Prediction And Application Of Neural Network Prediction For Corrosion Rate In A Multiphase Flow Submarine Pipe Section In The South China Sea

Among early domestic laying submarine pipelines, there are a number of pipelines' detection has never been implemented, and due to the complexity of pipeline internal situation, detection is also so difficult to be implemented. With the growth of service life, because of the deficiencies in the aspects of construction of the initial investment to pipeline building and the subsequent maintenance measures in the process of operation and management, lead to this pipelines to different levels of security risks, as well as the problems of low transmission efficiency.There is not a universal corrosion rate prediction model for subsea multiphase flow pipeline, because of the complexity of too many corrosion factors, and the factors interacted with each other, either. In response to this situation, the paper based on the nonlinear algorithm artificial neural network method, through the establishment of the model for corrosion rate prediction of multi-phase flow subsea pipelines, independent preparation of the "Prediction of corrosion rate in multiphase flow submarine pipelines" program, and the main contents are as follows:(1) Collection and analysis of basic parameters of multiphase flow pipeline and the selection of target pipeline. In view of the information of submarine pipeline transportation medium, conveying pressure and water content field investigation and collection are taken, and including but not limited to the design, construction, production, operation and maintenance. Based on the existing pipeline detection, pigging, operating conditions, etc., the selected SP76-EP76 pipeline as the research object, the parameters of the target pipeline data are collected and analysised.(2) Analysis of the internal operating conditions of the pipeline within the OLGA. Based on the route map of the submarine pipeline (SP76-EP76), the pipeline is divided into 9 sections, which is divided into 60 smaller sections. Through the simulation of the internal operating conditions of the pipeline, the distribution of the related parameters of the pipeline is determined, which provides an effective basis for the process of the "multi phase flow pipeline internal corrosion rate prediction"(3) The development of "Prediction of corrosion rate in multiphase flow submarine pipelines" program based on the neural network algorithm. In the base of the parameters (such as temperature, pressure, liquid holdup, acid content, etc.) calculated by OLGA the prediction model of artificial neural network was established.(4) The error analysis of the program of the "Prediction of corrosion rate in multiphase flow submarine pipelines". The comparison results show that the calculation results of the program are more practical.