Efficient long-term extreme response and fatigue analysis of offshore structures under stochastic wave, current and wind loads

Abstract

Offshore structures face complex and stochastic loads from waves, current, and wind. An accurate long-term analysis is crucial for reliability assessment for overloading and fatigue failures. This problem is challenging due to the small probabilities and numerous uncertainties, thus existing methods mostly rely on simplifications or empirical rules. To address these challenges, this paper presents a new method for long-term extreme response and fatigue analysis, incorporating seven long-term environmental variables following a prescribed joint distribution and short-term uncertainties from irregular waves. The method combines subset simulation (SS) for efficient reliability analysis of rare events and an advanced metamodel GE-NARX for predicting the time series response for a wide range of environmental inputs. A new design-of-experiments scheme is developed to train the metamodel effectively. Another novel aspect is the application of SS to efficiently evaluate not only the failure probabilities but also the mean damage. The proposed method is tested on a floating system and shown to accurately predict the long-term extreme response and cumulative fatigue damage when compared with a time-consuming benchmark method, while offering a substantial computational speedup. The proposed method is highly efficient, allowing the investigation of diverse scenarios for better insight. Among other things, the results reveal the critical role of wave, current and wind directionality, and assuming deterministic values for wave, wind and current parameters can be substantially erroneous, highlighting the limitations of design codes. The proposed method is an effective tool for design and potentially for real-time risk assessment of offshore structures.