Digital Twin for Power Generation Assets: Boiler and Turbine Integrity

An Outage Is a Data Reconciliation Problem With a Clock Running

A planned boiler outage at a coal or combined-cycle plant runs on a scope that's decided months in advance and a schedule that gets more expensive by the day once the unit is offline. The single biggest lever an integrity engineer has over that schedule and scope is how well the plant actually knows the condition of its boiler tube circuits, headers, and turbine components going into the outage — and that knowledge is only as good as how well decades of grid UT thickness mapping, creep life calculations, and borescope findings have actually been reconciled into one asset picture rather than sitting in separate spreadsheets from separate outage cycles.

This is where a digital twin earns its place in a power generation integrity program: not as a visualization novelty, but as the layer that turns “we have fifteen years of tube-wall thickness data across four boiler circuits” into an actual remaining-life picture the outage planning team can use to decide scope before the unit ever comes offline.

Boiler Tube Circuits and Where They Actually Fail

A pulverized coal boiler's pressure parts break down into circuits with genuinely different failure modes, and a useful digital twin has to keep them conceptually separate rather than treating “boiler tube thickness” as one undifferentiated dataset:

• Waterwall tubes — the furnace's primary heat absorption surface, most exposed to fireside erosion from ash particles and, in units burning high-sulfur coal or running reducing combustion conditions near burners, hydrogen damage from localized under-deposit corrosion that decarburizes and embrittles the tube wall from the inside. Hydrogen damage is particularly dangerous because it can produce a tube failure with comparatively little external wall loss visible — the damage mechanism attacks the steel's microstructure, not just its thickness, which is exactly why replication metallography and not just UT gauging matters in suspect zones.
• Superheater and reheater tubes — operating at the highest metal temperatures in the boiler, these circuits are where creep damage dominates: long-term, temperature- and stress-driven deformation that eventually produces intergranular cavitation and cracking, typically at bends, welds, and support attachment points where local stress concentrates. Creep is a time-at-temperature phenomenon, which is exactly why remaining-life assessment for these circuits depends on operating history, not just current-state inspection.
• Economizer tubes — generally cooler and less creep-exposed, but subject to fly-ash erosion, particularly at tube bends and in high-velocity flue gas zones, and to oxygen-pitting corrosion during shutdown periods if layup practices aren't followed.

Each circuit's dominant failure mode drives a different inspection technique emphasis: grid UT thickness mapping for erosion and general wastage trending across waterwalls and economizers, replication metallography and sometimes UT-based creep-void detection for superheater/reheater creep assessment, and targeted hydrogen damage screening (which may include hardness testing or metallographic sampling at suspect under-deposit corrosion sites) for waterwall zones with a known scaling or chemistry history.

Grid UT Mapping: Building the Thickness Picture

Grid UT thickness mapping — taking systematic thickness readings across a tube panel or a defined grid pattern rather than at isolated spot-check points — is the backbone dataset for waterwall and economizer wastage trending. Done consistently across outage cycles, at the same grid locations, it produces a genuine wall-loss-rate trend rather than a single-point condition snapshot. The practical challenge is exactly the one that shows up in every asset-heavy NDT program: unless the grid locations and the resulting thickness data are tied to a persistent, consistent reference system across outage cycles — ideally five, ten, fifteen years of history — the trend calculation degrades into comparing readings that may not actually be from the same physical location tube-to-tube.

A digital twin built around the boiler's actual tube panel geometry, with each grid point tied to a fixed coordinate reference rather than a loosely described location like “waterwall panel 3, lower furnace,” is what makes multi-outage trend analysis reliable instead of approximate. When wall-loss rate at a specific elevation band is trending toward a calculated minimum required thickness, that's the input an integrity engineer needs to decide whether the next outage needs tube replacement in that zone or whether one more operating cycle is defensible — a decision governed by ASME Boiler and Pressure Vessel Code Section I design minimums for new construction and increasingly by Section XI and jurisdictional requirements (many states adopt the National Board Inspection Code, NBIC, for in-service inspection and repair of existing boilers) for the in-service assessment itself.

Creep-Fatigue Life Assessment: Where EPRI Guidance and Fitness-for-Service Meet

High-temperature headers and superheater/reheater tube circuits accumulate creep damage as a function of operating temperature and stress over time — and increasingly, as plants cycle more aggressively to follow renewable generation on the grid, creep-fatigue interaction from thermal cycling adds a second damage mechanism on top of steady-state creep. EPRI has published extensive guidance on creep-fatigue life assessment methodology for fossil boiler components, and API 579/ASME FFS-1 fitness-for-service provides the structured assessment levels (Level 1 screening through Level 3 detailed analysis) that translate inspection findings — replication metallography grade, UT-detected creep void indications, measured wall thickness against design minimum — into an actual remaining-life number or a defensible run/repair/replace decision for a specific header or tube run.

The data reconciliation problem here is genuinely multi-source: creep life fraction consumed depends on operating temperature and pressure history (from the plant's DCS/historian data), current metallurgical condition (from replication metallography, typically performed at accessible header locations and select tube samples), and current wall thickness (from UT). A digital twin that ties these three data streams to the same physical component — this specific header, this specific tube bend — lets the fitness-for-service calculation actually reflect that component's real history rather than a generic assumption applied uniformly across a tube circuit that may have genuinely different local operating conditions at different points along its length.

Combined-Cycle HRSGs: Faster Cycling, Different Failure Emphasis

Combined-cycle plants replace the coal-fired furnace with a heat recovery steam generator (HRSG) — a boiler that recovers heat from combustion turbine exhaust rather than burning fuel directly, still organized into economizer, evaporator, superheater, and reheater sections, but built and operated differently enough that the digital twin's failure-mode weighting needs to shift. Fireside erosion and hydrogen damage, the dominant coal-boiler concerns, are largely absent in an HRSG because there's no coal ash stream and no direct-fired waterwall. What replaces them as the leading integrity concern is cycling-driven fatigue: combined-cycle units, particularly those running two-shift or load-following duty to balance intermittent renewable generation, see far more frequent startup and shutdown thermal transients than a baseload coal unit, and every transient cycles the thick-walled high-temperature headers and drum through a stress range that accumulates low-cycle fatigue damage independent of, and often faster than, straight creep.

This changes what the inspection program needs to track and, by extension, what the twin needs to model. Attemperator spray nozzle and downstream piping thermal fatigue cracking, HRSG harp tube-to-header weld cracking from differential thermal expansion during fast starts, and drum nozzle cracking are HRSG-specific failure locations that don't have a direct coal-boiler analog, and they're driven by cycle count and ramp-rate severity more than by cumulative operating hours at temperature. A twin that only tracks creep life fraction as a function of hours-at-temperature — the right model for a coal unit's superheater — will systematically under-weight the fatigue damage accumulating on an HRSG header that starts and stops daily.

Steam Turbine Integrity: A Different Inspection Discipline Entirely

Turbine rotor and blade integrity inspection runs on methods and failure modes distinct from the boiler side of the plant:

• Rotor bore UT — volumetric ultrasonic examination of the rotor's central bore (where one exists) and forging body, looking for fatigue cracking or forging-originated indications that could propagate under the high cyclic stress a rotor sees, particularly critical at high-pressure and intermediate-pressure rotor sections operating at elevated temperature where creep-fatigue interaction is also a factor.
• Phased array UT on blade roots and dovetails — the blade attachment point is one of the highest-stress-concentration locations in the turbine, and phased array's ability to sweep beam angles electronically makes it well suited to inspecting the complex dovetail or fir-tree root geometry for fatigue cracking without needing a separate probe angle for every scan position.
• Replication metallography on casing and rotor surfaces — used to assess creep damage progression at high-temperature casing and rotor locations, the same fundamental technique used on boiler headers, applied to turbine components operating at comparably high metal temperatures.
• Borescope inspection — for internal blade path, seal, and nozzle condition assessment without full turbine disassembly, often run during shorter outages specifically to decide whether a full major overhaul is warranted at the next planned outage.

These findings live in a genuinely different data structure than boiler tube thickness maps — turbine components are individually identified, individually tracked (a specific rotor forging has its own service history independent of the unit it's currently installed in, since rotors are occasionally moved between units during major overhauls), and the inspection findings need to travel with the component, not just the unit. A digital twin model for turbine integrity has to account for this component-level identity in a way a boiler tube grid map doesn't need to.

Bringing Boiler and Turbine Data Into One Outage Planning Picture

The actual payoff of integrating both sides of the plant into one asset model shows up at outage scope decision time. A defensible scope decision for a major boiler outage depends on knowing, across every circuit: which zones are trending toward minimum thickness and need tube replacement now versus next cycle; which headers have creep life fraction consumed past whatever threshold the plant's fitness-for-service policy sets for mandatory repair versus continued monitoring; and on the turbine side, which rotor and blade findings from the last borescope or UT campaign require this outage's disassembly level versus deferral. Scoping that conservatively — replacing more than necessary — wastes outage days and capital. Scoping it too thin risks a forced outage mid-cycle, which carries real operational cost through lost generation revenue and, depending on the unit's role in the grid, potential reliability penalties — costs that are well understood qualitatively across the industry as consistently exceeding the cost of the additional planned-outage work that would have avoided them, without needing to invent a specific dollar figure for any individual plant's situation.

A twin that holds tube-wall thickness maps, creep life fraction data, and turbine borescope/UT findings against the same asset model — with consistent component and location referencing across outage cycles — is what lets an outage planning team make that scope call from an actual aggregated risk picture instead of separately reviewing four different circuit reports and a turbine inspection binder and trying to mentally merge them into one prioritized list under schedule pressure. This is the same structured-data discipline that makes Atlantis NDT ERP useful for managing the inspection program itself — technician scheduling, CML/finding tracking, procedure compliance — feeding into a digital twin platform where the condition data becomes an actual asset-level integrity picture rather than a stack of individually correct circuit reports.

NBIC and Section XI: Who Requires What

It's worth being precise about which code governs what, since plants sometimes conflate them: ASME Section I sets design and construction rules for new power boilers. ASME Section XI governs in-service inspection primarily for nuclear power plant components, and its methodology has influenced but doesn't directly govern most fossil plant in-service inspection. For existing fossil boilers, the National Board Inspection Code (NBIC) is the standard most jurisdictions actually reference for in-service inspection, repair, and alteration requirements, and it's the NBIC — not Section I — that typically governs how a plant documents and gets jurisdictional acceptance for a repair or a fitness-for-service-based run decision on an existing pressure part. A digital twin's reporting output needs to align with whichever of these actually applies to the plant's jurisdiction and equipment, not a generic “boiler code compliance” label that doesn't survive a jurisdictional inspector's review.

Remaining Life Assessment as an Ongoing Process, Not a One-Time Study

Remaining life assessment studies — often commissioned as standalone engineering projects every five to ten years for major units — are valuable, but they're snapshots. The inspection data collected in every intervening outage either feeds forward into updating that remaining-life picture or it doesn't, and in practice, without a structured model to feed into, it usually doesn't in any rigorous way; the next remaining-life study essentially starts over with fresh data collection rather than building on the accumulated trend. A digital twin that's actively maintained across outage cycles turns remaining life assessment from a periodic expensive study into something closer to a continuously updated output — the underlying trend data (thickness loss rates, creep life fraction consumption) is already sitting in the model between formal studies, so the formal study becomes a validation and deeper-analysis exercise rather than a from-scratch data collection effort.

Where This Fits an Integrity Program Practically

None of this replaces the fitness-for-service engineering judgment that API 579/ASME FFS-1 assessments require, and it doesn't substitute for qualified NDT technicians performing the UT, replication, and phased array work to begin with. What it does is make sure that work compounds in value across outage cycles instead of resetting to zero every time. Atlantis NDT supports this with ASNT Level III consulting for plants building or auditing their boiler and turbine NDT program, ASNT SNT-TC-1A training for technician qualification, and the combination of structured NDT reporting software and the digital twin platform for turning outage after outage of inspection data into an asset model that actually gets more valuable with every cycle instead of just larger.