Pipe Flange Welding: Process, Fit-Up, and Inspection

The flange goes on, the bolts go through, and it looks finished—then the hydrostatic test fails, and the radiograph shows a root flaw nobody saw. That moment is where most flange weld stories begin, and it usually traces back to a decision made hours earlier: a bevel angle that was rushed, a root gap that was guessed instead of measured, or a procedure that was never really qualified for the job. Pipe flange welding looks simple because the joint is only a few inches wide, but the pressure boundary it creates is unforgiving.

Pipe Flange Welding: What It Really Is and Why It Fails

So what is pipe flange welding? At its simplest, it is joining a pipe end (or a vessel nozzle) to a flange ring so the two become one continuous pressure boundary. The pipe end is prepared with a groove, the flange ring slides over it, and the gap is filled with weld metal that becomes the new wall of the pipe. It looks like a job you could clamp with a wrench. It is not.

One Joint, Two Load Paths

A flange joint is uniquely unforgiving because it does two completely different jobs at once, and the weld has to be good enough for both.

• The weld metal carries the pressure and thermal load. It is a pressure-boundary element, like the pipe wall but thinner, and it may see cyclic service, vibration, or thermal cycling on top of that.
• The bolt and gasket system carries the clamped joint force. That load keeps two faces pressed together and the fluid out. It reacts through the flange ring, bolts, and gasket, and feeds back into the flange face.

Those load paths meet inside the flange. Pressure load ends where structural load begins, and the weld metal sits at that transition. A weld adequate for sealing can still be a poor structural fit, and a joint strong in the bolt circle can still leak through the root.

The Geometry That Punishes You

A few terms define every flange weld, and each changes the result:

• Root — where the weld metal meets the joint face, the fusion boundary.
• Bevel — the angled preparation on the pipe end that creates the groove to fill.
• Land or root face — the unmachined flat at the bottom of the bevel, giving the root pass something to sit on.
• Root gap — the distance between the pieces before welding. Too little and the root burns through; too much and you get incomplete penetration.
• Reinforcement — the extra weld metal standing proud of the joint. It helps seal but can also hide defects.

Here is what matters most: stress concentration in a welded flange joint peaks at the root of the groove and at the toe of the fillet, where the section changes abruptly. A crack or lack of fusion at the root does more damage than the same size of porosity at the cap. The worst-looking defect is often not the dangerous one, and the defect nobody noticed at the bottom of the groove usually is.

When the Flange Face Lies

A flange that is mismatched, out-of-square, or gapped after bolting up does not just make an ugly weld. It forces uneven preload on the bolting, which loads the weld metal eccentrically. That is a common root cause of flange leaks no amount of rewelding will fix, because the problem is the joint geometry, not the weld. Getting the flange square and concentric before you strike an arc is the cheapest quality control in the job.

The Four Things That Decide the Outcome

1. Material and geometry. Match the pipe and flange specification, check wall thickness against the minimum, confirm groove dimensions match the procedure.
2. Procedure qualification. A qualified welding procedure specification (WPS) covering the process, filler metal, thickness range, and position you are in.
3. Fit-up. Root gap set by measurement rather than by feel, root face maintained, flange square and concentric, clamping that does not distort the joint.
4. Inspection. Visual first, then whatever nondestructive examination the governing code and owner specification require.

That last point has teeth. Flange joint strength in ASME B31.1 and ASME B31.3 is tied to the joint efficiency assigned to the weld, so weld quality directly changes the allowable stress of the pipe run feeding into it. A pipe designed for a certain pressure and temperature can be illegal to operate simply because the weld in the flange missed its qualified joint efficiency.

This article covers the industrial and process side of flange welding, the kind that ends up in a hydrostatic test or in service under pressure, not decorative work. The governing standard must be checked every time: ASME B16.5 covers pipe flanges and flanged fittings up to NPS 24, large-diameter flanges fall under ASME B16.47, and compact flanges under ASME B16.36. Pipeline work runs under API 1104 and API 5L. Always confirm the edition in force and the one cited on the drawings.

Getting the process right matters as much as the geometry, so next we look at flange families, how to read a flange stamp, and the process choices that fit a flange joint. If you want a refresher on TIG welding fundamentals, or want to think through welding positions and electrode selection for out-of-position work, those primers will help.

Flange Types, Ratings, and the Joint You Are Actually Welding

A flange drawing looks cryptic: a size, a class, a material spec, and a couple of letters naming the type. Read those four items and you know how many welds the joint needs, where the heat lands, and what the inspector will look for. That vocabulary comes before any grinder touches steel.

The flange families you will actually meet

Weld neck (WN) is the reference standard for industrial piping. Its tapered neck is butt welded to the pipe end with a full penetration groove weld, so the weld sits in line with the pipe wall rather than out on the OD. That geometry gives the strongest, most leak-resistant connection and the highest allowable stress rating of the common types. It is the friendliest case for a welder: one clean, symmetrical groove weld that can be inspected visually and radiographed with confidence.

Flange type	How it joins	Weld and heat location	What the welder needs to know
Weld neck (WN)	Butt welded to the pipe end through a tapered neck	One full penetration groove weld, heat in line with the pipe wall	Strongest, most leak-resistant joint and the highest allowable stress rating of the common types
Slip-on (SO)	Slides over the pipe end	Single fillet at the pipe OD, heat on the thin wall	Lower rating than a weld neck of the same size and class, so it fits lower pressure, non-critical service; fillets read poorly on film and the flange-to-pipe crevice is a corrosion and leak path
Socket weld (SW)	Pipe slides into a machined socket	Fillet at the socket OD, low heat input	Compact for small bore, high pressure work in instrument and utility piping; fillet rather than groove, so inspection leans on visual and liquid penetrant
Threaded	Screwed onto threaded pipe	No weld in the joint	Utility steam, water and drain lines where disassembly matters; your concern shifts to thread condition, sealant and thread corrosion
Lap joint	Stub end welded, threaded flange bolted on	Full penetration butt weld on the pipe	Two connections and only one of them welded; radiography common
Blind	Bolted to a mating flange	Usually none, unless a weld neck blind	Caps a line or vessel nozzle with no flow path; check face flatness and bolts
Plate	Set over the pipe end	Fillet around the pipe OD	Flat disc, often field-fabricated for temporary or low-pressure service

Reinforcing pads are thick plates welded around a nozzle or branch opening to spread the load into the parent metal instead of letting it concentrate at the weld. Treat a pad as a separate job: it adds heat into an already restrained area, and its perimeter weld is a second potential leak path needing its own inspection.

Raised face, flat face, and the ring joint groove

Most ASME B16.5 flanges are raised face (RF), meaning a machined ring stands slightly proud so a gasket can be compressed evenly. Flat face (FF) flanges have no raised sealing surface and are used for weld-seal joints or where bolting is structural only. Check the drawing carefully, because a face detail changes how you fit, torque, and inspect the joint.

Ring type joint (RTJ) flanges have a machined groove that seals with a solid metal ring rather than a gasket. The weld is therefore not the sealing element, and the groove must be protected from spatter, gouging, and weld overlap. Weld the flange before final machining or ring installation, keep the torch and grinder off the groove, and treat the groove face as a machined surface.

Pressure class and pressure-temperature ratings

ASME B16.5 covers pressure classes from 150 through 2500, and the number on the flange is a class, not a pressure. The rating also falls as temperature rises because allowable stress drops. A Class 150 flange is commonly rated near 265 psig at ambient, so in a hot service line it is nowhere near a 150 psig joint at operating temperature. Pull the pressure-temperature table for the governing material and the edition in force.

Class 150 and Class 300 are temperature and pressure rated, and the rating derates with heat. Class 600 and higher are rated primarily by allowable stress at room temperature, so they hold up far better when hot, but they carry limited gasket seating capability and are not intended for ordinary raised face gasket service. That is why high-pressure, high-temperature service runs on RTJ grooves.

Size also changes the governing standard. ASME B16.47 covers large diameter pipe flanges from NPS 26 and above, ASME B16.36 covers compact high-pressure flanges, and API 6A governs wellhead and Christmas tree equipment flanges at 20 inches and larger.

Materials and what they mean for the arc

Forged carbon steel flanges commonly reference ASTM A105/A105M, low-temperature service commonly references ASTM A350 LF2, and stainless flanges commonly reference ASTM F304/F316. Stainless changes the whole job: clean, unpainted, oxide-free metal, dedicated grinding media, shielding and backing gas with very low oxygen, and a purge. The wrong gas or a contaminated surface turns a stainless flange into a rust-streaked repair job.

Choosing a Welding Process for Flange Joints

There is no single best process for a flange joint, and anyone who tells you otherwise is selling consumables. The right pick depends on base metal, service, access, and inspection. But one pattern holds on nearly every flange job: the root pass deserves a different process than the fill and cap passes. The root is where defects concentrate, where the pressure boundary forms, and where most radiographic rejections show up. So the question is really two: what controls the root, and what fills the gap quickly.

Ask five questions before you open a machine

• Material: carbon steel, stainless, or alloy? Stainless changes everything, from heat tint to the ban on carbon contamination.
• Service: sour, cryogenic, high temperature, or ordinary utility duty? Sour service constrains consumable hydrogen content.
• Criticality: a code joint qualified to ASME Section IX, or a structural detail?
• Position and access: a flange on a vertical vessel is a different weld than one on a skid-mounted pipe run.
• Inspection: radiography, ultrasonic, penetrant, or visual only? That decides how much slag and profile deviation you can tolerate.

SMAW (stick): the most tolerant process on site

Stick is the default for field and confined-space flange work because it needs no shielding gas, tolerates dirty, rusty, or shop-primed surfaces that would stop a gas-shielded process, and runs off a machine carried to the work. It dominates field pipe and structural work, as the guide to advantages of stick welding shows. The tradeoffs are slag to chip between passes, slower deposition, and a root profile shaped by electrode angle and drag rather than filler placement, leaving irregular profile and trapped slag for the examiner.

GTAW (TIG): the root pass specialist

TIG is generally favored for the root pass on stainless and high-criticality carbon steel flange joints because the welder controls heat input and filler addition pass by pass, producing a clean, low-porosity root with no slag and a smooth profile that reads easily on a radiograph or under penetrant. Purging the back side with inert gas decides it on stainless, where the root must stay bright. TIG costs more in labor, runs slower than the other processes, and needs clean, unpainted, oxide-free metal. The same fundamentals appear in aerospace TIG welding, at far tighter tolerances.

GMAW and FCAW: production fill and cap

FCAW offers a high deposition rate and good out-of-position capability, so it is widely used for fill and cap on structural and heavy pipe, and it can run self-shielded where no gas supply exists. GMAW with solid wire and gas is fast and clean on carbon steel, with the basics laid out in the practical MIG guide. Both produce slag and spatter that must be removed before inspection. Two GMAW problems repeat on flanges: wire feeding over a hot flange that causes stubbing and erratic feed, and porosity on stainless when shielding coverage or cleanliness is marginal.

The standard industrial answer: manual root, fast fill

Most industrial flange welding is a two-process job. The root is laid manually with TIG or stick, where the welder controls fusion, then fill and cap use FCAW or GMAW for deposition rate.

Consumables: the short version

• Carbon steel solid wire: ER70S-6 for general purpose work, ER70S-8 where notch toughness or cold-temperature service matters.
• Flux-cored: E71T-1 for carbon steel fill and cap in the field.
• SMAW: E7018 low-hydrogen for restrained and thick-section joints.
• Stainless: ER308L for 304 base metal, ER316L for 316. The “L” denotes low carbon content, which is what protects corrosion resistance.

Selection details are worth reading in the all position welding rod guide, with AWS filler metal specifications as the reference at AWS standards. Two rules cover most mistakes. Match or overmatch the base metal: undermatching stainless filler on stainless base metal can produce a corrosion-resistant weld that is weaker and behaves differently from the parent metal, and carbon steel filler on stainless is worse. Control diffusible hydrogen: low-hydrogen electrode and wire classifications matter most for thick-walled carbon steel and for sour service, where hydrogen promotes delayed hydrogen-assisted cracking.

How the process changes the inspection

Your process choice sets the inspection burden. A TIG root is smooth and slag free, so penetrant and radiographic interpretation are straightforward. An SMAW root is prone to irregular profile and trapped slag, pushing more burden onto cleaning and the examiner. FCAW and GMAW caps need slag and spatter removed before any NDE, and arc strikes outside the weld are common. Acceptance criteria come from the governing code, not the process.

Process	Best fit	Root	Fill and cap	Main drawback
GTAW (TIG)	Stainless, critical carbon steel, thin wall	Excellent	Good, slow	Cost, speed, clean metal
SMAW (stick)	Field, confined space, dirty steel	Fair to good	Good	Slag, slow, rough root
FCAW	Structural, heavy pipe, out-of-position	Acceptable if qualified	Excellent	Slag and spatter removal
GMAW (MIG)	Production fill and cap, carbon steel	Not for code roots	Excellent	Spatter, stainless porosity, hot flange wire feed

The Code Framework: ASME B16.5, Section IX, and API 1104

Here is the thing most people get wrong about flange welding: they think there is one rulebook. There is not. A single flange joint on a refinery piping spool is shaped by half a dozen standards, and each one answers a different question. If you know which document answers which question, the whole compliance picture stops being intimidating.

Who Defines What

ASME B16.5 is a dimensional and pressure-temperature rating standard for pipe flanges and flanged fittings. It defines what the flange is: its dimensions, face geometry, bolt circle, gasket seating surface, material specifications, tolerances, and the pressure-temperature ratings for each class. Large diameter applications are covered by ASME B16.47. Notice what B16.5 does not do. It tells you nothing about how to weld the flange to the pipe. It is a product standard, not a welding standard.

That welding requirement comes from somewhere else, and the division of labor looks roughly like this:

Standard	What it actually governs
ASME B16.5 / B16.47	Flange dimensions, ratings, materials, tolerances. What the flange is.
ASME B31.1 / B31.3	How the piping system is designed, including the joint efficiency factor applied to the welded joint in pressure calculations.
ASME BPVC Section IX	How welders, procedures, and operators are qualified.
API 1104 / API 5L	Pipeline welding and line pipe.
NACE MR0175 / ISO 15156	Material and consumable requirements for sour service.
ASME Section V / ISO 9712	Nondestructive examination methods and personnel qualification.

ASME B31.1 covers power piping; ASME B31.3 covers chemical process and petroleum refinery piping. Both require a joint efficiency factor for welded joints, which is how weld quality ends up living inside the pressure design equation. API 5L covers the pipe product itself, while API 1104 governs the welding of pipelines and related systems with its own complete qualification, testing, and acceptance framework.

The Section IX Triangle: WPS, PQR, WPQ

Section IX of the ASME Boiler and Pressure Vessel Code is the qualification backbone used across the B31 piping codes. It requires three distinct documents, and mixing them up is a classic audit finding:

• Welding Procedure Specification (WPS): the written instructions. It states the variables: process, base metal, thickness range, position, preheat, filler metal, gas, and the essential limits within which the procedure may be used.
• Procedure Qualification Record (PQR): the evidence. It documents a test weld made under the procedure and the results of testing it, proving the procedure produces an acceptable weldment.
• Welder Performance Qualification (WPQ): the individual proof. It shows a specific welder can make acceptable welds within the qualified ranges, tied to that person’s name and stamp.

A WPS without a supporting PQR is just a wish list. A PQR without a WPQ does not authorize a person to weld. You need both, and you need them in the right ranges.

Essential vs. Supplementary Essential Variables

Essential variables are the ones that can invalidate a qualification. Change the welding process, the base metal specification, the thickness range, or the welding position, and the previous qualification no longer covers the work. Supplementary essential variables are changes that require review but do not automatically break the qualification. Travel speed and electrode angle fall in this second group, as do several others listed in each code.

Consumables are governed by their own classification system, which is worth understanding before you order filler metal. The AWS A5 series covers welding electrodes and filler metals, and the alloy steel welding electrode guide is a useful primer on how classification numbers relate to base metal and service conditions.

How Qualification Welds Are Evaluated

Qualification welds are not accepted on appearance alone. They are typically evaluated by visual inspection plus a supplementary method such as bend testing, macroetch, or fracture testing, with acceptance criteria set by the governing code. Once a procedure is qualified, production welds made to that WPS are subject to quality control inspection, and the code requires that the WPS be followed.

API 1104 in the Field

API 1104 runs its own parallel system. It requires qualification testing, and it insists that each welder’s work be separately qualified and traceable. Production welds must be identified, and welders apply their identification stamp so a weld can be tied back to the person who made it. That traceability requirement is why you see stamping discipline enforced so hard on pipeline work. A weld that cannot be traced to a qualified welder is a weld nobody owns.

How to Read a WPS, PQR, and WPQ

Read a WPS top to bottom and you are looking at the rules of the job. Read the PQR beside it and you see the test that proved those rules work. Read the WPQ and you are checking the welder against both. Most importantly, check the range-of-validity limits: the thickness bracket, the diameter range, the positions, and the process on the WPQ. A welder qualified in 6G on Schedule 40 pipe has not automatically qualified for 6G on Schedule 80. The all position welding rod guide is a good reminder that position, not just process, defines what a welder is allowed to do.

Finally, remember that none of this replaces the owner’s specification. Many owner specs are stricter than the code, and the stricter document governs. The code tells you the floor. The specification tells you the ceiling.

Preparation, Fit-Up, and Tack Welding

The hour before the arc is struck decides more flange joints than the weld itself. A rushed flange fights you the whole way, and the evidence shows up later as undercut, porosity, or an unfused root.

Two rules govern: the Welding Procedure Specification (WPS) controls joint preparation and root gap, and the code controls inspection hold points. ASME codes and standards set those requirements, and your WPS is the version applying to this joint. A drawing calling for 45 degrees where the WPS calls for 37.5 degrees needs engineering, not a grinder.

Verify the material first

Match the flange stamp to the drawing and the mill certificate before grinding. Read nominal size, pressure class, material specification, and heat number off the ring, then locate that heat number on the certificate. Three mismatches derail jobs: wrong wall thickness, a carbon steel flange delivered into a stainless package, and a heat number on the stamp that matches no paperwork.

Confirm pipe wall against flange bore too. A weld neck flange has a bore matching the pipe inside diameter and a weld end sized to the pipe wall. If the pipe wall is thinner than the flange minimum, the joint cannot deliver design thickness however good the weld is. Measure with a caliper or ultrasonic gauge, not a schedule marking on reamed pipe.

Verification matters most where two metals meet: welding stainless to carbon steel is one of the most expensive flange mistakes in the shop, and it usually starts with a mislabeled flange in the yard.

Cut the bevel and hold the land

A common carbon steel bevel for butt-welded flange joints is 37.5 degrees with a 1/16 inch land, though the WPS and code govern the value. Stainless often uses a smaller land and a purging requirement. Thin wall may go square butt; heavy wall may need a compound bevel when one pass cannot fill the volume.

Land the bevel evenly. A pipe end ground in one pass comes out concave one side and convex the other, and that mismatch drags the root pass toward the high side.

Clean the metal back

Remove paint, mill scale, rust, oil, and grease from a defined distance either side of the joint, typically the weld width plus any travel length the WPS permits. Many quality plans call for cleaning to bright metal within an inch of the joint edge. Anything left there melts into the deposit and returns as slag, porosity, or lack of fusion.

Grinding wheels and wire brushes used on stainless must be dedicated or cleaned, because iron contamination causes rust staining that gets mistaken for a weld defect and can initiate corrosion. A carbon steel brush on stainless leaves an orange bloom within days that QA groups flag. Aluminum needs its own prep: Aluminum Cleaner for Welding: How to Prep Aluminum Right.

Fit-up: gap, land, squareness, clamping

Root gaps are normally specified in a narrow band around 1/8 inch for most GTAW pipe work. Set the gap with a feeler gauge or gauge block, not by eye. Too wide gives a high-loose root pass with poor penetration; too tight burns through. Keep the root face or land intact for the first pass.

Check squareness with a straightedge across the flange face and a square on the flange bore. A face out of square preloads the bolting unevenly, and that preload surfaces as bolt fatigue rather than a weld defect. Clamp the flange square without distorting the joint, and never clamp so the pipe goes out of round.

Closed systems need a vent path

A flange bolted to another with no gap creates a closed cavity with no vent path. Welding into one traps expanding gas, raises internal pressure, and can blow out the root or force porosity into the deposit. Provide a deliberate root gap or a vent and bleed path. OSHA’s welding standard requires hollow spaces, cavities, and containers be vented before preheating, cutting, or welding, and piping or vessel connections disconnected or blanked, with inert gas purging recommended.

Tack welding

Most codes limit tack welds in size and number because each is an uncontrolled, rapidly cooled deposit that can carry defects into the final weld. Common limits cap tack size at roughly a quarter inch or a third of the wall thickness, whichever is smaller, and restrict tack length to a few inches. Use the same filler metal and process as the root pass, and place tacks where the finished weld will melt them completely. Poorly placed or oversized tacks drive rework and trapped porosity.

Purge stainless before striking an arc

Stainless welding requires purging with an inert gas, and industry practice targets headspace oxygen in the tens of ppm before the arc is struck to keep the root free of oxidation and color. Nitrogen is the standard purge gas; hold it with a flow meter plus a sealed purge dam or a backing ring. Never start an arc on a dirty or unpurged root, because the first pass sets the color of the whole weld. Gas and cylinder handling: Air Gas Welding Supply: Gases, Cylinders and Safety Guide.

Fit-up checklist

1. Stamp, class, material, heat number verified.
2. Pipe wall measured, within minimum.
3. Bevel and land per WPS, cut evenly.
4. Cleaned to bright metal, specified distance.
5. Root gap gauged at three points.
6. Flange square, concentric, clamped without distortion.
7. Closed system vented, blanked, or purged.
8. Tacks within code limits, correct filler.
9. Stainless purged to low oxygen, verified.

Welding Technique: Bead Profile, Heat Input, and Interpass Control

The WPS on the clipboard is only a plan. What actually decides whether a flange joint holds pressure is what happens in the sixty seconds the arc is on. A flange weld is short, sometimes a single girth weld only a few millimetres of depth at the root, but it is deposited into a thick, rigid, heavily restrained geometry. That leaves the welder far less room for error than someone running a long fillet weld on a structural member, and it is why technique, more than almost anything else, decides whether a marginal fit-up gets saved or turns into a cracked joint.

What a Good Flange Weld Actually Looks Like

Begin with the geometry, because it is what the inspector sees first. A sound weld here has a smooth, slightly convex cap, even sidewall fusion into the groove faces, no undercut, no overlap, no trapped slag, and a consistent profile all the way around the circumference. The bead should be run on and run off so it starts and stops cleanly, using run-on and run-off tabs or a controlled start and stop that leaves a smooth taper rather than a crater or a blown-out end. Reinforcement is limited by the governing code rather than by taste.

Undercut deserves its own emphasis because it is the most common surface defect in flange work. It is the groove along the toe of the weld where fusion is incomplete, and it shows up when travel speed is too high, the arc is too wide, the electrode or torch angle is wrong, or the amperage is too high for the arc length you are holding. Codes commonly cap undercut depth at roughly 1/32 in (1 mm) on thinner wall and allow a little more on heavy wall, with limits on total undercut length. An undercut you can spot with the naked eye is usually an undercut that has already cost you a repair.

The Two-Lay Technique: Root, Hot Pass, Cap

Most flange joints are welded in two or three passes:

• Root pass makes the penetration and establishes the fusion to the joint faces. This pass carries all the risk of burn-through, incomplete penetration, and cold lap.
• Hot pass is a quick second pass over the root, sometimes called a wash or scrub pass. It burns off diffusible hydrogen, smooths an irregular root, and leaves a clean surface for the next layer. On a root made with SMAW it is effectively standard practice.
• Fill and cap passes build up the remaining wall thickness, with the cap pass laid to leave a smooth, slightly convex surface with the correct reinforcement.

Each pass is cleaned before the next one. You are stacking sound metal on sound metal, and a pass that looks fine in isolation can still ruin the joint if slag, spatter, or a loose oxide scale sits under the next one.

Root Control: GTAW and SMAW

On GTAW, a DCEN root with filler added to the puddle gives the cleanest penetration you can buy, which is why TIG roots dominate stainless and high-integrity work. The techniques, shielding, and temperature control behind that root pass are covered in detail in our aerospace TIG welding guide, and the same clean-root logic applies to a flange. Watch the puddle size. Push too much amperage and you burn through into the bore; back off and you get a cold, poorly fused root instead.

On SMAW, the root profile is set by electrode angle and drag technique. A drag (rather than push) angle, held consistently, gives a more uniform root than a wandering electrode. The trade-off is that an SMAW root is inherently lumpier than a TIG root, which is exactly why the hot pass exists. Our look at why SMAW still wins covers the case for stick on field work where setup speed matters more than cosmetic profile.

Heat Input: The Biggest Lever You Have

Heat input is a function of power, arc time, and travel speed. In practice it is reported as energy per unit length, calculated from voltage, amperage, and travel speed, and it is the single biggest controllable variable in the whole procedure. More heat input gives deeper penetration, but also a wider, flatter bead, more distortion, and a slower cooling rate that raises hydrogen cracking risk. Less heat input keeps the bead narrow and the distortion low, but if you go too low you invite lack of fusion and a cold lap at the sidewall.

Arc energy calculations and heat input limits for a given joint are set in the procedure, with the governing requirements published by standards bodies such as AWS and the relevant ASME codes and standards. The practical takeaway for the welder is simpler than the formula: if the bead looks flat and wide, you are running too hot; if you are stacking beads to chase penetration, you are running too cold.

Preweld and Interpass Temperature Control

Preweld preheat raises the starting temperature of the joint. That slows the rate at which the metal cools through the hydrogen-sensitive range, which sits below roughly 200 °C (390 °F), where hydrogen becomes mobile and can combine with a susceptible microstructure to initiate cracking. It also lowers residual stress in a joint that wants to move but cannot. In thick, restrained carbon steel sections, preheat is usually the difference between a sound weld and a latent crack that surfaces months later during a hydrostatic test or in service. Pipeline work has its own well-established preheat and interpass framework under API 1104.

Interpass temperature is the maximum temperature the weld metal may cool to between passes, and it matters for reasons that are easy to miss. Reheating the heat-affected zone repeatedly can drive it into an over-tempered, lower-toughness condition, and every pass adds to the accumulated distortion in a joint that has nowhere to flex. Modern procedures therefore set both a minimum preheat and a maximum interpass temperature, and both are non-negotiable limits rather than guidance.

Arc Etiquette: Stringers, Weave, and a Steady Arc

Stringer beads, one pass at a time, are generally preferred for a flange joint because they give a more consistent profile and less heat input than a weave. Weaving buys you speed on structural work but widens the bead, increases heat input, and, on SMAW, traps more slag. Hold a steady arc length (roughly 1/8 in, or 3 mm, is a common GTAW target) and a steady travel speed. A wandering arc is the root cause of most surface problems: undercut on the sides where the arc lingers, porosity where the puddle is disturbed, and lack of fusion where the arc is too far from the sidewall.

Porosity has its own diagnostic logic. In GTAW it is most often contaminated filler metal or base metal, a leaking gas line, moisture, or a dirty root. In SMAW it is most often trapped slag, a damp electrode, or excessive travel speed.

Inspection and NDE: What Is Actually Required

Not every flange joint earns the same level of examination. A utility reconnect may get a visual check and a hydrostatic test. A sour-service pipeline flange may get 100% radiography and a hold-point inspection. The governing code sets the floor; the owner specification, purchase order, and inspection and test plan set everything above it.

The order never changes: visual first, the code’s supplemental methods next, then the pressure test that proves the assembly holds. Most people meet the middle layer first and end up unsure about what was actually required.

Visual inspection is the first and most productive step

Visual examination is fast, free, and reliably identifies profile defects, undercut, overlap, arc strikes, and visible cracking. It happens after the joint is cleaned of slag, spatter, and paint. The examiner is looking for:

• Face profile: reinforcement, width, smoothness, and taper into the base metal
• Undercut, overlap, and excessive reinforcement
• Surface porosity, cracks, and trapped slag
• Arc strikes and grinding gouges outside the weld toe, both classic crack starters
• Spatter, oxidation, and cleanliness
• Fit-up results such as flange alignment and bolt hole concentricity

The four methods you will actually meet

Method	Applies to	Finds	Watch out for
Liquid penetrant (PT)	Clean, non-porous metals: stainless, aluminum, carbon steel	Surface-breaking cracks and flaws	Misses subsurface flaws; fails on rough or coated surfaces
Magnetic particle (MT)	Ferromagnetic materials only, such as carbon steel	Surface and near-surface defects, very sensitive	Cannot be used on stainless or aluminum
Radiographic (RT)	Most metals, including bimetallic joints	Volumetric coverage of the full joint	Cost, radiation safety, film handling, pipe shadow
Ultrasonic (UT)	Thick sections and restricted access	Internal flaws with depth and location	Operator- and procedure-sensitive; needs calibration

PT is the workhorse for stainless flange welds: quick, inexpensive, and sensitive to fine surface cracks on a clean, uncoated surface. MT is the fastest, most sensitive surface method on carbon steel, but only where the material holds a magnetic field. RT remains the reference method for full-joint examination and the acceptance basis for many code joints, and the most expensive route: licensed operation, radiation controls, film handling, and geometry traps like pipe shadow that can hide a root flaw. UT reaches into thick sections where depth matters, but it is the hardest method to qualify and the easiest to argue about.

Acceptance criteria come from the code, and from the code in force

Criteria are not universal. ASME B31.1, ASME B31.3, and API 1104 each set their own limits for undercut, reinforcement, and porosity, and they do not match one another. Radiographic acceptance is tied to the governing code or the owner specification, so confirm the edition in force before you accept a film. The owner specification is often stricter than the code: more joints tested, tighter repair limits, added methods such as UT. If the purchase order adds a requirement, it governs.

Who is allowed to do the work

ASME Section V sets the qualification requirements for nondestructive examination personnel, and ISO 9712 is the widely used international certification standard for NDT technicians. AWS separately offers Certified Welding Inspector (CWI) and Certified Associate Welding Inspector (CAWI) credentials, plus a Pipeline (API 1104) endorsement. A CWI is not automatically an NDT technician; the inspector verifies the job, the Level II performs and interprets the test. In practice the inspector checks material certificates, the WPS and PQR against the joint, the welder’s WPQ range, and consumable certificates. Rod classifications and low-hydrogen designations are worth refreshing in our all-position welding rod guide.

Hold points and the rule against welding over an indication

The inspection and test plan lists hold points and witness points. A typical flange job holds at fit-up sign-off, pre-weld verification, before NDE, and before release. Give the examiner the required notice period in writing, with the time window stated. One rule prevents most bad days: nobody grinds, gouges, or welds over an indication until it has been evaluated and dispositioned by the responsible inspector.

When the weld fails: repairs

A repair is a new weld. It must be made by a qualified welder using a qualified procedure, obey the same heat input and interpass limits, and be re-examined to the same acceptance criteria as the original weld. Codes also cap repeat repairs at the same location, commonly two attempts, before the next one needs formal engineering review rather than another weld. Track repair counts per weld location, not per joint: a third pass at one spot is a code event, not a maintenance decision.

The next section turns the microscope on the failures themselves: porosity, undercut, lack of fusion, burnthrough, and the cracks that come with all of them.

I’ll rewrite this section, tightening the prose while preserving every fact and link.

Common Flange Welding Defects and How to Prevent Them

Flange weld defects are rarely exotic. The same handful of problems show up on job after job, for the same handful of reasons. The real skill is not memorizing defect names. It is learning to read a symptom and work backwards to the condition that produced it.

Begin with the uncomfortable truth about flange joints: they hide defects. The root sits inside a bore, the cap rests on a narrow land, and the flange face is the only part of the weld anyone sees in service. Visual inspection catches a great deal, but some of the worst problems are tucked underneath a cap that looks perfectly sound.

A Symptom, Cause, and Fix Reference Table

Use this as a first-pass diagnosis. When the field symptom does not match the row, dig deeper into the cause column, because the visible evidence and the real cause often disagree.

Symptom	Likely cause	Fix
Porosity: pits, voids, a weeping fillet	Contamination or moisture on the joint faces, damp consumables, incorrect shielding gas flow, travel speed outrunning the puddle, trapped moisture in a closed cavity	Clean to bright metal, store consumables dry, set and verify gas flow at the torch, slow the travel speed, purge the root
Undercut: a groove along the toe of the bead	Amperage too high for the electrode diameter, arc held too long, travel speed too fast, wrong work angle	Lower amperage, shorten the arc, switch to stringers, reposition the torch so the arc points into the joint
Excess penetration or root burn-through	Excess heat input, root gap wider than the WPS allows	Reduce heat input, close the root gap, use a backing bar or consumable insert, add a second root pass
Slag inclusions between passes	Poor interpass cleaning, wrong electrode angle, weaving too fast for the slag to float out	Chip and clean every pass, keep a consistent electrode drag, slow the weave down
Root concavity (a root that dips inward)	Root pass traveled too slowly, filler added at the wrong moment so the puddle collapsed	Control filler timing, steady the travel speed, keep the arc moving
Excessive root reinforcement (a root that bulges into the bore)	Filler deposited too early, torch held over one spot too long	Delay the filler addition, keep traveling, trim only if the procedure allows
Arc strike outside the weld area	Grounding or electrode change made on the flange face or pipe wall	Grind out completely, blend smooth, have the area re-examined before acceptance
Sharp grinding gouges	Aggressive grinding leaving an abrupt notch in the parent metal	Blend with a smooth radius and gradual transition, then re-examine
Flange face out of square or bore off center	High heat input, unbalanced welding sequence, insufficient or poorly placed tacking	Balance the sequence, use back-step technique, more numerous smaller passes, better tacking
Root or toe cracking	Hydrogen-assisted cracking, lack of fusion, stress corrosion cracking in sour service	Stop work, evaluate, remove the crack and rewelder, or correct material and consumable selection

Lack of Fusion: The Defect That Hides Under a Good Cap

Lack of fusion is the defect most often missed by visual inspection alone, because a sound-looking cap can cover it. The causes are insufficient heat input, contaminated or oxidized surfaces, or a bead laid in the wrong position. The weld looks fine from the outside while the previous pass never actually melted into the layer beneath it.

The fix is unglamorous: clean every pass back to bright metal before welding over it, raise heat input within the limits of the qualified procedure, and place the bead where it can fuse with the surface it is joining. Once found, lack of fusion cannot be welded over or bridged. It has to be gouged out, re-prepared, and rewelded to a procedure that meets the qualification requirements in the ASME code and standards program. Control of the root pass matters most here, and the filler-addition discipline used in aerospace TIG welding translates directly to a stainless flange root.

Distortion: When the Flange Face Goes Out of Square

Distortion in a flange joint typically shows up as the flange face going out of square or the bore moving off center as the joint cools. That is not only a dimensional problem. An out-of-square face means the bolts do not load evenly, and uneven bolt preload is exactly the kind of condition that turns into a leak path at the gasket later.

Welding sequence control is the primary non-equipment method of controlling distortion in heavy flanged assemblies. Balance the shrinkage by choosing a sequence that spreads heat around the joint, use the back-step technique so the leading edge of the bead sits in cooler, already-shrunk metal, and divide a long joint into shorter sections instead of running one continuous bead around the circumference. More numerous smaller passes at lower heat input shrink less than a few heavy ones. Tacking helps too, but put tacks where they control shrinkage, not where they are easiest to reach.

Cracking: Three Mechanisms, Three Responses

Not all cracks are the same, and the correct response depends on which mechanism produced the crack.

Hydrogen-assisted cracking in carbon steel is time dependent. It may not appear for hours or even days after the arc is extinguished, which is precisely why code-required examinations and hold periods exist before a system is returned to service. Prevention is preheat, low-hydrogen consumables, and a clean, dry joint.

Lack-of-fusion-type cracking follows the unfused boundary and usually shows up as a root crack. It cannot be welded over. It has to be removed and the joint rewelded.

Stress corrosion cracking behaves differently again. In sour service, hydrogen sulfide, chlorides, tensile stress, and temperature combine to crack a joint that passed every other test. Sour service standards limit both base metal hardness and the diffusible hydrogen content of the consumable, and heat input is the variable a welder controls most directly to stay inside those limits, because excess heat drives hardness upward. Material selection, low-hydrogen consumables, and controlled heat input all have to be right at the same time.

Arc Strikes and Grinding Gouges

Arc strikes made outside the weld area are a known crack initiation site and are generally prohibited on pressure-retaining components. If a strike lands on a flange face or pipe wall, grind it out completely, blend the area with a smooth gradual transition, and have the ground-out strike re-examined before it is accepted. The same logic applies to sharp grinding gouges. A notch left sharp is a crack waiting to start. Slag control and electrode handling habits are covered in more detail in the all-position welding rod guide.

Safety, Distortion Control, and a Field-Ready Workflow

A flange joint that passes every radiographic check is still a failed job if the welder spent the afternoon coughing.

Ventilation First, Respirators Second

Under OSHA’s respiratory protection standard, the primary objective is preventing atmospheric contamination through engineering controls such as enclosure and ventilation, and respirators supplement those controls where engineering cannot do the job.

Local exhaust ventilation first: a fume extraction arm or hood close to the arc, plus forced ventilation inside a vessel or spool. OSHA’s welding ventilation standard requires general ventilation adequate to keep fume concentrations within limits, with local exhaust where general ventilation cannot keep pace. A respirator covers the residual: a half mask, right cartridge for the base metal, fit tested.

Stainless Steel and the Cr(VI) Problem

OSHA identifies hot work on stainless steel and other chromium-bearing alloys as a major source of worker exposure to hexavalent chromium. Cr(VI) is a known carcinogen that attacks the respiratory system, kidneys, liver, skin, and eyes. It forms during hot work, so every TIG pass on a 316L flange produces it. Control it at source with local exhaust, keep the area clean, and never dry-grind stainless without dedicated media and a dust respirator.

Manganese, Confined Spaces, and Manganism

Carbon steel fume carries manganese. NIOSH lists a recommended exposure limit of 1 mg/m³ as an 8-hour TWA and 3 mg/m³ short-term; OSHA’s permissible exposure limit is 5 mg/m³ as a ceiling, and ACGIH threshold limit values are far tighter at 0.02 mg/m³ respirable and 0.1 mg/m³ inhalable. NIOSH also notes that welding in confined spaces can significantly increase exposure to manganese fumes, and that prolonged exposure can produce a Parkinsonian-like syndrome called manganism.

PPE and Filter Shade Selection

OSHA’s PPE standard requires the employer to assess the workplace for hazards, select and provide appropriate PPE, train employees on its use and limitations, and provide it at no cost to the employee. Keep it on file. The welder needs the right filter shade, flame resistant clothing, process appropriate gloves, and hearing protection where extraction adds noise.

OSHA’s filter shade guide: shade 10 for SMAW with 1/16 to 1/8 inch electrodes, shade 12 for gas-shielded arc welding on ferrous metals, and shade 11 for gas-shielded arc welding on nonferrous metals. OSHA 1910.252 requires helmets or hand shields during all arc welding and arc cutting operations. A flange stays hot after the arc dies, so the arc shade can be too light for slag knocking. ANSI Z49.1, from the AWS standards program, is the consensus standard OSHA references.

Hot Work Permits and the Fire Watch

Any flange job involving hot work needs a permit: area inspected, combustibles cleared, extinguishing equipment staged, signed before the arc strikes. OSHA requires fire watchers when appreciable combustible material is within 35 feet of the point of operation, and a fire watch for at least a half hour after welding or cutting is complete.

Confined Space Rules

OSHA requires gas cylinders and welding machines to be left outside a confined space, heavy portable equipment blocked, and torch valves closed with the gas supply positively shut off outside the confined area whenever the torch is not in use. For stainless, purge with nitrogen and verify oxygen content before striking a root pass. Our welding equipment guide covers that equipment.

Keeping the Flange Flat

A flange warped out of flat will not gasket properly, and the crew that compensates by over-torquing the bolts ends up with a leak. Sequence the weld to balance shrinkage, keep passes small and consistent, use back-step technique out of position, and check the flange face with a straightedge and the pipe centering with a square before you tack. A joint pulled out of round by clamping fails the hydrostatic test.

An End-to-End Field Workflow

1. Document review: drawing, WPS, code edition, inspection hold points.
2. Material verification: flange stamp and heat number against the certificate, before grinding.
3. Preparation: bevel to the specified angle and land, clean to bright metal, remove paint, scale, oil, grease.
4. Fit-up: root gap gauged, root face held, flange square and concentric, tacks per procedure.
5. Purge and shielding: verify oxygen content on stainless, confirm gas flow and coverage.
6. Welding: root first, control preheat and interpass temperature, log parameters.
7. Visual inspection while the joint is accessible and the weld is warm enough to clean.
8. Notify the inspector and wait for the hold point to clear.
9. Final NDE, repairs only after written disposition, close out documentation.

Final Checklist

• Hot work permit signed, fire watch posted, 35-foot clear zone, 30-minute watch.
• Local exhaust verified at the joint, not just switched on at the wall.
• Respirator fit tested, right cartridge in hand.
• Correct filter shade, FR clothing zipped, hearing protection on.
• Cylinders and machines outside the confined space, torch valves closed, gas shut off.
• Consumables dry, electrodes and filler rods protected from moisture. Our electrode selection and safety guide covers handling and storage.
• Weld map, WPS reference, inspection records, and NDE reports filed.

Safety and workmanship are the same conversation: ventilate the fume, control the heat, verify the material, respect the hold points. That closes the technical arc.

Frequently Asked Questions

What is the best welding process for pipe flanges?

Most industrial shops use GTAW (TIG) for the root pass, especially on stainless steel and high-criticality carbon steel, because it gives the welder precise control of the puddle and produces a clean, easily inspected root. FCAW or GMAW is then commonly used for fill and cap passes to recover deposition rate, and SMAW remains the standard for field and confined-space work where gas-shielded equipment is impractical.

Do I need to qualify a welding procedure before welding a flange?

For code work, yes. ASME BPVC Section IX requires a qualified Welding Procedure Specification supported by a Procedure Qualification Record, plus a Welder Performance Qualification for each welder doing the work. API 1104 carries its own qualification framework. For non-code repairs, qualification is still good practice because it is the only way to know your heat input and technique produce an acceptable joint.

What is the correct root gap for a flanged joint?

Most GTAW pipe work runs around 1/8 inch, but the value must come from the WPS, and anything outside the qualified range means the procedure must be requalified. Too wide a gap gives a high-loose root with poor penetration; too tight a gap burns through and leaves a mismatched, concave root.

Why does my flange weld leak even though it passed visual inspection?

Visual inspection catches shape and surface problems but not subsurface or root flaws, so a leak usually points to porosity in the root pass, lack of fusion, a root concavity, or a joint that went out of square and preloaded the bolting unevenly. Leaks in pressurised flange joints also frequently come from the gasket and bolting rather than the weld itself, which is why hydrostatic testing and proper bolting procedures matter as much as the weld.

Do I need to purge the inside of a stainless flange joint?

Yes for stainless and other sensitive alloys, because the root is exposed to air and would oxidize without an inert gas purge. Industry practice targets weld headspace oxygen in the tens of ppm before striking the arc, and the purge should continue long enough to keep the root bright behind the arc. Purge gas must also be managed so it does not create a flammable or oxygen-deficient atmosphere in the confined space.

What is the most common defect in a flanged joint?

Undercut is the most common surface defect, driven by high amperage, a wide arc, or too much travel speed. Among the most serious, lack of fusion is the one that most often gets missed visually, because the cap can look sound while the joint is not fully melted. Prevention is mostly about clean bright metal, correct heat input, and a consistent technique.

How hot does welding a flange get, and is preheat always required?

Preheat is not automatically required for every flange. It is specified by the WPS and code based on material specification, thickness, and service conditions, and it is generally driven by hydrogen cracking risk in thick carbon steel and by the need for slower cooling in restrained joints. Whenever preheat is specified, verify the actual interpass temperature with a temperature-indicating pencil or thermocouple rather than trusting the torch colour.

Can you weld a flange that is already bolted up?

Not without creating problems. Welding a flange that is bolted to a live system locks the joint, so any shrinkage pulls on the existing bolts and the new weld is made under uncontrolled restraint, often with no root gap and no ability to purge. A flange also has to be disconnected, depressurised, drained, cleaned, and vented before hot work, with piping connections disconnected or blanked.

What inspection do most flange welds actually require?

Visual inspection is universal and is where most defects are caught. Beyond that, requirements come from the governing code, the owner specification, and the service: liquid penetrant testing for surface defects on stainless, magnetic particle testing on carbon steel, and radiography for critical or high-pressure joints. API 1104 work typically calls for radiography, and API 5L line pipe has its own additional requirements.

How do I control distortion when welding a heavy flange joint?

Distortion comes from unbalanced shrinkage in a rigid, restrained joint, so the tools are sequencing, weld size, and symmetry. Use multiple smaller passes instead of one large one, balance the weld around the circumference, keep heat input as low as the code permits, and consider a back-step technique. After the joint cools, check the flange for squareness and bore alignment before bolting up, because out-of-square flanges preload the bolting unevenly.