VIV fatigue reliability analysis of marine risers with uncertainties in the wake oscillator model. (1st January 2016)

• Record Type: Journal Article
• Authors: Low, Ying Min; Srinil, Narakorn
• Is Part Of: Engineering structures. Volume 106(2016:Jan. 01)
• Page Start: 96
• Page End: 108
• Publication Date: 2016-01-01
• Subjects: VIV -- Marine riser -- Fatigue damage -- Uncertainty -- Reliability analysis -- Wake oscillator; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2015.10.004
• Languages: English
• ISSNs: 0141-0296

Abstract

Highlights: Empirical coefficients in wake oscillator model have wide scatter in experimental data. Construction of joint probability distribution of empirical coefficients from data. New fast approach for including variability in coefficients for VIV fatigue analysis. Case studies show that new approach agrees well with Monte Carlo simulations. Randomness of wake coefficients leads to large variability in riser fatigue damage. Abstract: Uncertainties are rife in the fatigue life prediction of marine risers subjected to vortex-induced vibration (VIV). Industry deals with this issue by imposing large factors of safety that may not be properly justified, resulting in over-conservative riser designs in general. One important source of uncertainty arises from the VIV prediction models. This paper focusses on identifying the uncertainties of a wake oscillator model which approximates the fluctuating hydrodynamic force coupled with the riser equation of motion for nonlinear fluid–structure interaction analysis. This van der Pol-type oscillator relies on two wake coefficients which are described deterministically by empirical equations obtained via curve-fitting. However, the underlying data exhibit wide scatter; thus, it is proposed to model the two key coefficients as random variables. Based on experimental data, the joint probability density function of the variables is approximated. A new fast reliability approach is proposed for the VIV fatigue reliability analysis, while Monte Carlo simulations are performed for comparisons. Case studies of a vertical riser in a uniform flow show that the proposed method compares favorably with Monte Carlo in terms of predicting the failure probability as well as safety factors conforming to prescribed reliability levels. Moreover, this study reveals that the randomness of wake coefficients leads to large variability in the riser fatigue damage. The correlation between the coefficients should be properly incorporated as it affects the fatigue reliability of risers experiencing VIV.