ABSTRACT
The principal challenge in reliability assessment of wind turbines is rooted in the fact that a small tail probability, f(z > lT ), in the order of 10−7 for T = 50 years, needs to be estimated. To accurately estimate this type of small probability requires a sufficient number of high-z load response values. If one opts to collect enough high-z values from physical turbine systems, it takes tens of years, as the high-z values, by definition, are rare events. Adding to the challenge is that hardly have any commercial wind turbines been installed with strain sensors, due to cost concerns. Physically measured bending moments are typically obtained on a few test turbines and only for a short duration of time, which is the reason behind the need for an extrapolation and the modeling of the conditional load density, as explained in Chapter 10. Wind engineers have been developing aeroelastic simulators that can produce reasonably trustworthy bending moments response under a wind force input. The availability of these simulators lends a degree of convenience to load analysis, as a simulator can be steered, at least in principle, towards the region of high load responses so as to produce more high-z data points. For this reason, using the aeroelastic simulators could expedite and enhance the estimation of extreme load distribution and facilitate the reliability assessment of wind turbines. Of course, running aeroelastic turbine load simulators can be computationally expensive. Data science methods are much needed to make the simulator-based load analysis efficient and practical.
