摘要
对解释模型的正确识别是获得合理可靠的油气井试井解释结果的前提条件。为了提高油气井试井解释结果的准确性,基于油气藏数值模拟技术和随机反演理论,构建多模式下模型参数—试井曲线样本集合;以数据驱动为基础,基于多模式随机分析理论来识别试井解释模型,采用集合卡尔曼滤波方法(简称EnKF方法)进行试井曲线拟合,将数据驱动技术应用于试井模型识别—参数解释的全过程;进而提出了一套基于数据驱动技术的智能试井解释方法,选取X有水气藏1口产水气井进行了现场应用实验。研究结果表明:①所形成的智能试井解释方法可以实现对复杂渗流、复杂边界问题的试井解释,避免了常规试井模型的过度简化,降低了由于模型简化而产生的解释结果误差;②采用多模式EnKF方法,根据拟合误差最小化原则,可以识别出与气藏实际情况相匹配的水侵模式,从而准确把握气藏水侵动态特征。结论认为,基于数据驱动的智能试井解释方法实现了试井模型识别和参数自动解释;该方法适应性强、解释精度高,具有良好的应用前景。
Correct identification of interpretation model is the prerequisite to obtain reasonable and reliable well test interpretation results.In order to improve the accuracy of well test interpretation results,this paper constructs a model parameter-well test curve sample set under multiple modes according to the numerical reservoir simulation technology and the stochastic inversion theory.Then,based on data-driven,a well test interpretation model was identified according to the multi-mode stochastic analysis theory,the well test curve was fitted using the ensemble Kalman filter(EnKF)method,and the data driven technology was applied into the whole process of well test model identification-parameter interpretation.Finally,a set of intelligent well test interpretation method based on data driven technology was put forward.Besides,one water-producing gas well in X water-bearing gas reservoir was selected for field application test.And the following research results were obtained.First,the intelligent well test interpretation method proposed in this paper can provide well test interpretation in the situations with complex flow and boundary,avoid the excessive simplification of conventional well test model and reduce the interpretation error caused by model simplification.Second,by using the multi-mode EnKF method,and combined with the principle of fitting error minimization,the water invasion model matching the real situations of the gas reservoir can be identified,so as to accurately clarify the dynamic characteristics of water invasion in the gas reservoir.In conclusion,by virtue of using the intelligent well test interpretation method based on data driven technology,well test model can be identified and parameters can be interpreted automatically.What's more,this method is of strong adaptability and high interpretation accuracy and has a promising application prospect.
作者
糜利栋
顾少华
薛亮
赵林
MI Lidong;GU Shaohua;XUE Liang;ZHAO Lin(Sinopec Petroleum Exploration&Production Research Institute,Beijing 100083,China;College of Petroleum Engineering,China University of Petroleum,Beijing 102249,China)
出处
《天然气工业》
EI
CAS
CSCD
北大核心
2021年第2期119-124,共6页
Natural Gas Industry
基金
中国石油化工股份有限公司基础性前瞻性科技开发项目“基于机器学习的水侵气井试井解释方法研究”(编号:P18086-5)
中国石油大学(北京)前瞻导向及培育项目“基于大数据和机器学习的裂缝性油藏产能预测研究”(编号:2462018QZDX13)。
关键词
油气井试井
数值模拟
集合卡尔曼滤波方法
试井解释
智能试井
有水气藏
底水
Well test
Numerical simulation
Ensemble Kalman Filter(EnKF)
Well test interpretation
Intelligent well test
Water-bearing gas reservoir
Bottom water
