Probabilistic fatigue estimation of ageing wind turbine blades using a Bayesian network framework

This paper presents ongoing research towards a probabilistic remaining useful life (RUL) methodology for offshore wind turbine rotor blades. Building upon previous work that introduced the Bayesian Network (BN) framework for evaluating the RUL of wind turbine structural components, this study expands the methodology to address the fatigue assessment of composite rotor blades. The BN serves as the core framework for performing probabilistic RUL calculations and is designed with modular components that can be tailored based on functionality and available data. This paper discusses the development of two key modules: load characterisation and finite element (FE) modelling. Specifically, the paper outlines the methodology behind the load modelling module, a physics-based simplification model. The paper also addresses how the results from this module will be used in the following FE blade model through the application of a Gaussian process regression-based response surface model for mapping loads to strains. These approaches demonstrate how modular components can be integrated into the BN framework to estimate the RUL of composite wind turbine blades, with potential for further expansion in future research to include updating with SCADA data. This research addresses key challenges in current RUL assessment methods for rotor blades, including leveraging the limited information available at end of life for older turbines and accounting for uncertainties in RUL predictions.