A novel Kriging-model-assisted reliability-based multidisciplinary design optimization strategy and its application in the offshore wind turbine tower

Abstract

In Reliability-based Multidisciplinary Design Optimization (RBMDO), the key performance functions of wind turbine are usually implicit, which means the performance response can only be obtained through time-consuming Physics Experiment (PE) or Finite Element Analysis (FEA). However, for practical engineering, the computational cost of repeatedly using PE or FEA is prohibitive. To tackle this challenge, in this study, an adaptive Kriging-model-assisted RBMDO strategy is proposed. The novel updated-strategy for performance function in RBMDO is discussed to find effective training samples of active learning for Kriging model. Also, a powerful decoupling strategy of RBMDO is introduced and combined with the proposed method to enhance computational efficiency further. Two case studies, including a mathematic example and a hydraulic turbine rotor bracket design example, are utilized to illustrate the advantage of the given strategy. Finally, the proposed method is applicated into an engineering design of 5 MW offshore wind turbine tower to ensure its reliability and safety.