Two words, one line in the sand — and the code draws it. A discontinuity is anything in the weld that interrupts its expected structure. It becomes a defect only when the governing code or specification says it exceeds what the weld can tolerate.
The foundation definition: a discontinuity is any interruption in the expected structure of a weld — a pore, a slag pocket, a toe notch, a crater, a convex bead, a misalignment. Nearly every weld contains discontinuities; most are harmless. Discontinuity is a description of what exists, not a verdict on whether it matters.
The fabrication terminology sharpens it: a flaw is a discontinuity that is relevant to serviceability — one the inspector evaluates. A defect is a discontinuity that exceeds the acceptance criteria of the governing code or specification. The same discontinuity is a defect under one code and perfectly acceptable under another. The code decides.
Discontinuity, imperfection, flaw, defect classification.
The physical evidence never changes — only the ruler does. The same indication is judged by different limits under different documents.
What turns a harmless discontinuity into a defect — the six factors that move the verdict.
The verdict is not a property of the weld — it is a decision applied to the weld by a document. It forms at the intersection of what exists (the discontinuity), what is demanded (the code or specification), and what is at stake (the service). Run the path in this order and the answer stops being an argument:
Inspect by the method the code requires — VT, MT, PT, UT, RT, dimensional.
Name the discontinuity — porosity, slag, undercut, crater, misalignment, crack.
Find the table and the quality level for the type, size, and location you found.
Acceptable, repairable, rejectable, or scrap — per the numbers, not the mood.
The outcome classes are the vocabulary of every inspection report: acceptable — within limit, weld as-is; repairable — beyond limit but fixable by excavation and re-weld within code repair rules; rejectable — must be removed or repaired before acceptance; scrap — beyond repair by the code, no fix permitted. The same discontinuity can land in any class depending on which document governs the contract.
This page exists to stop two reflexes: "any imperfection = reject" and "don't worry, it's fine." Both cost money — and one of them costs structures.
Know the governing document before the weld: which code, which quality level, which tables. Weld to the standard, not to "good enough" — the discontinuities you prevent now are the ones nobody argues about later.
Weld to the limits the code will measure: profile, size, cleanliness, stop technique. The acceptance table is the spec for your technique.
Document. A clean inspection record — dates, methods, results, code edition — beats a debate about interpretation every time.
Both reflexes, on both sides of the table. The inspector who rejects everything acceptable wastes the customer's time and money. The welder who waves everything off risks the structure. The code is the referee — use it.
When a discontinuity is found, the correction is a decision before it is a weld. Run the classification first.
| Step | Action | Why |
|---|---|---|
| 1 | Identify the discontinuity type — by the required inspection method. | You can't check a limit you can't name. |
| 2 | Measure it against the code table — size, depth, location, frequency. | The numbers decide, not the emotion. |
| 3 | Classify: acceptable, repairable, rejectable, or scrap. | Each class has a defined next step. |
| 4 | If repairable — excavate to sound metal, re-weld to the WPS, re-inspect. | Repairs must meet the same standard as the original weld. |
| 5 | If rejectable or scrap — stop, quarantine, and involve engineering. | Re-working a scrap component costs twice and fools no one. |
The reflex judgment is the defect here. Rejecting an acceptable weld wastes money; accepting a rejectable one risks the structure. When there is doubt, the inspector and the engineer — not the loudest voice on the floor — settle it, against the governing document.
Each process produces its signature discontinuities — the ones its welds are judged on.
Porosity and lack of fusion — both governed by size and count limits that vary with quality level. Clean, fast MIG welds rarely leave anything for the code to measure.
Tungsten inclusions, crater issues, and porosity. Small tungsten inclusions are often acceptable; the stop and the crater are where TIG welds get rejected.
Slag inclusion — the classic "the code decides" discontinuity. A single slag line is a rejectable defect; scattered fines may be acceptable. The table, not the photo, settles it.
Slag and porosity, plus profile issues from over-welding. Self-shielded wire adds its own acceptance quirks — check the governing document's treatment of each indication type.
Slag, porosity, and cracking — driven by flux condition. SAW welds are judged by RT more often than any other process, so the limits are always in front of you.
Tungsten inclusions and porosity from worn nozzles or contaminated gas. The narrow, deep weld shape makes location-sensitive limits matter more than usual.
The code's parameters are the quantities it measures. These are the numbers the verdict is built from.
| Parameter | Check | Where the Limit Lives |
|---|---|---|
| Discontinuity size / depth | Measured against the code's table for the type | AWS D1.1, ISO 5817, API 1104 acceptance tables |
| Location | Toe, root, surface, or internal — same size, different verdict | Code tables and figure classifications |
| Frequency / spacing | Count and spacing limits for porosity and inclusions | ISO 5817 levels, project specifications |
| Length | For cracks, slag lines, and lack of fusion | Code tables per joint category |
| Weld profile dimensions | Reinforcement height, convexity, misalignment | Code dimensional requirements |
| Quality level | The severity grading the contract picked | ISO 5817 B/C/D; contract and project spec |
How the common discontinuities are governed:
| Discontinuity | Typical Example | Typical Governing Documents |
|---|---|---|
| Porosity | Gas pockets, pinholes, clusters | ISO 5817, AWS D1.1, API 1104, ASME VIII |
| Slag inclusion | Flux trapped between passes | ISO 5817, AWS D1.1, ASME VIII |
| Undercut | Groove at the weld toe | ISO 5817, AWS D1.1, API 1104 |
| Crater | Underfilled end-of-weld dish | ISO 5817, AWS D1.1, project specs |
| Convexity | Excess weld reinforcement | ISO 5817, AWS D1.1 profile limits |
| Misalignment | Offset plates at the joint | ISO 5817, AWS D1.1, project specs |
| Arc strike | Accidental arc outside the joint | AWS D1.1, project specs |
The discontinuity you produce is a function of the equipment state — the code doesn't care which machine made it.
The same machine, set and fed differently, produces different discontinuities: a worn contact tip escalates porosity, a bad gas path escalates it further, and a damp electrode escalates it past the code limit. Equipment care is discontinuity control — it keeps the indications inside the tables.
The method is chosen by the code, not by preference — and the result only means something against the code's tables.
Acceptance depends on the code and project: AWS D1.1, ISO 5817, API 1104 and project specifications each set their own limits for this discontinuity. Check the governing document before judging a weld acceptable — there is no universal pass/fail rule.
The questions welders actually ask about discontinuity and defect — answered straight.
No. A discontinuity is any interruption in the weld's expected structure — a pore, a toe notch, a crater, a slightly convex bead. It becomes a defect only when the governing code or specification says it exceeds the acceptable limit. Most welds contain discontinuities, and most of those are harmless.
The governing code or specification, applied by the inspector and, where required, the engineer. The welder's eye, the foreman's opinion, and the customer's worry are not acceptance criteria — the code tables are. When there is disagreement, the engineer and the applicable standard settle it.
Yes, routinely. A weld with a row of small surface pores can pass ISO 5817 quality level D and fail level B; a root gap accepted by one project specification can be rejectable under API 1104; an AWS D1.1 fillet may not meet a client's tighter requirements. The verdict belongs to the governing document on the contract.
B is the strictest level — for fatigue, pressure, and high-service conditions. C is intermediate, D is the most permissive, for moderate service. The contract picks the level, and the level sets the discontinuity limits for every indication type — size, depth, count, and location.
No. Porosity is acceptable up to the limits of the governing code — size, count, and spacing per the quality level. A few small pores within the limit pass; clusters or large voids fail. The inspection film and the code table decide, not the word "porosity" on its own.