摘要
基于CO_(2)腐蚀速率预测的注入管柱选型是保障咸水层安全稳定地注入及封存CO_(2)的基础工作。已有的腐蚀速率预测模型尚未考虑咸水层注入CO_(2)及管柱材质为s-13Cr的情况,同时也未考虑流速及腐蚀产物膜对腐蚀速率的影响,因此有必要增加对上述因素的考虑,创建面向s-13Cr的CO_(2)注入管柱腐蚀速率预测模型,以便更准确地预测其服役寿命。首先借助室内测试获得了不同环境及条件下s-13Cr不锈钢材质的CO_(2)耐蚀性特征及腐蚀速率,并借助测试数据初步建立了适用于s-13Cr的CO_(2)腐蚀速率半经验模型;进一步考虑流速及腐蚀产物膜等相关因素的影响,利用机器学习方法对模型进行修正。基于均方根误差(RMSE)、确定系数(R-squared)、平均绝对百分比误差(MAPE)的可行性评价结果显示,RMSE=0.0017,R-squared=0.9793,MAPE=0.026,在分别改变压力、温度及流速情况下,应用预测模型得到的不同条件下的腐蚀速率与试验测试集中对应条件下的腐蚀速率具有良好的一致性。该模型性能良好,能够较准确地预测不同环境及条件下CO_(2)对s-13Cr材质的腐蚀速率,从而为咸水层碳封存工程中注入管柱的选型提供相关理论与决策支持。
The selection of injection tubing based on CO_(2)corrosion rate prediction is fundamental to ensuring the safe and stable injection and storage of CO_(2)in saline aquifers.Existing corrosion rate prediction models have not yet considered the situation of CO_(2)injection into saline aquifers and the tubing material being s-13Cr,nor have they taken into account the impact of flow velocity and the corrosion product film on the corrosion rate.Therefore,it is necessary to increase the consideration of the above factors and create a CO_(2)injection tubing corrosion rate prediction model oriented towards s-13Cr,so as to more accurately predict its service life.First,the CO_(2)corrosion resistance characteristics and corrosion rate of s-13Cr stainless steel material under different environments and conditions are obtained through indoor testing,and a semi-empirical model of CO_(2)corrosion rate suitable for s-13Cr based on the test data is preliminarily established,the impact of factors such as flow velocity and corrosion product film is further considered,and the model is corrected using machine learning methods.The feasibility evaluation results based on RMSE,R-squared,and MAPE show that RMSE=0.0017,R-squared=0.9793,MAPE=0.026.Under the conditions of changing pressure,temperature,and flow velocity,the corrosion rates obtained by the prediction model under different conditions are in good agreement with the corrosion rates under the corresponding conditions in the test dataset.The model performs well and can more accurately predict the corrosion rate of CO_(2)on s-13Cr material under different environments and conditions,thus providing relevant theoretical and decision support for the selection of injection tubing in saline aquifer carbon storage projects.
作者
刘玲
刘彩卓
余芳
沈飞扬
李昀泓
LIU Ling;LIU Caizhuo;YU Fang;SHEN Feiyang;LI Yunhong(China University of Petroleum(East China);Downhole Services Company,CNPC Bohai Drilling Engineering Company Limited)
出处
《油气田地面工程》
2025年第8期61-67,共7页
Oil-Gas Field Surface Engineering
关键词
s-13Cr材料
注入管柱
腐蚀速率
腐蚀产物膜
机器学习
咸水层碳封存
s-13Cr material
injection tubing
corrosion rate
corrosion product film
machine learning
saline aquifer carbon sequestration
作者简介
刘玲,副教授,博士研究生,2014年毕业于中国石油大学(华东)石油工程管理专业,从事能源与碳储工程管理研究工作,18669881420,liuling@upc.edu.cn,山东省青岛市黄岛区长江西路66号中国石油大学(华东),266580。
