Life-extension classification of offshore wind assets using unsupervised machine learning

Abstract

The objective of the present study is to conduct a techno-economic life-extension analysis for fixed offshore wind turbines for the purpose of classification and certification. The early installed offshore wind turbines are approaching the end of their design service lives. The life extension of these structures is under serious consideration as the support structures are still in a condition for further use. Hence, it is technically possible and economically beneficial to investigate roadmaps for potential life-extension decisions. The methodology developed to analyse the life-extension projects merges a structural integrity assessment of a monopile structure with a corrosion-induced crack development with an economic analysis of the return on assets accounting for the likelihood of obtaining the estimated return. The methodology commences with preprocessing the structural health monitoring data using a Gaussian kernel for denoising, followed by a time-domain crack growth analysis accounting for retardation performed on a cycle-by-cycle basis. The corrosion-related failure mechanisms are growing concerns for currently operating monopile offshore wind turbines. The present study introduces a novel nonlinear corrosion model to address the emergent issue, which is developed considering the spatial and temporal changes in the environmental and operational parameters and the reinforcing effect of fracture on the corrosion. The failure assessment diagram identifies the threshold based on the maintenance cost calculated with a confidence level. The economic analysis combines revenue estimates and operational expenditures, considering the life-extension duration and appropriate discount rate. The results of the life-extension assessment, presented through a risk-return diagram, are used to classify the life-extension projects using unsupervised machine learning k-means clustering algorithm.