| Major accident risk on offshore platforms has increased significantly due to the long-term service in harsh operating environments.In order to prevent major accidents and mitigate the consequences,a large number of safety critical equipment(SCE)with preset safety functions are installed on offshore platforms.Preventive maintenance(PM)is an important activity to prevent the failure of SCE and to ensure the process safety of offshore platforms.However,the existing PM decision-making of SCE relies on the experience of equipment manufacturers and operators,ignoring the dynamic impact of major accident risk and maintenance cost in both operational and PM phases of SCE.In view of the flaws of the existing PM optimization methods,the following work is carried out in this study.The human error probability(HEP)model for the PM of SCE is established based on a two-layered risk influence factor(RIF)model in which the related human error modes and RIFs are identified.On the basis of Bayesian network method,the probability of human error is calculated.The related control measures for human error are proposed through the importance analysis,and the effectiveness of the measures are evaluated.The results reveal that the twolayered RIF model is helpful to consider the impact of operational management on HEP.The proposed HEP model provides a useful tool for the quantitative analysis of human error.A dynamic risk assessment methodology is proposed by integrating the data-driven approach with dynamic Bayesian network to capture the time dependencies.Based on this methodology,the dynamic risk assessment model for the operation and PM of SCE is established.The importance analysis is further conducted to study the impact of basic events on dynamic risk,and the corresponding risk reducing measures are proposed and evaluated.Finally,the uncertainty of the probabilistic model,time dependencies and logical dependencies are discussed respectively for the dynamic risk assessment process.The results reveal that the proposed methodology is effective for the dynamic risk assessment of SCE,and reduces the uncertainty in the dynamic risk assessment process.It is found that the personnel risk increases over time in both the operational and PM phases of SCE,and the personnel risk in the PM phase is higher,which verifies that the PM work of SCE may lead to a significant increase in personnel risk.Therefore,the PM-related risk should be taken into account when conducting PM decisionmaking.A dynamic multi-level early warning methodology is proposed for the SCE degradation on the offshore platform.Based on the test data of SCE,the warning indicators are identified,and the state updating models of warning indicators are established.In combination with the industry guidelines,the dynamic multi-level warning thresholds for SCE degradation is determined.Finally,the warning criteria are established,including the early warning signals and corresponding response plans for different warning situation.The results reveal that the proposed methodology provides a useful tool for the early warning of SCE degradation,and contributes to the PM decision-making of SCE.A multi-objective optimization model is established for the PM intervals of SCE,which integrates the dynamic risk and maintenance cost in both operational and PM phases of SCE.Based on the particle swarm optimization algorithm and Pareto rules,the Pareto set of PM interval is obtained.On this basis,a multi-attribute decision making method of PM interval based on loss function was proposed to further determine the optimal PM interval of SCE,which takes into account the optimization efficiency of different PM intervals on the objective functions.The results show that the multi-objective optimization model of PM interval can comprehensively consider various factors affecting the PM interval of SCE,which provides an effective tool for the optimization of PM for SCE on offshore platforms.The optimal PM interval achieves different degrees of improvements on all objective functions,in comparation with the traditional periodic PM interval. |
