Optical 3D Scanning in Subsea Weld Repair: A Visualisation Aid, Not an NDT Method

A subsea weld crack repair is one of the higher-consequence interventions an offshore operator will sanction, and it is governed accordingly. Non-destructive testing sizes and clears the crack; ultrasonic thickness gauging owns the remaining wall; the welding procedure specification governs acceptance. None of that is in question here, and none of it is replaced by anything below.

What we want to put on the table is narrower, and we think useful. During the excavation of a crack and the dressing of the groove, the people responsible – the diving supervisor, the inspection engineer, the onshore welding and integrity engineers – work largely without a full-field view of the three-dimensional shape they are creating. They have the diver’s commentary, a single camera looking down a soda-straw field of view, spot measurements and replicas. An underwater optical 3D scanner – an ordinary geomatics tool – can add to that a visual, full-field picture of the visible geometry as the work proceeds.

That is the whole of the claim. This is a visualisation and record aid that sits alongside an NDT-led workflow. It is not non-destructive testing, it does not detect or size cracks, and it produces no number that anyone should act on. We state it plainly up front because it is the single most important thing to understand: the scanner draws a picture; NDT and UT make the measurements.

Who owns what

The boundary only works if it is explicit.

Quantity	Physical nature	Owner
Crack depth and extent in parent metal	Sub-surface	NDT (ACFM / ACPD / ToFD / PAUT)
Remaining wall under the excavation	Sub-surface; back wall unseen optically	UT thickness gauging
Visible groove and cap surface shape	Open surface, where in line of sight	Optical scanning (this aid)

The line we will not cross: a groove depth read optically from the (itself-degraded) outer surface is neither remaining wall nor crack depth. Every safety-relevant number stays with NDT and UT.

Why the output is a picture, not a measurement

It is worth being concrete about why we keep the claim this low, because the reasons are physical, not modesty.

The features that matter most are usually the ones the scanner cannot see. The root and sharp bottom corners of a deep, narrow weld-prep groove – the geometry that governs re-weld quality – self-occlude. Optics needs line of sight and standoff to the feature being measured, and a tight groove root denies both. Those features are handed to visual inspection, replica or NDT; the scanner records the flanks and the open surface it can actually resolve, and nothing more.

Freshly ground steel is a hostile surface for optical scanners. Structured-light and laser scanners want a diffuse, matte surface; bare, just-dressed metal is closer to a mirror, and underwater it is wet and may carry a thin film. Specular returns drop out or bloom. That matters doubly, because the bright ground metal we actually capture behaves nothing like the corroded, rough surface anyone would normally qualify a scanner on.

The water itself degrades the geometry. Refraction at the housing and water interface, a refractive index that shifts with temperature, salinity and depth, turbidity and backscatter, and thermal layering near a hot repair all bend and blur the measurement. Underwater optical metrology is a real discipline, but the honest consequence here is simple: treat any dimension the scanner reports as indicative – a feature of the picture, not a quantity to gauge against a tolerance.

Taken together, these are not defects to be tuned away. They are the reason the output is a qualitative, full-field geometric record – a very good picture of the visible shape – and not a measurement. If a number off the mesh ever appears to matter, that is the cue to go to UT or NDT, not to trust the mesh.

Where it earns its place

Within those limits, the value is real, and it is the kind of value simple tools cannot give.

• Seeing the excavation, full-field. Instead of inferring the developing groove from a narrow camera and a verbal description, the supervisor and the onshore engineers share a three-dimensional view of the visible shape as it develops. On a complex or irregular excavation that is a genuine step up in situational awareness.
• A record of awkward geometry that spot tools handle badly. For irregular, complex or repeatedly-repaired shapes, a full-field mesh is simply a better archival record than a handful of spot gauges, a replica, or a few camera frames. Captured once, quantified-as-visible, and kept.
• Communication and handover. A shared 3D picture travels across shifts and between diver, ROV and onshore team far better than words. It shortens the onshore engineering review and reduces the misunderstandings that creep in over a long repair.
• Lessons and training. The same record feeds procedure improvement and crew familiarisation in a way a written paragraph never will.

None of these needs the data to be acceptance-grade, because none of them is a safety call. They are about seeing, recording and communicating the visible – which is exactly where the tool is strong. The scope follows honestly: this is worth doing on complex, irregular or high-consequence repairs, and where the record has value. On a routine repair with simple geometry, it is not warranted.

Fitting it in without getting in the way

The discipline that keeps this useful rather than disruptive is to stay off the critical path.

Capture at the clean hold points the repair already has – when grinding is complete and the front is clean and still, and again after welding – not while gouging or grinding is live, and never with scanning and active gouging co-located. The safety choreography does not change: the diver and ROV are sequenced as the supervisor dictates, and the scan waits its turn or is skipped.

And it is not free. A scan adds some diver or ROV time, some coordination, and some kit, and on a saturation-diving spread that is expensive, weather- and safety-limited time. If the visualisation ever competes with the repair for the critical path, the repair wins and the scan is dropped. The aid is justified only while it stays subordinate – to the supervisor, to NDT and UT, and to the schedule.

What it is explicitly not

Because the failure mode for any new tool is quiet overclaim, we will say the boundaries directly.

This is not non-destructive testing. It is not a crack detector or a crack sizer. It is not a remaining-wall gauge. It is not an acceptance tool, and it carries no acceptance authority of its own. And it is not an evidentiary measurement: a 3D mesh is editable, processed data, and on its own it does not stand as a metrology record. The crack belongs to NDT, the wall belongs to UT, acceptance belongs to the WPS, and the record of record is the NDT, UT and repair documentation. The scan sits alongside all of it as a visual and geometric supplement, and defers to all of it in every safety-relevant call.

The point: an under-used geomatics capability

The reason for writing this is not to bolt a step onto a weld repair procedure. It is to make a quieter point that the inspection and repair world does not hear often enough: underwater 3D scanning is a mature geomatics capability, and it has uses well beyond the survey and as-built work it is normally booked for. Combined with – never in place of – established NDT methods, it adds a visual and record dimension to a subsea repair that a single camera cannot, at a cost and within limits that are easy to state honestly.

That combination – a geomatics sensing method working in support of non-destructive testing – is the idea worth knowing exists. Used for what it is, and never for what it is not, an optical scanner is a useful extra pair of eyes on a job where seeing clearly is hard and the stakes are high.