Radiographic interpretation is a skill built on pattern recognition, and like all pattern recognition it looks like intuition from the outside and is actually structure from the inside. Every common weld defect has a characteristic appearance on a radiograph, and that appearance follows from the physics of how it absorbs radiation and the geometry of how it formed.
This is a guide to the defects you will meet most often, what they look like, why they form, and what typically decides whether they are acceptable.
First, why anything shows up at all
A radiograph is a shadow. Radiation passes through the component and darkens the film or the detector in proportion to how much gets through. Less material in the path means more radiation arrives, means a darker image.
That single fact governs interpretation. Nearly every weld defect is a void or an inclusion of lower density than the parent metal, so nearly every weld defect appears darker than the surrounding weld. The exceptions — tungsten inclusions being the obvious one — are denser than steel and appear lighter, which is why they are so instantly recognisable.
It also explains the biggest limitation of the method. A defect shows up in proportion to how much material it removes along the beam path. A tight planar crack lying perpendicular to the beam removes almost no material along that path and can be invisible. The same crack, with the beam aligned along its plane, shows clearly. Radiography is excellent at volumetric defects and orientation-dependent on planar ones — which is exactly the opposite of ultrasonics, and why the two methods complement rather than replace each other.
Porosity
Appearance: rounded dark spots, sharply defined, varying in size. May be isolated, clustered, distributed through the weld length, or aligned in a row. Linear porosity aligned along the weld root is a distinct sub-type.
Cause: gas trapped in the solidifying weld pool. Usually shielding gas problems — a draught blowing the shield away, flow rate too low, a leaking gas line — or contamination on the joint face: moisture, oil, rust, paint. Damp electrodes are a classic source in manual metal arc.
Interpretation notes: porosity is the friendliest defect to interpret because its shape is unambiguous. Rounded, sharp-edged and dark almost always means gas. The judgement is in quantity and distribution rather than identification, and most acceptance criteria treat clustered or aligned porosity more severely than scattered porosity of the same total volume, because clusters concentrate the section loss.
Slag inclusions
Appearance: dark, irregular in shape, usually elongated along the weld direction, with less sharply defined edges than porosity. Frequently found in lines along the fusion face or between passes.
Cause: slag from a previous pass not fully removed before the next pass was laid, or slag flowing ahead of the arc and becoming trapped. Almost exclusively a process defect in slag-producing processes — MMA and submerged arc.
Interpretation notes: the irregular outline is the tell. Where porosity is round and crisp, slag is ragged. Elongated slag lines running parallel to the weld can be mistaken for lack of fusion; the distinguishing feature is usually that slag has visible width and internal density variation, while lack of fusion is a tighter, more uniform line.
Lack of fusion
Appearance: a dark line, usually straight or gently curved, following the fusion face between weld metal and parent material, or between passes. Sharper and narrower than a slag line. Often shows only along part of the weld length.
Cause: insufficient heat input, incorrect electrode angle, travel speed too high, or a joint preparation that shielded part of the fusion face from the arc. The weld metal has touched the parent material without melting it.
Interpretation notes: this is a planar defect, and therefore the one radiography is most likely to miss. Side-wall lack of fusion in a narrow-gap or steep-sided preparation lies close to parallel with the beam only if the beam is angled to match, which a standard single-wall single-image shot is not. If side-wall fusion is your principal concern — and in thick-section narrow-gap welding it usually is — ultrasonics, and specifically phased array, is the more reliable method.
Severity: almost all codes treat lack of fusion as a planar defect and therefore severely. It is rarely acceptable at any size.
Lack of penetration
Appearance: a dark, straight, continuous line running centrally along the weld — following the root rather than the fusion faces. Usually of very consistent width, which is what distinguishes it.
Cause: the root of the joint was not fully filled. Root gap too small, root face too large, current too low, or travel speed too high. In single-sided welds without a backing, it is a root technique problem.
Interpretation notes: the giveaway is the straightness and the position — dead centre, running with the weld. Its regularity distinguishes it from a slag line, which wanders. Because the root is often the highest-stressed region and lack of penetration produces a sharp notch there, it is treated as a serious defect in almost every code.
Cracks
Appearance: a dark line, typically finer and more irregular than lack of fusion, often branching, and frequently changing direction. May run longitudinally along the weld, transversely across it, or radiate from a crater at the end of a run.
Cause: depends entirely on type. Solidification (hot) cracking forms as the weld pool freezes, driven by restraint and by impurities segregating to the centreline. Hydrogen-induced (cold) cracking forms hours or days after welding, in the heat-affected zone, and needs hydrogen, a susceptible microstructure and stress together — which is why preheat and low-hydrogen consumables exist. Crater cracks form where the arc was broken without filling the crater.
Interpretation notes: cracks are the hardest defect to find radiographically and the most serious to miss. A tight crack perpendicular to the beam may be completely invisible. Any suspicion of cracking on a radiograph justifies a complementary method — magnetic particle if it is surface-breaking in ferritic material, ultrasonics if it is not.
Severity: cracks are unacceptable. There is no size threshold in normal fabrication codes.
Undercut
Appearance: a dark, irregular band running along the edge of the weld cap or the root, at the weld toe. Density varies along its length, giving a wavy rather than uniform line.
Cause: excessive current, arc length or travel speed, or incorrect electrode angle, melting away the parent material at the toe and not refilling it.
Interpretation notes: undercut is a surface defect, which means visual inspection often finds it first and measures it more reliably than radiography does. Its radiographic signature — position at the toe, variable density — is distinctive, but depth is much better assessed with a gauge than an image.
Severity: normally assessed against a depth limit, sometimes as a proportion of wall thickness. It matters because it produces a stress-raising notch exactly where fatigue cracks like to start.
Tungsten inclusions
Appearance: small, sharply defined, light spots — the only common defect that appears brighter than the surrounding weld.
Cause: the tungsten electrode in TIG welding contacting the weld pool or the filler wire and depositing fragments into the weld.
Interpretation notes: unmistakable once seen, because tungsten is far denser than steel and the contrast reverses. Usually small and localised.
What determines acceptance
Identifying a defect is only half the job. Whether it is acceptable depends on the code the work was made to, and codes disagree with each other — sometimes substantially. A weld that passes ISO 5817 quality level C may fail ASME requirements, and vice versa, because they classify and threshold defects differently.
Two general principles hold across most of them. Planar defects — cracks, lack of fusion, lack of penetration — are treated far more severely than volumetric ones, because they concentrate stress at a sharp tip. And clustered or aligned defects are treated more severely than the same total volume scattered, because the section loss is concentrated.
The practical implication: the acceptance criteria must be agreed in writing before the shoot, not negotiated over an image afterwards. Interpretation against the wrong standard is one of the most common causes of a disputed report.
Where interpretation goes wrong
Three failure modes account for most disputes. Interpreting against the wrong acceptance criteria. Missing planar defects that were badly oriented to the beam and should have been looked for with ultrasonics. And accepting an image whose quality was never adequate — if the IQI does not demonstrate the required sensitivity, nothing seen on that image can be relied upon and nothing unseen can be ruled out.
That third one is the quiet killer, because a poor-quality radiograph looks like a clean weld.
Our radiographic interpretation is carried out by PCN / ASNT Levl 2 and Level 3 qualified interpreters, against the acceptance criteria agreed for your project, on film and digital images including work shot by others. If you have a set you would like a second opinion on, send it over.
