Reliability Analysis Of Corroded Pipeline Based On Bayesian Analysis And Markov Chain Monte Carlo Algorithm

Natural gas is currently the most clean and environmentally friendly fossil fuel It's the lifeblood of a country's economic development.Pipelines have become the most important transportation mode of gas in the past 100 years,and will still remain the most important transportation mode of natural gas in the next 50 years.According to the "Thirteenth Five-Year Plan" development plan,by 2020,China's total length of natural gas pipelines will reach 10.4×104 kilometers with the main-line gas transmission capacity exceeding 4000×108m3.While the service life of natural gas pipelines increases,corrosion becomes the main cause of pipeline failure,especially in China.According to statistics,39%of the failure accidents of gas transportation in Sichuan province between 1969 and 2003 were caused by pipeline corrosion.Once the natural gas pipeline accidents occur,the ecological environment around the pipelines as well as the lives and property of residents will both suffer huge losses.Thus,how to accurately predict the reliability of corroded pipelines is the key to ensure pipelines being operated safely and reliably.Based on the theory of Bayesian analysis and Markov chain Monte Carlo method as well as combined with the influence factors of corrosion of natural gas pipelines,this article establishes a corroded and degraded orbit model that adapts to the W-Q line pipe in order to establish a pipeline reliability model.Using the hybrid Gibbs sampling method,the model can calculate the current reliability of pipelines and predict the reliability of the pipeline in the next 30 years,thus providing a theoretical basis for pipeline maintenance.The main research contents are as follows:(1)The investigation analyzes the database of the natural gas pipeline failure in Europe,the United States,the United Kingdom,Canada,etc.as well as part of that in China.Corrosion is confirmed as the main cause of the nature gas pipelines failure.The article systematically analyzes the causes of pipeline corrosion failures and their influencing factors.It makes a systematic analysis of the cause that raises pipeline corrosion and its influencing factors.Thus,it lays a theoretical foundation for establishing the reliability model of pipeline corrosion.(2)According to SY/T 0029-2012 Test Methods for Corrosion Rate of Buried Steel Pieces,corresponding piece test is performed on No.47-56 pile section.Combined with the results of outside pipe inspection,the article gains corrosion data of W-Q line No.47-56 pile section.Besides,a renewable corroded power-rate degradation orbital model is established.Moreover,the prior distribution of unknown parameters in Bayesian analysis is obtained,and the curve of corrosion along with time is plotted.(3)The article investigates the basic parameters of W-Q line No.47-56 pile section.Combining the results of adaptability analysis of corroded pipeline pressure prediction,the modified B31G criterion(ASME B 31G-2009 criterion)is determined as the pressure prediction method in this paper.,the maximum allowable corrosion depth for the No.47-56 pile section is used as the threshold in the reliability calculation model of the pipeline.(4)Based on the power-rate degenerate orbit model,Bayesian analysis and the basic theory of Markov chain Monte Carlo(MCMC)method,the calculation formula of the posterior reliability is deduced and a renewable reliability-based calculation model is established.The test uses a mixed Gibbs sampling of 1×106times and it carries out the reliability estimation by using the 1001 sample points of smooth Markov chain from 9.99×105 to 1×106.With relevant calculations of MATLAB programming,the time-variation curve of reliability of pipeline corrosion is acquired.(5)For the W-Q line No.47-56 pile section,an economic decision-making maintenance model is established.It presents that under the restraint conditions of preventive maintenance cost,fault detection cost,and fault maintenance cost,maintenance time with minimum average cost per unit is the best.According to the model,difference in total cost caused by maintenance activities at various times can help determine the optimum maintenance time.