Fatigue reliability assessment of offshore wind monopile foundations is commonly performed at the component level, often neglecting system-level statistical dependence among fatigue-critical details. This simplification leads to conservative designs and limits the effective use of inspection information, particularly for components that cannot be directly inspected. Only a limited number of Bayesian updating frameworks have been proposed to address system effects, and these typically rely on assumptions that complicate the modeling of system-level correlations or on simplifications that reduce model accuracy. This paper proposes a multi-level Bayesian updating scheme (mBUS) as an alternative framework for system-level fatigue reliability assessment of monopile foundations. The method represents statistical dependence through physical, time-invariant parameters and incorporates inspection outcomes via a virtual observation mechanism, enabling system-level updating without optimization-based coupling and with linear computational scaling. The framework is demonstrated on a monopile supporting an offshore wind turbine, considering multiple circumferential welds subjected to fatigue loading. Results illustrate that accounting for system-level correlation leads to a stronger reduction of epistemic uncertainty in the deterioration model. For inspection scenarios with non-detection of cracks, this uncertainty reduction results in lower posterior probabilities of failure and increased estimates of remaining useful life for both inspected and uninspected components. From a design and decision-making perspective, the proposed approach supports less conservative fatigue design assumptions, such as design fatigue factors (DFF), while maintaining target safety levels.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号