Probabilistic fatigue reliability assessment of the wind turbine's structural components

Assessing the certain time to failure of a structure is impossible and such a problem is always a probabilistic exercise. In addition to this reality, high randomness in the wind loads, and complexity of the wind turbine’s structure introduce more uncertainty in the problem, both aleatory and epistemic. Thus, the design problem is inherently one of keeping the probability of failure below some acceptable level, which may be set by consensus standards or economic considerations. A probabilistic approach provides more information about margins and risk than a simple pass/fail assessment of meeting a fixed criterion.
The current research focuses on probabilistic fatigue reliability assessment of wind turbine’s structural components and presents insights and investigations on some aspects of such assessments. The research consists of four parts/studies: ’Effects of finite sampling on fatigue damage estimation’, ’Statistical extrapolation of the fatigue loads’, ’Sensitivity analysis of the fatigue reliability’, and finally ’Probabilistic assessment of the lifetime extension’.
The first two studies focus on decreasing the uncertainty in the fatigue load estimation by increasing the number of short-term data and anticipating the long-term data, respectively. The third study places the fatigue load into the fatigue reliability assessment framework and elaborates on the level and importance of uncertainty in the loads compared to other sources of uncertainty in such a framework. Finally, the last part of the thesis gathers the findings from the first three studies and puts them into application. The target application is the probabilistic assessment of the lifetime extension through re-assessing the semi-deterministic design-level evaluations.
The first three studies use the DTU 10MW wind turbine as the case study and the final study uses Siemens 2.3MW with site-specific inputs for the investigations related to Lillgrund wind farm. In all the studies, we use aeroelastic simulations and the concept of the DEL for assessing fatigue loads. The first and third parts of the thesis cover different structural components while the second and final parts are applied to blades.
The outcomes of the study are useful in providing more accurate assessments of reliability through a probabilistic approach. Most importantly, the current thesis mainly sheds light on the unavoidable need for further efforts on decreasing the uncertainty in the material properties to achieve more reliable and robust fatigue designs. Such efforts can include improvements in modeling the material fatigue strength and fatigue damage accumulation process and/or increasing the quality of manufacturing processes, tests, etc.