Condition monitoring data in the study of offshore wind turbines’ risk of failure

Segovia Garcia, Maria Del Carmen and Revie, Matthew and Quail, Francis (2013) Condition monitoring data in the study of offshore wind turbines’ risk of failure. In: Proceedings of the 19th AR2TS Advances in Risk, Reliability and Technology Symposium. ASME.

[img]
Preview
Text (Segovia-etal-AR2TS-2013-Condition-monitoring-data-in-the-study-of-offshore)
Segovia_etal_AR2TS_2013_Condition_monitoring_data_in_the_study_of_offshore.pdf
Accepted Author Manuscript

Download (766kB)| Preview

    Abstract

    Unplanned maintenance actions entail a period of inactivity of wind turbines and therefore a loss of revenues. This is even more pronounced in the case of offshore wind farms because of difficulties in access. To this end, condition monitoring (or health monitoring) systems have been implemented on wind turbines by manufacturer to support maintenance decision making by operators. However, a major concern with using condition monitoring is the creation of false positives. In this paper, we have considered how SCADA data may be used to support reliability and maintainability models. We have taken into account SCADA data to assess the level of degradation of a gearbox at any instant of time. Workshops with offshore engineers suggested that degradation is dependent on a small number of variables, e.g. turbulence intensity, wind speed, temperature, etc., and as such, these information sources have been considered in our model to support maintenance decision making. Dynamic Bayesian Networks allows for modelling multiple information sources (including expert judgement) and dynamic phenomena, e.g. system deterioration. They are also capable of representing dependencies between variables of interest. For these reasons, we developed a Dynamic Bayesian Network to assess the risk of failure of the component under study at any instant of time. We considered the information provided by SCADA about the factors affecting the degradation of the system. In this way, we obtained an estimation of the risk of failure of the system.