What a UT Level III Actually Approves, and What a UT Certificate Does Not Cover
A UT Level III owns the ultrasonic technique: the calibration block, the DAC or DGS sensitivity basis, transfer correction, scanning coverage and the sizing method used to report flaw height. Certification in ultrasonic testing does not by itself authorise phased array, TOFD or HTHA assessment. Those need separate documented training and a demonstrated procedure the Level III approves.
Port Arthur's ultrasonic workload is not generic weld inspection. It is heavy-wall hydroprocessing steel at Motiva, Valero and TotalEnergies, where 2.25Cr-1Mo reactor shells operating at high hydrogen partial pressure raise high temperature hydrogen attack as a live question under API 941. It is sour crude service, so hydrogen-induced cracking and blistering mapping under API 571. It is coarse-grained austenitic and dissimilar-metal welds that scatter a conventional 2.25 MHz shear beam into noise. It is API 653 tank floors across the Nederland and Port Arthur terminal farms, where magnetic flux leakage finds the indication and ultrasonics has to prove it up. Each of those is a technique decision, not a technician decision, and each one changes the essential variables in the procedure. The Level III chooses the frequency, the wedge, the block, the sensitivity basis and the sizing method, then owns the result when a fitness-for-service assessment depends on it.
Source: ASME Section V Article 4 for ultrasonic examination of welds and Article 5 for materials, with Section VIII Division 1 and Division 2 acceptance; ASME Section XI and its performance demonstration requirements for nuclear components; API 510, 570 and 653 for in-service thickness and examination requirements; API 571 for damage mechanisms; API 577 for welding inspection practice; API 579-1/ASME FFS-1 for fitness-for-service flaw characterisation; API 941 for high temperature hydrogen attack; API 1104 and AWS D1.1 acceptance criteria; ASNT SNT-TC-1A and ANSI/ASNT CP-189 for personnel qualification.
| Technique decision | What the Level III approves | Recurring audit finding | Consequence when wrong |
|---|---|---|---|
| Calibration block selection | Block material, thickness range, reflector type and size relative to the part | Block thickness outside the range the part requires; block material acoustically unlike the part | Sensitivity is set against the wrong reference and the whole examination is unverifiable |
| Sensitivity basis | DAC construction from the specified reflectors, or a probe-specific DGS diagram | DAC points taken from two reflectors and the curve drawn through by eye | Amplitude evaluation at depth is arbitrary; small reflectors near the far surface are missed |
| Transfer correction | Method of comparing block and part surface response, and the threshold for compensating | No transfer measurement recorded at all on rough, coated or as-welded surfaces | Real sensitivity in the part is lower than the recorded gain implies |
| Angle and skip planning | Refracted angles, leg strategy and coverage of the required volume | Single angle used on a thick weld; second-leg geometry echoes interpreted as flaws | Root or fusion-face flaws outside the beam are never interrogated |
| Examination temperature | Permitted surface temperature range and the correction rule for hot readings | Hot thickness readings logged with no temperature correction applied | Remaining wall is overstated, corrosion rate understated, remaining life overstated |
| Couplant selection | Couplant family and contaminant limits for the alloy being examined | General-purpose couplant used on austenitic and nickel alloy surfaces | Halide or sulphur contamination introduced onto crack-susceptible material |
| Sizing method | Amplitude drop technique or tip diffraction, and which is required for the reported dimension | Through-wall height reported from an amplitude drop and fed into an assessment | Flaw height under-reported; the fitness-for-service result is non-conservative |
Calibration is a Level III decision, not a technician habit
Under ASME Section V Article 4 the calibration block is part of the procedure, not an accessory the technician chooses from the truck. The block has to be acoustically similar to the part, within a defined thickness relationship to it, with reflectors of a size and depth distribution set by the code for the thickness range. Getting this wrong invalidates everything downstream, because every amplitude judgment made during the examination refers back to a reference level established on that block.
The failure mode is mundane and constant: a shop keeps two or three blocks, a job arrives with a wall thickness outside the range those blocks cover, and the nearest block is used because the unit is down and the crew is on the clock. Nothing looks wrong in the field. The finding appears months later when an auditor compares the block certificate against the component thickness recorded on the report, and the entire weld population examined that shift becomes questionable.
The Level III's job is to remove the field decision. That means specifying the block by identity in the procedure, controlling block certification and periodic verification, defining what happens when a component falls outside the covered range, and providing a documented route to authorise an alternative before work starts rather than after. On heavy-wall hydroprocessing components in Port Arthur, where thicknesses run well beyond what a general-purpose block covers, this is not an occasional problem but a scheduled one.
DAC and DGS are not interchangeable sensitivity bases
A distance-amplitude correction curve is built empirically: the operator maximises the response from the specified reflectors at their several depths, plots the peaks, and connects them. It compensates for attenuation and beam spread across the sound path and gives an evaluation reference that varies with depth. It is specific to the transducer, the wedge, the instrument settings and the block. Change any of them and the curve is no longer valid.
A distance-gain-size diagram is theoretical rather than empirical. It relates the response of a reflector to the response of an equivalent flat-bottomed hole using a diagram derived for that specific probe. It is efficient, it needs only a single reference reflector, and it is heavily used in European practice and in imported procedures. It is also only as good as the probe's own diagram and the attenuation and transfer values entered, which is exactly where it is abused: default attenuation values entered because the true value was never measured.
The two are not substitutes and a procedure must state which basis governs, with the referencing code's permission for it. The recurring field failure with DAC is a curve built from two points instead of the specified set, drawn through by hand, producing an evaluation line that is fictional beyond the second reflector. The recurring failure with DGS is unmeasured material attenuation. In both cases the report looks complete and the sensitivity actually applied is unknown, which is worse than a missing report.
Transfer correction: the two-decibel check that nobody records
Sensitivity is established on a calibration block whose surface is machined, clean and dry. It is then applied to a component that may be as-welded, grit-blasted, painted, insulated-scarred or lightly corroded, and whose grain structure may attenuate differently from the block. The difference between the two, measured by comparing responses on block and component, is the transfer correction. Where the difference exceeds a small threshold it must be compensated in the gain.
This is one of the most frequently skipped steps in the entire method, and one of the easiest to catch in an audit, because the report has a field for it and the field is blank. The consequence is systematic rather than random: the examination is quietly less sensitive than the recorded gain implies, uniformly, across every weld examined on that surface condition. A whole shift of examinations can be under-sensitive without a single obviously wrong reading.
Gulf Coast conditions make it worse, not better. Salt-air exposure, coating systems, insulation removal damage and the surface state of equipment that has run for decades all produce components acoustically unlike a polished block. A Level III writing procedures for Port Arthur service should specify the transfer measurement explicitly, name the surfaces on which it is mandatory, define the compensation threshold, and require the measured value to be recorded on the report whether or not compensation was applied.
Hot-wall readings and the one percent per hundred degrees trap
Thickness monitoring on live circuits is a large fraction of ultrasonic work in a refinery, and much of it is taken on hot metal because the alternative is a shutdown. Sound velocity in carbon steel decreases as temperature rises. A gauge calibrated at ambient converts transit time using the ambient velocity, so the same physical wall thickness produces a longer transit time when hot, and the instrument reports a larger number. The reading is not noisy; it is biased, and biased in the unsafe direction.
The customary correction is approximately one percent of the indicated thickness for each hundred degrees Fahrenheit above the calibration temperature, applied as a reduction, with high-temperature couplants and delay lines used to make the measurement possible at all. On a circuit running several hundred degrees above ambient the correction is not trivial. Worse, the error compounds through the integrity calculation: an overstated thickness understates the measured metal loss, which understates the corrosion rate, which overstates remaining life under API 510 or API 570.
The arithmetic trap has a second stage that audits find regularly. Corrosion rates are computed from a current reading and a previous one. If the two were taken at different metal temperatures and neither was corrected, the computed rate can come out near zero, or negative, and the circuit is quietly reclassified as non-corroding. The procedure must therefore require the surface temperature to be recorded with every reading, and the correction to be applied and shown, so the next engineer can see what was done.
Amplitude sizing versus tip diffraction, and why the assessment cares
Two questions are routinely confused: is there a flaw, and how tall is it. Amplitude-based techniques answer the first well. They locate and evaluate reflectors against a reference level, and codes with amplitude-based acceptance criteria are built around exactly that. Length is usually estimated by a drop technique. Through-wall height, however, is a different measurement, and amplitude response is governed by orientation, roughness and reflectivity as much as by extent.
When a planar flaw is smooth, tight or unfavourably oriented, an amplitude drop technique will report it as shorter and shallower than it is. That is tolerable if the acceptance decision is a simple amplitude comparison. It is not tolerable when the number is going into an engineering assessment, because a fitness-for-service evaluation under API 579-1/ASME FFS-1 is acutely sensitive to through-wall height. Under-reporting height produces a non-conservative result that reads as a comfortable margin.
The Level III's obligation is to make the procedure state, unambiguously, which technique is used for which dimension, and to require a tip-diffraction or equivalent height-measurement technique whenever the reported height will support an assessment or a run-repair-replace decision. It is also to make sure the report says which technique produced the number, so the integrity engineer receiving it knows what confidence to attach. A height with no stated method is an input nobody should use.
What a UT certificate does not cover
Ultrasonic testing is one method in the written practice, but it contains several techniques with materially different skill requirements. Straight-beam thickness and lamination scanning, angle-beam weld examination, corrosion mapping, phased array, time-of-flight diffraction, encoded and mechanised scanning, and advanced techniques used for damage mechanisms such as high temperature hydrogen attack are not the same job. A certificate that says ultrasonic testing, Level II, says nothing about which of those the holder has been trained and examined on.
Most owner supplements now close that gap explicitly, and any competent written practice should as well. The mechanism is a technique-specific training block, a technique-specific portion of the specific examination, and a practical demonstration on specimens representative of the configurations the technician will actually meet. The Level III writes the content, sets the specimens, grades the demonstration and recommends the endorsement. It is more work than issuing a general certificate and it is the difference between a defensible programme and a hopeful one.
Two boundaries are absolute. Where a code or code case imposes a performance demonstration, for example the demonstration requirements applying to ultrasonic examination of nuclear components under ASME Section XI, an employer's Level III cannot certify anyone into that qualification; it is achieved through the prescribed demonstration programme and nothing in your written practice substitutes for it. And where ultrasonic examination is substituted for radiography on new construction, the procedure, equipment and often the individual operator all require demonstrated performance before the substitution is valid.
Port Arthur metallurgy: sour service, hydroprocessing and dissimilar welds
The refineries on this stretch of the Texas coast were built and expanded to run heavy, sour crude. That crude slate drives wet hydrogen sulphide damage in the appropriate circuits: hydrogen blistering, hydrogen-induced cracking and its stress-oriented variant, described in API 571 and inspected largely by ultrasonics. These are volumetric, laminar and often mid-wall phenomena, and they are found by straight-beam mapping with a technique tuned for them, not by a weld procedure repurposed on the day. Getting the frequency, the element size and the scan index right determines whether small stepwise cracking is found at all.
Hydroprocessing is the second metallurgical driver. Reactor shells and heavy-wall piping in low-alloy chromium-molybdenum steels run at hydrogen partial pressures and temperatures that place them in the region where high temperature hydrogen attack has to be considered under API 941. Early HTHA does not present as a clean reflector; it is diffuse micro-fissuring and decarburisation. Assessment normally combines multiple ultrasonic measurements, and the case for or against damage rests on interpreting them together with the operating history. The Chemical Safety Board's investigation of the 2010 Tesoro Anacortes heat exchanger rupture is the reference case for why this is treated as a technical authority question rather than a scanning task.
The third is austenitic and dissimilar metal welds, common at cladding boundaries, in stainless piping and at transition joints. Their coarse, anisotropic structure scatters and skews a conventional shear beam and buries indications in structural noise. The technique answer is normally lower frequency and a dual-element longitudinal-wave transducer designed for the material, calibrated on a representative standard, with the reduced sizing accuracy stated in the procedure and repeated in the report. Pretending a standard weld procedure works on these joints is the most consequential ultrasonic shortcut taken in this market.
Tank floors on the Nederland and Port Arthur terminal farms
Storage capacity along the Sabine-Neches corridor is enormous, and API 653 out-of-service inspections put ultrasonics into a role that differs from weld examination. Floor scanning is normally done with magnetic flux leakage, which is fast and covers area but does not tell you which side of the plate the loss is on, nor its true remaining thickness. Ultrasonic prove-up does both. The scanning method finds candidates; the ultrasonic technique converts them into numbers an engineer can use.
That prove-up needs its own procedure and its own discipline. Small-diameter, high-frequency transducers, a defined prove-up pattern around the indication rather than a single spot reading, an explicit rule for distinguishing topside from underside loss, surface preparation requirements, and a recording convention that lets a later inspection be compared to this one. Where the prove-up is done casually, the minimum remaining thickness reported is whatever the technician happened to hit, and the settlement or corrosion rate derived from it is fiction.
Shell course thickness, critical zone examination and, on the Gulf Coast, the consequences of hurricane exposure add further scope. Flooding and wind damage drive water into insulation and coating systems and produce corrosion under insulation on shells and connected piping long after the storm. Deciding when ultrasonic profiling is the right answer, and when a screening method should come first, is a procedure-level decision. The Level III owns that choice and should be documenting it in advance of the next storm season, not improvising after one.
Engaging a UT Level III with Atlantis
We start by reading your ultrasonic procedures against the work you actually perform: thickness monitoring, weld examination, corrosion mapping, prove-up, and any advanced technique your clients request. Most procedure sets in this market are technically sound for generic welds and silent on the things that matter here, which are transfer correction, temperature correction, sizing method and technique-specific qualification. Those gaps are closed by amendment wherever possible, so the existing document history stays intact.
From there the work is continuing authority: procedure approval and requalification when an essential variable changes, calibration block control, technique selection for specific damage mechanisms, training content and examination for ultrasonic endorsements within your written practice, and independent review of ultrasonic data where an owner or an integrity engineer wants a second reading before a run-repair-replace decision. During turnaround windows the commitment is a defined response time for procedure variations, because that is when the signature is worth the most.
The engagement is affordable, accessible and fully customisable, and it can be scoped as a defined procedure project or as ongoing named authority in your written practice. We supply NDT technical authority only: we do not sign as the API inspector of record and we are not a process safety management auditor. Request a consultation with your current ultrasonic procedure set and a list of the units and circuits you work on, and the first review will tell you where the exposure is.
Does an ultrasonic Level II certificate cover phased array and TOFD?
Not automatically, and treating it as though it does is a common audit failure. Phased array and time-of-flight diffraction are distinct techniques with their own training content, their own setup and calibration disciplines and their own image-interpretation skills. Most owner supplements require documented additional training and a practical demonstration on representative specimens before an ultrasonic technician may apply them, plus a procedure the Level III has qualified specifically for the technique.
Why does a hot thickness reading overstate remaining wall?
Sound velocity in steel falls as temperature rises. A gauge calibrated at ambient uses the ambient velocity, so a longer transit time through hot metal is converted into a thickness larger than the true one. The industry rule of thumb is roughly one percent of indicated thickness per hundred degrees Fahrenheit above the calibration temperature. Uncorrected readings on a hot circuit therefore understate corrosion rate and overstate remaining life.
When is amplitude sizing not acceptable for flaw height?
Whenever the height is going into an engineering assessment. Amplitude drop techniques respond to reflectivity and orientation as much as to size, and they systematically under-report the through-wall extent of planar flaws. A fitness-for-service evaluation under API 579-1/ASME FFS-1 is sensitive to flaw height above almost anything else, so the procedure must specify a tip-diffraction or equivalent height-measurement technique when the result feeds an assessment.
Why do dissimilar metal and austenitic welds defeat a standard shear-wave technique?
Coarse, columnar and anisotropic weld structure scatters and skews the beam, so a conventional shear-wave setup returns high structural noise, distorted beam paths and a poor signal-to-noise ratio. The Level III's answer is normally a lower frequency, a dual-element longitudinal-wave transducer designed for the material, a matched calibration standard, and an acceptance of reduced sizing accuracy stated openly in the report rather than hidden in it.
What makes high temperature hydrogen attack a Level III problem rather than a technician problem?
Because early HTHA is diffuse micro-fissuring and decarburisation rather than a discrete reflector, and no single conventional ultrasonic parameter reliably finds it. Assessment normally combines several measurements, and interpretation depends on knowing the alloy, the hydrogen partial pressure and the operating temperature history against API 941. Choosing the technique combination, setting the acceptance logic and reading the result is technical authority work, not a scanning task.
Is API 510, 570 or 653 inspector training part of this offer?
No. Atlantis provides ultrasonic technical authority: procedure development and qualification, calibration and sensitivity basis, technique selection, personnel training content and examination within your written practice, and independent review of ultrasonic data. API inspector certification runs through API's own programmes, and the API inspector of record signs the in-service inspection. We are also not your process safety management auditor. The two responsibilities stay clearly separated in every document we sign.