Advances in Engineering Advances in Engineering features breaking research judged by Advances in Engineering advisory team to be of key importance in the Engineering field. Papers are selected from over 10,000 published each week from most peer reviewed journals.
Significance
Offshore wind turbines operate in demanding marine environments, where strong winds, high waves, earthquakes, and continuous mechanical loading can threaten their reliability. These large structures contain interconnected mechanical, electrical, and control systems and a failure in one component can disrupt electricity generation and lead to prolonged shutdowns and costly offshore repairs.
Understanding how different hazards and equipment faults affect a turbine is not straightforward. Extreme winds place additional loads on towers and foundations, influence blade operation, and increase demands on control equipment. Earthquakes can also threaten structural stability. Mechanical wear, electrical faults, lubrication problems add further risks and make it difficult to determine which failures have the greatest influence on overall reliability.
Traditional assessment methods, such as fault tree analysis, can trace how individual equipment faults lead to larger failures. However, accounting for dependencies between different hazards becomes more difficult when operating conditions change. These methods also have limited ability to update failure probabilities when new information becomes available.
Deciding which components deserve the most attention during inspection and maintenance is another challenging factor to consider. The importance of a particular fault depends partly on the stresses acting on the turbine and a component that is critical during normal operation may have a different influence during severe weather.
Bayesian networks can provide a method to examine these relationships. They represent connections between events using probabilities, which can allow engineers to estimate how changes in operating conditions affect reliability. They can also work backward from an observed failure to identify its likely causes.
In a recent research paper published in Ocean Engineering, PhD student Jia-Xuan Li and Professor Da-Gang Lu from Harbin Institute of Technology, together with Dr. Hao Liu and Dr. Wenbing Wu from China University of Geosciences, developed a Bayesian network model to assess offshore wind turbine reliability under multiple environmental hazards. Their approach examines how equipment faults and external loads contribute to turbine failures, helping identify critical vulnerabilities and guide maintenance decisions.
Methods and Findings
The turbine was divided into five subsystems: the support structure, pitch system, gearbox, generator, and auxiliary system. Together, these comprised 24 components and 65 potential failure causes. The researchers used existing fault tree models to establish relationships between failure events and represented them in the Bayesian network. Failure probabilities were estimated using offshore equipment reliability records, published studies, and expert assessments where sufficient data were unavailable. Forward probability calculations, backward fault diagnosis, and sensitivity analysis were then applied to evaluate reliability under normal operation, typhoons, earthquakes, and combined hazards.
The model predicted an annual turbine failure rate of approximately 8%, consistent with earlier field observations and fault tree analysis. When less favorable operating conditions were considered, the estimated failure probability increased to 8.6%, with the support structure showing a greater relative increase.
The researchers also investigated possible causes when turbine failure was treated as a known event. Working backward through the network, they estimated which component faults were most likely to be associated with that outcome. Tower failure and structural collapse caused by environmental conditions were identified among the most probable contributing events.
Sensitivity analysis helped determine which faults had the greatest influence on turbine reliability. Three measures were used: Risk Achievement Worth, Risk Reduction Worth, and Birnbaum Importance. These assess how the condition of individual components affects the wider system and where improvements in reliability might have the greatest benefit.
For the support structure, tower failure, abnormal stresses, environmental loading, and faults in connected equipment were among the most influential factors. These vulnerabilities pointed to the importance of structural integrity, welding quality, corrosion protection, and foundation stability.
In the pitch system, hydraulic failures and malfunctioning warning equipment were identified as important risk factors. Problems with hydraulic components can prevent blades from adjusting their angle, making it harder to control loading during strong winds. Faulty warning devices may delay protective responses, and power interruptions can disable essential pitch functions.
The gearbox analysis identified abnormal gear operation, inadequate lubrication, bearing damage, and gear cracking as important factors affecting reliability.
For the generator, fluid leakage and measurement equipment failures were identified as important risk factors. Leakage can interfere with cooling or lubrication, and malfunctioning sensors may delay the detection of developing faults. The auxiliary system was more sensitive to electrical component failures and problems with the yaw system, which keeps the rotor aligned with the wind. If this mechanism malfunctions, uneven tower loading can increase fatigue stresses. The researchers then compared typhoon and earthquake scenarios, both separately and under combined loading. The relative importance of several faults changed when both hazards were considered.
For the support structure, the Risk Achievement Worth analysis showed that tower failure had approximately 8.3% lower relative importance under combined hazards than under earthquake-only conditions. In the pitch system, the relative importance of hydraulic failure, measured using Risk Reduction Worth, decreased from 51.5% under typhoon-only loading to 32% under combined loading. Warning equipment and power supply failures, however, became more influential relative to the earthquake-only case.
The researchers described this behavior as an external masking effect. Severe environmental loads increase the overall vulnerability of the turbine through several possible failure pathways. Under these conditions, the relative contribution of an individual component can decline even when the equipment itself has not become safer. This explains why the ranking of critical faults can change between different hazard scenarios.
Significance and Implications
The new study shows why identifying the components most likely to fail is not sufficient for assessing offshore wind turbine reliability. A relatively uncommon fault may still have serious consequences for turbine operation, and make its contribution to overall system risk an important consideration when planning maintenance and protective measures.
The external masking effect adds another dimension to this assessment. During severe environmental events, several failure pathways can become influential at the same time, changing the relative importance of individual components. Maintenance priorities based on normal operating conditions may be less appropriate when a turbine faces a typhoon, earthquake, or combination of hazards.
The ability to work backward from an observed failure also has practical value. Offshore inspections are difficult and expensive, and identifying the most probable contributing faults can help engineers focus their investigations and sensitivity analysis here provides complementary information by showing which components have the greatest influence on reliability and where preventive measures may offer the most benefit.
The Bayesian network brings these capabilities together in a single probabilistic model. Its predictions depend on the available reliability data and assumptions about relationships between subsystems, but the approach provides a quantitative basis for assessing different operating scenarios and making more informed decisions about offshore turbine safety.
Reference
Jia-Xuan Li, Da-Gang Lu, Hao Liu, Wenbing Wu, A Bayesian network approach to multi-hazard risk assessment of offshore wind turbines, Ocean Engineering, Volume 351, Part 1,2026, 124340,
