AK‐DA: An efficient method for the fatigue assessment of wind turbine structures. Issue 5 (29th January 2019)

• Record Type: Journal Article
• Authors: Huchet, Quentin; Mattrand, Cécile; Beaurepaire, Pierre; Relun, Nicolas; Gayton, Nicolas
• Is Part Of: Wind energy. Volume 22:Issue 5(2019)
• Page Start: 638
• Page End: 652
• Publication Date: 2019-01-29
• Subjects: adaptive design of experiments -- damage assessment -- fatigue prediction -- Kriging metamodel -- wind turbine; Wind power -- Periodicals; 621.312136
• DOI: 10.1002/we.2312
• Languages: English
• ISSNs: 1095-4244

Abstract

Abstract: Lifetime damage estimation is a complex and demanding task that needs to be performed during the design of offshore wind turbine structures. A general damage analysis framework is proposed by the certification bodies. Therein the total lifetime of the structure is considered as a series of elementary situations combining structural and environmental states. For a given structural state, the loading environment is described using statistical parameters such as the wind mean speed at hub height or the peak spectral period of the sea. An estimation of the structural response is to be computed for each of the environmental combinations of parameters, therefore leading to tens of thousands simulations. The cost of a single simulation makes this process often unfeasible for engineers who are usually forced to reduce the number of simulations considering industrial feedback with risks of potential lack of representativity of results. This paper aims at presenting a novel method for the reduction of the simulation costs relative to the long‐term damage estimation (relative to a design load case) and based on the so‐called adaptive Kriging approach. From on a reduced set of observations (multiphysics simulator runs), a Kriging metamodel is here used to approximate the damage model response for all the nonsimulated sets of environmental parameters. The latter are subsequently used to assess the long‐term damage as presented in the standards. The statistical measure of the metamodel error of prediction is used into an iterative structure in order to progressively enrich the design of experiments with informative sets of environmental parameters. This allows us minimizing the global uncertainty of the approximation. The proposed algorithm, hereafter called the AK‐DA for "Adaptive Kriging Damage Assessment, " is illustrated with two industrial cases of fatigue analyses for the NREL 5MW reference monopile structure and its direct application