PAUT, TOFD or radiography for weld inspection?

All three look inside the wall rather than at its surface. That is where the similarity stops. Radiography measures how much radiation got through. Ultrasonic methods measure what sound bounced off, or diffracted around. Those are different physics and they fail differently.

Radiography — the one everybody trusts

Radiation passes through the weld onto film or a digital detector. Anything that changes the amount absorbed shows as a density change: a void, a slag inclusion, a gap.

What it is genuinely good at: volumetric flaws. Porosity, slag, lack of fill, excess penetration, burn-through. These have volume, they absorb less radiation than steel, and they appear plainly. It also produces a permanent image that a third party can review years later, which is why clients like it and why it survives in specifications.

What it misses: tight planar flaws that are not aligned with the beam.

This is the crucial limitation and it is worth being precise about it. A crack is a gap of a few microns. For the radiation to notice it, the beam has to travel along the gap so that the missing material accumulates over the path length. A crack lying across the beam presents almost no extra path through air and produces almost no density change.

The other cost is operational. Radiography means a radiation source and an exclusion zone, which on a live unit usually means clearing the area and working nights. That cost often exceeds the inspection itself.

Phased array ultrasonics — the one that sizes

An array of small elements fired with controlled delays so the beam can be steered and focused electronically. The result is a real cross-sectional image of the weld volume.

What it is good at: planar flaws — cracks, lack of fusion, lack of penetration — and, critically, giving them a through-wall height. That number is what a fitness-for-service assessment under API 579 Part 9 needs, and radiography cannot supply it at all.

Because the beam sweeps through a range of angles, it covers flaw orientations that a single fixed-angle probe would miss. It is also fast once set up, encoded for repeatability, and needs no exclusion zone — so it runs alongside other work.

What it needs: a competent operator and a procedure written for the flaws you expect. A beam angle wrong for the flaw orientation finds nothing, and the equipment will report that cheerfully as a clean weld. PAUT is more capable than radiography and more dependent on the person running it.

It also needs a scanning surface: reasonably smooth, accessible, with couplant. Rough as-welded caps and awkward geometry both degrade it.

TOFD — the one that measures height accurately

Two probes straddle the weld, one transmitting and one receiving. Instead of listening for a reflection off the face of a flaw, TOFD listens for the weak signal diffracted from its tips. The time between the tip signals gives the through-wall height directly.

What it is good at: accurate height sizing, and it is largely insensitive to flaw orientation — because tips diffract in all directions, unlike faces, which only reflect back if you hit them squarely. For monitoring whether a known flaw is growing between outages, nothing beats it.

What it misses: flaws in the dead zones. There is a band at the scanning surface where the lateral wave swamps the signal, and another at the back wall. Those zones are typically a few millimetres each, and surface-breaking flaws hide in them.

Which is why TOFD is almost never used alone. The standard combination is TOFD for the body of the weld plus PAUT or a surface method for the near-surface band.

Side by side

How to choose

New fabrication, workmanship acceptance

Radiography is defensible and often specified. The flaws it finds well — porosity, slag, lack of fill — are the ones workmanship criteria are written around, and the image satisfies a client who wants to see it.

Anything where cracking is credible

Ultrasonic. If the service produces environmental cracking, if the weld is thick, if a lack of fusion would be consequential, radiography is the wrong primary method. Many modern codes now permit or prefer PAUT plus TOFD in place of RT for exactly this reason.

In-service inspection of an operating plant

Ultrasonic, almost always. No exclusion zone means the inspection happens without clearing the area, and the flaws you are hunting in service — SCC, fatigue, HIC — are the planar ones.

A flaw you have found and now have to assess

PAUT and TOFD together. Part 9 needs length, height and position. Radiography gives you a shadow; it does not give you a number you can put into an assessment.

A flaw you are leaving in and monitoring

Encoded TOFD. Repeat the same scan next outage and compare. This is the most sensitive way to answer whether a flaw has grown, which is the question that decides whether it stays.