Writing a UT Procedure That Survives Erosion, Heat and an Alberta Audit

What the referencing code actually controls

A UT procedure is not a method statement. Under ASME Section V, Article 4, Table T-421 divides the procedure's contents into essential and nonessential variables, and that division is the whole game. Change an essential variable — search unit frequency or element size, the wedge angle, the surface the scan is made from, the material product form, the thickness range, the couplant, the calibration block, the reference sensitivity method, the scan overlap — and the procedure must be requalified and reissued. Change a nonessential variable and the document is simply revised under document control. Auditors ask which variables you called essential before they ask anything else.

Thickness measurement lives somewhere else in the same code. Article 23 adopts SE-797 as the practice for contact ultrasonic thickness gauging, and it is a different discipline from weld examination: no DAC curve, no flaw sizing, but a set of accuracy constraints that eroded oil sands service violates routinely. Article 5 covers examination of materials rather than welds. A single site procedure that tries to carry corrosion mapping, weld flaw detection and material examination in one document is usually the one that fails an audit, because the essential variables of the three are not the same set and the document ends up silent on most of them.

Above the ASME layer sits the referencing code that actually owns the equipment: ASME B31.3 for process piping, Section VIII Division 1 for vessels, API 510 and 570 for in-service work, CSA Z662 where the scope crosses into pipeline. Each imposes its own acceptance criteria and, in some cases, its own examination extent. The procedure has to name which one it serves and which acceptance criteria apply to each component class. Per ASME is not a referencing code, and an experienced reviewer will say so on the first page.

The damage mechanisms that decide the technique

Oil sands hydrotransport, froth and tailings service is abrasive slurry moving at velocity. The wear it produces is not general thinning; it is localised scarring downstream of elbows, tees, reducers and pump discharges, often with a sharp gradient across a few tens of millimetres. API RP 571 catalogues it as erosion and erosion-corrosion, and the practical consequence for ultrasonics is blunt: a thickness reading is only as good as the position it was taken from. A technique that works beautifully on uniform general corrosion in a sweet gas line is close to useless here.

Upgrader hydroprocessing and vacuum service adds a different family. High-temperature sulfidic corrosion and naphthenic acid attack on the crude side, high-temperature hydrogen attack in hydrotreaters and hydrocrackers, and wet hydrogen sulphide damage — blistering, HIC and SOHIC — in sour water and amine circuits. Each demands something different from the procedure. HTHA is not a wall-loss problem at all, so a thickness grid will never see it; it wants advanced backscatter or velocity-ratio techniques, and a Level III who has read the API RP 941 exposure history of the circuit before choosing anything.

SAGD adds once-through steam generator tubing, high-temperature steam and produced-fluid lines carrying sand, and — because Alberta insulates everything — corrosion under insulation on lines that cross the dew point twice a day for eight months of the year. Delayed coker drums add low-cycle fatigue cracking at the skirt-to-shell weld and bulging that changes the geometry the procedure assumed when it was written. A procedure drafted for uniform-wall carbon steel piping addresses none of these. Each needs its own technique sheet inside the procedure, or its own procedure entirely.

Straight beam thickness in erosion service: position, spot size and the arithmetic trap

The most common error in oil sands thickness work is not instrument error, it is position error. A five-point TML rosette on an elbow extrados can sit entirely between erosion scars and report a healthy wall on a component with forty percent loss thirty millimetres away. Where the damage mechanism is localised, the procedure has to specify coverage rather than points: B-scan along the flow path, or an encoded corrosion map with a stated index and overlap, with the minimum retained rather than the average of a handful of readings.

The trap that follows is arithmetic, and it is the one that quietly sets inspection intervals across the industry. Suppose a TML reads 12.7 mm in one survey and 12.2 mm three years later, and the two readings were taken twenty-five millimetres apart on the same erosion scar. The half-millimetre difference is position error, but the API 570 calculation does not know that. It becomes a long-term corrosion rate of roughly 0.17 mm per year, remaining life shortens, and the next inspection interval is set on a number that describes the technician's hand rather than the pipe. The fix lives in the procedure — permanently marked and photographed TML locations, same probe type, same wear-face condition, same couplant — not in the spreadsheet downstream.

Probe selection belongs in the procedure for the same reason. Dual-element transducers are the default in corroded service because they tolerate rough backwalls, but they carry a V-path error that is worst on thin wall, they can double on very thin sections, and on a steeply sloped erosion scar the beam reflects away and the instrument simply holds the last valid reading. A procedure that does not require the operator to confirm the A-scan rather than trust the digital readout is a procedure that will one day report a comfortable thickness on a hole.

Hot service: velocity, couplant and duty cycle

Longitudinal velocity in carbon steel falls as temperature rises, by roughly one percent for every 55 °C above ambient. Because the instrument was zeroed on a cold calibration block, the time of flight through hot steel is longer and the displayed thickness reads high. On a 320 °C transfer line that is about a five percent over-read: a genuine 10.0 mm wall displays as roughly 10.5 mm. Applying the correction is trivial. Forgetting it is common, and the resulting error is not conservative — it is optimistic, which is the direction that hurts.

Couplant is an essential variable and hot service makes it a real one rather than a box to tick. Standard gel boils off, and high-temperature couplants have both a stated ceiling and a shelf life. ASME Section V also constrains couplant chemistry — sulphur content where nickel alloys are involved, total halides on austenitic stainless and titanium — and upgrader hydroprocessing circuits are full of both alloy families. The certificate of analysis for the drum on the back of the truck is, reliably, the document nobody can produce when an auditor asks for it.

High-temperature probes have a contact duty cycle, typically a few seconds on the surface followed by a cooling interval, and the delay line expands while it is hot, so the instrument zero drifts during the reading itself. A credible procedure states the couplant and its temperature limit, the probe type and its rated surface temperature, the contact and cool intervals, the method of re-establishing zero at temperature — on a heated step wedge, not on the cold block in the truck — and requires the surface temperature at which each reading was taken to be recorded on the report so the correction can be audited later.

Angle beam work: DAC and DGS are not interchangeable

For weld flaw detection and sizing, the reference sensitivity method is an essential variable and it changes what your recorded data means. A DAC curve is built empirically from side-drilled holes in a basic calibration block whose material, product form, heat treatment and curvature represent the component, and whose thickness falls inside the block selection rule — nominal component thickness plus or minus one inch or twenty-five percent, whichever is greater. DGS derives sensitivity from a single reference reflector and published curves specific to the probe, which means the probe's own certified DGS diagram becomes part of the qualification package.

Swapping between the two mid-programme, or running a DGS-capable instrument with a probe whose diagram was never verified, survives for years until a client's technical authority asks to see the probe certificate. Recording sensitivity and transfer correction is not paperwork either. Transfer loss across a coke-fouled, weld-spattered or grit-blasted oil sands surface is frequently several decibels, and those decibels are the difference between recording an indication and never seeing it at all.

Sizing is where in-service oil sands work diverges sharply from new construction. If the answer feeds an API 579-1/ASME FFS-1 assessment — and for a cracked coker skirt weld or a hydrogen-blistered drum it will — then through-wall extent, not signal amplitude, is the number that matters, and amplitude-based DAC sizing is not fit for that purpose. The procedure has to say plainly which techniques produce sizing data of assessment quality and which produce detection only, because the engineer running the assessment will otherwise assume the former.

What the qualification demonstration has to show

Where the referencing code requires demonstration — encoded phased array or TOFD used in place of radiography, or ultrasonic examination in lieu of RT under Code Case 2235 — the procedure is proved on a qualification block rather than argued on paper. That block has to contain flaws representative of the ones the procedure claims to find: root-connected, mid-wall, and near the far surface, positioned at the extremes of the thickness and angle range the procedure covers. A flawless mock-up proves the instrument is switched on and nothing else.

A defensible demonstration records considerably more than pass or fail. It records who ran it, on which instrument and probes, at what sensitivity, with what scan plan and what coverage was actually achieved. It demonstrates the sizing tolerance claimed, not detection alone. And it includes at least one blind element, because an operator who already knows where the flaw sits will find it every time. If the demonstration was run by the Level III who wrote the procedure and never repeated with the technicians who will actually use it, what was demonstrated was the author.

For corrosion mapping there is an equivalent demonstration, and it is routinely skipped. Take a specimen with machined wall loss of the shape and width you expect to find, and prove that the chosen scan index and probe diameter actually resolve it. Ten minutes of that work tells you whether a twenty-five millimetre grid can see a twenty millimetre scar. Usually it cannot, and it is far better to learn that on a test piece than in a failure investigation.

Findings that recur when this procedure is audited

Across owner-user audits and client qualification reviews the same items come back. Calibration blocks with no material traceability, or of a different heat treatment or curvature than the component. No temperature correction clause, or a clause present but no record of the temperature at which readings were taken. Instrument calibration verification not performed at the required interval — Section V expects a check at the start and end of each examination, on any change of personnel, and at least every four hours — and, worse, no defined action for the work performed since the last good check when a verification fails.

Then come the document control findings, which are usually more serious than the technical ones. A procedure revised by the Level III but never reissued to the field, so two revisions are in simultaneous use on the same unit. A revision that changed an essential variable without a new demonstration. Technicians certified under a written practice whose stated scope does not include the technique they are running — a UT Level II limited to thickness measurement signing angle beam weld reports is by far the most common version. And a procedure approved by a Level III whose own certification does not cover ultrasonic testing at all.

None of these are found by watching a technician scan a weld. They are found by reading three documents side by side: the procedure, the written practice, and one completed examination report. That is precisely how an ABSA reviewer, an owner-user integrity engineer or a client's technical authority will read them, and it is exactly how the procedure should be tested internally before any of those people arrive.

Who owns the document, and who does not

A UT procedure is approved by a Level III certified in ultrasonic testing. In the oil sands the practical questions are who that person is, what their authority covers, and how quickly they can act. A procedure that cannot be revised inside a turnaround window will be worked around on the deck, and the workaround will be undocumented, which converts a procedural problem into a records problem that surfaces months later.

The Level III's role here is narrow and specific: approve the technical content, approve the qualification, approve the personnel who apply the procedure, and defend all of it in an audit. It is not the API 510 or 570 inspector's role — that inspector authorises and signs the in-service inspection and sets the interval. It is not the technician's role either; the technician is certified under the employer's written practice to apply the procedure exactly as written. Conflating those three roles is the most common conceptual error in oil sands programmes, and it produces documents signed by people who have no standing to sign them.

Atlantis supplies the first of those three. Procedure development and qualification, Level III approval, technique selection and demonstration design, and independent review of the resulting data. We do not act as the API inspector of record and we do not audit your process safety management system. Where an engagement needs both, the separation of duties should be written down before either starts.

How the Alberta regime changes the answer

Alberta is not a light-touch jurisdiction. Pressure equipment falls under the Safety Codes Act and the Pressure Equipment Safety Regulation, with ABSA as the delegated authority, and an owner-user operating an integrity management programme under AB-506 is audited on that programme as a system — including how NDE procedures are approved, who is authorised to approve them, whether personnel are certified to a controlled practice, and whether the resulting data actually feeds the integrity assessment rather than sitting in a folder. CSA B51 governs construction and registration on the way in.

Where scope crosses into pipeline, CSA Z662 applies instead, and it brings a different personnel certification expectation than ASME work does. On a single oil sands site you can have B31.3 piping examined by technicians certified under an employer's written practice and Z662 pipeline examined by technicians who need national certification through the NRCan-administered CGSB scheme. The UT procedure has to state which regime it serves, and the certification records behind each report have to match that statement. Auditors check this pairing specifically, because on integrated sites it fails so often.

The winter adds one more variable that southern procedures never contemplate. Field readings taken at ambient temperatures far below the calibration temperature shift velocity in the opposite direction, gel couplants stiffen or freeze, and access scaffolding and heat tenting change what the technician can physically reach. If your procedure has a stated temperature range, check that the low end of it is a real number and not an assumption imported from a template written for the Gulf Coast.

Request a consultation