乍得Bongor盆地花岗岩潜山裂缝型储层有效渗透率计算方法  

Determination of effective permeability of granitic buried-hill fractured reservoirs in Bongor Basin,Chad

作　　者：郭海峰 肖坤叶 程晓东 杜业波 杜旭东 倪国辉 李贤兵 计然 GUO Haifeng;XIAO Kunye;CHENG Xiaodong;DU Yebo;DU Xudong;NI Guohui;LI Xianbing;JI Ran(International Branch,China National Logging Corporation,Beijing 102206,China;Research Institute of Science and Technology Co.,Ltd.,CNPC,Beijing 100083,China)

机构地区：[1]中国石油集团测井有限公司国际公司,北京102206 [2]中国石油集团科学技术研究院有限公司,北京100083

出　　处：《岩性油气藏》2023年第6期117-126,共10页Lithologic Reservoirs

基　　金：中国石油集团科研项目“乍得花岗岩潜山油藏高效开发关键技术研究”(编号:2018D-4407);中国石油集团测井有限公司科研项目“海外稳油控水技术体系研究”(编号:CNLC2022-10D02)联合资助。

摘　　要：基于乍得Bongor盆地花岗岩潜山油藏的钻录井、测井和试油资料,提出了一种新的储层有效渗透率计算方法,将试油结果转换为视有效渗透率来作为样本库的标签数据;以领域知识和机理模型驱动为主,机器学习为辅,建立特征曲线;采用XGBoost+KNN作为双重预测模型参与视有效渗透率计算,并利用SHAP值对模型进行了可解释性分析。研究结果表明:①将储层的测井视波阻抗和孔隙度作为渗透率指示曲线,分别与生产指数进行交会,从建模数据一致性指示交会图中挑选出19口井共26个有效井段,将试油结果转换为视渗透率(0.01~1601.50 mD),共建立了51348个深度点数据,14条输入曲线,基本覆盖主要潜山带,包含了不同岩性、不同储层品质和不同试油产量井段,使得整个样本库具有足够的代表性。②XGBoost模型充分利用了测井视波阻抗曲线、辅助表征潜山储层纵向分带特性的归一化垂深曲线、密度曲线、自然伽马窗口均值曲线、声波时差曲线、补偿中子测井曲线、视中子-密度孔隙度差窗口均值曲线、深浅电阻率曲线和自然伽马窗口标准差曲线信息,其计算结果与潜山储层品质的定性认识一致,预测精度较KNN模型更高。③乍得Bongor盆地花岗岩潜山油藏中有效渗透率大于1.00 mD的储层为有效储层,有效渗透率大于50.00 mD的储层为好储层,该方法的计算结果与试油结果一致。A new effective permeability calculation method was proposed for the granitic buried-hill reservoirs in Bongor Basin of Chad based on drilling,logging,and oil testing data.Apparent effective permeability was derived from well testing results,serving as labeled data for the sample dataset.The method relies primarily on domain knowledge and mechanism-driven models,and is supplemented by machine learning to establish feature logs.The dual-prediction model XGBoost+KNN was employed to calculate apparent effective permeability,with SHAP values used for model interpretability analysis.The results show that:(1)Permeability indicator logs,namely apparent acoustic impedance and porosity,were utilized to select 26 effective testing intervals from 19 wells,based on their intersection with production index.The well testing results were converted into apparent permeability values,ranging from 0.01 to 1601.50 mD,resulting in a dataset comprising 51348 depth data points and 14 input logs.The sample dataset adequately covers the main buried hill zones and incorporates input log response characteristics from diverse lithologies,reservoir qualities,and well testing production performance,thus achieving sufficient representativeness.(2)The XGBoost model effectively uses various logs,inclu-ding the apparent acoustic impedance log,normalized depth log representing the vertical zoning characteristics of buried-hill reservoirs,density log,window average of natural gamma ray log,acoustic log,neutron log,window average of neutron-density porosity difference log,deep and shallow resistivity log,and window standard deviation of natural gamma ray log.The model’s predictions exhibit consistency with qualitative understanding of buried-hill reservoir quality and demonstrate higher accuracy compared to the KNN model.(3)The granitic buried-hill reservoirs of Bongor Basin,Chad,with effective permeability greater than 1.00 mD,are classified as effective reservoirs,while those exceeding 50.00 mD are considered good reservoirs.The calculated resul

关 键 词：花岗岩潜山 裂缝型储层 视有效渗透率 机器学习 样本库构建 XGBoost 储层参数建模 Bongor盆地 

分 类 号：TE122[石油与天然气工程—油气勘探] P618.13[天文地球—矿床学]