基于LHS-Kriging-DW的页岩气集输管道内腐蚀预测模型研究

Research on Internal Corrosion Prediction Model of Shale Gas Gathering Pipeline Based on LHS-Kriging-DW

为了准确预测页岩气集输管道的内腐蚀速率,对DW(DeWarrd)模型进行修正,运用LHS(拉丁超立方抽样)优化的Kriging(克里金方法)对修正DW进行训练以建立模型并验证。通过高清漏磁检测器检测某页岩气集输管道内腐蚀速率,基于PIPESIM模拟该管道现场工况,运用多元非线性回归修正DW模型。运用LHS-Kriging对修正DW模型进行训练并优化,运用建立的模型对另一相邻管道进行了腐蚀速率的预测和现场开挖验证。现场内检测的结果显示:管线在几处低洼积液处发生了严重的腐蚀,因此引入了持液率作为修正因子修正DW模型。选择样本数为500用于LHS-Kriging-DW模型时误差较小。在实例运用中,10个开挖检测点有9个都在预测结果的95%置信区间内,这说明建立的腐蚀速率模型可用于指导现场内腐蚀直接评价等相关工作。

In order to accurately predict the internal corrosion rate of shale gas gathering pipeline,the DW(DeWarrd)model was modified,and the modified DW was trained by Kriging(Kriging method)optimized by La LHS(latin hypercube sampling hypercube sampling)to establish the model and verify it.The internal corrosion rate of a shale gas gathering pipeline was detected by high-definition magnetic flux leakage detector.Based on PIPESIM,the field condition of the pipeline was simulated,and the Ninfit function was used to modify the DW model by multivariate nonlinear regression.The modified DW model was trained and optimized by LHS-Kriging,and the interval estimation of corrosion rate of another adjacent pipeline was predicted and verified by field excavation.The results of field detection show that the pipeline has serious corrosion at several low-lying effusion sites,so the liquid holdup is introduced as a correction factor to modify the DW model.The sample number of 500 is selected for LHS-Kriging training to correct the DW model with small error.In the application of the example,nine of the 10 excavation detection points are within the 95%confidence interval of the predicted results.It indicates that the established corrosion rate model has certain accuracy,which can be used to guide the direct evaluation of in-situ corrosion and other related work.