Transformer Intelligence for Offshore Wind Farms

Figures are planning benchmarks drawn from offshore operating experience and current lead time quotations. They vary by site, water depth, vessel market, and turbine platform. Use a fleet-specific model before committing any of them to a business case.

The Offshore Wind Operator's Transformer Risk Profile

Offshore wind imposes requirements on transformer assets that onshore utility practice does not prepare an operator for. Utility fleets are built with maintenance infrastructure, spares programs, and designed redundancy. An offshore wind farm depends on single-point-of-failure assets mounted inside individual turbine nacelles, plus one main transformer at the offshore substation carrying the entire project output. A transformer failure offshore does more than interrupt local generation. It starts a marine logistics chain the operator cannot compress. Access requires a specialized vessel, qualified technicians, and a weather window that may not open for weeks. The sections below set out the risk drivers in order, from the physical layout of the fleet to the specific failure modes VIE detects.

1.1 Three Transformer Populations, One Access Constraint

An offshore wind farm carries transformers in three places, and they present different problems.

Nacelle step-up transformers. One per turbine position, typically 2,000–5,000 kVA, mounted inside the nacelle or at the tower base. Many low-value units, widely dispersed, each individually affordable to lose and collectively impossible to inspect on any useful cadence.

The offshore substation main transformer. One or a small number of units, typically 100–400 MVA, stepping array voltage up to export voltage. High value, no on-site spare, and a failure that stops the entire project rather than one turbine.

The onshore grid connection transformer. Conventionally accessible and conventionally maintained. It is the only unit in the chain where standard utility practice works without modification.

The first two share one constraint. Reaching them requires a vessel and a weather window. That constraint sets the real diagnostic interval, regardless of which tests are technically available on the unit.

1.2 Decision Latency: The Root Cause of Preventable Failures

The greatest risk in transformer asset management is not the failure. It is decision latency. Operational data exists, but usable insight arrives too late to act on. Schedule-based asset management relies on periodic manual inspection with long blind intervals between visits. Threshold alarms sound only after an internal fault is already underway. By the time a standard measurement confirms a fault, most of the P-F interval has been consumed.

The P-F interval is the window between the point at which degradation becomes detectable and the point of functional failure. For an operator facing a 2–3 year procurement lead time, a compressed P-F interval removes every planning option that matters. The unit fails before a replacement can be ordered, built, shipped, and lifted.

Key InsightEvery month of advance warning carries measurable financial value when replacement cycles stretch 2–3 years. Continuous fleet-wide intelligence extends the actionable lead time from days to months.

1.3 Two Insulation Systems, Two Diagnostic Gaps

Nacelle transformers are built two ways, and the offshore population includes both. Cast resin dry-type units are common because they carry no fluid and remove a fire load from the nacelle. Liquid-filled units are also widely used, frequently with synthetic or natural ester fluid selected for its high fire point and environmental profile. An operator may run both across a single portfolio, and in some cases across a single project.

Each construction leaves a different diagnostic gap. Neither gap is closed by the chemistry the utility industry relies on.

The conclusion holds for both populations. The mechanical and thermal degradation pathways that dominate nacelle service produce no chemical signature until they have already progressed to abrasion or arcing. On a dry-type unit there is no chemistry to wait for. On a liquid-filled unit the chemistry exists but arrives on a vessel schedule. In both cases the pathway that fails the asset is a mechanical one, and it is observable continuously from outside the enclosure.

1.4 Marine Environment, Thermal Cycling, and Dielectric Breakdown

Temperature fluctuation accelerates internal degradation in both constructions, by different mechanisms.

Dry-type. The insulation system is cast resin or vacuum pressure impregnated, laminated against the core and windings. Wind velocity fluctuation and variable generation load drive rapid heating, forcing thermal expansion across the solid section. Rapid cooling from load drops or wind cut-outs reverses the path. Cast resin and Grain-Oriented Electrical Steel have different coefficients of thermal expansion, so each cycle works the interface between them. Micro-fissures and delamination develop over time. Salt-laden humidity penetrates the fissures and introduces tracking paths across the dielectric surface. This lowers the effective breakdown voltage and opens a window for partial discharge, micro-arcing, and flashover.

Liquid-filled. The fluid provides both dielectric strength and thermal transport, and a nacelle unit carries a small volume of it relative to a power transformer. That reduces thermal mass, so temperature excursions are faster and larger for the same load step. Cycling drives expansion and contraction of the fluid, which works seals and gaskets in a salt atmosphere. Moisture that reaches the paper accelerates cellulose aging and lowers dielectric strength, and moisture partitions between paper and fluid as a function of temperature, so a sample drawn on a cool day understates what the paper holds.

Point-in-time sampling misses transient, temperature-driven events in both cases. VIE evaluates high-frequency acoustic and micro-vibrational signatures alongside localized thermal behavior under live load, which captures the onset of partial discharge, micro-arcing, and asymmetric heating while the unit is in service rather than during a scheduled visit.

1.5 Nacelle Mechanical Fatigue, Aerodynamic Vibration, and Core Warpage

Wind turbine transformers operate in a vibrational environment that has no equivalent in a substation. Thermal cycling, wind turbulence, blade rotation imbalance, and drivetrain dynamics generate mechanical shear across the core and winding assembly. The magnetic core is built from thin, tightly stacked sheets of Grain-Oriented Electrical Steel. Where the windings are encased in solid resin, the expansion mismatch between resin and steel adds localized shear on every load and ambient cycle. Where the windings are paper insulated in fluid, the same cycling works the clamping structure that holds them in place.

Compounded by continuous low-frequency structural vibration inside the nacelle, this cumulative fatigue produces core warpage, winding looseness, fastener degradation, and, in cast resin units, resin tearing. Once geometric distortion compromises the assembly, the unit can no longer absorb routine electromagnetic transients without further damage, and the degradation becomes self-accelerating.

VIE targets this mechanism directly. Sensors attach non-invasively to the outer enclosure and capture the mechanical vibration fingerprint of the unit under live load. A matrix of triaxial vibration features isolates the acoustic signatures associated with winding looseness, loose core clamps, and lamination warpage, which identifies physical distortion well before it presents as an electrical fault.

1.6 Financial and Marine Operational Exposure

A nacelle transformer failure costs $2–5 million USD under normal equipment replacement benchmarks. Offshore logistics push that figure higher. Remote location, limited specialty vessel and jack-up availability, offshore labor premiums, and narrow weather windows drive emergency response to 5–10 times the cost of an equivalent planned onshore intervention.

Direct costs include heavy-lift vessel mobilization, secondary marine support, emergency transport, and procurement premiums. Indirect costs include extended turbine downtime, lost generation revenue, and exposure under the offtake agreement. An uncontrolled electrical failure inside a nacelle also carries a fire risk, which brings structural safety consequences and regulatory reporting obligations that can exceed the equipment cost by a wide margin.

1.7 Workforce Attrition and the Knowledge Gap

Technical expertise in transformer condition assessment is leaving the industrial workforce. Personnel retire or move faster than operators can replace them, and specialized institutional knowledge leaves with them. Operators lose the ability to interpret field test results, recognize early mechanical degradation signatures, and place a single reading in the context of fleet-level trends across identical turbine models.

Traditional asset management depends on specialist inspection and site visits. That dependency does not survive contact with a distributed offshore fleet. What the work requires is a system that encodes assessment expertise into the platform and operates without an on-site specialist in the loop.

1.8 Siloed Assets in a Distributed Turbine Fleet

Most operators manage nacelle transformers as individual units inside separate turbines, each with its own maintenance history and isolated risk profile. That structure misses fleet-level patterns. A manufacturing or stress-driven failure mode appearing across multiple units under similar wind load profiles stays invisible until individual assets cross a threshold one at a time. A degradation trend that accelerates under a specific load profile goes undetected. A position-dependent thermal stress pattern produces no warning signal. These patterns are only visible in a cross-asset comparison.

1.9 Electrical, Mechanical, and Thermal Failure Modes in Nacelle Service

Nacelle step-up transformers experience a failure mode profile that differs from utility service. The table below maps that profile to the VIE diagnostic that observes it. Every mechanism listed applies to both dry-type and liquid-filled construction.

Wind turbine generation blocks rely on solid-state converters to match grid frequency and synchronize voltage output. These non-linear systems can inject a small direct current component into the low voltage winding through switching asymmetry and DC link voltage imbalance. Where a winding neutral is earthed, that current also finds a return path through the turbine earthing system, the tower and foundation steel, and the array cable screens that bond each turbine back to the offshore substation earth. Either path creates a DC magnetic bias in the core.

Winding configuration matters here. A delta low voltage winding blocks the ground return path by design, which leaves direct injection into the winding as the remaining mechanism. Neither path is externally measurable on a nacelle unit without continuous magnetic sensing.

The bias shifts the alternating operating flux into the saturation region of the GOES curve. That magnifies core magnetostriction, raises core losses, produces asymmetric heating, and generates characteristic even-harmonic mechanical vibration. Manual electrical test methods do not detect this pathway until structural damage has occurred. On a dry-type unit there is no gas signature to detect it either, and on a liquid-filled unit the gas signature appears only after the thermal consequence is advanced. VIE identifies core saturation as a leading indicator, from the magnetic and vibrational signature, before any lagging-indicator confirmation is measurable.

VIE Continuous Fleet Intelligence

VIE converts individual transformer condition data into fleet-level intelligence. Non-invasive sensors attach externally to the transformer enclosure. Deployment requires no outage and no integration with the operator's IT infrastructure. The platform uses long-life wireless sensors with a battery life exceeding 10 years, operates with end-to-end encryption, and supports up to 128 sensors per gateway. Nothing in the sensing method depends on the presence or absence of insulating fluid, which is why the same deployment applies to dry-type and liquid-filled units without modification.

Deployment AdvantageSensors install in minutes. The process requires no downtime, no outage, and no integration work. Continuous data collection begins at first installation.

2.1 From the P-F Interval to Long Lead Time

The P-F interval describes the window between potential failure and functional failure. Manual electrical testing and visual inspection detect degradation late in that interval, which leaves a short lead time for offshore planning. Continuous vibration-based sensing detects structural and electrical degradation earlier on the P-F curve, identifying micro-vibrational and acoustic patterns before thermal breakdown occurs. That extends the available planning window from days or weeks to months.

The timing difference is what matters for an operator facing a 2–3 year procurement lead time. It separates an orderly lifecycle decision and a scheduled vessel lease from an emergency marine logistics response.

2.2 How Continuous Intelligence Works

Sensors capture the vibration fingerprint of the transformer's internal activity under live operating load. The platform observes winding electromagnetic behavior, core lamination dynamics, core structure convection patterns, and partial discharge transients. A cloud-based engine processes these signals and correlates them against public marine weather data, turbine load profiles, and transformer metadata.

The engine uses a matrix of triaxial vibration features that map specific vibration signatures to specific failure modalities across eleven diagnostic dimensions, including core looseness, DC bias, insulation loss, overheating, partial discharge, arcing, and winding deformation. Surface temperature at multiple sensor positions adds further variables. The engine compares temperature readings within a single transformer to identify asymmetric heat distribution, which indicates internal tracking or hotspots, and compares readings across units on identical turbines at the same site to identify outliers. That cross-asset comparison is only possible when every unit is measured continuously and simultaneously.

2.3 The Three-Tier Indicator Framework

VIE structures diagnostic output into three tiers that map to decision urgency.

‍

The leading tier carries the highest operational value offshore, because it identifies conditions for accelerated deterioration before electrical breakdown or thermal runaway, which is where the lead time that vessel and procurement cycles demand comes from. The coincident tier identifies active faults that require rapid intervention, including the conditions that precede a nacelle fire. Lagging tests then confirm the finding and close the loop.

The lagging tier is where the two insulation systems diverge. A liquid-filled unit has both electrical and chemical confirmation available. A dry-type unit has electrical confirmation only. In both cases confirmation requires a nacelle visit, which is precisely why the leading tier has to carry the planning decision.

2.4 Fleet Intelligence, Not Isolated Asset Management

VIE functions as a continuous learning layer across the whole turbine transformer fleet. Cross-asset pattern recognition builds failure signature libraries that improve with each evaluated unit. Risk ranking updates continuously, so operators can align vessel deployment with actual fleet risk rather than an inspection calendar. The platform accumulates maintenance records, electrical test data, and turbine load profiles over time and learns what degradation looks like under the operating conditions of each turbine position.

This addresses the workforce problem directly. Assessment logic encoded in the model persists through personnel turnover, and accuracy improves as the fleet history grows rather than resetting when an experienced engineer leaves.

2.5 Wind Converter Harmonics and Variable Current Loading

Offshore turbines use solid-state frequency converters and switching rectifiers to adapt variable rotor speed to fixed grid frequency. These non-linear electronics inject high-order current harmonics back into the nacelle step-up transformer windings. Harmonic current raises winding eddy current losses and stray losses in the structural steel enclosure, which creates localized thermal hotspots and accelerates dielectric aging. Converter-induced harmonics also drive mechanical resonance in the core laminations, which accelerates structural loosening.

The thermal consequence differs by construction. A dry-type unit has no fluid convection to move heat away from a stray-loss hotspot, so local temperature rises faster for the same harmonic content. A liquid-filled unit moves the heat but pays for it in accelerated fluid and paper aging. Harmonic exposure is also decoupled from useful output, so a turbine running at low output can carry close to the full harmonic burden while carrying very little load. A load-based thermal estimate does not see that, because it assumes losses scale with output.

VIE captures harmonic loading signatures in the vibration record, provides a continuous account of the mechanical stress from converter loading, and flags when that stress moves outside healthy operating bounds.

2.6 The Industry Transition

Infrastructure operators across sectors are moving from schedule-based asset management to condition-based continuous intelligence. In offshore wind the pressures driving that move are distributed fleets, workforce and logistics constraints, grid code obligations, capital constraints, and uptime commitments to offtakers and lenders.

Applying Continuous Intelligence Offshore

Offshore operation adds geographic dispersion across a marine array, constant salt and humidity exposure, and electrical stress modes specific to converter switching. These conditions set the execution principles below.

3.1 Zero-Downtime Deployment in Active Nacelle Environments

Sensors deploy in minutes. Installation causes no turbine interruption, requires no access to internal electrical compartments, and demands no integration with operational IT or turbine SCADA. The system operates independently, transmitting encrypted data through its own wireless gateway infrastructure. Shutting down generating turbines to install diagnostics is not a realistic option offshore, which makes a non-invasive deployment profile a prerequisite rather than a convenience.

3.2 Distributed Remote Fleet Management Across Marine Arrays

An offshore wind farm carries a transformer at every turbine position, one or more at the offshore substation, and one at the onshore grid connection. No two of them are reachable on the same trip. Managing transformer condition across that footprint through manual inspection creates blind intervals measured in months, because every visit depends on vessel availability and a marine weather window.

Continuous remote monitoring closes these blind intervals on every monitored asset. The platform delivers fleet risk visibility across all instrumented units at once, so operations teams can prioritize field intervention by actual asset stress and thermal outlier ranking rather than by calendar. When a vessel does go out, the trip is planned around units that are actually showing a developing issue.

3.3 The Offshore Substation Transformer

The nacelle fleet and the offshore substation transformer present opposite monitoring problems, and an operator has both.

The nacelle population is many low-value units where inspection cannot economically scale. Losing one curtails one turbine. The offshore substation main transformer is a single high-value unit where inspection is affordable but a miss is unrecoverable. It carries the full project output, it has no on-site spare, and replacement requires a heavy-lift vessel, a purpose-planned lift, and a manufacturing queue. Large power transformer lead times averaged around 128 weeks as of the second quarter of 2025, with some units beyond four years.

A failure there does not curtail generation. It stops the project, and it does so for a period measured in years rather than weeks. That exposure typically reaches the offtake agreement, the availability guarantee, and the lender covenants at the same time.

VIE addresses both populations with the same sensing method and the same platform. On the nacelle fleet the value is ranked visibility without a truck roll or a boat trip to every position. On the offshore substation transformer the value is continuous, asset-specific condition data on the single least redundant piece of equipment in the project.

3.4 Capital Planning Under Procurement and Vessel Constraints

An extended detection window addresses the 2–3 year replacement lead time directly. When the platform identifies a unit entering an elevated risk trajectory, the operator has time to initiate procurement and lease specialized vessels through normal channels rather than the spot market. Rising winding looseness trends, increasing structural heat flux, and insulation degradation markers are all detectable early enough to avoid emergency charter premiums.

Avoided emergency mobilization, extended transformer life, and deferred capital spend typically produce a payback period in the range of 9–15 months on the VIE deployment.

On The Payback FigureThe 9–15 month range is an output of the VIE ROI calculator, not a fixed result. It moves with fleet size, unit rating, vessel day rate, capacity factor, contracted or merchant energy price, and the avoided-mobilization assumption applied. Request a fleet-specific model from your VIE Technologies representative before using the figure in a business case.

3.5 Risk Stratification Before Adverse Weather Windows

Continuous leading-indicator evaluation identifies the highest-risk units in a fleet before an extreme weather event arrives. These units often perform acceptably under moderate conditions while carrying structural or insulation vulnerabilities that a high-stress event exposes. Surfacing that latent risk in advance gives operators a defined window to schedule maintenance, targeted testing, or planned replacement while logistics are manageable and costs are predictable.

Risk-Based Action Framework for Offshore Wind Operators

VIE translates continuous condition metrics into prioritized action triggers. The framework accounts for long procurement lead times, offshore vessel logistics, and nacelle risk exposure. Thresholds below are the default starting points and are calibrated per fleet during deployment.

The loop improves with customer data. As operators share field test results and maintenance records, VIE calibrates turbine-specific thresholds and improves alert specificity. Providing existing baseline insulation resistance records is the first data exchange cycle, delivers the largest single accuracy improvement, and is the recommended first step after deployment.

Evidence from Deployed Assets

The mechanisms described above are not theoretical. The cases below are drawn from live VIE deployments and are presented in anonymized form. Each was corroborated by an independent method the customer controlled.

Scope Of This EvidenceThese deployments are on oil-filled power and distribution transformers in utility and data center service. They are not nacelle-mounted, and they are not offshore. They demonstrate that the failure mechanisms this paper describes are detectable from outside the tank and that the detections hold up against independent testing. They are not measurements taken at an offshore wind farm. Section 6 states where the evidence base does not yet extend.

5.1 Four Months of Lead Time, Confirmed by Two Independent Methods

A global data center operator deployed VIE across 50 oil-filled transformers rated 2,000–5,000 kVA. That is the same rating range as a nacelle step-up transformer, on a comparable duty profile of continuous load with switching-derived harmonic content.

Working from vibration and surface temperature alone, VIE identified the units four months before invasive testing confirmed severe degradation. Two independent methods, electrical and chemical, agreed with the original ranking. For an offshore operator, four months is the difference between a planned vessel booking and a spot charter.

5.2 Mechanical Degradation on Units That Passed Conventional Testing

Four 220/66 kV power transformers at a single transmission substation, each from a different manufacturer. All four passed insulation resistance testing, rated above the applicable threshold. Seven sensors and two gateways per unit, monitored continuously from February 2025.

This case matters offshore for a specific reason. Insulation resistance is the test an offshore operator is most likely to have on file, because it is the quickest to run during a short nacelle visit. A pass or fail insulation test returned the same verdict for all four units here. Continuous mechanical monitoring separated them and singled out Unit C for developing core looseness, which is the same degradation pathway nacelle vibration drives.

5.3 Partial Discharge Detected Externally, Corroborated by Gas Trend

A 40 MVA, 132/11 kV substation transformer, five sensors, continuous collection. Approximately two months into monitoring, VIE identified high-frequency transients at very high density on one high voltage phase sensor, rising localized flow vibration around the same location, and increasing heat flux at two adjacent sensor positions. Partial discharge activity was flagged around that location and the customer was asked to investigate the high voltage bushings and that winding set.

Independent corroboration came from the customer's own dissolved gas program. Across eight samples, hydrogen rose from 360 to 1,141 ppm and methane from 21 to 73 ppm, with ethane low but rising, and no ethylene or acetylene detected. Duval Triangle analysis was consistent with the VIE finding, indicating possible partial discharge without arcing and potential low temperature thermal hotspots.

Both methods reached the same conclusion. The difference is what each could do with it. Each individual gas sample read as a healthy transformer against threshold, and only the trend across eight samples was concerning. That trend was visible in retrospect at quarterly resolution. The continuous record localized the activity to a specific phase and sensor position and showed it developing in real time. An offshore unit would not have produced eight samples in that period, and a dry-type unit would not have produced any.

5.4 Condition-Dependent Stress Invisible to Point-in-Time Testing

Across a fleet of eight 1960s-era substation transformers monitored continuously, one unit produced a core and structure stress signal that appeared only once surface temperatures rose above 30 degrees Celsius. It was not present during winter monitoring. On a separate unit in the same fleet, the mechanical impact metric trended steadily upward toward the concern threshold, indicating structural loosening. Oil testing on that unit showed advanced cellulose degradation, but furan analysis reports thermal cellulose loss and has no equivalent measure for the mechanical pathway.

A third unit in the fleet illustrates an honest limitation. It ran at low load throughout the monitoring period, which suppresses the measured signals. Its underlying risk was assessed as higher than its score suggested, and the assessment said so explicitly. This case transfers directly to wind, where a unit at a low-yield turbine position or in a low wind season will produce suppressed signals for the same reason. Condition data read at actual operating conditions is more useful than a nameplate assumption, and it is also conditioned by those operating conditions.

Scope and Boundaries

VIE sits alongside an operator's existing testing and compliance programs rather than replacing them. The boundaries below keep the positioning honest and specific.

VIE has no deployed evidence base on dry-type nacelle transformers. The physics of the sensing method does not depend on insulating fluid, and the mechanical and thermal degradation pathways are the same. But the enclosure coupling on a cast resin unit in a ventilated nacelle housing differs from a filled tank, and VIE has not yet published detections from that configuration. VIE will state that clearly in any dry-type engagement and will treat an initial deployment as a baselining exercise.

VIE does not replace dissolved gas analysis on liquid-filled units. It supplies continuous mechanical, thermal, and magnetic data alongside it. Where a nacelle unit is liquid-filled and a sample can be drawn, that sample remains the definitive chemical measure of insulation condition.

VIE does not replace offline electrical testing. Insulation resistance, polarization index, power factor, tan delta, SFRA, and short-circuit impedance remain the confirming measures. VIE tells an operator which unit to test and when, not what the test will say.

VIE does not compute winding hottest spot temperature. The winding gradient is a manufacturer design constant and is not externally observable. VIE supplies measured surface temperature and continuously estimated thermal behavior at the asset, in place of an inference from regional ambient and load.

VIE does not test or certify converter, inverter, or grid code performance. That certification belongs to the turbine and converter manufacturer and is independent of anything VIE measures.

VIE does not monitor the turbine. It measures the transformer. Blade, gearbox, generator, and pitch system condition are outside its scope, though nacelle vibration from those sources appears in the record as an input to the transformer's mechanical environment.

VIE does not integrate directly with turbine SCADA, EMS, or plant historian platforms. Data reaches the customer's systems through the REST API.

VIE does not guarantee against failure or guarantee offtake, availability, or lender covenant compliance. It supplies a continuous documented condition record an operator can use to support those commitments. The decision and the determination remain the operator's.

VIE Technologies is not a certifying body and does not audit or attest to a customer's regulatory, grid code, or standards conformance.

Conclusion

Offshore wind operators face a transformer asset management problem that differs structurally from onshore practice. Failure consequences are concentrated, procurement lead times are long, the marine operating environment is demanding, and redundancy options are limited. Above all, access is gated by vessels and weather rather than by scheduling.

The industry's periodic inspection tools do not close that gap for either nacelle population. A dry-type unit offers no chemical diagnostic at all. A liquid-filled unit offers one that arrives on a vessel schedule, in small fluid volumes that move quickly between samples, in fluids whose interpretive limits differ from the mineral oil tables the industry grew up on. Point-in-time testing with blind intervals set by vessel access is not compatible with a 2–3 year replacement cycle.

VIE is built for that environment. What the deployed record supports:

• Leading-indicator detection. The platform has identified core and winding anomalies months ahead of lagging-indicator confirmation, documented in Section 5.1 at four months against two independent methods.
• Detection where conventional testing returned a pass. Section 5.2 documents four units that all passed insulation resistance and were separated only by continuous mechanical monitoring.
• Fleet-wide cross-asset pattern recognition. Correlated thermal and structural degradation trends across identical units, which single-asset inspection does not surface.
• Zero-downtime, non-invasive deployment. Sensors install in minutes without a turbine outage, without entry to internal electrical compartments, and without IT integration.
• One method across both insulation systems and both asset populations. The same sensing applies to dry-type and liquid-filled nacelle units, and to the offshore substation transformer.

The transition from schedule-based asset management to condition-based continuous intelligence is underway across the infrastructure sector. Offshore wind has the strongest case for it, because it is the sector where the cost of arriving late is set by a shipyard queue and a weather window rather than by a maintenance budget.

The transformer is the asset in the nacelle that cannot be reached, cannot be sampled on any useful cadence, and cannot be replaced inside a planning cycle. It should be the one that is measured continuously.