Surrogate-based limit state modeling for reliability analysis of stationary marine operations

Abstract

Safe execution of marine operations (MOs), such as lowering a jackup leg, depends on its reliability. Given the stochastic nature of dynamic responses, the randomness of actual operational limits, and practical information available onboard, e.g., roll, pitch, and heave Root Mean Squared (RMS) values. This paper introduces a simple but robust method for reliability analysis of an MO using a surrogate limit state equation. For a location in the North Sea, 37 years of hindcast directional wave spectra are used to compute the covariances of the roll, pitch, heave, and leg impact velocity of a jackup barge. The covariances are input into a Differential Evolution Optimization Algorithm to optimize a variance-based model and find the coefficients of a surrogate limit state equation, whose performance is the best for the month- and season-based models. Reliability indexes are calculated using the First Order Reliability Method, and are used to find workable weather windows for a single-leg and the complete four-leg lowering operation. The operator can assess the probability of failure of the entire leg lowering process by simply using RMS values of the jackup motions. The method is practical, efficient, robust, and can be used to develop alternative real-time onboard decision-making systems.