A case study for the optimization of moment-matching in wind turbine blade fatigue tests with a resonant type exciting approach

Abstract

In order to effectively test the performance of the wind turbine during the blade fatigue test with a resonant type of exciting approach, it is important to moment-match the test moment to the target moment. The distribution of the moments on the critical sections can be adjusted by using additional masses that are reasonably located on these sections, but that have different location, number and size options. Achieving this through the traditional trial-and-error method however, which attempts to find the best additional mass variables combination for optimal moment-matching, requires an extremely heavy artificial workload, as the combination options for the additional mass variables are many. To tackle this dilemma, FEA (Finite Element Analysis) associated with PSO (Particle Swarm Optimization) method is now commonly used for moment prediction and optimal combination searching. Nonetheless, the enlargement of the blade scale has meant that using FEA poses some disadvantages, which in turn means an increased amount of time needed for integrated-parameters computing. Moreover, pre-computing work, such as physical object modeling, grid generation & checking, solution setting & parameter adjusting, may also require lots of time and a heavy reliance on professional workers to carry out processing tasks – requirements that are adverse to realizing an automatic dealing procedure. Fortunately, in the blade design and factory leaving stages, those important blade parameters are already defined in a lumped manner. This therefore leads to an applicative use of TMM (Transfer Matrix Method) as when outputting the section data, including moment, it only requires these section parameters for the modeling and the moment-matching process. It is reasonable to believe that FEA usually offers better computing accuracy, while it is also understandable that for the present condition cloud and after describing the lumped definition manner, TMM also presents sufficient accuracy. For the present case study, TMM shows satisfactory prediction accuracy when compared with the test results, with a relative error of no more than 1% for any critical blade section.Additionally, moment-matching optimization work has been conducted by combining with the PSO procedure. The optimal variables combination is finally confirmed with two additional masses of 4000 kg and 605.6 kg located 25m and 48m from the blade root section respectively, which obtains the minimum standard of 4.25%, a higher matching precision compared with that of the test which had 5.6%. Thus, it seems that the TMM-associated PSO method introduced in this paper provides a more efficient way to figure out the optimal variables combination for moment-matching while maintaining sufficient computational precision. In addition, since the strict requirements for moment-matching accuracy have not yet been officially issued, the present case study does not treat the additional mass number as a variable but as the constant 2. Compared with the test however, this paper's result still shows improvements in moment-matching accuracy while also reducing the installing workload needed for the 2 additional masses.