ABSTRACT
Fatigue assessment of offshore wind turbines (OWTs) is of utmost importance due to their continuous subjection to cyclic loads caused by wind and wave excitations. Fatigue assessment includes estimation of the remaining lifetime and accumulated fatigue damage. In order to do so, the complete distribution of stresses along the structure with main emphasis at the fatigue hot-spots is to be known. Stress time histories are directly linked to the corresponding strains which can normally be measured using conventional or optical strain gauges. For the case of OWTs on monopile foundations, though, most of the fatigue hot-spots are located be-neats the water level, thus making it impossible to install strain sensors at these locations for existing turbines. This measurement restriction motivates the use of virtual sensing technique to estimate strains at fatigue critical and inaccessible locations and subsequently accurately assess their fatigue. The proposed virtual sensing scheme is based on the concept of modal expansion and accounts for both quasi-static and dynamic loading using adaptive sensor/mode settings in different frequency bands and environmental and operating conditions (EOC). The paper presents the results of the validation of the proposed virtual sensing technique for a representative set of EOC over a long period of measurements, result from a short term event as well as response prediction for an indicative timestamp of a day during summer storm with strong wave activity.
