Data-driven Probabilistic Design of Wind Turbines

Description

Wind turbine manufactures and wind farm owners and operators typically have a large amount of data available both on materials and loads. Data on materials are collected through tests and measurements, and from validation studies, information of the model uncertainties may be obtained. Similarly on the loading side, data are available from wind and wave measurements, wind tunnel tests, full-scale tests and proto-type tests, model validations etc. The aim of the ProbWind project is to develop a probabilistic basis for using these data to obtain more cost-effective wind turbine designs which have the same required reliability level as obtained by traditional semi-probabilistic design sing partial safety factors as in IEC 61400-1 [1]. The probabilistic / reliability-based design approach corresponds to level 2 in the design levels in ISO 2394 [2]: 1) semi-probabilistic, 2) reliability-based and 3) risk-informed.
Further, a new international standard / technical specification ‘Probabilistic design measures for wind turbines’, IEC 61400-9 [3] is being developed and a CD version is available. The project will develop and demonstrate the implementation of the new probabilistic design methods for wind turbines and thereby providing a reliable design at a reduced LCOE. The technical specification will provide minimum requirements to the use of probabilistic design measures in order to ensure the structural and mechanical integrity of wind turbines. The technical specification will provide 1) the basic requirements with respect to reliability in terms of target probabilities of failure to be used as design criteria; 2) stochastic modelling of uncertainties including both aleatory (physical) and epistemic (model, statistical and measurement) uncertainties; 3) probabilistic modelling of the design load cases in [1]; 4) computational procedures to estimate the reliability; and 5) use of probabilistic methods in site suitability. Recommendations for stochastic models and demonstration of the application of probabilistic methods are also being developed, also in the ProbWind project with participation of major wind turbine manufacturers and wind farm operators / owners demonstration studies are performed.
The reduction in LCOE can be achieved by using the very large amount of data available by wind turbine manufacturers and operators to reduce uncertainties compared to the uncertainties used for calibration of partial safety factors in deterministic design codes such as [1] and thereby avoid unnecessarily conservative designs, but at the same time obtain the required target reliability requirements. The more accurate probabilistic designs will result in lower amounts of material needed for towers, rotor/nacelle structures, drivetrain components, blades, offshore substructures, etc. Thereby, the advances will also contribute to increased sustainability. This presentation will describe the above developments.
Reference:
1. IEC 61400-1:2019 - Wind energy generation systems - Part 1: Design requirements. International Electrotechnical Commission, Switzerland.
2. ISO 2394:2015: General principles on reliability for structures.
3. IEC CD 61400-8:2023 - Wind turbines - Part 9: Probabilistic design measures for wind turbines, International Electrotechnical Commission, Switzerland.

Period	25 May 2023
Event title	Wind Energy Science Conference 2023
Event type	Conference
Location	Glasgow, United KingdomShow on map
Degree of Recognition	International