ارزیابی قابلیت اطمینان ساختار شناور دریایی با استفاده از شبکه بیزی Reliability assessment of marine floating structures using Bayesian network

1. Introduction Marine floating structures are widely used in the oil and gas industry, marine transportation and exploration areas, and renewable energy applications. Conceptual design scenarios for each of these structures are based on environmental loads such as wave, wind and currents. Due to the stochastic behaviour of the sea environment, different types of failures are expected to occur, however it is necessary to improve the safety of marine structures during their lifetime. In the past few years, there has been an increasing focus on analysis of the extreme loads on oil and gas platforms [1–3], and have growing investigations on human reliability assessment in marine harsh environment [41,42]. To explain the complexity of the problem and the various factors involved in the field of marine engineering, a review of marine reliability analysis adopted from previous research is schematically illustrated in Fig. 1. Previously, in order to conduct mooring failure analysis, traditional reliability methods were applied, such as the first order reliability method (FORM) and second order reliability method (SORM) applied by Gao [4], and Frosing and Jansson [5], and Nazir et al. [40]. Siddiqui and Ahmad [6] suggest that failure probability of a mooring system may increase when one mooring system has to be replaced or repaired due to partial or complete damage. With emphasis on the importance of progressive failure, or the entire collapse of the floating system, they investigated reliability of the mooring system of a Tension Leg Platform (TLP). Li et al. [7] analysed the effect of downstroke on the reliability of tendon unlatching using FORM and SORM, rather than considering the loss of tendon tension. Although there are a number of methods in the literature for reliability analysis of marine structures, Bayesian statistics is recommended by Sørensen [8]. An extensive review of BN and probabilistic tools including a wide range of BN applications are provided by Nielsen et al. [57]. Among the current probabilistic models for risk and reliability analysis, Bayesian approach is a promising tool that allows reflection of available knowledge on the process (Abaei et al. [9], Groth et al. [52], Khakzad et al. [54], Musharraf et al. [55], Montewka et al. [56], Trucco et al. [59]). Since Bayesian approaches are capable of considering continuous variables in a discrete format [9–11], it is possible to conduct the inference of more complicated stochastic relationships among random variables in the network, i.e. each variable may have more values than true or false (such as different level of storm conditions), and not all the dependencies have to be deterministic (such as utilities for decision making [33]). In comparison, other probabilistic models such as FORM and SORM are not well suited to conduct risk and reliability analysis efficiently [10]. Recent research has applied BN to engineering fields such as corrosion on steel structure and conditionmonitoring [11–13]. Wang et al. [14] used Object Oriented Bayesian Network (OOBN) to investigate the failure probability of different types of Australian bridges in terms of both structural reliability and conditional-based reliability. Morales-Napoles et al. [15] applied BN as a tool for assessing the failure risk of earth dams providing a conceptual framework for implementation of continuous stochastic variables in BN.