UT on Boilers, HRSGs and Headers: Training Built for the Outage

Why power generation ultrasonics is not refinery ultrasonics

A technician moving from fixed equipment inspection into a boiler house arrives with real skills and the wrong instincts. In refining, the working mental model is a corrosion loop: nominal thickness, a measured wall, a corrosion rate, a remaining life. That model assumes a component whose acceptable wall is broadly stable and whose damage is progressive and general. A boiler breaks all three assumptions. Metal temperature varies dramatically over a few feet of tubing, the allowable stress moves with it, and the damage mechanisms are local and mechanism-specific rather than uniform.

The second difference is the population. A refinery unit has a manageable number of monitored circuits. A boiler has thousands of tubes, and no outage inspects them all. What the crew inspects is decided by a damage mechanism review, by failure history and by where the last outage found trouble. That means the value of the work depends on hitting the right locations, not on volume of readings. A crew that produces ten thousand readings from the wrong panels has produced nothing.

The third difference is time. The unit is off, and every hour of scanning sits inside a critical path that also holds tube replacement, welding, post-weld heat treatment and hydrostatic testing. Technique decisions that can be deferred elsewhere have to be made in advance here, which is why so much of the training weight sits on preparation rather than on scanning skill.

The acceptance basis: minimum required wall, not a percentage

ASME Section I sets the required wall thickness of tubes and piping by formula. The inputs are design pressure, outside diameter, the maximum allowable stress value at the design metal temperature, and the appropriate coefficients. The output is a number specific to that component at that temperature. Nothing in that calculation refers to how thick the tube was when it was installed, which is why the industry rule of thumb about a percentage of nominal loss is not an acceptance criterion and never was. It is a screening habit that survived into places where it does not belong.

For power piping outside the boiler proper, ASME B31.1 does the same job with its own formula and its own allowances. Repairs and alterations to the boiler are governed by the National Board Inspection Code, which controls the repair method, the welding procedure, the heat treatment and the examination that follows, all under the jurisdictional inspector. Where an owner wants to run a component that no longer meets the design calculation, the route is a fitness-for-service assessment under API 579-1 and ASME FFS-1, not a judgement call on the scaffold.

The training consequence is direct. A technician has to understand that the number they report is an input to a calculation somebody else performs, and that the value of the reading is determined by whether it is accurate, located and attributable. The technician does not decide whether the tube stays in. They decide whether the engineer who does can trust the data.

Flow-accelerated corrosion, and how it is actually scanned

Flow-accelerated corrosion is the dissolution of the protective magnetite layer on carbon steel by flowing water or wet steam, and it is the mechanism that has caused some of the industry's most serious feedwater line failures. It runs in single-phase and two-phase forms, it is governed by water chemistry, temperature, velocity and geometry, and it is strongly suppressed by even small amounts of chromium in the steel. That last point drives inspection planning: replace a susceptible spool with a low-alloy or stainless component and the mechanism stops there, which means the inspection grid must know what material it is sitting on.

On the scan, the damage is broad, smooth and shaped by the flow. It concentrates downstream of elbows, tees, reducers, orifices and control valves, and the internal surface develops a scalloped or orange-peel texture rather than pits. A technician looking for the sharp, localised losses that dominate other sectors can bracket the thin region with a coarse grid and never land on it. Grid density, and the placement of the grid relative to the flow disturbance rather than relative to the weld, is what finds the minimum.

Most utilities run these programmes on a predictive model that ranks components by susceptibility and feeds measured data back in. That gives the inspection an obligation beyond finding thin metal: the reading must be attributable to a component, a grid position and a material, so the model can be updated. Readings that arrive as a spreadsheet of numbers with no component identity are data the programme cannot ingest.

Oxide scale on superheater and reheater tubing

Steam-side magnetite grows on the internal surface of superheater and reheater tubing as a function of metal temperature and time. Because it insulates, it also accelerates the process. Measuring its thickness ultrasonically is one of the few practical ways to estimate what metal temperature a tube has actually run at, which then feeds a creep life estimate. The measurement is made from the outside with a high-frequency delay-line probe capable of resolving a layer that may be only a few thousandths of an inch thick.

The practical requirements are unforgiving. The tube must be at ambient temperature. The outside surface must be prepared to clean metal, because scale and coating on the outside destroy the resolution the technique depends on. The probe, the instrument and the software must be matched, and the technician has to know the resolution floor of that combination. Below it, no honest number exists. Spalled or laminated internal scale returns multiple confused echoes, and reporting the first one as the scale thickness produces a life estimate that is wrong in the optimistic direction.

This is the measurement that most often gets reported without qualification. A responsible report states the technique, the resolution limit, and which readings fell below it. A report that gives three decimal places on every tube in a pendant, with no failures to resolve anywhere, is telling you about the contractor rather than about the boiler.

Creep and Type IV cracking in Grade 91 headers

Creep-strength-enhanced ferritic steels, of which Grade 91 is the most widespread, allow higher operating temperatures and thinner sections, and they brought a damage mechanism that a great many technicians were never trained for. Type IV cracking develops in the fine-grained heat affected zone of a weld, a narrow band a few millimetres outside the fusion line, where the microstructure produced by the welding thermal cycle has lower creep strength than either the weld metal or the parent. It initiates subsurface and it runs roughly parallel to the weld.

Every part of that description is a problem for a conventional weld inspection setup. The target is not in the weld. It is small, it is oriented awkwardly, and the surrounding material is coarse-grained and attenuating, so the obvious response of adding gain raises structural noise faster than it raises the signal. A phased array focal law or a TOFD setup has to be designed deliberately for the heat affected zone at the expected depth range, and it has to be demonstrated on a representative mock-up before anyone believes a clean result.

There is a second failure mode that has nothing to do with ultrasonics: the header may not be Grade 91 at all, or may be Grade 91 that was welded or heat treated incorrectly during construction, which changes both the damage expectation and the repair route. Positive material identification and hardness data belong alongside the ultrasonic data, and technicians who understand why are far more useful on an outage than those who only scan.

HRSG geometry: harps, fins, and the access that decides the technique

A heat recovery steam generator is a very different physical object from a conventional boiler, and combined-cycle plants cycle hard, which loads it with thermal fatigue rather than the steady creep of a baseload unit. Damage concentrates at tube-to-header attachment welds, in the duct burner region, at the top and bottom of harps where restraint is greatest, and in the low pressure evaporator where flow-accelerated corrosion finds the right temperature window.

The obstacle is access. Tubes are finned, and a contact probe needs a coupling footprint on bare metal, so the technician is working in the narrow gaps between fins or waiting on fin removal. Harps are deep, and the outer rows shield everything behind them. Scopes that were written assuming a probe could be placed anywhere on a tube get rewritten on the scaffold, which is the worst place and the worst moment to rewrite them.

Training for this industry therefore has to include the unglamorous half: what is reachable, what requires fin removal, what requires tube removal, and what has to be covered by an alternative method or by a different access route entirely. A technician who can say on the walkdown that a given row cannot be inspected as specified has saved the outage more than one who discovers it two days in.

Thin wall, tight curvature and the probe that will not work

Boiler tubing is small in diameter and thin in wall, and both properties fight a conventional contact setup. A standard contact probe rocks on a two-inch outside diameter tube, and inconsistent coupling produces readings that wander in a way that looks like corrosion. Contoured wear faces, small-footprint transducers and delay lines exist for this reason, and calibration blocks with matching curvature are not an optional refinement.

Thin wall brings its own limit. Where the wall is short relative to the near field of the transducer, or where the interface ringing has not decayed before the backwall echo arrives, a low-frequency contact probe simply cannot resolve the backwall. Higher frequency and a delay line move the problem out of the way, at the cost of penetration and of sensitivity to surface condition. Knowing which compromise applies to which component is exactly the knowledge that a generic method course does not deliver.

Angle beam work has an analogous problem: on small-diameter tube, the beam skips more quickly and the geometry echoes from the far wall crowd the screen. Procedures that were qualified on plate do not transfer, and technicians who assume they do produce reports full of confidently identified indications that are the tube's own shape.

The location scheme, and the failure that costs the next outage

The most expensive mistake made on power outages is not a missed indication. It is data that cannot be trended. A thickness reading is only useful twice: once now, against the minimum wall, and once in three years, against itself. The second use is worth more than the first, because it is what converts inspection into a replacement plan. It requires that the next crew can put a probe on the same square inch of steel.

That means an agreed location scheme before the outage starts, tied to unit, pass, panel or harp, row, tube number and elevation, with a datum somebody can find with a tape measure. It means the scheme survives into the report, and that the report is delivered in a form the plant's data system can absorb. Photographs of a chalked grid on a tube are not a location scheme, though a great many reports treat them as one.

Technicians resist this because it feels like clerical work stacked on top of technical work under time pressure. The counterargument is simple: the utility is paying for a trend, and a reading without a reproducible location is not part of a trend. It is a number.

What a technician needs before badge-in

The certification baseline is the same as anywhere else in the United States: employer-based certification under a written practice, written to a recommended practice such as SNT-TC-1A, with the certifying Level III named and the training, experience, examination and recertification rules stated. For ultrasonics that means the documented training hours and method experience for the level, general, specific and practical examinations, and current vision records.

What is specific to power generation sits on top. Owners commonly require demonstration on representative mock-ups before a technician is allowed on a scope, particularly for header weld examination, dissimilar metal welds and oxide scale. Procedures must be qualified for the material, geometry and damage mechanism, not adapted on site. Repair work brings the jurisdictional or National Board inspector into the picture, and examination that supports a repair has to satisfy that inspector as well as the owner.

Then there is site access: confined space for drum and duct entry, work at height, energy isolation, and the plant's own contractor orientation. None of it is nondestructive testing, and all of it decides whether a certified technician is actually able to work on the day the unit comes off. Crews that treat this as an afterthought lose shifts they cannot recover.

Request a consultation