Phased array flaw detectors (PAUT)

Choosing a phased array flaw detector

What an instrument can do is set first of all by its channel configuration, written as receive channels over transmit channels. A 32/128 unit drives a 128-element probe but fires 32 elements at a time. More receive channels mean a larger active aperture, so deeper focusing and fewer probe positions on the weld. The second number is the pulse repetition frequency: it sets how fast the scanner may travel before data lines are lost.

The whole Doppler family is below. Every model records encoded data and builds C-scans; they differ in how far the configuration goes and how much imaging they compute in real time.

When TFM is actually needed

Conventional phased array sums the signals inside the instrument, following focal laws set before the inspection. FMC records every transmit–receive combination separately: a 64-element probe yields 4096 waveforms at each position. Only then does the TFM algorithm reconstruct an image focused at every point of the grid, instead of at one chosen depth.

The gain is real for small discontinuities, coarse-grained materials and geometries where the echo returns from an awkward angle. So is the cost: processing that much data slows scanning down. On pipeline girth welds the zonal technique runs at 50–100 mm/s, while FMC/TFM systems typically work at 25–30 mm/s (figures from a published paper on pipeline AUT systems). In practice TFM is switched on where ordinary PA leaves doubt, not instead of it.

TFM imaging, in the FMC-TFM and PWI-TFM variants, is available on every Novascan we sell: Novascan L3, Novascan V1, Novascan V2 and Novascan V3. PCI goes one step further, building the image from phase coherence instead of amplitude — it pulls small, awkwardly oriented flaws out of material noise. FMC-PCI and PWI-PCI are available on Novascan V3 only, the newest and most capable model in the family. Phascan II computes TFM as well. FLEXSCAN works in phased array, TOFD and conventional UT — a conventional UT channel being something every instrument on this list has.

TFM or PCI — what each image shows

Both techniques compute the image from the same data: the matrix of raw A-scans captured with FMC or PWI. FMC fires the elements one by one and records every combination — 4096 waveforms with a 64-element probe. PWI sends a plane wave at several angles instead, so acquisition is several times faster and the image refreshes more often. TFM and PCI only part company at the processing stage: TFM computes the amplitude of the summed signal for each pixel, PCI computes the phase coherence factor and ignores echo height altogether.

How this works in practice

The techniques do not compete, they complement each other. TFM is what goes into the acceptance report: it has standardised criteria and renders the joint geometry faithfully. PCI runs on the same data as a probing tool — wherever the flaw scatters the wave instead of reflecting it: porosity, early stages of high-temperature hydrogen attack, branched stress corrosion cracking in austenitic steel. On lack of sidewall fusion, TFM shows the reflection off the fusion face while PCI brings out the diffraction signals from both of its tips, which makes height sizing easier.

Instruments that compute both also build a hybrid image: TFM amplitude weighted by the PCI coherence factor. Evaluation stays amplitude-based and standard-compliant, while structural noise in coarse-grained material fades out. In our range that combination comes from Novascan V3 — the only model with PCI — and from the ROBUST F1 board in automated systems.