-around inspection.

Lack of fusion: the defect that hides best

Lack of fusion is a bond failure between weld and base metal. It shows as a dark seam along a fillet weld toe, a dull or gritty root pass, or a groove the arc ran past. On butt welds it hides under an acceptable cap pass, which is why ASME B31.3 requires examining both the cap and root side of the joint.

• Cold joint. Arc struck on solidified weld metal. Grind back and re-weld.
• Wrong groove angle or root gap. Check the bevel against ASME B16.25 first.
• Travel speed too fast. The puddle runs ahead of the arc. Slow down and weave.
• Mill scale, rust, paint, moisture. Clean 1 in. back from the prep edge.
• Not enough heat. Raise preheat or amperage and slow the travel.

Porosity is a shielding problem before it is a wire problem

Porosity shows up as pinholes or worm-like voids, and in a pressure joint it is a leak path. The cause is usually gas or cleanliness, not filler metal: contaminated shielding gas, a wet cylinder, rust or oil on the joint, wind stripping the shielding envelope off the puddle, or dirty wire and rod.

Fixes are unglamorous: check gas flow with a flowmeter at the nozzle, check cylinder moisture, keep filler metal dry in a rod oven, clean to bright metal, and use a windscreen outdoors. Still chasing holes? Check the electrode and filler selection guides first.

Undercut and excess reinforcement

Undercut is a groove melted into the base metal at the weld toe, and a stress raiser. ASME B31.3 limits undercut to 1/32 in. and bars encroachment on the minimum required section thickness. Causes: excessive travel speed or amperage, the wrong electrode drag angle, or too long an arc. Fix with stringer technique, a 15 to 20 degree work angle, and less heat input.

Excess reinforcement has its own B31.3 limits:

Base material thickness	Maximum reinforcement
1/4 in. and under	1/16 in.
Over 1/4 in. to 1/2 in.	1/8 in.
Over 1/2 in. to 1 in.	5/32 in.
Over 1 in.	3/16 in.

Grind a proud crown down before it reaches the inspector.

Burn-through

Burn-through happens when too much heat hits thin wall or a wide root gap. The puddle sags and drops through, leaving a bead on the underside. Correctives: a backing bar or ring, a consumable insert, reducing the root gap to specification, backing gas, a chill ring, and lower amperage with a choppier weave. For equipment, see our guide to air arc welding and carbon arc gouging.

Root suck-back and incomplete penetration

Root suck-back is the concave root profile on open-root welds, driven by three competing forces: surface tension pulling the molten metal inward, gravity pulling the puddle down, and arc force blowing the pool away from the root. A smaller, lower-amperage root pass reduces the profile, so keep the root bead thin.

Incomplete penetration behind the root means the first pass was too fat or too fast, the arc strayed off the root, or the bevel was too narrow to reach. Gouge back, re-prep the groove, and re-weld. Skipping the gouge is why repaired joints fail again.

Cracking

Hydrogen-assisted cracking appears hours or days after welding, in high-strength or restrained joints, as root or toe cracks. Drivers are moisture on consumables or base metal, wrong preheat, excessive restraint, and residual stress. Fixes include low-hydrogen consumables, correct preheat and interpass temperatures, fewer passes with smaller beads, preheat for stress relief, and controlled slow cooling. The alloy steel welding electrode guide covers hydrogen classification.

Distortion, misalignment and stray strikes

Angular misalignment and distortion come from excessive heat input, uneven weld sequencing, poor tacking, or a forced fit-up. Correct with sequenced welding, back-stepping, strong fixturing, and balanced welding of opposing fillets on socket joints. Arc strikes outside the weld area leave hard, untempered heat affected zone metal and are prohibited in ASME and AWS acceptance criteria. Ground them out and re-inspect.

Why looking good is not the same as being sound

A weldment that passes visual inspection can still fail a bend test, so the codes require destructive testing at qualification and nondestructive testing in the field. Bend testing is the most sensitive indicator of weld ductility problems: hydrogen assisted cracking, solidification cracking, slag inclusions, porosity and wrong filler metal selection.

Very thick joints get special treatment: a 2 in. thick electroslag butt weld cannot practically be face or root bend tested, so side bend testing is used instead. Pipe specimen requirements sit in the AWS D1.1 Table 4.2 criteria, and similar logic applies under ASME Section IX. Most of these defects are cheap to prevent at fit-up and expensive to fix in service, so treat the pre-weld check as the real quality gate.

Inspection and Testing: Proving the Joint Is Sound

A weld can look like a weld and still hide a root crack. Inspection on pipe fitting welds is layered rather than optional: each layer catches a different class of defect. Visual and surface methods run first because they are cheap. Volumetric methods follow, because they alone can see inside the weld. Pressure testing comes last, and it proves the narrowest thing of all.

Visual Inspection: What the Inspector Actually Looks For

Visual inspection (VT) is the baseline examination on nearly every weld, and far more than a glance. The inspector works against written acceptance criteria, typically ASME B31.3, and checks the finished weld for a specific list of conditions. VT is also the only examination that covers the whole weld, including areas a surface method may not reach cleanly.

• Undercut — a groove melted into the base metal at the weld toe, removing load-carrying material.
• Overlap — weld metal lapping over base metal without fusing; always a defect.
• Excess reinforcement — weld metal proud of the surface beyond the permitted height, often masking poor fit-up.
• Concavity — a root surface dipping below the pipe or socket bore. On a single-welded joint it may not reduce total joint thickness, including reinforcement, below the nominal thickness of the thinner component; beyond that, insufficient effective throat.
• Cracks — any crack found visually is rejectable; there is no acceptable crack length.
• Arc strikes — craters left outside the weld; usually grounds for rejection on high-alloy material, being decarburized, oxidized zones.
• Surface porosity — gas bubbles broken at the surface, visible as pits or blowholes.
• Incomplete fusion at the toe — the bead failed to bond to the side of the joint, often a dark line beside the weld.
• Transition smoothness — an abrupt change of direction along the weld is a stress raiser and a reject.

Surface Methods: MT and PT

Magnetic particle testing (MT) applies to ferromagnetic materials, in practice carbon steel and low-alloy steel. The part is magnetized and powder collects where flux leaks out of a discontinuity, making surface and near-surface cracks visible. MT is fast, inexpensive, and handles complex geometry such as elbows and branch connections well. It says nothing about the weld interior.

Liquid penetrant testing (PT) is the equivalent for stainless steel, nickel alloys, and other non-ferromagnetic material. Penetrant soaks into an open flaw, excess is cleaned off, and a developer draws it back out, leaving a bright indication of the crack’s size and shape. PT finds surface-breaking discontinuities in any material on a clean, non-porous, non-coated surface.

Together, MT and PT are the standard NDE for socket welds, where overlapping pipe and socket walls make radiography and ultrasonic testing meaningless: a socket weld gets VT plus MT or PT and nothing else.

Volumetric Methods: Radiography and Ultrasonic Testing

Radiography (RT) puts a radiation source on one side of the joint and a film or digital detector on the other. Sound weld metal shows consistent density, while porosity, slag, cracks, and lack of fusion show as dark indications. The film or digital equivalent must cover at least 1.5 times the wall thickness so the source can penetrate the joint, and RT leaves a permanent record. The costs are real: ionizing radiation, a qualified radiographer, controlled access, and an examiner reading a two-dimensional shadow of a three-dimensional joint.

Ultrasonic testing (UT) sends a sound wave into the joint with a transducer and displays reflected echoes on a screen. There is no radiation and no film handling, and UT is considerably faster than radiography, a real advantage on long runs of butt welds. It can walk a full 360 degrees around a joint in a single pass and directly measures flaw depth. Interpretation depends on technician skill, and the technique needs a qualified reference block with representative reflectors to calibrate the instrument and prove detection capability. Without that block and a qualified operator, a UT report is close to worthless.

Pressure Testing

Hydrostatic testing uses water at controlled pressure; pneumatic testing uses air or nitrogen. For B31.3 the hydrostatic test pressure is typically 1.5 times the piping design pressure, a deliberately low figure. Hydrostatic is preferred wherever practical because water is far less compressible than gas, so a small leak shows as a measurable drop on the gauge. Pneumatic testing carries a stored-energy risk and is usually limited to systems that cannot be filled with water.

Be clear about what it proves: leak-tightness, not weld strength, and not that the weld survives design loading. A weld with serious subsurface porosity or a deep crack can pass a 1.5 times design hydrostatic test and still fail months later at design pressure, because the crack has not propagated far enough yet. Pressure testing is never a substitute for volumetric examination.

Progressive Sampling: The Scheme That Protects Both Sides

Requiring radiography on 100% of welds in a large fabrication is often unrealistic on cost and schedule, so codes allow progressive sampling. Under the ASME B31.3 radiographic examination requirements, examination starts by selecting 5% of the welds produced by a given welder within a defined lot. Many codes also allow progressive sampling to be applied to required visual examination.

1. Examine that 5% of the lot radiographically.
2. If all pass, accept the lot for that welder.
3. If any rejectable defect appears, pull and examine two more welds from the same lot and welder.
4. If those two pass, accept the lot with the defective weld repaired.
5. If either fails, every weld in that lot from that welder escalates to 100% radiography.

One bad weld is often an isolated defect, and two more passing welds support that conclusion. A failure in the additional pair means the welder’s process is not in control, so everything that welder made in the lot has to be checked. That protects the owner from a bad lot and the fabricator from a blanket rejection over one weld.

Documentation, Repair, and Rework

Every welded joint should be traceable, which on paper means three things: a weld map showing joint number and location, a welder identification stamp on or near the joint, and a WPS reference identifying the qualified procedure used. Without them, a failed joint becomes an investigation with no starting point.

A confirmed defect follows a fixed sequence: excavate the defective metal out to sound parent metal; re-prepare the joint to original geometry, a fresh bevel per B16.25 for a butt weld or a prepared and re-cleaned socket for a socket weld; re-weld using a qualified procedure, remembering that repeated repairs carry code limits and a joint repaired beyond them is rejected outright; then re-test with the method that found the defect and document the result. Repair welds are welds: they need their own welder qualification coverage, WPS, and inspection record.

Why This Matters More Than It Sounds

Most field failures in piping are found by inspection before service, not in service. A flaw caught in the shop costs a grinder, some filler metal, and an hour. The same flaw found during a thermal cycle in service costs an unplanned shutdown, a night shift, and often the credibility of the whole fabrication package. Each layer pulls its own weight, and no single layer is sufficient.

Safety, Practical Habits, and How to Build a Fitting Weld That Lasts

A weld that passes inspection and a shift that hurts nobody come down to the choices made before the arc is struck. Ventilation, a clean joint, a qualified procedure, and honest identification of your work are not paperwork. They are the craft.

What Welding Fume Actually Does to You

Welding fume is metal oxide particles, gases, and vapors that depend on what you are welding. OSHA’s welding hazards fact sheet spells out the pattern. Short exposures irritate the eyes, nose, and throat and cause coughing, chest tightness, nausea, and headaches. Metal fume fever appears as flu-like symptoms with fever, chills, and body aches, usually within hours and clearing within a day or two.

Chronic effects are why regulators take this seriously. Prolonged fume exposure can damage the lungs and raise the risk of lung, larynx, and urinary tract cancers. Manganese, in nearly all welding fume including carbon steel fume, is linked to manganism, a Parkinson-like condition with tremor, rigidity, and gait disturbance. The OSHA hazards and solutions page treats these as long-term disease.

Hexavalent Chromium: The Stainless Steel Problem

Stainless and other chromium-bearing alloys can generate hexavalent chromium, a confirmed carcinogen. Under OSHA’s exposure control standard, the permissible exposure limit for Cr(VI) is 5 µg/m³ as an 8-hour time-weighted average and the action level is 2.5 µg/m³. Where initial monitoring hits the action level, the employer must institute exposure monitoring at least every six months, plus medical surveillance where exposures reach the PEL.

Most shops get this wrong: OSHA requires feasible engineering and work practice controls as the primary means of reduction, with respiratory protection supplementing them, not replacing them. A respirator is the last layer, not the first. Stainless surfaces must be free of paint, oil, and solvent residue before welding, because coatings and solvents generate additional toxic fume. The NIOSH hexavalent chromium resources cover control options.

Ventilation, Gases, and Confined Spaces

Local exhaust ventilation with hoods and extractors near the plume works. Welding outdoors helps, but still air, wrong wind direction, or an extractor on the far side of the fitting will not capture fume. Inside a vessel or between pipe runs, welding becomes confined space work. Helium, argon, and carbon dioxide displace oxygen and can cause asphyxiation, and carbon monoxide can form in shielded welding. Cylinder handling deserves the same attention as wire feed; the AllGas welding supply guide to gases, cylinders, and safety covers that.

Personal Protective Equipment

Item	Why it matters
Helmet with correct shade lens	Auto-darkening hoods suit short bursts; check the lens rating
Respirator	Where ventilation cannot hold exposure below limits; galvanizing or stainless fume
Fire-resistant clothing and gloves	Sparks and spatter run downhill on vertical joints
Hearing protection	Plasma cutting and gouging damage hearing
Chloride-free, sulfide-free cleaner	Prevents toxic fume and corrosion on stainless and alloy piping
Stainless-dedicated wire brushes	Carbon steel brush contaminates stainless and causes rust streaking

Consumable handling matters too, and alloy steel electrode selection and safety covers the storage and low-hydrogen handling rules that keep fume down and cracks out.

The Pre-Weld Checklist

1. Confirm the material and schedule match the line class on the isometric.
2. Confirm the end condition: butt-weld, socket-weld, or threaded.
3. Set the 1/16 inch withdrawal gap on a socket joint, uniform around the circumference.
4. Inspect the bevel against ASME B16.25: angle, root face, root gap, clean ground surface, no dross.
5. Clean the joint and pipe end to bright metal.
6. Confirm the WPS covers this joint, position, thickness, and material.

After the Arc

• Clean and dress the weld, remove slag and spatter, check toe and cap.
• Stamp or mark your welder identification where code requires.
• Mark the joint for the weld map so inspectors can find it.
• Support the joint so weight does not sit on a hot weld, and protect it from impact.
• Never weld over a threaded fitting without a qualified procedure.

Match the joint to the service, respect the 1/16 inch socket gap, prep the bevel to B16.25, clean to bright metal, weld to a qualified procedure, and prove it. Fabricators earn their value in the unglamorous margins: the gap nobody measured, the root properly gouged, the stamp that makes traceability possible, the refusal to weld a joint that was never fit up right. Years later, that is the difference between a fitting in service and a line brought down.

Frequently Asked Questions

What is the difference between socket weld and butt weld fittings?

A butt weld fitting has ends bevelled to match the pipe wall, so the joint is a full-penetration groove weld through the entire wall with a smooth, continuous bore. A socket weld fitting has a machined recess; the pipe slides in and is closed with a fillet weld around the outside, leaving an internal crevice at the socket bottom. Butt welds are stronger and can be radiographed; socket welds are faster to fit up but are limited to smaller sizes and surface NDE only.

Why does ASME require a gap between the pipe and the socket before welding?

The pipe expands longitudinally faster and further than the heavy fitting body when heated. If the pipe is welded hard-bottomed with no gap, that expansion drives the pipe against the socket shoulder and forces the fillet weld into severe bending and tensile shear, which causes root cracking, toe cracking, or rupture. ASME B31.3 therefore requires the pipe to be withdrawn about 1/16 inch (1.5 mm) before welding, even though the gap does not need to be verified afterward.

What size fillet weld is required on a socket weld fitting?

The fillet leg dimension Cx must be at least 1.09 times the nominal pipe wall thickness (or the socket wall thickness, whichever is smaller). This produces a theoretical throat of 0.707 x Cx, which must be at least 0.77 times the nominal pipe wall thickness. Slight undersizing is permitted under B31.3 within defined limits.

Can you radiograph a socket weld joint?

Not in any meaningful way. The pipe wall inside the socket and the fitting wall around it overlap along the beam path, so the resulting film superimposes two thicknesses and cannot distinguish a sound root from lack of fusion. That is why socket weld joints are normally inspected by visual inspection plus magnetic particle or liquid penetrant testing, and why specifications requiring radiography effectively dictate butt weld joints.

Which welding code applies to pipe fitting welding?

It depends on the service. ASME B31.3 governs process piping, B31.1 governs power piping, B31.4 and API 1104 govern pipeline systems, and AWS D1.1 governs structural and tubular work. All of them defer to ASME Section IX for procedure and welder qualification. The governing code and the line class specification together decide the joint type, filler metal, inspection extent, and acceptance criteria.

What is the groove angle for a standard butt weld joint?

ASME B16.25 specifies 37.5 degrees ± 2.5 degrees on each side, a 75 degree included groove, for wall thickness from 5 mm to 22 mm, with a root face of 1.5 mm ± 0.8 mm and a root gap of 1.5 mm to 3.2 mm. Above 22 mm, compound bevels or J-grooves are typically used to limit weld metal volume and distortion.

Why is my socket weld cracking right at the ends of the fillet?

Toe cracking at the ends of a socket weld is the classic signature of a missing withdrawal gap. It is also worth checking preheat, hydrogen level, and joint restraint. Another common cause is a fitting whose wall thickness does not match the pipe schedule, so the heat input suited to one will overheat the other.

Does a groove weld qualification also qualify me to weld fillet welds?

Under AWS D1.1, a welder qualified on groove welds is also qualified for fillet welds in the positions indicated by the qualification table, but the reverse is not true. The same general principle applies under ASME Section IX. Note that many owner specifications and project standards will not accept plate-based qualifications to support pipe procedures, regardless of what the base code permits.

What causes porosity in a pipe fitting weld?

Almost always contamination or inadequate shielding: oil, rust, moisture, or cutting slag on the joint, a damp shielding gas cylinder, wire that has absorbed moisture, or wind disrupting the shielding gas around a GTAW or GMAW root pass. On stainless steel, back-purging with nitrogen is needed to prevent oxidation at the root, and the purge must be established before welding because the root is not accessible afterward.

Is post-weld heat treatment always required on pipe fittings?

No. It is typically required for heavy-wall chrome-moly and low-alloy piping in power and high-temperature service, and it is sometimes specified for stress relief in high-pressure process lines. For stainless and duplex stainless, PWHT must be controlled carefully because long holds in the wrong temperature range can sensitize the material and destroy corrosion resistance. The line class specification and the governing code decide.