Recovery Boiler Tube Failure: When the Failed Tube Is Not the Root Cause

01

The failed tube was visible. The mechanism was not.

A leak is found in a recovery boiler circuit. The unit comes down, scaffold goes up, the section is opened, the damaged length is cut out and replaced, the repair is tested, the boiler is dried out and returned to service. The work order closes. Nothing in that sequence is wrong, and nothing in it answers the only question that matters commercially: what removed the metal.

Pitting is a symptom with several possible authors. Oxygen in the feedwater is one. Under-deposit concentration of caustic or phosphate is another. Flow-accelerated corrosion, off-line lay-up corrosion, erosion, fireside attack, localized overheating from deposit or circulation problems and fabrication or weld defects are all capable of producing wall loss that looks similar in a photograph and is not similar at all in the laboratory.

If the mechanism is not identified, the repair is a payment against a recurring liability. The tube is new. The condition that consumed the last one is still in service, still running at the same chemistry, the same load profile and the same firing behaviour. In a recovery boiler, where every unplanned entry carries cost far beyond the weld itself, that distinction is the difference between a maintenance expense and an asset problem.

A repaired tube may restore operation. An unresolved dissolved-oxygen mechanism can still destroy the asset.

02

Why recurrence, not failure, destroys the economics

Investors and lenders tend to price a single tube failure correctly. They rarely price recurrence correctly, because the diligence file usually shows the repairs and not the pattern behind them.

Each event in a recovery boiler carries a stack of cost that has very little to do with the length of tube replaced: forced outage and lost production, black liquor handling and disposal during the outage, cooldown and isolation, scaffold and access, inspection and NDE, welding and post-weld requirements, hydrostatic testing, boil-out or chemical cleaning when indicated, refractory and casing restoration, dry-out, ignition and firing stabilization, tuning, and the elevated risk window around every restart. Replacement steam or replacement power may be purchased through the whole of it.

Three or four of those events per year will change the availability assumption a model was built on. Sustained, they change the financing case, then the maintenance budget, then the decision whether to invest in the asset at all. Plants are not usually lost to one dramatic failure. They are lost to a mechanism nobody closed out.

• A repaired tube restores capacity. It does not restore confidence in the remaining circuits.
• Repeat failures at different elevations in the same circuit indicate a systemic condition, not a local defect.
• Repeat failures at the same location indicate a design, flow or support condition, not a chemistry condition.
• An outage taken to fix one leak almost always finds more when the inspection scope is honest.
• The cost of the repair is rarely the cost of the failure.

03

What is physically found, and how it is characterized

The physical evidence phase is not an inspection walk. It is the construction of a measured record capable of supporting or excluding each candidate mechanism, and of showing rate of loss rather than a single reading.

Damage morphology is the primary discriminator. Deep, discrete, hemispherical pits with tubercles differ from broad gouging under deposit, from the orange-peel scalloping typical of flow-accelerated corrosion, from thinning on the fireside crown, and from cracking at attachments. Location within the circuit, orientation relative to flow, and whether the attack is waterside or fireside narrow the field before any laboratory work begins.

Visual and Outage Inspection

Systematic examination of accessible waterwall, generating bank, superheater, economizer, floor and spout areas, with damage mapped by elevation, circuit and orientation rather than described narratively.

Videoscope and Borescope

Internal examination of drums, headers and opened circuits to see waterside surface condition, deposit distribution and whether attack is general or localized.

UT Thickness Mapping

Conventional and phased array thickness grids and C-scan mapping over defined areas, repeated against prior surveys so the output is a loss rate, not a number.

Thickness Trending

Comparison against historical inspection data to establish when loss started and whether it accelerated after a chemistry excursion, a load change or a modification.

Tube Sample Removal

Removal of representative failed and unfailed sections from the same circuit, so the laboratory has both damaged material and a reference condition.

Deposit Evaluation

Deposit weight density and deposit chemical analysis, which govern under-deposit mechanisms, metal temperature elevation and the need for chemical cleaning.

Metallurgical Examination

Sectioning, metallography, hardness traverse, and SEM with EDS on corrosion products, to identify the mechanism and the species driving it.

Weld, Support and Restraint Review

Examination of attachments, buckstays, seals and penetrations where thermal expansion is restrained, which is where fatigue rather than corrosion tends to appear.

Header, Drum and Circuit Review

Assessment of distribution, feeder arrangement, drum internals and separation, and whether any circuit is receiving less flow than design assumed.

History Reconstruction

Inspection reports, prior repair locations, hydrostatic test records, chemical cleaning history, and the startup, shutdown and upset log across the same period.

04

Establishing the damage mechanism by elimination

A defensible investigation carries every credible mechanism forward until the physical, chemical and operating record eliminates it. Adopting the first plausible explanation is the most common failure of boiler root cause work, and the easiest to attack later in a claim or a dispute.

Several mechanisms can also be present at once. Oxygen attack during idle periods and under-deposit corrosion in service are not competing hypotheses if the evidence supports both. The investigation has to be able to say which initiated the loss and which accelerated it.

• 01. Oxygen pitting. Discrete pits with characteristic corrosion products, often concentrated in the economizer and in circuits exposed during lay-up or startup. Supported by feedwater dissolved oxygen history and iron transport data.
• 02. Under-deposit corrosion, caustic gouging or acid phosphate attack. Loss beneath deposit in high heat flux zones, correlated with deposit weight density and with the boiler water treatment regime.
• 03. Flow-accelerated corrosion. Smooth, scalloped wall loss in the feedwater train and economizer, governed by pH, temperature, oxidizing potential and local velocity or geometry.
• 04. Corrosion fatigue and thermal fatigue. Cracking initiating at waterside pits or at restrained attachments, driven by cycling, rapid startup or repeated upsets rather than by chemistry alone.
• 05. Circulation starvation and localized overheating. Reduced flow in a circuit, restriction, or deposit-driven metal temperature rise, producing overheat morphology and microstructural change rather than corrosion products.
• 06. Fireside mechanisms. Sootblower erosion, ash and alkali attack in superheater and generating bank, and lower furnace wall thinning, all of which are fireside in origin and must not be attributed to feedwater chemistry.
• 07. Composite tube and penetration cracking. Cracking at floor tubes, spout openings and air port openings, a recovery boiler specific concern with its own inspection approach.
• 08. Lay-up and startup corrosion. Damage accumulated when the unit was off line, which will not appear in any operating chemistry trend because the boiler was not operating.
• 09. Fabrication, weld and material defects. Dissimilar metal welds, original fabrication defects, prior repairs, and material substitution, established by records and by examination rather than assumed.

05

Why elevated dissolved oxygen matters

Dissolved oxygen attacks the protective magnetite layer and drives localized pitting, preferentially in the economizer and in the feedwater path where temperature rises and the gas is least soluble. It is a mechanism that can run for years at levels that look unremarkable on a daily log and are severe in cumulative terms.

The relevant engineering expectation is design-based. BLRBAC guidance for recovery boiler feedwater treats the deaerator as equipment designed to deliver feedwater below roughly 7 ppb dissolved oxygen without relying on chemical treatment, and treats sustained operation above that expectation as a condition capable of damaging the economizer. That gives an investigation a reference point that does not depend on any individual plant's internal target.

The distinction that matters is between deaeration and scavenging. An oxygen scavenger is a polishing step for the residual the mechanical deaerator was not designed to remove. When scavenger dosing rises to hold a residual, the plant is not controlling oxygen. It is compensating for a deaerator that is no longer performing, and it is doing so downstream of part of the equipment that oxygen damages first.

Evidence that supports or weakens the oxygen hypothesis is rarely the dissolved oxygen reading alone. It is the combination of pit morphology and corrosion products, iron transport trends, scavenger consumption per unit of feedwater, the correlation between chemistry excursions and later discovered damage, and whether the sample system was representative enough for the readings to mean anything.

06

The deaerator as the key suspect

If the feedwater carries oxygen, the deaerator is where the question goes next, and the interrogation is mechanical and operational before it is chemical. A deaerator can hold pressure, show a normal outlet temperature and still deaerate badly.

Approach to Saturation

Outlet water temperature compared against saturation at operating pressure. A persistent approach below saturation is direct evidence that non-condensable removal is incomplete.

Steam Pressure Stability

Pressure swings, pegging steam behaviour and response to load change. Rapid depressurization can release dissolved gas back into solution downstream.

Vent Setting and Vent Flow

Whether the vent is actually passing the continuous flow the design requires. Vents throttled to reduce plume, noise, icing or apparent steam loss are a recognised route to poor deaeration.

Plume Behaviour

Vent plume character as an indirect indicator of vent flow and of whether the vent condenser or orifice arrangement is functioning as intended.

Internal Condition

Tray condition, displacement or collapse, spray valve and nozzle condition, spring tension, fouling and corrosion products in the storage section.

Level and Storage Retention

Level control stability and residence time in storage, which affects both scavenger reaction time and re-aeration risk.

Air In-Leakage

Condensate return contamination, vacuum condensate paths, pump seals and makeup water introduction points, any of which can load the deaerator beyond its design duty.

Instrumentation Reliability

Calibration and representativeness of dissolved oxygen analyzers, sample coolers, sample line material and flow rate. Unrepresentative sampling produces confident, wrong data.

Scavenger Demand

Chemical consumption trended against load and makeup. Rising demand at constant duty is one of the earliest available indicators of deteriorating mechanical deaeration.

Operating Practice

Startup sequence, makeup surges, deaerator operation during low load and off-line periods, and whether the equipment was operated as designed or worked around.

07

Chemistry cannot substitute for mechanical deaeration

There is a pattern that recurs across plants with recurring tube damage. Oxygen appears in the feedwater. Scavenger dosing is increased. The residual returns to target. The daily log looks correct. The mechanical problem is never corrected, and the damage continues at a slower rate in a place nobody is measuring.

This compensation trap is visible in the data if the right series are examined together: scavenger consumption against feedwater flow, iron transport against time, dissolved oxygen variance rather than average, and the relationship between off-line periods and subsequently discovered pitting. Averages hide excursions, and it is excursions that remove metal.

It also has a claims and diligence consequence. A plant that has been chemically compensating for years has an equipment problem with a documented paper trail showing normal chemistry. Records of that kind are frequently presented as evidence that chemistry was controlled. Read correctly, they are evidence of how much chemical was required to make it look that way.

08

Combustion, heat flux and stress as parallel lines

Restricting a recovery boiler investigation to water chemistry is the second common failure. Tube life is set by the combination of waterside environment, metal temperature and mechanical stress, and the fireside controls two of those three.

In a black liquor fired unit, liquor solids, spray pattern and gun condition, char bed height and stability, air system balance between primary, secondary and tertiary, and furnace temperature profile all shape where heat flux is highest and where it moves. Combustion instability, frequent blackouts and unstable bed behaviour produce thermal cycling in the lower furnace and in the circuits that surround it. In a wood or biomass fired unit the equivalent variables are fuel moisture and sizing variability, bed agitation, fines carryover and alkali-driven fireside attack.

On the waterside and pressure-part side the parallel variables are deposit-driven metal temperature rise, circulation adequacy at operating pressure, header distribution, restraint and differential expansion at attachments, sootblower operation and lance alignment, attemperation behaviour where superheater metal temperature is being managed, and the number and severity of startup and shutdown cycles the unit has accumulated.

Where the record does not support a conclusion on these, they are reported as open diligence lines, not findings. The distinction between what the evidence establishes, what it is consistent with, and what could not be excluded is the part of the report that survives cross examination.

• Liquor solids, spray pattern, gun condition and swirl, and their effect on where heat is released.
• Char bed profile and stability, bed temperature and blackout or upset history.
• Air distribution and combustion uniformity across the lower furnace.
• Deposit weight density and its effect on waterside metal temperature.
• Circulation ratio, feeder and header distribution, and any circuit restriction.
• Sootblower coverage, erosion pattern and lance condition.
• Cycle count, ramp rates and the thermal fatigue duty actually imposed on the unit.

09

What correction looks like

Correction has three parts, and the common error is to deliver only one of them. Chemistry control is tightened, monitoring is made representative, and control limits are set on variance rather than on daily averages. Operating practice is changed where the practice is part of the mechanism, including startup sequence, lay-up regime, load ramping and how the deaerator is run at low load. Then the equipment itself is put back into the condition its design assumed.

Where venting had been restricted because the vent plume created condensate, icing and nuisance conditions on the exterior of the building, the engineering answer is not to accept degraded deaeration. It is a site-specific mechanical design that preserves the vent flow the deaerator requires while managing the discharge: vent routing and termination, silencer sizing on acoustic and flow duty, condensate capture and drainage, and freeze protection appropriate to the climate. Described accurately, this is a site-specific mechanical redesign, not a novel product.

The engineering logic is the point. If the deaerator has to vent to function, and venting creates an exterior hazard, then the hazard is the design problem to solve. Throttling the vent solves the visible nuisance and transfers the cost into the pressure parts, where it is invisible until a tube leaks.

10

Planning the repair or alteration for least total cost

Diagnosis without execution planning leaves most of the available savings on the table. The repair strategy is an engineering and cost problem in its own right, and the largest variable is how much of the scope is decided before the outage rather than discovered during it.

Scope sequencing comes first: which circuits are opened, in what order, and what the decision criteria are for extending replacement beyond the failed length. Access and scaffold strategy is usually the single largest discretionary cost, and it is set by the inspection plan, so the inspection plan has to be finalized before scaffold is designed rather than after. Isolation, drain, cooldown and boil-out or chemical cleaning are scheduled against the deposit evidence, not by habit.

The NDE plan is sized to risk: full coverage where the mechanism is active and the consequence is high, sampling elsewhere, with acceptance criteria agreed in advance so findings do not stall the critical path. The repair route is selected and documented against ASME Section I construction requirements and the applicable repair and alteration authority, with ASME PCC-2 and National Board practice used where they apply, including weld procedure, preheat and any post-weld requirements. Hydrostatic test strategy is chosen deliberately, since test scope, fill water quality and drain and dry-out practice can themselves introduce the corrosion the outage was called to stop.

Return to service is part of the engineering scope, not a handover: dry-out, lay-up until startup, ignition and firing stabilization, tuning against the combustion evidence gathered during the investigation, and a defined post-repair chemistry and monitoring watch period so the correction is verified rather than assumed.

• Inspection plan before scaffold design, so access is built once for a known scope.
• Pre-agreed acceptance criteria and repair decision rules, to remove outage-day engineering delay.
• Long-lead tube, panel and header material identified from the damage mapping, not from the first leak.
• Chemical cleaning decided on deposit weight density evidence rather than on elapsed time.
• Hydrostatic and lay-up water quality specified, so the test does not seed new damage.
• Startup, ignition and tuning treated as engineered steps with data capture, not as operations recovery.
• A defined post-repair verification period with chemistry, iron transport and thickness re-survey points.

11

The diligence read

A conventional technical diligence review reads a maintenance history and reports that tube repairs were completed, inspections were performed and the unit passed its tests. Every one of those statements can be true while the asset carries an unpriced recurring liability.

The engineering read is different. It asks whether each failure was root-caused or simply repaired, whether the same mechanism appears in different circuits, whether thickness trends show a loss rate that will require capital intervention inside the holding period, whether chemistry data is representative, and whether the deaerator, combustion system and operating regime that produced the damage have actually changed.

That difference is directly financial. It moves the expected availability, the maintenance CAPEX line, the outage frequency assumption and, in a levered structure, the covenant headroom. It can also determine whether an insurance programme responds to recurrence or treats it as a maintained condition.

12

What a buyer, lender or insurer should ask

These questions are answerable from records a competent operator already holds. Difficulty in answering them is itself a finding.

• What is the three to five year history of feedwater dissolved oxygen, including excursions rather than monthly averages?
• Were previous tube failures root-caused with laboratory examination, or repaired and closed?
• What do tube samples, deposit weight density and thickness trends show about rate of loss?
• Is deaerator performance consistently within its design expectation, measured as approach to saturation as well as outlet oxygen?
• Are oxygen scavenger consumption trends rising at constant duty, and if so, why?
• Is the chemistry sample system representative, calibrated and independently verified?
• What is the repeat outage and forced outage pattern over the last five years, and what did each event actually cost including lost production?
• What inspection, replacement or chemical cleaning CAPEX is credibly due in the next three to five years?
• Are liquor firing, spray pattern, char bed stability or fuel variability conditions contributing to thermal cycling and tube stress?
• What is the lay-up practice when the unit is off line, and is it documented and followed?
• Is the current operating regime, including cycling and load profile, sustainable for the pressure parts as they exist today?

13

Tools, instruments and analysis stack

The credibility of the conclusion depends on the evidence behind it. These are the categories of capability an engagement of this type uses, selected for the question being answered rather than applied indiscriminately.

Plant Diagnostic

Conventional and phased array UT thickness mapping, IRIS and remote-field or eddy current tube inspection, videoscope and borescope access, infrared thermography, online dissolved oxygen analyzers with representative sample coolers, specific and cation conductivity, pH and iron transport monitoring.

Specialist Failure Analysis

Tube sample sectioning, metallography and hardness traverse, SEM with EDS on corrosion products and deposits, deposit weight density determination, PMI for material verification, and laboratory water and deposit chemistry.

Engineering Evaluation and Modeling

Boiler circulation and thermal-hydraulic assessment, heat flux and metal temperature evaluation, finite element stress and thermal fatigue analysis where geometry and cycling justify it, acoustic and flow sizing for vent and silencer design, CAD and structural design for the modification, and CFD only where plume dispersion or furnace flow genuinely requires it.

Data and Reliability Method

AVEVA PI or equivalent historian correlation of chemistry, load, firing and upset data against failure dates, logic tree root cause analysis, FMECA and criticality ranking to set inspection and capital priority, and recurrence analysis across circuits and outages.

Records and Diligence

Inspection and NDE report review, work order and CMMS history, chemical consumption records, hydrostatic and repair documentation against ASME Section I and the applicable repair authority including ASME PCC-2 practice, modification and design change records, and operator log reconstruction.

14

Fixing the tube restores service. Fixing the mechanism protects the asset.

A specialist will usually identify one symptom correctly. A metallurgist will characterize the pit. A chemist will describe the excursion. A boiler contractor will execute a sound repair. None of those, on their own, tells an owner or a buyer whether the condition is over.

BlackOut Power Group integrates the evidence across the full system: tube condition and NDE, damage mechanism and metallurgy, feedwater and boiler water chemistry, deaerator and feedwater equipment performance, combustion and firing behaviour, circulation and stress, and the repair or alteration strategy that closes it out at the lowest defensible total cost. The output is not a list of observations. It is a statement of what caused the damage, what remains at risk, what it will cost to correct, and what happens to availability if it is not corrected.

For an owner, that is a reliability decision. For a buyer, a lender or an insurer, it is a valuation decision, and it is better made before capital is committed than after the third outage of the year.