Writing a UT procedure that holds up on a BSEE-regulated facility

A procedure is a controlled document, not a technique sheet

The word procedure gets used for two different things offshore, and conflating them causes most of the trouble. One is a technique sheet: probe angles, gain, a sketch, enough for a technician who already knows the job. The other is a controlled quality document that names its referencing code and acceptance criteria, defines its qualified ranges, states every variable that would invalidate it if changed, and carries a qualification record proving it works.

Only the second survives contact with an integrity management system. Under a safety and environmental management system built on API RP 75 and the federal offshore rules, inspection and testing sits inside mechanical integrity, and contractor personnel and documents fall inside contractor management. That means the procedure has a revision number, an approval by a named Level III, a distribution record, and a link to the qualification evidence. A technique sheet emailed to a vessel has none of those.

The practical consequence is timing. Procedure development has to finish before mobilisation, because the qualification demonstration needs specimens, a laboratory or shop, and time. Procedures written during mobilisation get approved on the strength of a signature and nothing else, and that is exactly the document an auditor pulls first.

Essential variables: what freezes the procedure and what does not

Ultrasonic codes divide procedure variables into those that require requalification when changed and those that only require a revision. The essential list runs to material and product form, thickness range, joint configuration and surface condition, couplant, search unit frequency and element size, wedges and shoes, refracted angles, the instrument, the calibration block and reference reflector, scanning directions and extent, the scanning technique, the method of discriminating geometric responses from flaws, the sizing method, data acquisition and scan overlap where automated, and personnel performance requirements.

Offshore, the variable most quietly violated is couplant. A procedure qualified with a proprietary gel is executed subsea or in a wet splash zone with water, or with whatever the vessel has aboard when the drums run out. Acoustic coupling efficiency changes, sensitivity changes, and nothing in the report records it. The second most violated is thickness range: a procedure demonstrated on one wall thickness gets applied across a riser string with several.

The discipline is unglamorous. List the essential variables explicitly in the procedure, state the qualified range for each, and give the technician a pre-scan checklist that maps to that list. Two minutes of checking against a printed range prevents an entire campaign of data being ruled inadmissible months later.

What the referencing code actually demands

The referencing code is chosen by the work, not by the procedure author, and it changes what the same ultrasonic examination means. Pressure equipment and vessels pull ASME Section VIII with the examination performed to Section V Article 4. Pipeline and flowline girth welds pull API 1104, which offers both workmanship criteria and an alternative annex based on fitness for purpose. Fixed structures and their tubular joints pull AWS D1.1 tubular provisions and, for offshore structural fabrication specifically, API RP 2X, which also sets out expectations for how ultrasonic technicians should be qualified for that work.

In-service structural inspection is governed differently again. API RP 2SIM, layered over API RP 2A-WSD, sets survey levels running from an above-water visual through to detailed underwater examination with cleaning and nondestructive testing of selected joints. The ultrasonic work at the top level is not a general survey but a targeted examination of preselected welds, chosen by a structural engineer from a fatigue and consequence assessment. The procedure has to be written for those specific joints.

Mixing codes is the classic error. A procedure that quotes ASME acceptance criteria for a pipeline girth weld, or applies workmanship criteria from one standard while claiming the sizing benefits of another, produces results that nobody can accept. The referencing code should be named on the first page and the acceptance criteria quoted from it verbatim.

Damage mechanisms choose the technique

Upstream and offshore production does not corrode the way refinery process piping corrodes, and a procedure ported from a refinery will look for the wrong things in the wrong places. Wet gas lines suffer top of line corrosion, driven by condensation on the upper internal surface, so damage concentrates near the twelve o'clock position. A thickness grid laid out for the convenience of the technician, typically at the accessible sides and bottom, is capable of returning a full set of healthy readings from a line that is nearly perforated at the top.

Carbon dioxide driven sweet corrosion produces mesa attack and localised flow-induced wastage, and where the weld metal or heat affected zone corrodes faster than the parent metal the result is preferential weld corrosion: a narrow circumferential groove at the root that a spot reading either steps over or clips. Sour service brings hydrogen induced and stepwise cracking, which is mid-wall, planar and parallel to the surface, and therefore missed by an angle beam technique aimed at weld flaws and found only by deliberate straight beam mapping.

Erosion is geometric and predictable. Sand production concentrates wear on the extrados of bends, on tees, and immediately downstream of chokes and control valves, over an arc rather than at a point. Structural steel offshore does not corrode so much as crack, with fatigue initiating at the brace toe of tubular joints under wave loading. Each of these dictates the scan pattern, the probe, and where the examination should physically be performed.

Structural work: node welds, survey levels and flooded member detection

Tubular node welds are the hardest ultrasonic geometry in routine offshore use. The local dihedral angle between brace and chord changes continuously around the intersection, from a very acute angle at the heel to near perpendicular at the toe, so the weld profile, the fusion face orientation and the required probe angle all vary with position around the same weld. A single angle and a single scan direction cannot cover it, and a procedure that specifies one has not understood the joint.

That is why offshore structural practice treats these welds separately, with technician qualification expectations to match. A technician who is competent on plate butt welds has not demonstrated anything relevant to a K joint heel. The qualification specimens should be tubular, they should include the acute region, and the demonstration should be scored on coverage and location accuracy, not only on whether the flaw was called.

Flooded member detection sits alongside this as a screening tool for in-service structures. Detecting water inside a member that was sealed at fabrication is strong evidence of a through-thickness crack, and it is enormously cheaper than cleaning marine growth off every node for close visual and magnetic particle examination. Its limitations belong in the procedure: it will not see part-through cracks, it is meaningless on members that were never sealed or were flooded during installation, and grouted members cannot be assessed this way at all.

What the qualification demonstration has to show

A demonstration proves capability against the specific thing you intend to find. That starts with specimens: production material, production bevel, production surface condition, and flaws of the type, orientation, size and location the damage mechanism produces. Side-drilled holes calibrate sensitivity; they do not demonstrate detection of a fatigue crack at a brace toe or a stepwise crack at mid-wall. If the specimens contain only machined reflectors, the demonstration proved the equipment worked, not that the technique detects real damage.

It must then show coverage. A beam coverage plot or scan plan, worked at the thinnest and thickest ends of the qualified range, showing the root, both fusion faces and the cap being swept, with the physical scanning surface available in production and not the generous access of a workshop bench. Where the production joint has a clamp, a stiffener or a riser guide inside the required skip distance, the demonstration should reproduce that restriction rather than ignore it.

Where the acceptance criteria depend on flaw height, the demonstration must establish sizing accuracy as a number: a measured error against known flaw heights across a set of specimens, with the tolerance carried forward into the criteria. And it should record misses. A demonstration that reports only successful calls has thrown away the most useful information it generated.

Findings that recur when this procedure is audited

The first is a procedure written to the wrong code, or to a code the client never invoked, usually because it was copied from a previous project. The second is a qualification record that does not match the procedure it supports: different thickness, different bevel, different probe, sometimes a different revision entirely. The third is calibration blocks with no traceability, or blocks whose material does not acoustically match the production pipe, with no transfer correction requirement anywhere in the document.

Then come the operational findings. Calibration interval checks missed across a shift change, with no re-examination of the intervening work. Scan speeds or encoder resolutions exceeded on automated setups so the data density falls below what the procedure assumed. Temperature compensation absent on warm flowlines. Reports listing indication length only, under acceptance criteria that require height. And technicians qualified in the method generally, with no record of a demonstration on the geometry they were scanning.

Two document-control findings appear almost every time. Field copies of a superseded revision still in the container, and no archive of raw automated ultrasonic data, so a disputed call cannot be re-analysed. Both are trivially preventable and both are treated seriously, because they undermine every result the campaign produced rather than a single weld.

Getting a procedure written that survives the campaign

The work runs in a fixed order. Establish the referencing code and acceptance criteria for each scope, with the client and the certifying authority agreeing them in writing before anything is drafted. Take the expected damage mechanisms from the corrosion or integrity engineer and let them dictate technique, scan pattern and location. Write the procedure with the essential variables and their qualified ranges stated explicitly. Then design the demonstration around the specimens that will prove it, and run it before mobilisation.

Personnel qualification is planned in the same pass, because a procedure and a technician qualification that were developed separately rarely fit together. The practical specimens for the technicians should come from the same family as the demonstration specimens, and the record should show performance on the production geometry rather than on plate.

If you are preparing a Gulf of Mexico campaign, revising procedures after an audit finding, or moving to alternative acceptance criteria and need the sizing capability demonstrated first, ask for a consultation at info@atlantisndt.com. The opening deliverable is a written assessment of the existing procedure set against the codes the scope actually invokes, and a demonstration plan sized to the joints you will meet.

Request a consultation