Quantifying uncertainties for Risk-Based Inspection planning using in-service Hull Structure Monitoring of FPSO hulls

Abstract

In order to obtain valuable information from an Hull Structure Monitoring system, a large data set and consistent analysis of that data is required. The monitoring requires significant efforts over multiple years and as a result, uncertainties obtained from in-service measurements are rarely published. Instead, researchers have to rely on numerical simulations and conjecture to quantify certain parameters. In this article, two years of continuous monitoring data is used to quantify several sources of uncertainties of the hull structure of an FPSO. These sources include uncertainty related to the future extrapolation of loads and statistical uncertainty of the long-term sea states which is quantified using a Bayesian re-sampling scheme. Next, the uncertainty introduced through the use of analytical load distribution models is addressed. Finally, the uncertainty in the calculation method is quantified. These data are then used in a case study for the particular FPSO which has been monitored to demonstrate their practical application using a simple reliability model. Multiple stochastic models for the long-term description of loads are examined. Besides the traditional Weibull model, the less frequently used Pareto, Lognormal and Gumbel model were tested and compared against an uncertainty modal based on a spectral fatigue assessment. The Pareto and Weibull models are considered appropriate models and were compared against design stage analyses. Good design procedures adopt conservative parameters to describe the uncertainties. In the presented example, this was found to be true and therefore the inclusion of measurement data in Risk Based Inspection analysis for the presented case results in prolongation of the inspection interval.