Methodological Study On Reliability Monitoring And Analysis Of Offshore Structures

With more and more extensive applications of offshore structures in the fields of oil and gas exploitation and new energy,the demand for their reliability has been increasing.Thanks to condition monitoring and diagnosis,the product's situation is achieved during the whole life time,which is beneficial to fault detection and prediction at the early stage.And thanks to practical reliability analysis,it is possible to implement the reliability-based design,risk prevention measures.However,with the developments of large-scale,automated and process-oriented products,there are still many key technical problems involving the existing technology of monitoring and reliability analysis of offshore structures.In this paper,the methods of resilient monitoring and reliability analysis of offshore structures are investigated,aiming at improving the resilience of the monitoring system for better on-line reliability maintenance and improving the system reliability analysis methods for risk reduction and reliability improvement.Firstly,the exploration is carried on the characteristics and acquisition methods of reliability data of offshore structures to provide the basis for the subsequent methodology study.Then,when reliability data is obtained by monitoring,the resilient methods of signal recovery and failure diagnosis based on correlation are proposed to address the problem of sensor failure.Finally,based on reliability data obtained form various data acquisition methods,the system reliability analysis methods considering correlation and dynamics are studied.The contents of this paper are displayed as follows:(1)Analyze the methods of reliability data acquisition and processing for offshore structures in view of data diversity,database insufficiency and difficulty of data obtaining.As for different types and characteristics of system and elements,analyze the applicability of each data acquisition method.Especially,taking consideration of offshore environmental characteristics,analogy&correction method based on onshore equipment data is proposed for obtaining unreadily available data(2)In terms of a system concluding strong-correlated elements,as for the problem that the reliability of the monitoring sensor is not high enough and the database is incomplete,the idea of using the related normal sensor(s)as its virtual sensor is put forward.Based on multivariate ARMA model and time series of virtual sensor(s),the recovery of monitored data can be realized when individual sensor fails.The numerical examples show that this method can achieve real-time data recovery,and the results have a good agreement with the simulated data,which sufficiently meets practical engineering requirments.Based on the applications of virtual sensors,the proposed method makes it a reality that the system keeps operating for a time period after some sensor fails without replacement and maintenance of sensors and system shutdown.This presents the resilient data recovery method's ability to resist interference and insure data integrity.(3)In terms of a system concluding strong-correlated elements,aiming at the problem of improbable fault diagnostic,virtual sensors are applied to identify the fault features of components and perform real-time fault diagnostic.According to the characteristics of environmental components,a fault feature extraction method based on focused frequency wavelet analysis is proposed.And the dynamic alarm thresholds for time-varying environment based on artificial neural network is established.The integrated method has the ability to assess the health conditions of components and sensors in short time when individual sensor fails in multi-component system.The proposed fault diagnostic method is highly feasible and meets the requirements of precision.The resilient method can help avoid the operational risk of the inspectors and save the time for shutdown detection,which is very important for preventing and controlling significant risks.(4)For the offshore structure systems with a large number of correlated components and failure modes,Clustering Approximation method is proposed for system reliability analysis.By introducing the concept of relevancy,the intra-group correlation is calculated by approximation,and the correlations are ignored between poor related groups.Finally,the reliability of correlated system is calculated.An example demonstrates the favorable validity,correctness and efficiency of the methods' applications in analyzing offshore system reliability.Besides,the applicability of proposed method is analyzed.(5)There are characteristics of redundancy,function dependency and sequence of failure among offshore components,which are impossible to describe by traditional static reliability methods.Therefore,dynamic fault tree is used to solve this problem and obtain system reliability results.The method is despicted as follows,the dynamic fault tree is constructed based on function and structure based system grading,the bottom events' failure data is obtained by different failure data acquisition methods,and the events' occurrence probabilities of dynamic gates are estimated based on approximate calculation.Finally,the qualitative and quantitative system reliability analysis is completed.Through the application of data input,system grading and dynamic fault tree in floating offshore wind turbine system,a seires of reliability results are evaluated in depth.The analysis results are compared with the statistical references,which shows the feasibility and efficiency of the method application.