A recent study by DNV has shown that the failure rate of wind turbine blades has not decreased in recent years. Furthermore, the rapid scale-up of turbine capacity and the size of the associated blades require a fast succession of new blade models with lengthy testing campaigns. Although significant efforts have been made to control the manufacturing process more closely, the blades still exhibit manufacturing defects that cause damage during both the certification process and in commercial use, sometimes resulting in blade failures. This can have a major impact on the electricity production of a wind farm but also considerable consequences for the producers of these large blades. Ultimately, this may also affect the wind energy industry and, thereby, the energy transition.
In current practice, a blade is designed based on design loads and material properties derived from a large number of coupon tests. A new blade is designed, manufactured and tested for certification with key load factors and practical experience of the blade design team. However, recently, it has been shown that developing and implementing condition-based maintenance practices for wind turbine blades and introducing them into the rotor blade design process is more cost-effective.
By monitoring the progression of damage throughout the life of a rotor blade, from production to the end of its operational life, inspections and repairs can be carried out in an increasingly preventative manner, reducing the chance of complete blade replacement.
This ReliaBlade2-NL project builds on the developments of the ReliaBlade 1 projects in Germany (led by Fraunhofer-IWES) and Denmark (led by DTU Wind Energy). In addition, it leverages a multinational collaboration with Germany, Great Britain, Denmark and Greece for the reliability of wind turbine blades. During the execution of this project, there will be close cooperation with the consortia assembled in the countries mentioned above.
The ultimate goal of the ReliaBlade research programme is to reduce the operational expenditure of wind turbine blades. The chosen path is to improve the reliability of wind turbine blades by monitoring their structural health throughout their lifespan. To achieve this, we will adapt existing digital twin concepts to develop capabilities for predicting the remaining blade lifetime and supporting operation and maintenance decisions.
To that end, this project intends to adopt an integrated approach to three major aspects:
1. Blade modelling
An improved blade modelling scheme will be developed to enable blade structural behaviour predictions over time based on its updated condition. The scheme will be tailored to operational blades by linking deterministic models of defective/damaged structural details to a model of the full blade through a sub-structuring approach . Probabilistic analyses will be used to characterise the stochastic characteristics of blade materials and defects based on test results, and these properties will be taken into account in the predictions.
Furthermore, for the probabilistic remaining useful life predictions, it is necessary to develop a methodology that enables increasingly better estimates of blade material and defect/damage uncertainties based on periodic blade monitoring data from the operation. As more data is gathered during operation, the level of uncertainty about blade characteristics will decrease, and the updated models will become blade-specific. In the end, this allows the past performance of the blade structure to be taken into account when updating the lifetime predictions of the blade.
To validate the methods and models, testing at different levels includes material coupons, subcomponent level, full-scale blade and operational turbine. This project includes experiments on all these scale levels.
2. Structure monitoring for blade characterisation and damage diagnosis
Firstly, continuous blade monitoring is necessary to obtain data to diagnose damage for the most critically stressed parts of the blade, the so-called hot spots. To this end, we develop algorithms to combine, analyse and interpret data from different sensing systems, such as strain, acceleration and acoustic emission sensors, to extract the desired information about the presence, location, type and size of damage.
Secondly, monitoring will provide data for blade characterisation. On the one hand, it is used to make generic structural models of the blade design specific to the blade in operation. On the other hand, it allows for the detection of anomalies during blade operation, away from the instrumented hot spots. The data will be recurrently analysed and fed into the modelling framework to update the probabilistic residual lifespan predictions.
3. Decision support for operation and maintenance
We will develop a decision support tool that uses information from the digital twin. The tool will be required to integrate information on the multiple aspects that influence the planning of offshore maintenance activities, ranging from weather data (affecting both operational limits and power production), resource costs and availability (e.g. vessel charter, technician teams), up to asset-specific information and maintenance requirements. Considering the probabilistic nature of the remaining useful life predictions, the risk of moving maintenance response actions in time (e.g. postponing until the next calendar-based inspection) can be quantified, and tolerance limits can be defined. The decision support tool will help asset owners find the best period to perform maintenance interventions, looking for synergies between condition-based preventive tasks and calendar-based campaigns while keeping a quantitative overview of the risk-benefit profile.
In short, ReliaBlade2-NL extends the current state-of-the-art by bringing all aspects together in one integrated framework that can support decisions for the predictive maintenance of turbine blades. This framework uses a digital twin that combines structural health monitoring, modelling and probabilistic analysis for residual actionable life predictions. This will reduce the chance of blade failure as inspections and preventive repairs can be scheduled before catastrophic blade failure occurs.
The main result of this project will be a digital twin framework for the lifetime prediction of turbine blades. The following intermediate results will be obtained:
Finally, a decision support tool will be developed for the tactical scheduling of preventive maintenance activities throughout the year. The tool will compute and regularly update key performance indicators (KPIs) for different calendar weeks to quantify the risks and benefits of carrying out maintenance activities in each of those weeks. The KPIs will combine relevant weather, market, and resource parameters with quantitative risk metrics, defined based on the probabilistic remaining helpful life estimates made by the digital twin framework to schedule preventive maintenance activities optimally.
+31 6 2933 5590
This project is supported with a subsidy by the Dutch Ministry of Economic Affairs and Climate Policy. Find more project information at the TKI Offshore Energy website.