Pipe Welding Inspection: Codes, NDT Methods & Acceptance

The hydrotest gauge had been holding steady for twenty minutes when a thin bead of water appeared along a girth weld that had passed visual inspection two days earlier. Nobody argued about the welder’s steady hand — the argument was about the sheet of paper that said the joint had been examined, by whom, with what method, and against which acceptance criteria. That is the world of pipe welding inspection: less about catching bad welders and more about proving, in writing, that a joint is fit for the pressure and the product it will carry.

What Pipe Welding Inspection Really Means (and Why It Exists)

The hydrotest gauge had held steady for twenty minutes when a thin line of water traced along a girth weld that had passed visual examination two days earlier. The welder had not become careless. What failed was the chain of proof around the joint: who examined it, with what method, at which stage, and against which acceptance criteria. That is the real subject of pipe welding inspection. It is less about catching bad welders than proving, in writing, that a joint is fit for the pressure and product it will carry.

In plain terms, pipe welding inspection is the planned system of examination, testing, and documentation used to confirm that a welded piping or pipeline joint meets its governing code, project specification, and service requirements. The weld is made by a process. Inspection verifies that the process was controlled, then records the result so the next person can trust it.

The Words That Cause the Most Arguments

ASME B31.3 draws a distinction that trips up even experienced crews: examination is the act performed by the manufacturer or constructor, while inspection is the owner’s function. That split drives who signs what. The constructor examines; the owner inspects and accepts. Confuse the two and a completed weld can sit in limbo because the right signature never landed on the right form.

Non-destructive examination (NDE), also called NDT, covers methods that reveal conditions without cutting the joint apart: visual, radiography, ultrasonic, penetrant, magnetic particle, hardness, and positive material identification. Destructive testing bends or pulls a specimen until it fails, which is how procedures and welders are qualified rather than how production joints are accepted. Acceptance criteria are the written limits that separate acceptable from rejectable. Hold points are stages where work stops until a nominated person signs off.

Who Is on Site, and What Each Person Owns

Six people typically touch the same joint, and their duties overlap without being interchangeable:

• The welder makes the weld within the qualified procedure and flags fit-up problems before striking an arc.
• The welding supervisor controls crew, consumables, preheat, and sequence, and usually takes the first look at the root.
• The quality control inspector performs or coordinates examination and compiles the records.
• The NDT technician applies the specified method and interprets indications within their certification.
• The owner or third-party inspector verifies that the constructor’s system is working and accepts the finished work.
• The welding engineer resolves procedure questions, evaluates nonconformances, and decides whether a repair is sound.

One joint may be checked by three different people at three stages: fit-up before welding, the root before capping, the finished weld by NDT. Each stage answers a different question, which is why skipping one cannot be covered up later.

The Four Questions Every Inspection Program Answers

1. Is the joint geometrically and visually acceptable? Alignment, root gap, cap profile, undercut, surface condition.
2. Are the internal conditions acceptable? Fusion, penetration, porosity, inclusions, and cracking, as revealed by volumetric or surface methods.
3. Is the paperwork traceable? Can this weld be tied to a procedure, a qualified welder, the correct materials, and the right examination record?
4. Was the welder qualified for this joint? Same position, process, and thickness range. A good welder on the wrong qualification is still a nonconformance.

The five most common inspection points on a spool are fit-up and hi-lo, the root pass, the completed cap, the specified NDT, and the final documentation package. Only one of those is about looking at a finished weld.

The scope is broader than it looks. ASME B31.3 covers piping in petroleum refineries, chemical, pharmaceutical, hydrogen, textile, paper and pulp, power generation, semiconductor, and cryogenic plants. ASME notes that the 2024 edition runs 592 pages and remains one of its most requested codes. Under B31.3, visual examination is required on all completed welds, while volumetric examination in Normal Fluid Service is sample-based unless the owner specifies more. Pipelines follow a different family: API 1104 governs welding of pipelines and related facilities, while ASME B31.4 and B31.8 cover liquid and gas pipeline systems respectively.

The logic holds whatever the material. A procedure is a recipe, not a suggestion, and the discipline that governs a root pass on a spool applies to aluminum MIG welding, where cleaning and travel speed can change the outcome entirely.

What Inspection Cannot Do

Inspection cannot weld quality into a joint. It cannot turn a poorly fitted root into a sound one, restore lost preheat, or undo moisture that entered a cellulosic electrode. It verifies and documents what process control has already produced. That is why the most useful inspection conversation happens before the first arc. Built into the sequence, most findings become corrections while they are still cheap. Treated as a final gate, every finding becomes a repair.

The Code Map: ASME B31.3, API 1104, Section IX, and What Each One Actually Governs

Ask five people on the same job which code governs a girth weld, and you can get five confident, contradictory answers. That is not because any of them are wrong. It is because the standards stack is deliberately divided into jobs, and most confusion comes from expecting one document to do all of them. No single code qualifies your welder, sets your acceptance limits, tells the technician how to shoot a radiograph, and defines how hot the preheat must be. Four or five documents split that work.

The Division of Labor

• ASME BPVC Section IX qualifies welding procedures and welding personnel. It tells you whether a WPS is supported by a valid PQR and whether a welder has demonstrated the skill for a given process, position, thickness and diameter. It does not tell you whether a weld is acceptable.
• ASME B31.3 covers materials and components, design, fabrication, assembly, erection, examination, inspection and testing of process piping. It is the construction code for most refinery, chemical and pharmaceutical piping, and its Table 341.3.2 is where acceptance limits actually live.
• ASME B31.1 does the same job for power piping: steam, boiler external piping, and high-energy systems where thermal cycling is brutal.
• ASME B31.4 and B31.8 cover liquid pipelines and gas transmission and distribution piping respectively, largely by reference to other standards for the weld itself.
• API 1104 is the standard for welding pipelines and related facilities, and it is the reference most US pipeline operators and regulators use for girth welds. Unlike Section IX, it includes its own acceptance criteria, which is a key difference.
• ASME Section V provides the NDT methods themselves: how to shoot and interpret a radiograph, how to perform ultrasonic, penetrant and magnetic particle testing. It defines technique, not acceptance.
• AWS D10.10 is a recommended practice for local heating of welds in piping and tubing. It guides preheat and post-weld heat treatment control. It says nothing about whether a weld passes.

Standard	Scope	What it does not cover
ASME Section IX	Qualification of WPS, PQR and welders	Weld acceptance limits
ASME B31.3	Process piping design, fabrication, examination, testing	Welder qualification rules (refers to Section IX)
ASME B31.1	Power piping	Process plant service allowances
API 1104	Pipeline welding, inspection and acceptance	Process piping systems
ASME Section V	NDT methods and techniques	Acceptance criteria
AWS D10.10	Local heating for preheat and PWHT	Weld acceptance

The Contract Hierarchy Problem

In practice, more disputes come from document ordering than from metallurgy. The usual resolution order runs code first, then project specification, then owner specification, then drawings, with the more restrictive requirement generally governing. Read the contract’s order-of-precedence clause before you assume anything, because some owners flip it deliberately.

Here is the classic conflict. B31.3 permits a percentage of random radiography in normal fluid service. Your project specification demands 10% instead of the code minimum. The code does not forbid extra examination, so the stricter document wins and the job pays for the extra shots. The inspector’s job is not to argue on the floor but to document: raise a clarification, cite both clauses, note the owner’s written direction, and record the resolution in the examination plan. When the client later asks why 10% was shot, the answer is one page, not a phone call from memory.

Welder qualification positions are another place this bites. A welder tested in the 6G position on pipe covers a wide range, but a specification can still demand a fresh test on the actual diameter and wall thickness. Track that against the all-position welding coverage in the welder’s continuity record, not against what the crew remembers.

Outside the ASME World

If your project runs on ISO, the equivalents are ISO 9606 for welder qualification and ISO 3834 for quality requirements in welding fabrication. They occupy roughly the same seats as Section IX and a welding quality management system, but the tests, ranges and documentation differ enough that a Section IX welder cannot simply be assumed qualified. For NDT personnel, ASNT SNT-TC-1A is a recommended practice and ASNT CP-189 is a standard; both define Levels I, II and III. ISO 9712 is the international certification route and is commonly required outside the United States. Projects sometimes require both, which is a staffing problem worth spotting at bid stage.

The safest habit is simple: before the first arc, write down which document qualifies the welder, which document accepts the weld, which document defines the test, and which document you would hand an auditor if they asked why.

The Inspection Workflow: From Fit-Up to Final Sign-Off

Watch a single joint travel from the laydown yard to the turnover package and the whole discipline stops looking abstract. Every check has a place, a person, and a piece of paper attached to it. Skip a link in that chain and you don’t find out at the next meeting — you find out at the hydrotest.

Stage 1: Before the Arc Even Strikes

Most of the decisions that determine whether a weld passes were made before anyone struck an arc. This is the stage where inspection has the most leverage and costs the least.

• Procedure support. The WPS must be backed by a procedure qualification record produced by testing per ASME Section IX. No PQR, no production welding. A WPS that appears out of nowhere is a red flag, not a formality.
• Welder coverage. The welder’s performance qualification has to cover the process, position, pipe diameter, thickness and backing condition actually being welded. A 6G qualifier on 6-inch schedule 40 does not automatically cover a 2-inch socket or a 24-inch heavy wall.
• Material identity. Check mill certificates against the heat numbers stamped or tagged on the pipe and fittings. Mixed material is a silent failure — it welds fine and fails later.
• Consumables. Verify storage conditions, that low-hydrogen electrodes were held in a working oven at the specified temperature, and that lot numbers are recorded and traceable.
• Fit-up. Root gap, land, and hi-lo misalignment. Internal misalignment is commonly capped around 1/16 in. (1.6 mm) in B31.3-style work, but confirm the value in the applicable code edition and project spec rather than trusting memory.
• Purge and preheat. Stainless and duplex roots need oxygen content verified, not guessed. Preheat must be measured and recorded before the first pass.

Stage 2: During Welding

The root pass is the single most consequential few minutes of the job. Once it is capped, an entire class of defects — incomplete penetration, suck-back, burn-through, internal misalignment — becomes invisible to everything except radiography or ultrasonic testing, and by then the fix costs ten times as much.

Beyond the root, the inspector watches interpass temperature against the WPS ceiling, amperage, voltage and travel speed against the parameter range, cleaning between passes (wire brushing and slag removal, not a hopeful wipe), and weather protection. Wind stripping shielding gas or rain hitting a hot joint will produce porosity that no amount of technique recovers. If the crew is running cellulosic stick electrodes on a root, the sensitivity to moisture and travel speed is even higher — the same reasons stick welding still dominates pipeline work are the reasons it demands discipline.

Stage 3: After the Arc

Visual examination comes first, on a cleaned and cooled joint. Then the specified NDT at the specified time — some procedures require a delay before final examination to let hydrogen-assisted cracking reveal itself. Where ASME B31.3 calls for post-weld heat treatment based on material and thickness, that happens next, often followed by hardness verification as a project add-on to prove the treatment actually worked. Pressure testing follows at roughly 1.5 times design pressure, subject to the code’s temperature and material limits. Then the paperwork closes the loop.

Stage	What is checked	Who checks	Record produced
Pre-weld	WPS/PQR, welder quals, material certs, fit-up, preheat	QC inspector, welding supervisor	Fit-up report, preheat log, material trace
During weld	Root condition, interpass temp, parameters, cleaning	Supervisor, inspector (witness)	Parameter log, hold point release
Post-weld	VT, NDT, PWHT, hardness, pressure test	NDT technician, inspector, owner rep	VT report, RT/UT report, PWHT chart, test cert

Hold Points vs. Witness Points

A witness point means “we’ll tell you when we get there, and you’re welcome to watch.” Work continues whether or not the inspector shows up. A hold point means “work stops here until you sign.” Welding past a hold point is one of the most expensive mistakes on any job, because it converts a five-minute verification into either a cut-out or a negotiated concession nobody wants to sign.

The Root That Got Capped Too Soon

A crew on a stainless spool capped a root pass at shift change; the inspector arrived at 7 a.m. to find the joint complete. Options: cut the cap off to inspect the root (destructive, expensive, and it damages the qualified joint), radiograph the full joint to settle the question (fast but not equivalent to seeing root geometry), or accept it on the strength of the welder’s record and the RT result with a documented concession. Most owners take the third option and note the deviation. The lesson isn’t about blame — it’s that a hold point without a communication system is just a sentence in a document.

One more continuity detail worth checking on the pre-weld side: under ASME Section IX, a welder’s qualification lapses after six months without welding with the qualified process, unless extended under the code’s continuity rules. Verify it before the welder climbs the scaffold, not after the radiograph comes back.

Visual Examination: The Cheapest Test You Cannot Skip

Visual examination, or VT, is the least glamorous test in pipe welding inspection and the one that carries the most weight. ASME B31.3 requires visual examination on every completed weld, which makes VT the only method with universal coverage in normal process piping work. Radiography might check five percent of joints. Ultrasonic testing might be reserved for critical lines. VT sees all of them, and it costs little more than a flashlight and a trained pair of eyes.

That matters because surface conditions predict internal problems. A wandering cap, deep undercut, or a root that was never cleaned usually hints at something the welder was struggling with, and those struggles rarely stay on the surface.

What a Real Visual Examination Involves

A proper VT is not a glance on the way to lunch. The joint must be clean and free of slag, spatter, paint, and moisture. The inspector needs adequate lighting, a sensible viewing angle, and enough distance to see the whole profile rather than one small patch. Root access matters too. If the root is visible, look at it before the joint is capped or buried under a fitting.

Welding procedures commonly call for at least 100 foot-candles, roughly 1,000 lux, at the examination surface, with more light for fine detail. The exact number belongs in the project procedure, not in a guess. Calibrated gauges turn an opinion into a measurement, which is the difference between “that looks deep” and “that undercut is 1.2 mm.”

Tool	What it measures
Undercut gauge	Depth of metal removed at the weld toe
Reinforcement gauge	Cap height above the parent metal
Hi-lo gauge	Internal misalignment across the joint
Fillet weld gauge	Leg length and throat thickness
Magnifier	Fine surface cracks and toe defects
Borescope or mirror	Root condition in restricted access
Camera	Traceable records and before/after comparison

Before, During, and After the Arc

Before welding, VT is really fit-up inspection: root gap, land, bevel condition, hi-lo, and pipe alignment. During welding, the root pass deserves a look before anyone caps it. Root appearance reflects technique in every position, and rod selection and manipulation show up there quickly, as our all-position welding rod guide explains. After welding, check cap profile, toe tie-in, undercut, surface porosity, spatter, arc strikes, and overall finish.

Acceptance category	Typical treatment
Cracks	Prohibited under both ASME B31.3 and API 1104, regardless of size
Undercut	Limited by depth and length; API 1104 sets separate limits for cover and root bead
Reinforcement	Capped at a maximum; excess creates a stress concentration
Surface porosity	Small isolated indications may pass; clustered or open porosity usually fails
Internal misalignment	Limited as hi-lo; excessive offset is rejectable before strength is even considered

Discontinuity or Defect? The Difference Matters

A discontinuity is a legitimate variation in the weld. It becomes a defect only when it exceeds the acceptance criteria of the governing code or project specification. That distinction explains why the same undercut can pass on a low-pressure water line and fail on a line subject to cyclic service. The geometry did not change. The consequence of failure did.

Commonly Missed Visual Items

• Arc strikes outside the weld zone: these create untempered martensite and a potential cracking site, so they are commonly rejectable.
• Spatter bridging: spatter connecting a pipe to a support or across an expansion gap can restrain movement and initiate cracking.
• Weld splatter on threads: it damages fittings, prevents proper assembly, and can hide a crack beneath it.
• Root suck-back: a concave root reduces throat thickness and is easy to miss once the cap goes on.
• Cap underfill: a low cap may look neat while failing minimum thickness or reinforcement requirements.

None of these require a radiation source or a specialized technician to find. They require time, light, access, and the discipline to look before the next pass covers the evidence.

Volumetric Examination: Radiography, Ultrasonic Testing, and Phased Array

Surface inspection tells you what a weld looks like. Volumetric examination tells you what is hiding inside it. Two methods do nearly all of that work on pipe: radiography, which shoots radiation through the joint and records the shadow, and ultrasonic testing, which pumps high-frequency sound into the steel and listens for the echoes. They are not interchangeable, and choosing between them is one of the most consequential decisions in a welding procedure or an inspection plan.

Radiography: Reading Shadows Through Steel

Radiographic testing uses X-rays or gamma rays and produces an image based on differences in density and thickness through the weld volume. Where the metal is thinner, more radiation reaches the detector and the area darkens; where a slag inclusion or a void sits, the path is less dense and shows up as a distinct shape. The technique is direct and intuitive, which is why it remains the default on so many pipelines.

The source matters. X-ray generators produce adjustable, high-intensity radiation and switch off when de-energized, making them attractive for shop work. Gamma sources such as iridium-192 and cobalt-60 are compact, need no power, and reach tie-ins in the field, but they cannot be turned off — the isotope stays hot and must be cranked back into its shielded camera every time. Iridium is common on thinner walls; cobalt penetrates thicker sections.

Film is still used, but computed radiography and direct digital radiography are now widely used alternatives, offering faster review and digital archiving that can be attached to a weld number. Whichever medium is used, image quality indicators — often called penetrameters — are placed on the part to demonstrate that the technique achieves the required sensitivity. Density, geometric unsharpness, and source-to-object distance all have to be controlled to keep the image readable. The NDE Resource Center’s radiography primer is a good place to see how those variables interact.

Ultrasonic Testing: Sound That Comes Back

Ultrasonic testing sends high-frequency sound into the weld with a probe, typically using shear waves at an angle so the beam crosses the fusion face rather than running parallel to it. A calibration block proves the equipment is behaving, and distance amplitude correction compensates for the fact that echoes weaken with distance, so an indication at 40 mm can be compared fairly against one at 10 mm. The fundamentals of ultrasonic testing explain the physics in more depth.

UT is generally stronger than radiography for detecting planar defects such as lack of fusion and cracks, especially in thicker sections. But it is far more operator-dependent. Surface finish scatters the beam, couplant quality matters, and the technician has to interpret a moving A-scan in real time. The discipline demanded here is closer to prepping aluminum right before welding than most welders expect: a rough, painted, or spattered surface quietly ruins the result. The weldment applications page covers geometry and probe selection well.

Phased Array: Steering the Beam Electronically

Phased array ultrasonic testing uses a probe made of many small elements, firing them in timed sequence to steer and focus the beam without moving the probe. One pass produces a sector scan sweeping through a range of angles, and encoded scanners record positional data so the whole inspection can be replayed and audited later. That replayability is why PAUT has gained ground on pipeline girth welds, thick-wall joints, and crowded racks where access is tight.

Choosing Between Them

Factor	Radiography (RT)	Ultrasonic / Phased Array
Detection strength	Volumetric flaws: porosity, slag, voids	Planar flaws: lack of fusion, cracks
Blind spots	Tight planar defects aligned with the beam; thick sections	Coarse-grained austenitic welds, rough surface, complex geometry
Access needs	Both sides of the joint, exclusion zone	One side, but clean surface and couplant
Throughput	Fast once set up; slower on tie-ins	Slower per joint, faster to re-scan
Safety	Radiation controls, barriers, dosimetry	No ionizing radiation; conventional electrical safety
Record quality	Film or digital image, easy to archive	Encoded scans replayed later

What the Code Actually Requires

Coverage rules decide how much of the work gets examined at all. Under ASME B31.3, Normal Fluid Service piping typically requires random examination of about 5% of circumferential butt and miter groove welds, while Severe Cyclic service requires 100%. Random is not the same as full: a spot shot examines a short window of one weld, so a defect outside that window simply is not seen. Inspectors who treat a passing spot as proof of a sound joint are misreading the intent.

Neither method is a verdict on its own. They produce indications, and a qualified interpreter compares those indications against acceptance criteria before anyone cuts metal.

Surface, Hardness, and Specialized Checks That Catch What RT and UT Miss

Radiography and ultrasonic testing get the attention, but a complete inspection plan leans on a bench of supporting methods. These checks answer questions the volumetric methods never ask: is that crack open to the surface, is this really 316L, is the heat-affected zone soft enough after heat treatment, and will the finished system actually hold pressure? Skip them and you can pass every radiograph on a spool and still fail turnover.

Penetrant Testing: Finding What Breaks the Surface

Liquid penetrant testing (PT) relies on capillary action. Clean the surface, apply a coloured dye or fluorescent penetrant, let it dwell, wipe the excess, then apply developer. Any penetrant trapped in a surface-breaking discontinuity bleeds back out and shows up as a vivid indication. Fluorescent penetrant under UV light is far more sensitive than visible dye, which is why it dominates critical stainless and nickel-alloy work.

The whole method lives or dies on cleaning. Scale, paint, oil and machining smears can bridge a crack and hide it completely, and the dwell times in the procedure exist for a reason. PT only works on nonporous materials, and by definition it cannot detect a subsurface flaw — it is a surface examination, not a substitute for RT or UT.

Magnetic Particle Testing: Carbon Steel’s Surface Specialist

Magnetic particle testing (MT) magnetises the part and dusts it with iron particles, wet or dry. Flux leakage at a discontinuity attracts the particles and draws a visible indication. It finds surface and near-surface cracks very effectively, and it is faster and cheaper than PT on large carbon steel joints. The catch is the ferromagnetic limitation: MT works on carbon and low-alloy steels, not on austenitic stainless, aluminium or duplex.

Field direction matters as much as the equipment. A crack is only detected when it cuts across the magnetic flux lines, so the procedure requires two shots roughly 90 degrees apart — typically a yoke in two positions, or a prod and coil combination. One direction only is a common and quiet way to miss a longitudinal crack.

Hardness Testing: Proof That Heat Treatment Worked

Portable hardness testing uses rebound (Leeb) or ultrasonic contact impedance (UCI) instruments to give a quick reading on a weld cap, HAZ or parent metal. It is used as a proxy for post-weld heat treatment effectiveness: if the HAZ is still hard, the PWHT did not do its job, and the joint is a cracking candidate. In wet H2S service, ASME B31.3 and NACE MR0175/ISO 15156 drive the limits, with carbon steel commonly capped at 22 HRC. Surface preparation and curvature both affect readings, so a grinder pass and a calibrated instrument are not optional.

PMI and Ferrite: Catching the Wrong Metal

Positive material identification uses X-ray fluorescence (XRF) or optical emission spectroscopy (OES) to verify alloy chemistry. It is the cheapest defence against mix-ups between carbon steel, stainless and duplex, and it belongs on every alloy component that arrives on site — fittings, flanges, valves and filler wire. Duplex work adds a second check: ferrite content in the weld metal is measured because an out-of-range ferrite level hurts both corrosion resistance and toughness in opposite directions. Too little ferrite invites hot cracking and stress-corrosion issues; too much reduces toughness and phase balance.

Pressure Testing and the Rest of the Plan

Hydrostatic testing at 1.5 times design pressure is the classic final proof, with water as the medium precisely because it stores so little energy. Pneumatic testing stores vastly more energy at the same pressure, so it requires calculated standoff distances, exclusion zones and written controls — the kind of hazard OSHA’s welding and cutting standards expect you to plan for, not improvise. Vacuum box testing covers the gap where positive pressure is impractical: soap solution over a lap weld or tank bottom, a box with a transparent window, and a vacuum that pulls air through any leak.

Where the Technology Is Heading

Automated ultrasonic testing is now routine on pipeline spreads, digital radiography detectors are replacing film at a steady clip, and inspection data platforms are starting to flag weld-level trends — a welder whose repair rate climbs on one procedure, or a joint type that keeps generating the same indication. The methods are old; the ability to learn from them is new.

Reading the Weld: Defect Types, Causes, and Accept/Reject Decisions

An indication is not a defect. That single sentence separates a technician from an inspector. The technician reports what the film, the screen, or the probe revealed — a dark smudge, a blip, a shadow. The inspector decides whether that revelation matters, and the answer depends on what the flaw is, where it sits, how it is oriented, and which clause of which code governs the joint. Get that decision wrong in either direction and you either scrap a sound weld or accept a leak waiting for a hydrotest.

Defect Types: Appearance, Cause, and Typical Disposition

Most rejectable conditions on pipe welds fall into a short list. Learn the list well enough to name the cause, and the disposition usually names itself.

Defect	What you see	Usual cause	Typical disposition
Porosity / clustered porosity	Rounded dark spots on film; rounded indications on UT	Moisture, grease, paint, low amperage, long arc, bad gas coverage	Evaluate against size and count limits; clusters usually reject
Slag inclusion	Irregular elongated dark lines, often at pass boundaries	Poor interpass cleaning, wrong travel angle, low heat input	Compare to length and width limits; remove if exceeding
Incomplete fusion	Straight-sided planar indication at sidewall or pass line	Cold lap, travel too fast, insufficient penetration into the joint face	Reject; planar and crack-like
Incomplete penetration	Dark line along the root, straight and continuous	Root gap too tight, land too thick, insufficient root heat	Reject where full penetration is required
Root concavity / suck-back	Hollowed root profile, dark wide band on film	Excess purge, too much root heat, gravity on overhead root	Measure depth; reject beyond code limit
Icicles / burn-through	Drooping metal inside the bore, bright blobs on film	Overheating the root, excessive amperage, slow travel	Remove; burn-through usually rejects
Undercut	Groove melted into base metal at toe or root	Arc length, angle, travel speed, wrong polarity	Measured against depth and length limits; service-dependent
Overlap	Rolled-over metal at the toe without fusion	Low current, slow travel, poor angle	Reject; it is an unfused notch
High-low misalignment	Offset between internal surfaces	Fit-up, ovality, mismatched wall thickness	Compare to code alignment limit; often a fit-up fix, not a weld fix
Longitudinal / transverse cracks	Sharp, jagged, branching dark lines	Hydrogen, restraint, hot shortness, poor bead shape	Reject regardless of length
Crater crack	Star-shaped crack at arc stop	No fill on the crater, abrupt arc break	Reject; remove and repair
Hydrogen-induced delayed cracking	Crack in the heat-affected zone appearing hours to days later	Hydrogen from moisture, grease, or cellulosic electrodes meeting a hard HAZ	Reject; requires removal, re-weld, and process change
Arc strike	Small localized melt spot away from the weld	Striking on the pipe surface instead of a tab	Often reject or require dressing and PT on pressure parts

The Metallurgy Behind the Crack You Find Tomorrow

Hydrogen is the quiet one. Moisture in the electrode coating, grease on the wire, damp flux, or a humid night gets hydrogen into the arc, and from there into the weld metal and the heat-affected zone. On a hard, fast-cooling HAZ — high-carbon equivalent, thick wall, cold weather, no preheat — that hydrogen has nowhere to go. It collects, pressure builds, and a crack opens hours or days after everyone has gone home. This is why some project specifications require a delay before final inspection of high-strength or sour-service welds, and why low-hydrogen consumables and holding ovens are not paperwork theatre. The electrode selection principles covered in our guide to alloy steel welding electrodes apply directly here.

Why Orientation Decides What the Film Shows

Radiography is a shadowgraph, and a shadow only forms when the flaw has thickness across the beam. A tight planar lack of fusion lying parallel to the radiation beam may barely register, which is exactly why two methods exist. Angle-beam shear-wave ultrasonic testing crosses the joint at an angle and reflects off planar flaws that RT would miss — the mechanism is well described in these resources on ultrasonic testing of weldments and radiograph interpretation. It also explains why “it passed RT” is not the same claim as “it has no planar defects.”

Workmanship Criteria Versus Fitness-for-Service

Two philosophies sit underneath every accept-or-reject call. Workmanship criteria say a weld must look and measure a certain way, full stop — that is the logic of ASME B31.3 Table 341.3.2, which sets limits for cracks, undercut, porosity, incomplete fusion, incomplete penetration, misalignment, and reinforcement. Fitness-for-service asks a different question: will this flaw grow and fail in this service? A client specification can be stricter than the code and frequently is, so a defect that passes B31.3 may still be rejected on a project that forbids any undercut on cyclic service. API 1104 adds another layer, judging individual defects and the aggregate defect length within a defined weld length — a joint can fail on total length even when no single flaw looks alarming.

Three Decisions From the Field

1. The isolated rounded indication. A single small rounded indication well inside the porosity size limit, with no clustering, on a normal fluid service line. Disposition: accept, record it, and move on. Chasing it with a grinder costs more than the flaw does.
2. The crack. Any crack — crater, transverse, longitudinal, or HAZ — is rejectable under both ASME B31.3 and API 1104, with no size-based tolerance. Length does not rescue it. Remove to sound metal, verify by PT or MT, re-weld to a qualified procedure, and re-examine.
3. The undercut. Shallow undercut on a low-pressure water line may sit comfortably inside the code limit and be accepted. The same depth on a line with thermal cycling concentrates stress at the notch and gets rejected by specification. Same measurement, different verdict.

The best inspectors are not the strictest ones. They are the ones who can say, on the record and in plain language, what the flaw is, why it formed, and which clause makes it acceptable or not. That reasoning is what turns an indication into a defensible decision.

QA/QC in Practice: Procedures, Hold Points, and the Paperwork That Proves It Happened

In pipe welding inspection, a weld nobody can document is, to the owner, a weld that never happened. Ask for the radiograph on spool 14-B in a turnover meeting and watch the room go quiet. The joint may be perfect; without a paper trail, it is unproven.

The quality system turns good welds into an accepted system. It is built before the first arc, maintained during fabrication, and closed out after the last cap pass cools.

Documents That Must Exist Before Welding Starts

• Welding Procedure Specification (WPS): the recipe: process, filler, current range, preheat, position and technique.
• Procedure Qualification Record (PQR): the test evidence behind that recipe. No PQR, no WPS.
• Welder Performance Qualification (WPQ): proof the individual welder can produce sound welds to that procedure, in that position, thickness and diameter.
• Inspection and Test Plan (ITP): the map of what is checked, by whom, at which stage, with what record.
• Project Quality Plan: the umbrella document covering organisation, responsibilities, document control and reporting.

For process piping, these documents must satisfy ASME B31.3. The ITP translates that code into daily field activity.

The ITP Is a Living Document, Not a Poster

An ITP defines hold points, witness points and review points. A hold point stops work until the inspector signs off; welding past it is a serious nonconformance. A witness point invites the inspector but allows work to continue if they do not attend. A review point is a records check after the fact. A typical pipe spool ITP looks like this:

Stage	Activity	Point	Record
Pre-fab	WPS/PQR/WPQ verification	Hold	Document review sheet
Pre-fab	Material and heat number check	Hold	Material receiving report
Fit-up	Root gap, hi-lo, alignment	Hold	Fit-up checklist
Welding	Root pass before capping	Witness	Inspector log
Post-weld	Visual examination	Hold	VT report
Post-weld	RT / UT per coverage	Hold	NDT report
Post-weld	PWHT and hardness	Hold	Chart and hardness sheet
Close-out	Pressure test	Hold	Test certificate

The plan gets revised as the job changes. An ITP untouched since mobilisation usually means nobody is using it.

Nonconformance Reports: Raise, Disposition, Close

When a condition deviates from specification, raise an NCR. It needs a documented disposition: repair, replacement or use-as-is with engineering approval. Then it needs verified closure: the repair re-examined by the method that found the problem, or the replacement traced into the records.

An open NCR is a hook in the system. Most owners will not grant mechanical completion or accept a turnover package while one is unresolved, because accepting the package means accepting its risk.

Traceability: The Chain That Makes a Record Believable

Traceability links every joint to its WPS, welder, consumables, heat numbers and examination records. In practice that means a weld map with numbered joints; a weld log showing who welded what and when; heat numbers transferred from mill certificates onto pipe and fittings; consumable lot numbers recorded against the joint, the discipline covered in guides on alloy steel welding electrode selection and use; and welder stamps that match the log.

NDT records must join the chain. A radiograph with no weld number is close to worthless. Digital radiography and encoded phased array files carry metadata that can be stored, replayed and audited years later, which is why they are displacing film on major projects.

Turnover Packages and Digital Records

A turnover package typically holds the document index, WPS/PQR/WPQ set, welder list, weld log and map, material certificates, NDT reports, PWHT charts, hardness and PMI results, pressure test records, NCR log and as-built drawings. Paper packages get lost, photocopied badly and amended without trace. Digital systems with audit trails record who changed what and when, exactly what an auditor will ask five years later.

Welder Continuity and Repair-Rate Trends

Under ASME Section IX, a welder’s qualification lapses after six months without welding with the qualified process, subject to the code’s continuity provisions. Track expiry actively. A lapsed welder is unqualified no matter how good the bead looks.

Repair-rate tracking by welder and procedure is a leading indicator of process control problems, not just individual skill. One welder with a high rate needs coaching. A whole crew failing on one WPS usually means the procedure, consumables or fit-up is wrong. Use the data for training, such as a structured welding course or internal refresher, not punishment, or people will hide defects.

Audit Your Own System First

Before a client arrives, pull ten random weld numbers and trace each end to end: WPS revision, welder qualification currency, consumable lot, heat number, NDT report, NCR closure and final test record. If a link breaks, fix the system before someone else finds it.

Costs, Common Mistakes, and Practical Tips for a Smoother Inspection

Ask a project manager what pipe welding inspection costs and you will usually get the price of a radiography crew and a few days of an inspector’s time. Ask the same person after a bad month and the number looks completely different. The film, the technician and the report are minor compared with what follows a rejected joint on a live plant tie-in: scaffold that has to go back up, insulation cut away and reinstalled, a radiation boundary that shuts down three other crews, and days of schedule evaporating while a welder cuts out pipe he finished last Tuesday.

That is the money conversation most welding articles skip. The dominant cost of a failed weld is almost always rework and schedule delay, not the non-destructive examination itself. Every dollar moved from rework into prevention tends to come back with interest.

The Cost Drivers Nobody Puts in the Estimate

On new construction in a fab shop, inspection is cheap and predictable. On brownfield work, it is a logistics problem wearing a lab coat.

• Access and scaffolding. Getting an inspector and a radiation source to a weld at 40 feet usually costs more than the examination.
• Insulation removal and reinstatement. A single re-test may mean destroying and rebuilding finished cladding and thermal protection.
• Radiation exclusion zones. Radiography stops every trade inside the boundary. On a busy deck that is measured in lost crew-hours, not in film.
• Out-of-position radiography on tie-ins. Shooting a joint in a congested rack takes far longer than a shop shot, and often requires a source positioned by rope access.
• Repair cut-outs and re-welding. Removing a defect with carbon arc gouging and re-welding means re-inspection, re-heat treatment and a second chance for the same mistake.
• Schedule delay. The expensive part of a reject is rarely the weld. It is the sequence of downstream work waiting behind it.
• Documentation rework. Rebuilding traceability after the fact can absorb weeks of engineering time.

Note how few of those line items appear on an NDT quote. That is the point.

Ten Mistakes That Cost the Most

1. Welding without an approved WPS on the floor, signed and current.
2. Wrong filler or the wrong consumable lot — often the root cause behind a run of unexplained rejects.
3. Skipping root inspection before the cap goes on, which buries the evidence.
4. Poor fit-up and excessive hi-lo, which no amount of welder skill can fully recover.
5. Welding in wind or rain without adequate protection, inviting porosity and hydrogen pickup.
6. Ultrasonic testing over painted, scaled or spattered surfaces — surface condition directly affects UT quality, so the coating has to come off first.
7. Radiographing the wrong percentage or the wrong joints, then discovering the gap at turnover.
8. Post-weld heat treatment skipped or run incorrectly, which shows up later as hardness failures and delayed cracking.
9. No weld map, leaving radiographs that cannot be matched to a specific joint.
10. Treating a failed test as a paperwork problem instead of a process problem.

Practical Tips by Role

For welders. Clean your root before capping and look at it with a light. Use the gauge instead of your eye. Protect the joint from rain, grit and grease between passes. Report a bad fit-up before you strike an arc, not after — the fitter’s problem is far cheaper to fix than yours. Choosing the right rod for the position also matters more than most crews admit; an all position welding rod guide is worth ten minutes of reading.

For supervisors. Hold a pre-inspection meeting that confirms acceptance criteria, coverage percentages and hold points before the first weld — it prevents most later disputes. Run a mock-up on unfamiliar joints. Keep gauges calibrated and available. Control the weather instead of arguing with it. And watch repair rates as a process indicator: a rising rate usually points to a procedure, consumable, fit-up or training issue rather than random bad luck.

For inspectors. Define acceptance before welding starts, in writing, referencing the governing code such as ASME B31.3 and the project specification. Photograph everything, including the joints that pass. Write findings the same day, while the details are still true. Know the surface preparation limits for each method, as described in this nde-ed overview of ultrasonic weldment testing, and refuse to shoot through paint when the procedure forbids it.

Conclusion: Building Inspection Into the Weld Instead of Onto It

That thin bead of water along the girth weld did not appear because someone was careless with a torch. It appeared because a chain of small, ordinary decisions went unwitnessed. A root pass capped before anyone looked at it. A fit-up with more hi-lo than the procedure allowed. A radiograph interpreted against criteria nobody had written down before the arc was struck. A mature inspection culture would not have needed a hydrotest to find that joint. It would have found it at 8 a.m., with a mirror and a flashlight, for the price of ten minutes.

That is the whole argument of this guide, compressed into five working rules:

• Know your governing code before the first arc. Codes are revised on cycles, and the edition named in your contract governs the job, not the newest one on the shelf. Confirm the ASME B31.3 edition, project spec, and any owner addenda in writing.
• Control fit-up and the root pass. Most rejectable conditions are born in the first two inches of travel, long before a technician arrives with a detector.
• Choose the examination method for the defect you fear. Rounded volumetric flaws suit radiography; tight planar lack of fusion usually needs shear-wave ultrasonic testing. Surface-breaking cracks want penetrant or magnetic particle.
• Document as if the file will be audited in five years. It often is. Inspection records are frequently retained for the life of the piping system and pulled during audits, incident investigations, and turnarounds.
• Treat every failed test as process feedback. A rejected joint is data about preheat, fit-up, consumable storage, or technique. Fix the process and the rejections stop repeating.

Your 30-Day Action Plan

Nothing here requires a new department or a bigger budget. Start with the next job on the board:

1. Pull the governing specification and highlight every examination requirement, percentage, and hold point. One page, printed, in the gang box.
2. Review the WPS and welder qualifications side by side. Confirm process, position, diameter, thickness, and backing all match the joint being made.
3. Build a one-page ITP checklist for your most common spool type — fit-up, root, cap, NDT, test, sign-off. Keep it where the work happens.
4. Calibrate and organize your gauges. Undercut, reinforcement, hi-lo, and fillet gauges in one case, with current calibration stickers.
5. Set up a weld log template that links weld number, welder stamp, procedure, filler lot, and examination result. Future you will be grateful.

If You Want to Move From the Torch to the Clipboard

Plenty of good welders make excellent inspectors, because they already know what a bad root feels like through the stinger. The AWS Certified Welding Inspector credential remains one of the most widely recognized inspection qualifications in North America. ASNT Level II and Level III certifications are built around method-specific training, experience, and examination requirements. For pipeline work specifically, API 1169 is aimed squarely at construction inspectors. Get the field time first, though — classroom theory lands much harder once you have cut out your own defect. Welders who also work aluminum will find the same discipline of preparation and traceability in our aluminum MIG welding guide.

So: Obstacle or Advantage?

It is an advantage, and not a subtle one. Inspection is the mechanism that lets a welder’s work survive decades of pressure, thermal cycling, and corrosion — the record that says this joint was made by a qualified hand, to a qualified procedure, and was examined honestly. The hydrotest leak was never the inspector’s fault alone. It was what happens when examination is bolted on at the end instead of built into the weld from the first tack. Build it in, and the gauge holds steady.

Frequently Asked Questions

Is pipe welding inspection legally required?

It depends on the application and jurisdiction. Pipeline work in the United States is regulated under 49 CFR Parts 192 and 195, which reference welding standards such as API 1104, and workplace safety obligations come from OSHA. Process piping is usually driven by contract and adopted codes such as ASME B31.3 rather than a single federal inspection law.

What is the difference between examination and inspection in ASME B31.3?

Examination is the physical work of looking at or testing the weld, performed by the manufacturer or constructor with qualified personnel. Inspection is the owner’s oversight function, carried out by the owner’s inspector, which may include witnessing examinations and reviewing records. Keeping the two roles separate is what makes the documentation credible.

How much non-destructive testing is required on pipe welds?

For ASME B31.3 Normal Fluid Service, volumetric examination is typically random and sample-based, commonly around 5% of circumferential butt welds, while Severe Cyclic service requires 100%. Project specifications frequently raise these percentages. Always confirm the coverage in the contract, because the specification governs when it is more restrictive than the code.

Should I use radiography or ultrasonic testing?

Radiography is strong for volumetric defects such as porosity and slag and produces a permanent image that is easy to review. Ultrasonic testing, especially phased array, is generally better for planar defects like lack of fusion and cracks, thick walls, and joints where radiation cannot be used safely. Many critical projects use both.

Can a welder inspect their own welds?

A welder can and should perform visual checks on their own work, but code-required examination is normally performed by qualified personnel who are independent of the welding being examined. Project specifications commonly prohibit self-inspection for acceptance records.

What is the most common reason a pipe weld fails inspection?

Incomplete fusion, incomplete penetration, porosity and slag are among the most frequent rejectable conditions, and most trace back to fit-up, travel speed, current, cleaning between passes, or contaminated consumables. Cracks are less common but always rejectable and usually point to hydrogen, restraint or procedure problems.

Does a failed weld always have to be cut out?

No. Many defects can be removed by grinding and re-welded, provided the repair procedure is qualified and the repaired area is re-examined with the same method. Cracks and defects that extend deeply into the wall often require cutting out the joint, and the disposition should be documented on a nonconformance report.

What certifications should a pipe welding inspector hold?

Common credentials include AWS Certified Welding Inspector, ASNT Level II or III in the relevant NDT methods, API 1169 for pipeline construction inspection, and CSWIP or ISO 9712 certifications on international projects. The right combination depends on the code and the client’s specification.

Does paint or coating affect weld inspection?

Yes. Paint, scale, spatter and oxide interfere with visual examination, penetrant testing and ultrasonic testing, and couplant contact can be compromised. Most specifications require the weld and adjacent heat-affected zone to be clean and bare before examination, and coatings applied before inspection may need to be removed.

How soon after welding should a pipe weld be inspected?

Visual examination is usually performed as soon as the weld is cool enough to inspect safely, and the root should be checked before capping where access allows. Volumetric examination timing follows the specification, and high-strength or sour-service joints may require a delay of hours or days to allow hydrogen-induced cracks to develop before final testing.

Approach	Relative cost