Writing and qualifying a UT procedure for recovery boilers and digesters

What the referencing code actually asks of a UT procedure

The most common structural mistake in a mill UT procedure is treating ASME Section V as the specification. It is not. Section V is the method standard; it tells you how to examine. The referencing code tells you what has to be examined, what is acceptable, and who may accept it. For a kraft recovery boiler that referencing code is Section I as the code of construction, with the National Board Inspection Code governing repair and alteration in service. For a continuous digester built as a pressure vessel it is Section VIII Division 1. Get the hierarchy wrong and the procedure will quote acceptance criteria that no code actually imposes, which an auditor spots in the first five minutes.

Within Section V, three articles do the work in this industry. Article 4 covers ultrasonic examination of welds and flaw detection, and carries the table of essential and nonessential variables that a written procedure has to address. Article 5 covers ultrasonic examination of materials and thickness determination. Article 23 adopts the ASTM standard practices, including SE-797 for manual pulse-echo contact thickness measurement, which is the practice most recovery boiler tube surveys are actually run under whether or not the procedure says so.

On top of the code sits a second layer that mills routinely confuse with it: BLRBAC recommended good practice, the property insurer's inspection requirement, and the mill's own integrity programme. Those documents set scope and frequency. They do not set method. A procedure that cites an insurer bulletin as its technical basis, with no code reference behind it, has no defensible acceptance criteria at all.

Damage mechanisms in a recovery boiler that decide the technique

A recovery boiler is not a generic power boiler with a different fuel. The lower furnace runs a reducing, sulphidising atmosphere over a smelt bed, and the wall construction reflects that: co-extruded composite tubing, commonly 304L or Alloy 825 clad over SA-210 carbon steel, is used precisely because bare carbon steel wastes at an unacceptable rate. That construction is what makes the ultrasonics hard, because a composite tube returns more than one back wall.

The damage list that drives technique selection is short and specific. Fireside general and localised thinning of the clad layer. Cracking of the stainless clad, including thermal fatigue cracking on floor tubes and around openings, which can and does propagate into the base metal. Sootblower erosion producing narrow grooves with steep walls. Waterside hydrogen damage and caustic gouging under deposits where boiler water chemistry has been upset. Cracking at membrane and attachment welds where restraint is high. Each of these calls for a different beam, a different calibration, and a different recording rule.

The consequence of missing them is not a repair cost. A lower furnace tube leak on a recovery boiler puts water onto molten smelt, and a smelt-water explosion is the reason this asset has its own advisory committee, its own emergency shutdown procedure, and a mandatory annual inspection culture. A UT procedure written for this equipment is a safety document before it is a quality document, and it should read like one.

Digesters: cracking under overlay, and why shear wave lies

Continuous digesters present the opposite problem. The shell is carbon steel, frequently with a stainless weld overlay in the impregnation and cooking zones, and the dominant threat is caustic stress corrosion cracking in and adjacent to the heat affected zones of the shell seams, together with localised corrosion where the overlay is thin, missing, or has been ground away during a previous repair. Surface cracking is found reliably with wet fluorescent magnetic particle after blasting. The question the ultrasonics has to answer is how deep it goes, and whether there is anything under the overlay that never reached the surface.

Conventional refracted shear wave is the wrong tool here and a procedure that specifies it without qualification is producing false confidence. Austenitic weld metal is coarse grained and elastically anisotropic. A shear wave crossing it scatters, attenuates, and beam skews away from the nominal refracted angle, so the technician sweeps a region he believes is covered and it is not. The corrections are well established: low frequency longitudinal transmit-receive probes, typically at or below two megahertz, matched wedges, or a dual matrix array with a focal law set designed for the overlay thickness in question.

Whichever is chosen, the procedure has to state the coarse grain compensation, the probe separation and focal depth, and the fact that the technique was demonstrated on an overlaid coupon representative of this vessel. Depth sizing of caustic cracking then feeds directly into a fitness-for-service assessment, so the sizing accuracy claimed in the procedure has to be an accuracy the demonstration actually supported, not a manufacturer's brochure figure.

Calibration: DAC, DGS, transfer correction and the block that must exist

A calibration section fails audit in three predictable ways. The first is that the basic calibration block named in the procedure does not exist in the shop, or exists in a different material, thickness or heat treatment condition than the component. The block has to be of the same material specification, product form and heat treatment as the item, with a thickness within the range the code permits for the component thickness, and its side drilled hole positions have to give at least the three points that a usable DAC curve needs.

The second is transfer correction. The calibration block is machined and clean. A recovery boiler tube is scaled, curved, and often ground unevenly by the crew who prepared it. The difference in surface coupling and attenuation between block and component is real, is measurable by a simple two probe transfer measurement, and is routinely never done. A procedure that does not say how transfer loss is measured and applied is producing readings that are systematically optimistic on the very components with the worst surfaces.

The third is DGS. Distance gain size is a legitimate sizing method and many crews prefer it because it dispenses with block scanning, but it is probe specific, requires the correct diagram set for the exact search unit, and does not automatically satisfy a specification written around a DAC curve. If the owner specification calls for DAC and the crew intends to work to DGS, the procedure needs to record the equivalence demonstration and the reference reflector, and the Level III has to sign that reasoning, not simply the substitution.

The arithmetic traps: composite wall, temperature and curvature

The composite tube trap is the one that costs mills money. A co-extruded tube presents an interface echo at the clad-to-base boundary as well as a back wall echo. A dual element gauge in auto mode can lock onto either. If the crew reports total wall where the design calculation credits only the base metal as pressure retaining, tubes are left in service on a number that overstates the margin. If they report base metal only where the mill's own retirement criterion was written against total wall, perfectly good tubes are replaced. Neither error is visible in the data set; both are prevented by one sentence in the procedure that states which measurement is being reported and how the gate is set to obtain it.

Temperature is the second. Velocity in steel decreases as temperature rises, at roughly one percent per hundred degrees Fahrenheit, and an uncompensated instrument therefore reads thick on hot metal. On a nominal 0.300 inch lower furnace tube at 300F, that is about 0.006 inch of apparent wall, which is a meaningful fraction of the margin between the current reading and the Section I minimum thickness. Either the procedure requires a hot calibration on a block at temperature, or it states the correction factor and the temperature range over which it may be applied.

Curvature is the third and is specific to tube work. A flat contact face on a small diameter tube couples over a narrow line, and the practice adopted in Section V for contact thickness measurement addresses when curvature correction or a curved shoe is required. Below roughly two inches outside diameter this stops being an academic concern. A procedure that has been copied from vessel work and applied to tube banks usually has no curvature clause in it at all.

What a qualification demonstration has to show

Codes distinguish between a procedure that has been demonstrated to the satisfaction of the Level III and a formally qualified procedure with a documented performance demonstration. Pulp and paper work sits between the two: the code often asks only for the former, while the insurer, the client, or the mill's own integrity engineer increasingly asks for the latter. Build the procedure so it can satisfy the harder requirement, because retrofitting a demonstration onto a live outage is not possible.

A credible demonstration for this industry uses representative specimens, not convenient ones. That means a composite tube section with an intact bond line and, ideally, real thermal fatigue cracking; an overlaid shell coupon with caustic cracking or fatigue grown flaws; and erosion or wastage geometry that resembles the sootblower scars actually seen in that boiler. Side drilled holes and EDM notches have their place in calibration; they are poor proxies for detection capability because they are far more reflective than a tight, branched, oxide-filled crack.

The demonstration should be run blind by the technicians who will do the work, on the equipment they will use, with results recorded before the true state is disclosed and compared against sectioning or an independent measurement. The output is a record: specimens used, flaws present, flaws found, sizing error distribution, and the essential variables in force. That record, not the procedure cover sheet, is what convinces a client auditor that the number in the report means something.

Findings that recur when a mill UT procedure is audited

Across audits of mill and contractor UT programmes, the same findings appear. No recording level is defined, so indications below the reference level were seen, discussed on the scaffold and never written down, which leaves the mill with no trend data for the following outage. Calibration verification intervals are ignored, and when a check finally fails nobody re-examines the work performed since the last valid check, which is what the code actually requires and what makes an out-of-calibration event expensive.

Essential variables drift silently. The procedure was demonstrated with a 5 MHz dual element probe; the crew arrived with 2.25 MHz because that is what was in the case, and the coverage and near surface resolution changed with it. The procedure names a calibration block by number; the block on site is a different one with the same nominal thickness. The procedure was written to an older edition of the referencing code and nobody reconciled it when the mill's repair organisation updated its quality manual.

The last recurring finding is about people rather than paper. Technicians are certified to a written practice that never mentions the limited scope they actually work in, or their practical examination was taken on a flat butt weld coupon while their entire working life is spent on composite tube walls. The procedure and the written practice are one system. An auditor who finds a weakness in one will always go looking in the other.

How Atlantis develops and qualifies the procedure

The engagement starts with the referencing code and the damage, in that order. We establish which construction and repair codes actually apply to each item, what the insurer and BLRBAC regime adds, and what the mill's own history says about where it loses wall and where it cracks. Only then is the technique chosen, because a technique selected before the damage mechanism is understood is a guess with a procedure number on it.

We then write the procedure itself: scope, referencing code, personnel requirements tied to your written practice, equipment, calibration blocks and standardisation, transfer correction, scanning plan with coverage and overlap, temperature and surface condition limits, recording and evaluation levels, reporting format, and a clearly separated table of essential and nonessential variables so that future changes route correctly. Every procedure is issued under revision control with a signature and a date from the Level III of record.

Qualification follows: specimen design or sourcing, the demonstration itself, the sizing comparison, and the written record. Where the mill wants it, we run the same demonstration as a practical examination for the technicians, so one exercise produces both the procedure qualification and the personnel evidence an auditor will ask for. Talk to us about a procedure development and qualification consultation at info@atlantisndt.com.

Request a consultation