摘要
碳捕集、利用与封存(CCUS)技术证实了CO_(2)地质埋存可以有效缓解温室效应、提高油田经济效益,而评价CO_(2)的地质埋存能力,需要考虑CO_(2)在地层中与水的互溶情况。本研究利用高温高压反应釜等设备,测量不同温度、压力、矿化度下CO_(2)在水中的溶解度,明确CO_(2)在水中的溶解规律。基于实验测量结果,利用遗传算法优化核岭回归算法(KRR)参数,建立CO_(2)在水中的溶解度预测模型。结果表明:通过56组数据对模型进行精度评估,其皮尔逊相关系数为0.99,平均相对误差为2.98%。分别使用核岭回归模型、Chang模型和Furnival模型预测温度35~135℃、压力8~50 MPa下CO_(2)在地层水中的溶解度,Chang模型和Furnival模型的平均相对误差分别为8.77%和7.44%,核岭回归模型的预测值与实验测量值拟合程度最高。该模型为预测CO_(2)在地层水中的溶解度提供了一种新方法。
Carbon capture, utilization and storage(CCUS) technology confirms that the geological storage of CO_(2)can effectively mitigate the greenhouse effect and improve oilfield economic benefits.However, the evaluation of the geological storage capacity of CO_(2)needs to take into account the mutual solubility of CO_(2)with formation water.The solubility of CO_(2)in formation water under different temperatures, pressures, and mineralization degrees are measured using high temperature and high pressure reactors and other equipmentto clarify the dissolution law of CO_(2)in water.Based on the experimental results, the predictive model of CO_(2)solubility in water was developed by using genetic algorithm to optimize the parameters of kernel ridge regression(KRR) algorithm.The results show that the accuracy of the model was assessed by 56 sets of data, the Pearson’s correlation coefficient is 0.99 and the mean relative error is 2.98%.The solubility of CO_(2)in formation water under temperature of 35~135 ℃ and pressure of 8-50 MPa is predicted using kernel ridge regression, Chang model and Furnival model respectively, and the average relative error of Chang model and Furnival modelis 8.77% and 7.44% respectively.The predicted values of kernel ridge regression model are the best fit to the experimental measurement values, providing a new method for predicting the solubility of CO_(2)in formation water.
作者
龙震宇
王长权
石立红
刘洋
项小龙
LONG Zhenyu;WANG Changquan;SHI Lihong;LIU Yang;XIANG Xiaolong(Key Laboratory of Hubei Province for Oil and Gas Drilling and Production Engineering,Yangtze University,Wuhan,Hubei 430100,China;Artificial Intelligence College,China University of Petroleum(Beijing),Beijing 102249,China;Northeast Sichuan Gas District,Petrochina Southwest Oil and Gas Field Company,Dazhou,Sichuan 635000,China)
出处
《西安石油大学学报（自然科学版）》
CAS
北大核心
2023年第1期95-101,共7页
Journal of Xi’an Shiyou University（Natural Science Edition）
基金
国家自然科学基金“高温高压CO_(2)-原油-地层水三相相平衡溶解度规律”(51404037)。
关键词
核岭回归算法
遗传算法
CO_(2)溶解度
理论模型
kernel ridge regression algorithm
genetic algorithm
CO_(2)solubility
theoretical model
