Study On Inner Wall Corrosion Prediction And Evaluation Method Of Subsea Pipeline

The continuous development of inner wall corrosion of subsea pipeline will eventually lead to leakage of oil and gas,causing huge losses to the economy and society.However,frequent inspection and maintenance for pipelines would inevitably consume lots of unnecessary cost.Therefore,it has been an urgent demand for oil and gas companies to accurately predict the corrosion trend and failure probability of pipelines for making maintenance plans.In this paper,the corrosion defects of inner wall of subsea pipeline are regarded as research objects and corrosion trend prediction model and corrosion damage assessment method are studied.The main research work and conclusions are as follows.(1)The corrosion growth of pipelines is random,so there is a fluctuation in the growth rate of corrosion depth.In order to deal with that,this paper proposes a new prediction model for corrosion depth based on triple exponential smoothing method,and uses genetic algorithm to determine the optimal smooth initial value and smooth coefficient of the model.This model can use historical data series to predict corrosion trend in the future,and will make corresponding adjustment with the change trend of the data series.It is verified by examples and compared with the traditional gray model GM(1,1)for corrosion depth prediction.The results show that the newly proposed method is more suitable for the prediction of non-smooth corrosion depth series,which makes up for the shortcomings of the traditional prediction model.(2)In view of the complex nonlinear relationship between pipeline corrosion rate and environmental factors,the single prediction model has the disadvantages of low prediction accuracy and poor stability.Based on the idea of ensemble learning,a new prediction model for pipeline corrosion rate is established based on the gradient boosting decision tree algorithm,using grid search and cross-validation methods to optimize model parameters.The model is verified by corrosion data of an oil pipeline,and compared with the prediction results of the widely used BP neural network and support vector machine model.The results show that the gradient boosting decision tree model has higher prediction accuracy and generalization ability,and the model has the advantages of strong interpretability,so it can provide a more practical method for the prediction of pipeline corrosion rate in the future.(3)The parameters such as corrosion defects,pipeline size and material properties always contain a certain degree of uncertainty,so the calculation results of traditional deterministic assessment model has the disadvantage of conservatism.More practical predictive model for corrosion failure probability is established on basis of Markov chain theory and reliability theory.The results of parameter sensitivity analysis show that the radial corrosion rate is the most important factor affecting the failure of pipeline.Finally,a two-level corrosion damage assessment and remaining life prediction system is established based on the remaining wall thickness and remaining strength of corrosion pipeline,which can provide a powerful tool for pipeline integrity management.(4)According to the model researched above,a subsea pipeline inner wall corrosion prediction and evaluation software is developed on the Microsoft Visual Studio platform using C#,and the software function is tested using known data.The test results show that the software meets the design requirements and can predict and evaluate for pipeline corrosion based on the selected model and collected corrosion data.Therefore,it can provide a convenient work platform for pipeline evaluators and improve their work efficiency.