Session: 09-01-14 Fatigue and Structural Analysis

Paper Number: 79265

79265 - Probabilistic Assessment of the Effect of Bolt Pre-Load Loss Over Time in Offshore Wind Turbine Bolted Ring-Flanges Using a Gaussian Process Surrogate Model 

Achieving, and maintaining, a suitable level of bolt pre-load is critical to ensure required fatigue life for bolted ring-flanges in offshore wind turbine (OWT) structures. It is understood that bolt pre-load is likely to vary during operational lifetime. Pre-load relaxation following installation is common, and relevant guidelines for industry allow for a certain amount of relaxation over time. This allowance is utilised in fatigue design of bolted ring-flanges to ensure structures are resilient to potential relaxation. The allowance also includes a requirement for inspection and maintenance for bolt pre-load, where re-tensioning may be necessary if relaxation exceeds the guideline. Such inspection and maintenance regimes may be costly and time consuming. Therefore, an approach to assess the influence of varying bolt pre-load may be useful in the operational context. Recent work has demonstrated the suitability of a Gaussian Process (GP) surrogate model to simulate cumulative fatigue damage under uncertainty in the bolted ring-flanges. The GP surrogate model emulates the high-fidelity Finite Element Method (FEM) structural simulation model, predicting stress at critical locations on the bolted ring-flange in a computationally-efficient manner, given OWT tower shell load and bolt pre-load as inputs. The efficiency of the GP surrogate model, coupled with accuracy comparable to the high-fidelity FEM simulations, allow the possibility of probabilistic prediction of fatigue damage in the bolted ring-flange, under some load profile, and uncertain bolt pre-load. In this paper we detail the application of a Gaussian Process surrogate model to predict consumption of fatigue life over time for a bolted ring-flange, given a bolt pre-load that is assumed to vary. This prediction is probabilistic, delivering intervals of expected remaining fatigue life of the bolt at given points across operational lifetime. We explore the influence of differing assumptions on the variation of bolt pre-load over time, and suggest practical applications of this method to inform inspection and maintenance practices.

Presenting Author: Jack Jorgensen University of Western Australia

Authors:

Jack Jorgensen University of Western Australia
Melinda Hodkiewicz University of Western Australia
Edward Cripps University of Western Australia
Ghulam Mubashar Hassan University of Western Australia