Architecture-informed O&M modelling of modular X-Rotor configurations for offshore wind: a two- and three-module comparison

1 Introduction

Offshore wind farms are increasingly being deployed further from shore, where weather-restricted access, vessel availability, and maintenance logistics place growing pressure on operational expenditure. Under these conditions, O&M cost is not determined only by component failure frequency, but also by the interaction between failure type, vessel requirement, repair duration, weather windows, and lost production (Martin, et al., 2016; Carroll et al., 2017; Flannigan, et al., 2022). This makes turbine architecture an important factor in long-term economic performance, particularly where component replacement requires specialist offshore vessels.

The X-Rotor concept was developed to address several O&M limitations associated with conventional horizontal-axis wind turbines (HAWTs). Instead of locating the main drivetrain at hub height, the X-Rotor combines a vertical-axis primary rotor with modular secondary horizontal-axis power take-off units positioned closer to sea level (Leithead, et al., 2019; Leithead et al., 2025). In this architecture, the primary rotor and cross-arm form the main rotating structural system, while electrical power is extracted through the secondary horizontal-axis power take-off modules. The cross-arm/main-shaft bearing system is therefore treated as a shared turbine-level mechanical subsystem rather than as a duplicated module-level power take-off component. This distinction is important for the O&M mapping used in this study: secondary module components are scaled with module count, while shared primary-rotor and support-structure components are retained as turbine-level elements and treated separately where their reliability is uncertain.

The secondary-rotor power take-off principle has also been examined in recent X-Rotor studies (Morgan et al., 2024), which show that practicable secondary-rotor designs can provide high power-conversion efficiency while substantially reducing nacelle mass relative to conventional HAWT drivetrains. This architecture is intended to reduce reliance on jack-up vessels, improve accessibility, support detachable module replacement, and allow continued operation at reduced capacity following selected secondary-module failures (Leithead et al., 2019; Flannigan et al., 2022).

Previous X-Rotor O&M modelling has focused on the established two-module configuration, in which rated power is distributed across two 2.5 MW secondary power take-off modules (Flannigan et al., 2022). The StrathX-OM model was introduced to represent X-Rotor-specific O&M features, including detachable secondary modules, onshore repair, reduced-capacity operation, and modified vessel-assignment logic. The published 100 km offshore case showed that the established two-module X-Rotor could achieve higher availability and lower O&M cost than conventional 3-stage geared and direct-drive HAWTs under the same baseline assumptions.

In parallel, X-Rotor concept and performance/loading studies present both two- and three-bladed 5 MW primary-rotor variants. The three-bladed configuration uses reduced blade chord dimensions while maintaining a comparable overall turbine scale (Leithead et al., 2019; Morgan et al., 2024; Leithead et al., 2025). This provides the architectural basis for extending the published two-module O&M model to a three-module configuration, where the same approximate 5 MW rating is redistributed across three smaller secondary power take-off modules. The central question of this study is therefore: how does O&M performance change when the established two-module X-Rotor architecture is extended to three modules while keeping the overall turbine rating approximately constant?

This study addresses that question by adopting the StrathX-OM modelling philosophy and input assumptions reported by (Flannigan et al., 2022). The published 100 km case study is first reproduced as a model-consistency check, after which the framework is extended to evaluate a three-module X-Rotor configuration with three 1.67 MW secondary power take-off modules. The extension is implemented through transparent subsystem mapping, component exposure scaling, production impact assumptions, and repair cost adjustments. This allows the three-module configuration to be assessed not as an independent final design or field-validated prediction, but as a structured architecture-informed O&M comparison grounded in the same failure-mode methodology, repair categorisation framework, and vessel-strategy logic as the original StrathX-OM model.

The main contribution of this study is to quantify the O&M trade-off between increased component exposure and increased modular redundancy when the established two-module X-Rotor architecture is extended to three modules at approximately constant turbine rating. The analysis also makes the key architecture-dependent assumptions explicit. It tests their influence through sensitivity and bounding scenarios, including failure-rate scaling, reduced-capacity production impact, repair-cost scaling, and cross-member main-bearing replacement uncertainty.

Key terminology used to distinguish the conventional HAWT, the established two-module X-Rotor, and the three-module X-Rotor configuration is summarised in Table 1.

TABLE 1. Key terminology used to describe the conventional HAWT, the established two-module X-Rotor, and the three-module X-Rotor configurations, based on the X-Rotor concept descriptions in (Flannigan et al., 2022).

The two- and three-module X-Rotor configurations are illustrated schematically in Figures 1, 2, respectively.

2 Literature context and reliability basis

Offshore wind O&M models are widely used to examine how component failures, weather-restricted access, repair duration, vessel availability, technician resources, logistics strategy, and lost production combine to determine availability and cost. Early time-domain O&M modelling studies showed that offshore wind farm performance is strongly influenced by the interaction between stochastic failure events, metocean accessibility, vessel mobilisation, and maintenance-resource constraints (Dinwoodie et al., 2012). Subsequent drivetrain-focused analysis demonstrated that turbine architecture can lead to different availability and O&M cost outcomes, even under the same offshore operating environment (Carroll et al., 2017). Sensitivity studies have also shown that failure rates, repair durations, access limits, and vessel-related costs can substantially influence offshore wind farm availability and O&M cost estimates (Martin et al., 2016). These studies establish O&M modelling as a useful early-stage tool for assessing how architectural choices affect operational performance.

A central challenge in modelling novel offshore wind turbine concepts is the lack of long-term field reliability data. For conventional offshore HAWTs, component-level failure rates, repair times, repair costs, and technician-resource requirements have been reported using operational data from approximately 350 offshore turbines in Europe (Carroll et al., 2016). This dataset provides an empirical basis for early-stage O&M modelling, but it cannot be transferred directly to a novel architecture without explicit assumptions about component removal, duplication, relocation, accessibility, replacement strategy, and operational consequence. Reviews of O&M modelling for novel wind turbine concepts, therefore emphasise the need for architecture-specific mapping when conventional reliability data are applied to emerging turbine designs (McMorland et al., 2022). This issue is particularly important for modular concepts, where some subsystems are removed, some are duplicated, some are relocated closer to sea level, and others introduce new reliability uncertainties.

Recent O&M studies for future offshore wind turbines further reinforce the importance of uncertainty treatment when assessing emerging turbine concepts (Donnelly et al., 2024b). showed that failure-rate assumptions can strongly affect the O&M cost performance of next-generation offshore wind turbines, while (Donnelly et al., 2024a) demonstrated that drivetrain configuration, vessel accessibility limits, and major replacement times can alter the relative O&M cost performance of future 15 MW offshore turbine concepts. These findings are directly relevant to the present work because the three-module X-Rotor configuration has no long-term operational reliability record. Its O&M performance must therefore be inferred from adapted reliability data and engineering assumptions, rather than from directly observed field data. Transparent reliability mapping and sensitivity analysis are therefore essential for interpreting the results.

Recent reviews of wind energy prognostics, maintenance decision-making, and digital twins also show that future O&M modelling is moving beyond fixed deterministic failure assumptions (Li et al., 2026b). review wind turbine prognostics, remaining useful life prediction, artificial intelligence methods, and maintenance optimisation, while (Li et al., 2026a) discuss digital-twin frameworks linking physical assets, monitoring data, diagnosis, prognosis, and maintenance decision-making. These studies provide wider context for modular offshore wind architectures because they show that long-term reliability assumptions should ideally be updated as monitoring and operational data become available.

The X-Rotor concept provides a clear example of why architecture-informed O&M modelling is needed. The concept relocates major power take-off equipment closer to sea level and uses modular secondary horizontal-axis power take-off units to reduce dependence on high-elevation drivetrain access and heavy offshore intervention (Leithead et al., 2019; Leithead et al., 2025). The secondary-rotor power take-off mechanism has been examined by (Morgan et al., 2024), who showed that secondary rotors can provide an efficient power take-off route while reducing the need for conventional high-torque drivetrain arrangements. The updated X-Rotor feasibility study also discusses both two- and three-bladed 5 MW X-Rotor configurations and highlights potential O&M advantages, including removal of the gearbox and yaw system, lower-height power take-off equipment, and modular replacement (Leithead et al., 2025)These features motivate the present comparison between the established two-module configuration and an extended three-module architecture.

At the same time, aerodynamic and structural studies show that the physical behaviour of the X-Rotor should not be oversimplified in O&M assumptions. Experimental and numerical studies of the X-Rotor wake have shown that the coned blade geometry can produce three-dimensional induction, wake asymmetry, vertical flow effects, and sensitivity to blade pitch and model fidelity (Bensason et al., 2023; 2024; Ajay et al., 2024; Ajay et al., 2025). These findings are important for O&M modelling because reduced-capacity operation and module-level production-loss assumptions depend on how the remaining modules and the primary rotor behave after a module failure. Production-loss assumptions, such as 50% loss for a two-module configuration or approximately one-third loss for a three-module configuration, should therefore be treated as modelling assumptions requiring sensitivity analysis rather than as validated full-scale operational outcomes.

(Flannigan et al., 2022) introduced the StrathX-OM model as an extension of the StrathOW-OM framework to represent X-Rotor-specific O&M features, including detachable secondary modules, onshore repair, reduced-capacity operation, and modified vessel-assignment logic. Their 100 km offshore case showed that an established two-module X-Rotor configuration could achieve lower O&M cost and higher availability than conventional 3-stage geared and direct-drive HAWTs under the same baseline assumptions. This work provides the immediate modelling basis for the present study. However, the published X-Rotor O&M analysis focused on the established two-module configuration (Dinwoodie et al., 2012). The O&M effect of redistributing the same approximate 5 MW turbine rating across a larger number of smaller secondary power take-off modules has not yet been assessed within the same modelling structure.

X-Rotor techno-economic studies also show that cost estimates are scenario-dependent and sensitive to technical, logistical, and financial assumptions. The X-ROTOR (Devoy and Carroll, 2023) LCoE study developed a 5 MW X-Rotor LCoE estimate and applied sensitivity analysis to project lifetime, O&M strategy, distance from shore, financial assumptions, vessel costs, and CAPEX (Devoy and Carroll, 2023). The subsequent D6.3 comparison used updated X-Rotor CAPEX assumptions and benchmarked the X-Rotor against conventional HAWT drivetrain configurations under a common 2030, 40 km, 100 × 5 MW offshore wind farm scenario (McAuliffe et al., 2024). These studies provide important context for the present paper: X-Rotor O&M and LCoE results should be interpreted as scenario-dependent estimates rather than fixed deterministic predictions.

The present study addresses the remaining O&M modelling gap by extending the published StrathX-OM logic to evaluate a three-module X-Rotor configuration. This extension is not intended to represent a field-validated final engineering design. Instead, it provides an architecture-informed comparison of the redundancy-versus-exposure trade-off that arises when the same rated capacity is redistributed across three smaller secondary power take-off modules. Increasing the number of modules may increase component exposure and unscheduled corrective maintenance activity, while also reducing the production impact of an individual module failure. Because the reliability data, repair-cost scaling, repair-duration assumptions, and reduced-capacity operation logic remain uncertain, the results should be interpreted as early-stage comparative estimates. The study therefore uses the established two-module X-Rotor and conventional HAWT cases as reference points while explicitly identifying the assumptions that require sensitivity testing and future validation (Flannigan et al., 2022; McMorland et al., 2022; Anderson and Carroll, 2023; Campos-Gaona et al., 2021).

3 Materials and methods

3.1 Modelling framework and baseline reproduction

This study applies the StrathX-OM operational expenditure framework introduced by (Flannigan et al., 2022), which extends the StrathOW-OM time-domain simulation approach originally developed for offshore wind O&M assessment (Dinwoodie et al., 2012). The framework represents the interaction between stochastic component failures, weather-limited access, vessel availability, repair duration, technician resources, lost production, and direct maintenance expenditure. This structure is suitable for comparing turbine architectures because it links component-level failure behaviour to operational responses, resource requirements, downtime, and cost outcomes.

Component-level failure rates, repair-time categories, technician requirements, and unscheduled repair-cost data were taken from (Carroll et al., 2016) as the empirical reliability and repair-resource baseline. These data were not applied directly as conventional HAWT interventions (Flannigan et al., 2022). Instead, they were translated to the X-Rotor architecture using the modular replacement and vessel-assignment logic of StrathX-OM. This allowed architecture-dependent changes, including component removal, component duplication, subsystem relocation, improved accessibility, and altered replacement strategy, to be represented explicitly.

After reproducing the baseline case, the same environmental inputs, vessel-strategy assumptions, technician allocation, electricity price, cost structure, and operational simulation period were retained across all turbine configurations, following the modelling approach used by (Flannigan et al., 2022). Only architecture-dependent assumptions were modified, including component exposure, production loss per module failure, selected repair durations, and selected repair-cost assumptions. This approach ensures that the comparison isolates the effect of extending the modular X-Rotor architecture from two to three secondary power take-off modules, rather than reflecting differences in site conditions, vessel strategy, or economic inputs.

The O&M simulations in the present study were carried out over a 20-year operational lifetime, which was used as a common modelling period for all compared configurations. The analysis therefore focuses on comparative O&M performance over a consistent operating period and does not attempt to provide a full lifetime LCoE assessment.

The baseline cost inputs were taken from the published StrathX-OM and offshore wind O&M studies by (Carroll et al., 2016; Carroll et al., 2017; Flannigan et al., 2022). Where published cost outputs were reported in euros, the values were converted to GBP using a fixed reporting conversion factor of EUR 1 = GBP 0.8528 (European Central Bank, 2026). This factor was applied uniformly across the reproduced and extended cases to preserve internal comparability between the established two-module X-Rotor, the three-module X-Rotor, and the conventional 3-stage geared HAWT. The conversion was used only to express O&M cost outputs in a common reporting currency and does not represent year-by-year exchange-rate variation, inflation adjustment, purchasing-power correction, or modification of the original cost assumptions. Because the same conversion factor was applied to all cases, it affects only the reporting currency and not the relative comparison between configurations.

The overall architecture-informed modelling workflow is summarised in Figure 3. The workflow shows how conventional offshore wind O&M model inputs are combined with X-Rotor-specific architecture assumptions to define component mapping, failure exposure, reduced-capacity operation, vessel assignment, and the resulting O&M performance outputs.

The number of spare secondary rotors reported in Table 2 is used as an operational constraint in the modular replacement logic rather than as a full spare-inventory cost model. In StrathX-OM implementation, spare secondary rotors allow failed secondary power take-off modules to be removed and replaced offshore, after which the failed module is assumed to be repaired onshore. The reported O&M cost metric includes lost revenue, transport costs, staff costs, and repair costs, consistent with the reproduced StrathX-OM cost-output structure (McMorland et al., 2022). However, it does not include additional spare-inventory holding costs, long-term storage costs, procurement overheads, financing costs for spare modules, or inventory-management costs. This treatment should therefore be interpreted as an operational availability and repair-logistics assumption, not as a complete economic accounting of spare-module ownership.

TABLE 2. Summary of model inputs and strategies used for each turbine technology.

3.2 Technology cases and architectural assumptions

The technology cases considered in this study are the conventional 3-stage geared HAWT, the established two-module X-Rotor, and the proposed three-module X-Rotor extension. The conventional 3-stage geared HAWT is retained as the reference case used in the reproduced StrathX-OM baseline, while the established two-module X-Rotor represents the published modular X-Rotor O&M configuration described by (Flannigan et al., 2022)These two cases provide the baseline against which the three-module extension is assessed.

The three-module case is defined as a structured architectural extension of the established two-module X-Rotor. The overall turbine rating is kept approximately constant at 5 MW by redistributing rated power across three smaller secondary power take-off modules of approximately 1.67 MW each, rather than by increasing the total turbine rating (Flannigan et al., 2022; McMorland et al., 2022; Anderson and Carroll, 2023). This allows the O&M implications of changing the modular architecture to be assessed without introducing a capacity-scaling effect.

The architectural basis for the three-module case is informed by X-Rotor concept studies and performance/loading evaluations that present both two- and three-bladed 5 MW X-Rotor variants (Leithead, et al., 2019; Ajay et al., 2022; Leithead et al., 2025). In these concept studies, the three-bladed variant uses reduced blade chord dimensions relative to the two-bladed configuration while maintaining a comparable overall turbine scale. The present O&M analysis does not directly model aerodynamic loading, blade structural response, or wake behaviour. Instead, these studies are used to support the architectural definition of the three-module case and the redistribution of the 5 MW rated capacity across three secondary power take-off modules.

The purpose of the three-module case is not to define or validate a final engineering design, but to test how a change in modular architecture affects O&M behaviour under otherwise consistent modelling assumptions. The main architectural change is the redistribution of secondary power take-off capacity from two larger modules to three smaller modules. This increases the number of duplicated module-level subsystems and therefore increases component exposure. At the same time, it reduces the production impact of a single secondary-module outage from 50% in the two-module case to approximately one-third in the three-module case, where reduced-capacity operation is assumed to be possible (McMorland et al., 2022).

The main architectural distinction between the conventional HAWT and the X-Rotor cases is the location and distribution of the power take-off system (Anderson and Carroll, 2023; Campos-Gaona et al., 2021). The conventional HAWT concentrates the drivetrain at hub height, whereas the X-Rotor distributes power take-off across lower-height modular secondary units. The three-module case retains the same modular X-Rotor principle but increases duplicated module-level exposure while reducing the production consequence of an individual module failure.

The architecture-to-failure-rate mapping logic used for the three-module extension is summarised in Table 3. This table distinguishes shared turbine-level components, duplicated secondary power take-off modules, module-level auxiliary systems, removed HAWT components, blade-related categories, and rated-power redistribution. The corresponding implementation assumptions for failure-rate treatment, repair-time treatment, technician requirement, vessel/logistics assignment, and repair-cost treatment are summarised in Table 4. Together, these tables provide the transparent modelling basis needed to reproduce the three-module extension from the published two-module X-Rotor baseline (Flannigan et al., 2022; McMorland et al., 2022; Anderson and Carroll, 2023).

TABLE 3. Architecture-to-failure-rate mapping logic used to extend the established two-module X-Rotor case to the three-module configuration.

TABLE 4. Summary of component-level implementation assumptions used to translate the published HAWT reliability and repair-resource dataset into the three-module X-Rotor O&M model.

The scaling assumptions reported in Table 3 and Table 4 should be interpreted as architecture-informed modelling assumptions, rather than as direct evidence that all module-level failures are independent, additive, or fully represented by conventional HAWT reliability data. Since long-term operational reliability data for a three-module X-Rotor are not yet available, these assumptions define a transparent comparative modelling framework based on exposure mapping and sensitivity analysis, rather than a validated reliability model for a future commercial three-module X-Rotor (da Silva and Muskulus, 2021; Flannigan, et al., 2022; Ajay et al., 2022; McMorland et al., 2022; Leithead et al., 2025).

The module-count scaling reported in Table 3 should be interpreted as an architecture-informed exposure-based modelling assumption, rather than as evidence that all module-level failures are fully independent or strictly additive. It is used to represent the increased number of duplicated secondary power take-off components when moving from two to three modules. However, aerodynamic interaction, shared loading conditions, control behaviour, support-structure response, and common-cause mechanisms may alter the actual reliability of individual module-level components in a full-scale three-module design (Carroll et al., 2016; McMorland et al., 2022; Anderson and Carroll 2023).

In the absence of long-term operational reliability data for the three-module X-Rotor, these effects cannot yet be parameterised directly as component-specific failure-rate changes. The present study therefore applies transparent exposure-based mapping as the base-case assumption and tests the influence of more optimistic and more conservative reliability behaviour through the global failure-rate sensitivity analysis in Section 4.4.1.

While Table 3 defines the architecture-to-failure-rate mapping logic, Table 4 shows how this mapping is implemented in the O&M model through failure-rate scaling, repair-time treatment, technician allocation, vessel/logistics assignment, and repair-cost assumptions. This distinction is included to make the modelling assumptions transparent and reproducible.

The component-level implementation assumptions are further documented in the Supplementary Material to support reproducibility. Supplementary Figures S1, S2 show the failure-rate distributions by subassembly and repair category for the established two-module and extended three-module X-Rotor cases, respectively. Supplementary Figure S3 reports the average repair-time inputs used for the three-module X-Rotor case, grouped by subassembly and repair category.

Supplementary Figures S4, S5 show the corresponding repair-cost distributions for the two-module and three-module X-Rotor cases.

3.3 Failure-rate, repair-time, technician, and repair-cost data basis

The empirical reliability and repair-resource basis for this study is the offshore wind turbine dataset reported by (Carroll et al., 2016) and The X-Rotor-specific pitch-system and secondary-generator failure-rate and repair-cost inputs followed the derived values reported by (Flannigan et al., 2022). This dataset provides component-level failure rates, repair-time categories, technician requirements, and unscheduled O&M cost information for conventional offshore HAWT subassemblies. In the present study, these values are used as the starting point for failure and repair modelling, but they are not transferred directly to the X-Rotor without modification. Architecture-dependent modifications were applied through the mapping logic summarised in Table 3 and the implementation assumptions summarised in Table 4.

The component failure and repair-resource basis is primarily corrective-maintenance focused, because it is derived from unscheduled failure data in which failures are defined as turbine visits outside scheduled operations where material is consumed (Carroll et al., 2016). The reproduced StrathX-OM input structure retains the scheduled-maintenance allowance included in the baseline model; however, the present study does not optimise preventive-maintenance strategy or model detailed inspection scheduling. Differences between turbine cases should therefore be interpreted mainly through corrective-maintenance events, access constraints, vessel assignment, repair-resource assumptions, and lost-production effects. Spare inventory holding costs, storage costs, procurement overheads, and long-term spare-management costs are not included in the reported O&M cost metric and are treated as limitations when interpreting the economic implications of the modular X-Rotor configurations.

The X-Rotor architecture changes the practical meaning of several repair categories because major power take-off components are located closer to sea level and are represented as detachable modular secondary power take-off units. Repair duration, technician requirements, vessel requirements, and selected repair-cost assumptions were therefore adjusted where the intervention is expected to differ from an equivalent conventional HAWT repair. Selected secondary-module-level interventions were treated as more accessible than high-elevation nacelle-based drivetrain interventions. Selected jack-up-vessel requirements were also removed or reduced where the X-Rotor modular replacement logic allows intervention using lower-cost vessel strategies.

Failure-rate values were mapped using the architecture-informed logic described in Table 3. Components whose physical exposure increases with the number of secondary power take-off modules were scaled with module count, while shared turbine-level components were retained unchanged. No artificial reliability improvement factors were applied to individual components. This treatment does not assume that all duplicated components fail independently under all operating conditions. Rather, it provides a transparent exposure-based mapping in the absence of long-term operational reliability data for the three-module X-Rotor. Reductions in repair time, technician demand, vessel requirement, or repair cost therefore reflect changed accessibility and modular replacement assumptions, not an assumed improvement in intrinsic component reliability.

This approach combines the empirical failure and repair-resource basis of (Carroll et al., 2016), with the X-Rotor-specific operational logic introduced by (Flannigan et al., 2022). It preserves traceability to published offshore wind reliability data while allowing the repair process to reflect the lower-height, modular, and reduced-heavy-lift characteristics of the X-Rotor architecture. Where a conventional HAWT intervention would normally require high-elevation nacelle access or heavy-lift replacement, the equivalent X-Rotor module-level intervention was reassigned according to the modular replacement logic of StrathX-OM. Selected JUV- dependent repair assumptions were therefore removed or reduced for module-level X-Rotor components, while major structural interventions and blade-related replacements retained specialist vessel requirements where appropriate.

Under this structured mapping approach, the increase in total failure rate reflects additional component exposure arising from modular duplication rather than any assumed degradation in individual component reliability. The component-level implementation is further documented in the Supplementary Material to support reproducibility. Supplementary Figure S1 and Supplementary Figure S2 show the failure-rate distributions by subassembly and repair category for the established two-module and extended three-module X-Rotor cases, respectively. Supplementary Figure S3 reports the average repair-time inputs used for the three-module X-Rotor case, grouped by subassembly and repair category and based on the repair-time categories of (Carroll et al., 2016). Supplementary Figure S4 and Supplementary Figure S5 show the corresponding repair-cost distributions for the two-module and three-module X-Rotor cases.

The cross-member/main-shaft bearing system was treated as an X-Rotor-specific reliability uncertainty. It was not included as an additional deterministic failure-rate contribution in the base-case point estimate because representative long-term field data for large offshore vertical-axis main-shaft bearing systems are not yet available, and because the base-case comparison was intended to remain consistent with the reproduced established two-module X-Rotor baseline. This treatment should not be interpreted as assuming that the bearing system is operationally insignificant. Rather, it reflects the lack of component-specific reliability evidence for this X-Rotor-specific subsystem. Because this component may have high operational consequences if replacement requires jack-up-vessel access, its possible effect is examined separately through the cross-member/main-shaft bearing replacement bounding sensitivity in Section 4.4.3. Potential blade-arm or pitch-interface bearing effects were retained within the broader blade- and pitch-related failure categories to avoid unsupported parameterisation and potential double counting, consistent with the need for architecture-specific mapping when conventional reliability data are applied to novel turbine concepts (McMorland et al., 2022).

3.4 Reduced-capacity operation and availability calculation

Reduced-capacity operation was modelled as a key architecture-specific feature of the X-Rotor. In the established two-module configuration, the failure of one secondary power take-off module was represented as a 50% turbine production impact, with the remaining module continuing to operate where partial operation is permitted. In the three-module configuration, rated power is distributed across three smaller secondary power take-off modules. The failure of one secondary module was therefore represented as an approximately one-third turbine production impact, if the two remaining modules continue to produce subject to the operational constraints of the model.

This assumption does not imply that all failures lead to partial production. Failures associated with shared turbine-level systems, major structural components, safety-critical systems, or repair categories requiring full turbine isolation may still result in complete turbine shutdown depending on the component category and intervention requirement. The reduced-capacity logic was applied only where the failed component was associated with an individual secondary power take-off module and where continued operation of the remaining modules is consistent with the StrathX-OM operational logic.

Because the X-Rotor can operate at reduced output during selected module-level failures, availability was interpreted using a production-based availability concept rather than a simple binary up/down metric. This is important for modular architectures because a turbine may remain partially productive even when one secondary power take-off module is unavailable. Production-based availability was calculated as:where A E is the production-based availability, E actual is the simulated energy production after failure events, downtime, and reduced-capacity operation are applied, and E potential is the potential energy production under the same environmental conditions in the absence of failure-related production losses (International Electrotechnical Commission, 2014).

A E = E actual E potential × 100 % (1)

This definition encompasses both complete shutdown events and scenarios where operations are partially ongoing. It is therefore more appropriate for interpreting modular X-Rotor cases than a purely time-based availability metric, which does not distinguish between full turbine shutdown and reduced-output operation. Production-based availability is also consistent with offshore wind O&M simulation studies in which lost production and partial availability are key cost drivers (Carroll et al., 2017; Flannigan et al., 2022; Donnelly et al., 2024b).

The reduced-capacity assumptions should be interpreted as modelling assumptions rather than validated full-scale operational behaviour. In particular, the one-third production-loss assumption for the three-module configuration assumes an equal rated-power contribution from each secondary module and does not explicitly model aerodynamic interaction, load redistribution, power derating, or control-system constraints following a module outage. These effects are therefore treated as architecture-dependent uncertainties and are tested through the reduced-capacity production-impact sensitivity in Section 4.4.2.

3.5 Repair logistics and repair-cost treatment

Repair logistics were adjusted only where the X-Rotor architecture changes the practical intervention pathway relative to a conventional HAWT. The repair-time, technician-resource, and unscheduled repair-cost categories reported by (Carroll et al., 2016) were used as the empirical starting point. However, selected module-level X-Rotor interventions were reassigned to reflect lower-height access, detachable secondary power take-off modules, and modular replacement logic. Selected jack-up-vessel-dependent assumptions were therefore removed or reduced for module-level interventions where the StrathX-OM logic allows repair or replacement using lower-cost vessel strategies. Major structural interventions and blade-related replacements retained specialist vessel requirements where appropriate.

Repair-duration reductions were applied only to selected module-level interventions where lower-height access, detachable secondary power take-off modules, and reduced heavy-lift requirements change the intervention pathway relative to an equivalent conventional HAWT repair. Repair durations for shared turbine-level systems, major structural interventions, and blade-related replacements were retained unless a specific architecture-based reassignment was justified. Reduced repair durations therefore represent changed access and replacement logistics rather than an assumed improvement in intrinsic component reliability.

For selected replaceable module-level repair-cost categories, a size-related scaling factor was introduced as an engineering assumption. This factor was not taken directly from a published O&M repair-cost model. Instead, it was defined in the present study to reflect the redistribution of the same approximate 5 MW turbine rating from two 2.5 MW secondary power take-off modules in the established X-Rotor case to three approximately 1.67 MW secondary power take-off modules in the three-module case (Flannigan et al., 2022; Ajay et al., 2022; Leithead et al., 2025) This treatment is consistent with the wider X-Rotor techno-economic practice of applying component- and capacity-related cost assumptions when comparing X-Rotor and conventional wind turbine configurations, but the specific factor used here remains an O&M modelling assumption rather than an empirically validated repair-cost relationship (McAuliffe et al., 2024; Donnelly et al., 2024a).where S module is the module-size scaling factor, P 3 ‐ m o d u l e is the rated power of each secondary power take-off module in the three-module configuration, and P 2 ‐ m o d u l e is the rated power of each secondary power take-off module in the established two-module configuration.

S module = P 3 ‐ module P 2 ‐ module = 5 / 3 5 / 2 = 1.67 2.5 ≈ 0.668 ≈ 0.67 (2)

This factor was applied only to selected replaceable module-level repair-cost categories whose material, manufacturing, handling, or replacement cost is expected to scale partly with module size. It was not applied to shared turbine-level systems, fixed mobilisation costs, vessel costs, staff costs, scheduled-maintenance allowance, or major structural interventions, because these cost elements are not expected to scale linearly with the rated power of an individual secondary module. The repair-cost scaling should therefore be interpreted as a transparent engineering approximation intended to avoid assuming identical replacement costs for smaller secondary modules, rather than as a detailed techno-economic cost model.

Transport structure, vessel day rates, technician availability, weather time-series, electricity price, and the Fix on Fail–Fixed Charter jack-up strategy were otherwise retained from the reproduced 100 km baseline. This ensures that differences between the established two-module and extended three-module X-Rotor cases arise from architectural scaling, modular exposure, production-impact assumptions, and selected repair-treatment changes rather than from altered site, weather, or vessel-strategy assumptions.

Total O&M cost was reported using the same main cost categories as the reproduced StrathX-OM baseline: lost revenue, transport costs, staff costs, and repair costs. As discussed in Section 3.3, the reproduced input structure also retains the scheduled-maintenance allowance included in the baseline model, but the present study does not optimise preventive-maintenance scheduling or introduce a detailed inspection strategy. Spare secondary rotors are included as operational inputs to support modular replacement logic; however, spare holding costs, storage costs, procurement overheads, and long-term spare-management costs are excluded from the reported O&M cost metric.

Because the module-size repair-cost scaling factor is an engineering assumption rather than an empirically validated cost relationship, its uncertainty is considered in the interpretation of the results and discussed as a limitation and priority for future work.

4 Results

4.1 Baseline reproduction

The published 100 km offshore baseline case was first reproduced for the conventional 3-stage geared HAWT and the established two-module X-Rotor. This step served as a model-consistency check to confirm that the modelling setup reproduced the published StrathX-OM reference indicators before implementing the three-module X-Rotor extension. It is therefore treated as a baseline reproduction exercise rather than as an independent validation of the O&M model. The reproduced baseline results are compared with the published reference values in Table 5.

TABLE 5. Published and reproduced baseline performance indicators for the 100 km offshore O&M comparison case. The reproduced values are used as a model-consistency baseline before extending the StrathX-OM framework to the three-module X-Rotor configuration.

The reproduced values match the published StrathX-OM reference values to the reported numerical precision. The established two-module X-Rotor achieved a model-reported availability of 92.57%, a total failure rate of 6.97 failures per turbine per year, and a total O&M cost of £8.00/MWh. The conventional 3-stage geared HAWT achieved a model-reported availability of 91.54%, a total failure rate of 5.43 failures per turbine per year, and a total O&M cost of £13.76/MWh. These reproduced results provide the reference baseline for extending the modelling framework to the three-module X-Rotor case. The close agreement confirms that the reproduced implementation preserves the published modelling assumptions before introducing architecture-specific modifications for the three-module configuration (Carroll et al., 2017; Flannigan et al., 2022).

The 100 km offshore distance was retained because it corresponds to the published StrathX-OM O&M comparison case used as the baseline for this study. The objective of the present reproduction step was not to validate the model across different site distances, but to preserve comparability with the published 100 km O&M baseline before assessing the three-module architectural extension (Flannigan et al., 2022).

4.2 Three-module X-Rotor results

The extended three-module X-Rotor configuration produced a total failure rate of 7.73 failures per turbine per year. This increase compared to the established two-module X-Rotor is a direct consequence of the larger number of duplicated module-level subsystems, rather than an assumed degradation in intrinsic component reliability. The additional failure exposure is therefore an expected outcome of redistributing the same approximate 5 MW turbine rating across three secondary power take-off modules instead of two.

Despite the increase in total failure rate, the three-module configuration achieved a model-reported availability of 92.28%. This limited availability penalty is explained by the reduced production impact of an individual secondary-module failure. In the two-module case, the loss of one module corresponds to a 50% production impact, whereas in the three-module case, the loss of one module corresponds to approximately one-third of turbine production. The increase in component exposure is therefore partly offset by distributing the production consequence of a module-level outage across a larger number of secondary power take-off modules.

The three-module configuration resulted in a total O&M cost of £10.70/MWh and lost revenue of £145.0 million under the base-case 100 km offshore scenario. These results reflect the combined effect of increased module-level exposure, selected repair-cost scaling, reduced-capacity operation, and the retained 100 km offshore logistics assumptions. The results should therefore be interpreted as base-case model outputs for the three-module architectural extension, rather than as field-validated performance predictions. Table 6 summarises the base-case O&M performance indicators for the three-module X-Rotor.

TABLE 6. Base-case O&M performance indicators for the three-module X-Rotor under the 100 km offshore scenario. Lost revenue is reported as the cumulative model output over the simulated 20-year operational period.

4.3 Comparative O&M performance

The results show that total failure rate alone is not a sufficient indicator of O&M performance for modular turbine architectures. The total failure-rate comparison between the three cases is shown in Figure 4. The conventional 3-stage geared HAWT has the lowest total failure rate among the three cases, but it also has the lowest model-reported availability and the highest total O&M cost. This reflects the operational consequence of failures that require high-elevation access, heavy-lift intervention, specialist vessels, and longer weather-constrained downtime. This behaviour is consistent with previous offshore wind O&M studies showing that availability and cost are strongly affected by the interaction between failure type, vessel requirement, repair duration, accessibility limits, and lost production, rather than by failure frequency alone (Carroll et al., 2017; Flannigan et al., 2022).

By contrast, both X-Rotor configurations have higher total failure rates than the conventional geared HAWT, but lower total O&M costs under the same base-case 100 km offshore conditions. This occurs because the X-Rotor architecture changes the intervention pathway for key power take-off components. The lower-height modular secondary power take-off system reduces dependence on high-elevation drivetrain access and selected jack-up-vessel interventions, while allowing some failures to be managed through modular replacement and reduced-capacity operation. As a result, the operational consequence of selected module-level failures is lower than would be expected from the total failure rate alone.

The three-module X-Rotor does not outperform the established two-module X-Rotor in O&M cost. Its higher component exposure increases unscheduled corrective maintenance activity and raises the total O&M cost from £8.00/MWh to £10.70/MWh. However, the three-module X-Rotor remains below the conventional 3-stage geared HAWT value of £13.76/MWh under the base-case 100 km offshore scenario (Figure 5). This indicates that the O&M benefit of the X-Rotor architecture is not removed by adding a third module, although the reduced production consequence of a single module failure does not fully compensate for the increased subsystem exposure relative to the established two-module case.

The central interpretation is therefore that modularity changes the relationship between failure rate and O&M cost. Increasing the number of modules can improve redundancy distribution and reduce the production consequence of an individual module failure, but it also increases component exposure and corrective-maintenance activity. Whether additional modularity provides a net O&M benefit depends on the balance between redundancy, lower-height access, reduced production loss, repair-cost treatment, and the additional exposure introduced by duplicated subsystems. This finding should be interpreted within the base-case assumptions of the model, particularly the module-level failure-rate scaling, one-third production-loss assumption, repair-cost scaling factor, and 100 km offshore logistics scenario.

4.4 Sensitivity analysis of key architecture-dependent assumptions

Several assumptions required to extend the established two-module X-Rotor to the three-module X-Rotor remain uncertain, particularly the component failure-rate scaling applied to duplicated secondary power take-off modules, the reduced-capacity production-impact assumption, the module-size repair-cost scaling treatment, and the treatment of X-Rotor-specific cross-member/main-shaft bearing replacement events. The first, second, and bearing-replacement assumptions are tested through sensitivity analysis, while the repair-cost scaling treatment is retained as a transparent engineering assumption and discussed as a limitation because no empirical module-level repair-cost relationship is currently available.

A one-at-a-time sensitivity analysis was therefore performed to assess how the most important architecture-dependent assumptions influence the three-module X-Rotor O&M result. The purpose of this analysis is to provide a directional robustness check rather than a full probabilistic uncertainty assessment. This approach is consistent with previous offshore wind O&M sensitivity studies, where failure rates, repair durations, vessel accessibility, logistics assumptions, and major replacement events have been shown to affect availability and O&M cost outcomes strongly (Martin et al., 2016; Carroll et al., 2017; Donnelly et al., 2024b; 2024a). It is also consistent with previous X-Rotor O&M sensitivity work, where overall failure-rate uncertainty and main-bearing replacement assumptions were identified as important drivers of X-Rotor O&M cost (Anderson and Carroll, 2023).

The sensitivity cases retain the same 100 km offshore site, turbine rating, vessel strategy, weather input, electricity price, and baseline modelling structure used in the reproduced StrathX-OM comparison. The failure and repair-resource basis remains grounded in the offshore wind reliability dataset of Carroll et al. (2016), while the X-Rotor-specific modelling logic follows the reproduced StrathX-OM framework developed for the established two-module X-Rotor case (Flannigan et al., 2022). The analysis therefore isolates the influence of selected architecture-dependent assumptions rather than introducing changes in site conditions, logistics strategy, or economic inputs.

Because the sensitivity analysis varies one assumption at a time, it does not capture interactions between uncertain parameters. The results should therefore be interpreted as an indication of which assumptions most strongly affect the base-case cost margin, rather than as a complete uncertainty envelope. This limitation is important because failure-rate behaviour, reduced-capacity operation, repair-cost scaling, and bearing replacement logistics may interact in a full-scale three-module X-Rotor design.

4.4.1 Failure-rate sensitivity

The first sensitivity group examined the effect of changing the overall reliability level of the three-module X-Rotor. The same three-module X-Rotor base model was used in all cases, and only the global failure-rate factor varied from 0.9 to 1.3 (Table 7). All component categories, repair times, technician requirements, vessel assignments, repair costs, weather input, distance-to-port, power curve, and reduced-capacity assumptions were kept unchanged.

TABLE 7. Failure-rate sensitivity results for the three-module X-Rotor. The same three-module X-Rotor base model was used in all cases, and only the global failure-rate factor was varied.

The purpose of this sensitivity is to test whether the comparative O&M conclusion remains robust when architecture-dependent failure-rate assumptions are varied within a conservative range. The results are therefore interpreted as directional robustness checks rather than field-validated reliability predictions.

The failure-rate sensitivity shows that the three-module X-Rotor result is sensitive to the assumed reliability level. Increasing the failure-rate factor from 0.9 to 1.3 increased the model-reported total failure rate from 7.06 to 9.37 failures per turbine per year. Over the same range, availability decreased from 92.55% to 88.96%, lost revenue increased from £138.4 million to £217.9 million, and total O&M cost increased from £10.38/MWh to £14.30/MWh.

The model-reported total failure rate does not increase perfectly linearly with the input failure-rate factor because it is obtained from stochastic time-domain simulations rather than by direct deterministic multiplication of the base-case failure rate. For example, the FR-1.1 and FR-1.2 cases produced similar model-reported total failure rates, but the FR-1.2 case resulted in substantially lower availability and higher lost revenue. This indicates that the sampled failures in the FR-1.2 case had greater downtime, repair-resource, or vessel-access consequences.

The overall trend is clear: higher failure-rate factors reduce availability and increase lost revenue and total O&M cost. The conservative FR-1.3 case gives a total O&M cost of £14.30/MWh, which is higher than the reproduced conventional 3-stage geared HAWT reference value of £13.76/MWh. This indicates that the O&M cost advantage of the three-module X-Rotor is sensitive to higher component-failure assumptions. The base-case three-module X-Rotor result should therefore be interpreted as conditional on the assumed reliability mapping and modular repair logic, rather than as a guaranteed cost reduction under all reliability scenarios. The corresponding effect of the global failure-rate factor on total O&M cost is shown in Figure 6.

4.4.2 Reduced-capacity production-impact sensitivity

A second sensitivity group was conducted to test the reduced-capacity production-impact assumption used for secondary module-level failures in the three-module X-Rotor. In the base case, failure of one secondary power take-off module was assumed to reduce turbine production by approximately one-third, reflecting the equal distribution of rated power across three secondary modules. This assumption is consistent with the architectural definition of the three-module X-Rotor used in X-Rotor concept and performance/loading studies (Ajay et al., 2022; Leithead et al., 2025).

The actual production impact may be higher if aerodynamic interaction, control limitations, load redistribution, or operational derating reduces the output of the remaining modules after a secondary-module outage. X-Rotor aerodynamic and wake studies show that rotor behaviour can be sensitive to blade pitch, wake deformation, three-dimensional induction, and aerodynamic model fidelity, supporting the need to treat reduced-capacity operation as an uncertain modelling assumption at this stage (Bensason et al., 2024; Ajay et al., 2024).

To test this uncertainty, the turbine-level production impact assigned to one secondary-module failure was increased from the base value of 33%–40% and 50%. The 40% case represents a moderate derating assumption, while the 50% case represents a conservative condition in which the production consequence of a single module failure approaches that of the established two-module X-Rotor. All failure rates, repair times, technician requirements, vessel assignments, repair costs, weather inputs, distance-to-port, electricity price, and other cost assumptions were kept unchanged. The sensitivity therefore isolates the effect of changing the reduced-capacity production-impact assumption.The reduced-capacity production-impact sensitivity results are summarised in Table 8.

TABLE 8. Reduced-capacity production-impact sensitivity results for the three-module X-Rotor.

The reduced-capacity production-impact sensitivity shows that the three-module X-Rotor result is strongly influenced by the assumed production consequence of a secondary-module outage. Increasing the turbine production impact from 33% to 50% reduced production-based availability from 92.28% to 89.77%, increased lost revenue from £145.0 million to £194.7 million, and increased total O&M cost from £10.70/MWh to £13.52/MWh.

This result shows that the O&M benefit of the three-module X-Rotor depends not only on component reliability and repair logistics, but also on the extent to which the remaining secondary modules can continue to produce during a module-level failure. The conservative RC-50 case remains slightly below the reproduced conventional 3-stage geared HAWT reference cost of £13.76/MWh, but the cost margin becomes small. The reduced-capacity assumption should therefore be treated as an important architecture-dependent uncertainty rather than as a fixed performance guarantee.

These results do not imply that a 50% production impact is expected for the three-module X-Rotor. Instead, the RC-50 case is used as a conservative stress test of the reduced-capacity operating assumption. The results indicate that maintaining the base-case O&M advantage depends on the practical ability of the three-module X-Rotor to continue operating with limited derating following a secondary-module outage.

4.4.3 Cross-member main-bearing replacement bounding sensitivity

The cross-member main bearing was examined as a separate X-Rotor-specific bounding sensitivity because it is a potentially high-consequence component for a vertical-axis primary rotor located 100 km offshore. The base-case results presented in Sections 4.1–4.3 were retained to preserve consistency with the reproduced established two-module X-Rotor baseline. However, because excluding an explicit cross-member main-bearing replacement contribution could underestimate O&M cost, a separate bounding sensitivity was added to quantify its possible effect.

This treatment follows previous X-Rotor O&M sensitivity work, which identified main-bearing reliability as an important driver of O&M cost for the X-Rotor concept (Anderson and Carroll, 2023). The main bearing is particularly important because replacement may require jack-up-vessel access and can therefore have a disproportionate effect on downtime, transport cost, and lost revenue. The bounding range used in the present study follows the D6.1 sensitivity setup, in which the reference main-bearing replacement rate of 0.015 failures per turbine per year was varied from 0% to 250% (Anderson and Carroll, 2023).

The structural relevance of the cross-member main-bearing system is also supported by the X-Rotor detailed mechanical design work. The detailed design report includes the cross-arm, shaft, and bearings as part of the mechanical structure, and reports shaft and bearing design parameters, bearing loads, safety-factor checks, and estimated bearing masses. However, this structural design work does not provide long-term operational failure-rate data for these components. The cross-member main bearing is therefore treated here as a real design subsystem with unresolved reliability and replacement-access uncertainty, rather than as a negligible component (Correia et al., 2024).

For the present analysis, a separate cross-member main-bearing major-replacement bounding sensitivity was implemented to test the possible O&M consequence of this X-Rotor-specific uncertainty without changing the base-case architecture comparison. The minor- and major-repair modes of the cross-member main bearing were not separately parameterised because representative long-term operational data is not available. They were therefore set to zero in this bounding sensitivity so that the analysis isolated the high-consequence major-replacement pathway. Only the cross-member main-bearing major-replacement failure rate was varied.

The replacement event was modelled as a full-turbine-production-loss intervention requiring jack-up-vessel access. This is a conservative replacement pathway, consistent with previous X-Rotor cost modelling in which jack-up vessels are hired on a fix-on-fail basis for selected vertical-axis component replacements and main-bearing replacement is assumed to require jack-up-vessel intervention. Previous X-Rotor cost modelling also notes that future design-for-maintenance improvements, such as easier-removal mechanisms or SOV-based replacement strategies, could reduce or remove this additional jack-up-vessel requirement. In the present study, these improvements are treated as future design options rather than as the conservative reference assumption (Devoy and Carroll, 2023; McAuliffe et al., 2024).

The major-replacement failure-rate values were selected to follow the D6.1-style bounding range (Anderson and Carroll, 2023). The MB-0 case corresponds to the base-case point estimate with no additional cross-member main-bearing major-replacement contribution. MB-50 represents 50% of the reference value. MB-100 represents the reference value of 0.015 failures per turbine per year. MB-200 and MB-250 represent higher-consequence cases at 200% and 250% of the reference value, respectively. The global failure-rate factor was fixed at 1.0 for all cases, and all other model inputs were kept unchanged, including vessel strategy, technician allocation, weather input, distance-to-port, repair-cost assumptions for other components, and reduced-capacity production logic. The cross-member/main-bearing major-replacement sensitivity results are summarised in Table 9.

TABLE 9. Cross-member main-bearing major-replacement bounding sensitivity results for the three-module X-Rotor.

The results show that cross-member main-bearing replacement can materially affect the three-module X-Rotor O&M result. Increasing the main-bearing major-replacement failure rate from 0 to 0.0375 failures per turbine per year increased total O&M cost from £10.70/MWh to £15.66/MWh. Over the same range, availability decreased from 92.28% to 88.76%, while lost revenue increased from £145.0 million to £223.5 million. Transport and repair costs also increased, reflecting the conservative assumption that main-bearing replacement requires JUV access and full turbine production loss during the intervention. The MB-100 case gives a total O&M cost of £13.32/MWh, which is close to the reproduced conventional 3-stage geared HAWT reference value of £13.76/MWh.

The MB-200 and MB-250 cases exceed the conventional HAWT reference value. This indicates that the O&M advantage of the three-module X-Rotor is sensitive to the reliability and replacement strategy of the cross-member main bearing. The base-case result should therefore be interpreted as conditional on the assumed treatment of main-bearing reliability, rather than as a guaranteed cost reduction under all bearing-replacement assumptions.

Figure 7 shows the cost components most affected by the cross-member main-bearing replacement assumption. Lost revenue and transport cost dominate the increase across the sensitivity range, while repair costs remain comparatively smaller. This reflects the fact that the main economic consequence of this high-impact event is not only the component repair cost itself, but also the downtime and JUV mobilisation associated with major replacement.

This bounding sensitivity directly addresses the possibility that excluding an explicit cross-member main-bearing major-replacement contribution from the point-estimate result could underestimate O&M cost. The point-estimate result should therefore be interpreted as an architecture-consistent comparison of the two-module and three-module PTO layouts under the reproduced StrathX-OM baseline treatment, not as a claim that cross-member main-bearing replacements are negligible. If high-consequence main-bearing replacements occur at reference or upper-bound rates, the O&M cost advantage of the three-module X-Rotor can be substantially reduced or removed.

The revised interpretation is therefore that the base-case result represents a conditional architecture-informed comparison, while cross-member main-bearing reliability, replacement access strategy, JUV/SOV logistics, and structural design maturity remain key design-dependent uncertainties for future work.

5 Discussion

5.1 Failure exposure versus modular redundancy

The three-module X-Rotor case illustrates the competing effects of increased component exposure and distributed modular redundancy. Increasing the number of secondary power take-off modules from two to three increases the number of duplicated module-level subsystems and therefore raises the total failure rate from 6.97 to 7.73 failures per turbine per year. This increase is not associated with an assumed degradation in intrinsic component reliability. Instead, it results from the structured scaling of components whose physical exposure increases with module count.

The increase in failure exposure is distributed across several subsystem categories, including blade-related components, secondary hubs, generators, power converters, electrical and protection components, sensors, and lubrication/cooling systems. It is therefore not concentrated on a single high-impact component. At the same time, the production impact of an individual secondary-module outage is reduced from 50% in the established two-module configuration to approximately one-third in the three-module configuration. This reduced-capacity logic extends the modular-operation principle introduced for the established X-Rotor O&M model by (Flannigan et al., 2022). The distribution of production loss across three secondary power take-off modules helps explain why the three-module case maintains an availability of 92.28%, despite the higher total failure rate.

This result shows that modular redundancy changes the consequence of failure, but it does not remove the effect of increased component exposure. The three-module architecture therefore introduces a clear O&M trade-off: it reduces the production consequence of an individual module outage but increases the number of module-level components that may require corrective maintenance. The net O&M outcome depends on the balance between these two effects, together with the assumed repair logistics, module-level repair-cost treatment, and reduced-capacity operation.

5.2 Why does the higher failure rate not translate linearly into O&M cost?

The results highlight the non-linear relationship between total failure frequency and total O&M cost under long-distance offshore operating conditions. At the 100 km offshore site considered here, cost escalation is influenced not only by the number of failure events, but also by vessel strategy, mobilisation delay, weather-limited access windows, repair duration, and lost production. This is consistent with previous offshore wind O&M studies showing that cost and availability are driven by the interaction between failure type, access constraints, vessel logistics, and repair duration, rather than by failure frequency alone (Martin et al., 2016; Carroll et al., 2017; Flannigan et al., 2022). This explains why the conventional 3-stage geared HAWT has the lowest total failure rate but the highest total O&M cost among the three cases.

For the X-Rotor configurations, modular accessibility and reduced dependence on heavy-vessel intervention moderate the cost penalty associated with higher component exposure. Selected module-level interventions can be represented through lower-height access, modular replacement logic, and reduced-capacity operation rather than through conventional high-elevation drivetrain replacement pathways, consistent with the StrathX-OM modelling approach developed by (Flannigan et al., 2022). Consequently, the higher failure rate of the three-module X-Rotor does not lead to a proportional increase in downtime, lost revenue, or total O&M cost.

The three-module configuration therefore retains much of the O&M advantage of the X-Rotor architecture over the conventional geared HAWT in the base-case scenario. Its total O&M cost of £10.70/MWh remains below the 3-stage geared HAWT value of £13.76/MWh under the same 100 km offshore conditions, although it is higher than the established two-module X-Rotor value of £8.00/MWh. This supports the interpretation that the three-module architecture remains favourable relative to the conventional geared HAWT baseline under the base-case assumptions but does not improve on the established two-module X-Rotor in O&M cost.

This interpretation is conditional on the architecture-informed assumptions used in the model, particularly the module-count failure-exposure scaling, reduced-capacity production logic, selected repair-cost scaling, and treatment of cross-member main-bearing replacement uncertainty. The sensitivity results show that the cost margin relative to the conventional HAWT can narrow or disappear under more conservative reliability, derating, or bearing-replacement assumptions. The base-case result should therefore be interpreted as a conditional architecture comparison rather than as a field-validated prediction of long-term O&M performance.

5.3 Positioning of the three-module configuration

The three-module configuration occupies an intermediate O&M position between the established two-module X-Rotor and the conventional 3-stage geared HAWT. It is less cost-effective than the established two-module X-Rotor because the additional module increases component exposure, corrective maintenance activity, and cost exposure. However, it remains more cost-effective than the conventional geared HAWT in the base-case 100 km scenario because the key O&M advantages of the X-Rotor architecture are retained: lower-height power take-off, modular secondary units, reduced dependence on selected jack-up-vessel interventions and reduced-capacity operation.

This positioning supports the central interpretation of the study. Additional modular redundancy does not automatically improve O&M performance. It reduces the production consequence of individual module failures, but it also increases the number of components that can fail. The O&M value of adding modules therefore depends on whether the benefit of reduced production loss and improved intervention accessibility outweighs the cost of increased subsystem exposure.

The sensitivity results reinforce this interpretation. The base-case three-module result remains below the conventional geared HAWT reference cost, but this margin is reduced or removed under more conservative failure-rate assumptions, higher production-impact assumptions, and upper-bound cross-member main-bearing replacement scenarios. The three-module result should therefore be interpreted as conditional on the assumed reliability mapping, modular repair logic, reduced-capacity operation, and bearing replacement strategy.

5.4 Limitations and future work

The results should be interpreted as an architecture-informed comparative O&M assessment rather than a direct prediction of field performance. The three-module X-Rotor has not yet accumulated long-term operational reliability data. Therefore, the failure-rate, repair-time, technician-resource, and repair-cost assumptions were derived from conventional offshore HAWT datasets and adapted through structured architecture mapping. This preserves traceability to published reliability data, but it also introduces uncertainty in the representation of X-Rotor-specific components and intervention pathways (Carroll et al., 2016; Flannigan et al., 2022; McMorland et al., 2022).

Several architecture-dependent assumptions remain uncertain. These include the proportional scaling of duplicated module-level components, repair-duration reductions for selected accessible module-level interventions, the capacity-linked repair-cost scaling factor used for smaller secondary power take-off modules, the reduced-capacity production-impact assumption, and the treatment of X-Rotor-specific components such as the cross-member main bearing. The sensitivity analysis shows that these assumptions can materially affect the three-module O&M conclusion. In particular, the failure-rate sensitivity, reduced-capacity production-impact sensitivity, and cross-member main-bearing replacement bounding sensitivity demonstrate that the base-case cost advantage is not guaranteed under all reliability, derating, and replacement-access assumptions.

The capacity-linked repair-cost scaling factor represents a simplified engineering assumption. It was introduced to avoid assuming identical repair costs for smaller secondary power take-off modules, but it is not a detailed component-level cost model. It was applied only to selected replaceable module-level repair-cost categories and does not represent fixed mobilisation costs, vessel charter costs, handling costs, or wider logistics overheads. Future work should validate this assumption using industrial replacement data, manufacturing cost estimates, and detailed cost models for modular X-Rotor assemblies.

Distance-to-port was held constant at 100 km to preserve direct comparability with the reproduced StrathX-OM O&M baseline of (Flannigan et al., 2022) and to isolate the architectural effect of extending the X-Rotor from two to three secondary power take-off modules. The results should therefore be interpreted as representative of this reproduced 100 km far-offshore scenario, rather than as distance-independent cost rankings. Farm-to-port distance can affect vessel transit time, weather-window utilisation, charter strategy, technician access, lost production, and transport-related O&M cost. Published X-Rotor techno-economic work also shows that cost outcomes are scenario-dependent and sensitive to O&M strategy, logistics assumptions, vessel requirements, and offshore wind farm configuration (McAuliffe et al., 2024). Future work should therefore examine the interaction between module number, site distance, metocean accessibility, and vessel strategies across a wider range of offshore locations.

The conventional 3-stage geared HAWT should also be interpreted as a reproduced reference case rather than as a claim to represent the latest state-of-the-art offshore HAWT drivetrain. It was retained because it provides the published StrathX-OM comparison baseline against which the established two-module X-Rotor was previously assessed. This choice supports internal comparability, but it also means that the relative cost advantage of the X-Rotor configurations may change if future comparisons use more recent HAWT reliability data, improved drivetrain designs, different vessel strategies, or updated maintenance practices. Future work should therefore compare the modular X-Rotor configurations against updated conventional and direct-drive HAWT baselines.

The reported O&M cost boundary should also be interpreted carefully. The model reports lost revenue, transport cost, staff cost, and repair cost using the reproduced StrathX-OM cost structure. It does not include a detailed spare-inventory, storage, procurement, insurance, warehousing, or hold-cost model for secondary modules or major replacement components. The number of spare secondary rotors is therefore treated as an operational constraint within the modular replacement logic, rather than as a full spare-management cost assessment. This limitation is important because modular architectures may shift cost from offshore intervention to spare-module ownership, storage, and logistics planning. Future work should include spare-module inventory optimisation and lifecycle spare-management costs.

The present study focuses on comparative O&M performance rather than full levelised cost of energy. The site, climate time-series, vessel strategy, electricity price, wind farm rating, and operational simulation period were kept fixed to isolate the effect of architecture. This makes the comparison internally consistent, but it does not capture broader design-level trade-offs such as structural loading, substructure cost, aerodynamic performance, wake behaviour, capital cost, financing, insurance, transmission charges, or decommissioning. Translating the O&M results into LCOE would require additional CAPEX, financing, operational-phase cost, and end-of-life assumptions that are beyond the scope of the present study.

The maintenance-cost boundary is primarily corrective-maintenance focused because the component failure-rate, repair-time, technician-resource, and unscheduled repair-cost basis is derived mainly from failure events and repair interventions reported for conventional offshore wind turbines (Carroll et al., 2016). The reproduced StrathX-OM input structure retains the scheduled-maintenance allowance included in the published baseline model, but the present study does not optimise preventive-maintenance strategy, inspection frequency, condition-based replacement, or scheduled campaign planning (Carroll et al., 2016; Martin et al., 2016; Flannigan et al., 2022; McMorland et al., 2022).

This limitation is important because scheduled inspection and preventive-maintenance requirements may differ between conventional HAWTs and modular X-Rotor configurations. For example, lower-height access and detachable secondary power take-off modules may reduce some scheduled-maintenance burdens, while additional duplicated module-level components may increase inspection requirements. These effects are not explicitly resolved in the present model. Therefore, the reported O&M cost comparison should be interpreted mainly through corrective-maintenance events, access constraints, repair logistics, and lost-production effects, rather than as a complete representation of all lifetime maintenance planning.

Future work should extend the modelling framework to include scheduled preventive-maintenance campaigns, inspection intervals, spare-module management, and condition-based replacement strategies. This is particularly important for modular X-Rotor architecture because accessibility, redundancy, and detachable power take-off modules may change not only corrective repair logistics, but also the cost and timing of planned maintenance. Integration of condition monitoring, prognostics, and digital-twin updating could support more realistic maintenance scheduling as prototype or operational data become available (Li et al., 2026b; Li et al., 2026a).

6 Conclusion

This study evaluated how extending the established two-module X-Rotor configuration to a three-module architecture affects O&M performance under a reproduced 100 km offshore scenario. The analysis adopted the StrathX-OM modelling framework, first reproducing the published baseline for the conventional 3-stage geared HAWT and the established two-module X-Rotor, before extending the model using architecture-informed component mapping, failure-exposure scaling, reduced-capacity operation, and selected repair-logistics and repair-cost assumptions.

The results show that increasing the number of secondary power take-off modules from two to three raises the total failure rate from 6.97 to 7.73 failures per turbine per year. This increase arises from additional duplicated module-level exposure rather than from an assumed deterioration in intrinsic component reliability. However, the production impact of a single secondary-module outage is reduced from 50% to approximately one-third, which limits the availability penalty. The three-module configuration achieves an availability of 92.28%, only slightly below the established two-module X-Rotor value of 92.57%.

From a cost perspective, the three-module X-Rotor is less favourable than the established two-module configuration but remains below the conventional geared HAWT under the same base-case 100 km offshore assumptions. The total O&M cost is £10.70/MWh for the three-module X-Rotor, compared with £8.00/MWh for the established two-module X-Rotor and £13.76/MWh for the conventional 3-stage geared HAWT. This indicates that the O&M advantage of the X-Rotor architecture is partly retained after adding a third module, although the additional modular redundancy does not fully compensate for the increased subsystem exposure.

The sensitivity analysis shows that this conclusion is conditional on the assumed reliability mapping, reduced-capacity production logic, repair-cost treatment, and cross-member main-bearing replacement strategy. Under higher failure-rate assumptions, more conservative production-impact assumptions, and upper-bound main-bearing replacement scenarios, the O&M cost margin relative to the conventional geared HAWT is reduced or removed. In particular, the main-bearing bounding sensitivity shows that cross-member main-bearing reliability and replacement access strategy are important design-dependent uncertainties for the X-Rotor architecture.

Overall, the findings show that modular turbine architectures should not be assessed using total failure rate alone. For the X-Rotor, the operational consequence of failures, including reduced-capacity operation, repair duration, vessel requirement, accessibility, and replacement strategy, strongly influences availability and cost outcomes. The three-module configuration should therefore be interpreted as an architecture-informed O&M variant of the established two-module X-Rotor, rather than as a direct design recommendation or a field-validated prediction of long-term performance.

The results remain dependent on adapted reliability data and engineering assumptions because long-term operational data for three-module X-Rotor configurations are not yet available. The conventional 3-stage geared HAWT is also retained as a reproduced reference case, rather than as a claim to represent the latest state-of-the-art HAWT drivetrain. In addition, the analysis holds distance-to-port constant at 100 km to preserve comparability with the reproduced StrathX-OM baseline and to isolate the architectural effect of changing the number of secondary power take-off modules. The results should therefore be interpreted as representative of this reproduced 100 km far-offshore O&M scenario, rather than as distance-independent cost rankings. Future work should extend the analysis to a wider range of farm-to-port distances, metocean conditions, vessel strategies, and maintenance policies.

Further work should also develop more detailed cost models for modular secondary power take-off assemblies, spare-module management, cross-member main-bearing replacement, and design-for-maintenance access strategies. Broader design-level effects, including structural loading, substructure cost, aerodynamic performance, wake behaviour, capital cost, and full levelised cost of energy, should also be assessed before drawing final conclusions on the preferred X-Rotor configuration.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The authors used ChatGPT for language editing and manuscript-structuring support. The authors reviewed and edited all AI-assisted text and take full responsibility for the content of the manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenrg.2026.1891630/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Zifei Xu, University of Liverpool, United Kingdom

Reviewed by: Mingxin Li, The University of Tokyo, Japan

Muhammad Moman Shahzad, National University of Sciences & Technology, Pakistan